LIBRARY 
 
 OF THE 
 
 University of California. 
 
 RECEIVED BY EXCHANGE 
 
 Class 
 
Digitized by the Internet Archive 
 
 in 2007 with funding from 
 
 Microsoft Corporation 
 
 http://www.archive.org/details/empiricalstudyofOOwhitrich 
 
AN EMPIRICAL STUDY OF 
 
 CERTAIN TESTS FOR INDIVIDUAL 
 
 DIFFERENCES 
 
 BY 
 
 MARY THEODORA WHITLEY 
 
 / 
 
 (submitted in Partial Fulfilment of the Requirements for the Degree of Doctor of Philos- 
 ophy, in the Faculty of Philosophy, Columbia University. 
 
 Reprinted from Archives of Psychology, No. 19. 
 
 NEW YORK CITY 
 AUGUST. 1911 
 
I I %/ I 
 
 WSf~ 
 
 Press op 
 
 •he New Era printing cohpant 
 
 Lancaster. Pa. 
 
 * '• i • 
 
CONTENTS 
 
 Page 
 
 I. History of the Interest in Individual Differences . . 1 
 
 1. The work of various investigators 1 
 
 2. Representative lists of tests 7 
 
 3. Aim of the present study 13 
 
 II. Experimental Work with Several Groups of Tests . . 15 
 
 1. On association 18 
 
 2. On memory 44 
 
 3. On perception 61 
 
 4. On discrimination 75 
 
 5. Discrimination and motor 79 
 
 6. Motor 84 
 
 7. Miscellaneous 91 
 
 III. Changes with Practise 98 
 
 1. Methods of measuring such changes 98 
 
 2. Results from a special series of tests 110 
 
 IV. Conclusions 137 
 
 APPENDIX 
 
 Keys and Material Used in some of the Tests Described . . 139 
 
 i23304G 
 
AN EMPIRICAL STUDY OF 
 CERTAIN TESTS FOR INDIVIDUAL DIFFERENCES 
 
 HISTORY OF THE INTEREST IN INDIVIDUAL 
 DIFFERENCES 
 
 1. The Work of Various Investigators 
 
 The history of scientific inquiry into the nature and amount of 
 individual differences dates back only about twenty-five years. Be- 
 fore that time experimental psychology had concerned itself chiefly 
 with investigations into typical mental functions, especially those 01 
 perceiving the external world. For this purpose long and detailed 
 tests were made upon a very few, or perhaps a single subject. * 
 
 Galton in England was the first who devised and applied a series 
 of tests, both physical and mental, to large numbers of subjects with 
 a view to determining norms and studying the amount, causes and 
 kinds of variation. Since the publication in 1883 of Galton 's " In- 
 quiries into the Human Faculty and its Development," the work 
 done in this field in England has been chiefly confined in its applica- 
 tion to school children; witness Bryant's experiments in 1886 in 
 testing the character of school children 1 and the more recent work of 
 Winch, 2 Spearman, 3 W. G. Smith, 4 Wimms, 5 and Burt. 6 
 
 In Germany there is the general work of Munsterberg in 1891, 7 
 Kraepelin, 8 Aschaffenberg, 9 and Oehrn in 1896, 10 Cron in 1897, 11 
 
 1 Journal of the Anthr. Inst, of Gr. Britain and Ireland. 
 
 2 Brit. Jour, of Psych., 1, 1904. 
 
 3 Am. Jour, of Psych., 15, 1904. 
 'Brit. Jour, of Psych., 1, 1905. 
 5 Brit. Jour, of Psych., 2, 1907. 
 *Brit. Jour, of Psych., 3, 1909. 
 
 7 "Zur Individual Psychologie, " Centralblatt f. Nerv. in Psychiatrie, 14, 
 1891. 
 
 8 "Der Psychologische Versuch in der Psychiatrie," Psych. Aro., I, 1896. 
 
 8 ' ' Experimentelle Studien iiber Associationen, ' ' Psych. Aro., 1, 1896. 
 
 10 il Experimentelle Studien zur Individuellen Psychologie," Psych. Aro., 
 1, 1896. 
 
 u "Ueber die Messung der Auffassungsf ahigkeit, " Psych. Arb., 2, 1897. 
 
 1 
 
2 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 Cohn in 1898, 12 Stern in 1900, 13 and Wiersma in 1902. 14 In these 
 cases experiments, if made at all, were usually in the form of a few 
 carefully prepared tests given to a few subjects either with a view to 
 studying their individual variations in detail or else for the sake of 
 discussing the question of method of administration. There is also 
 the other method of work, that of testing large groups of school chil- 
 dren, as for instance the work of Ebbinghaus in 1897, 15 Netschajeff 
 in 1900, 16 Lobsien in 1901, 17 and Meumann in 1905. 18 
 
 In France under the influence of Binet and his publications in 
 L'Annee Psychologique, there has been an enormous amount of work 
 done, especially with children — investigations into normal and ab- 
 normal conditions, both mental and physical, culminating in 1905 
 and 1908 in the Binet and Simon sets of graded tests of intelligence 
 adapted to children of all ages from three years up. In 1904, Tou- 
 louse, in his "Technique de Psychologie experimentale," gave, as 
 the result of nearly ten years' work, a full and detailed exposition 
 of the methods of giving certain tests, and of computing the results 
 gained. 
 
 In America following the publication in Mind, 1890, of "Mental 
 Tests and Measurements" by Cattell with comments by Galton there 
 was a rapid development of the work represented by that of Bolton 
 in 1892, 19 Gilbert in 1893-94, 20 Shaw in 1896, 21 Griffing in 1896, 22 
 Macdonald in 1897-98, 23 Kirkpatrick in 1900, 24 Bagley in 1901, 25 
 Seashore in 1901, 26 Smedley in 1901, 27 Swift in 1903, 28 and others 
 
 12 ' ' Experimentelle Untersuchungen . . .," Zeitschr. fur Psych., 15, 1897. 
 
 13 ' ! Ueber Psych, der Individuellen Diff erenzen. ' ' 
 
 ""Die Ebbinghausche Combinationsmethode, ' ' Zeitschr. f. Psych., 30, 1902. 
 
 15 ' ' Ueber eine neue Methode zur Priif ung geistiger Fahigkeiten und ihre 
 Anwendung bei Schulkindern, ' ' Zeitschr. f. Psych., 13, 1897. 
 
 18 ' ' Exp. Untersuchungen liber d. Gedachtnissentwickelung bei Schulkin- 
 dern, " Zeitschr. f. Psych., 24, 1900. 
 
 11 ' ' Exp. Untersuchungen iiber d. Gedachtnissentwickelung bei Schulkin- 
 dern," Zeitschr. f. Psych., 27, 1901. 
 
 18 ' ' Intelligenzpriif ungen an Kindern der Volksschule, ' ' Die Exp. Pad., 1, 
 1905. 
 
 19 i t rpjjg Q row th f Memory in School Children, ' ' Am. Jour, of Psych., 
 3, 1892. 
 
 ' M Studies from the Yale Psychological Laboratory, 1, 2, 1892, 1893. 
 
 21 Ped. Sem., 4, 1896. 
 
 22 Psych. Bev., 3, 1896. 
 
 23 " Experimental Study of Children," in Beport United States Comm. of 
 Ed., 1898. 
 
 24 Psych. Bev., 7, 1900. 
 
 20 A m. Jour. Psych., 12, 1901. 
 28 Ed. Bev., 22, 1901. 
 
 "Beport Dept. of Child-study, 3, 1900-01 (Chicago Public Schools). 
 
HISTOBY OF INTEREST IN INDIVIDUAL DIFFERENCES 3 
 
 on school children j that of Jastrow in 1893, 29 Thompson in 1903, 30 
 and Ternan in 1906, 31 on laboratory subjects (in the last instance 
 children who came to the laboratory regularly), and further work of 
 Cattell in 1893 32 -96, 33 and Jastrow in 1893, 34 on college students. A 
 study of method and a somewhat extended inventory of seven sub- 
 jects has also been made by Sharp. 35 
 
 Columbia appears to be the only university still making tests 
 upon the freshmen. An inquiry among the universities and larger 
 colleges of the United States and Canada has resulted in fifteen 
 replies in the negative. 
 
 This by no means exhausts the list, since a large proportion of 
 recent investigations of whatever topic include a treatment or state- 
 ment of individual differences in method of work or degree of 
 achievement, and since, too, some treatises on the psychology of 
 individual differences, Stern's for example, are largely reviews of 
 other investigators' general work from this particular standpoint. 
 
 There are, aside from the questionaire method so largely used by 
 Stanley Hall and others by which large quantities of crude, descrip- 
 tive material are amassed from untrained observers, two customary 
 methods of experimental procedure which have already been indi- 
 cated. One is to use a few specialized tests upon a limited number 
 of subjects, with a sufficient number of repetitions to establish the 
 reliability of the reaction or to induce fatigue or practise. Oehrn, 
 Kraepelin, Ternan, Wimms, and Binet make use of this method. 
 The second method, scoffed at by Stern and criticized by Binet in his 
 review of Wissler's work, is to use very simple tests, many of them 
 physical, upon large numbers of subjects, usually without repetition. 
 Cattell 's tests for freshmen, Galton's tests and the many tests of all 
 kinds on school children are of this nature. This latter method is the 
 predominant one in this country to-day. 
 
 That this should be the case, is not surprising since the first 
 laboratory work directly concerning itself with individual psychol- 
 ogy was instituted by Cattell whose early work in individual differ- 
 ences has been noted. Already in the eighties his experiments on 
 himself and others 36 on the time taken to recognize colors, letters of 
 the alphabet, to see and name the same, and on three groups of as- 
 
 29 Ed. Rev., 5, 1893. 
 
 30 < ' The Mental Traits of Sex. * » 
 81 Fed. Sem., 13, 1906. 
 
 32 Phil. Rev., 2, 1893. 
 
 33 Psych. Rev., 3, 1896. 
 
 84 Am. Jour. Psych., 4, 1893. 
 
 85 Am. Jour. Psych., 10, 1899. 
 
 38 ' • Psychometrische Untersuchungen, ' ' Phil. Stud., 2, 3, 1895-6. 
 
4 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 sociation tests anticipate much that has since become part of the 
 regular stock in trade of those who use the methods of simple mental 
 tests of the higher psychic processes. His list of ten tests employed 
 upon all freshmen and other volunteers in the University of Penn- 
 sylvania published in 1890, 37 was the first definite psychological 
 inventory in this country. 
 
 In 1896 following Baldwin's suggestion at the annual meeting of 
 the American Psychological Association a committee of five was 
 formed consisting of himself, Jastrow, Sanford, Witmer and Cattell 
 to consider the feasibility of cooperation among the various psycho- 
 logical laboratories in the collecting of mental and physical statistics. 
 A suggestive but indefinite report was made by this committee 
 through Witmer the next year. 
 
 In 1907 the Association again appointed a committee of five con- 
 sisting of Angell, Judd, Pillsbury, Woodworth, and Seashore to de- 
 termine a series of group and individual tests with reference to 
 practical applications, and to determine standard experiments of a 
 more technical character. Their first report appeared in Decem- 
 ber, 1910. 
 
 Not the least interesting feature of the development of the work, 
 has been the fluctuating of opinion with regard to its value, and the 
 criticism of the methods used in accordance with the aim in view, 
 and the evident influence of parallel work in general psychology. 
 For instance in Germany there is first the intensive work on some of 
 the higher mental processes by Kraepelin and his school in the early 
 nineties, contemporaneously with extensive work in America on 
 simpler processes with emphasis on the accompanying physical 
 measurements — the subjects being sometimes children — and with 
 characteristic French investigations into abnormal and criminal 
 types as well as into the thinking powers of school children. 
 
 The long article in Volume 2 of L'Annee PsycJiologique, 1895, by 
 Binet and Henri, is notable in that it formulates two distinct prob- 
 lems of individual psychology, definitely favors the use of tests com- 
 plex in content and therefore less capable of precise treatment, and 
 suggests a grouping of appropriate tests under ten functions. In 
 this article the preceding work of Cattell, Miinsterberg, Jastrow, 
 Kraepelin, and Gilbert is illustrated and criticized. The lists of tests 
 given by the first three men are termed too simple, incomplete and 
 too partial — that is confined too entirely to tests of memory, sensa- 
 tions and physical abilities. Kraepelin 's are criticized as being not 
 only partial but impractical since the tests require five hours for 
 completion, necessitating several visits to the laboratory. Gilbert's 
 
 87 Mind, 15, 1890. 
 
HISTOBY OF INTEBEST IN INDIVIDUAL DIFFEBENCES 5 
 
 are said to show the difference in degree but not in kind between the 
 thinking powers of the child and the adult. Their own list of tests 
 could be given in from one to one and a half hours. In describing 
 them only vague directions for administration are given, and oc- 
 casional illustrative results from some tests already used with school 
 children. They conclude by saying that their tests probably need 
 modification, and might not disclose the finer mental differences be- 
 tween individuals similarly trained and belonging to the same social 
 group. The work is fruitful in suggestions, though with a -sketchy 
 indefiniteness rather than a diagrammatic precision. 
 
 Further progress, especially in the application of the tests to 
 school children, was made in each country but along lines already 
 indicated. Ebbinghaus 38 devised and applied a new sort of test 
 ■since known as his ' ' combination ' ' or completion test, which aroused 
 no little interest and discussion. 
 
 In 1899 Sharp 39 took up the question of method. The first half 
 of her work is largely a review of the theses of Binet and Henri, 
 while the remainder is a careful study of some of the tests suggested 
 by them, as applied to seven college students. She considers the re- 
 sults unsatisfactory except that they show that a single trial of any 
 of the tests, made in the suggested hour and a half among single 
 trials of many other tests, would be practically valueless and most 
 unreliable, especially in the case of the tests of a complicated nature. 
 
 The following year appeared Stern's work, "Uber die Psychol- 
 ogic der Individuellen Differenzen." This contains a review of 
 methods, but not of results to date, and criticisms which are largely 
 destructive. Thus in pointing out the dangers of extensity and the 
 probable resulting superficiality, he makes some enlivening remarks 
 on the American fondness for the questionaire method, comparing 
 it to the questions concerning favorite author, color, food, etc., com- 
 piled in the autograph books of the Backfisch of the day, which re- 
 sults in what he elsewhere calls ' ' pseudostatistics. ' ' He would place 
 no reliance on the results of any series of tests which could be com- 
 pleted in an hour and a half, and considers the individual differences 
 found in sensation and perception to be due to lack of experience 
 with the material, since practise reduces those differences. He also 
 says that tests on memory should seek to discover ways of memo- 
 rizing and length of retention rather than content, and that as a 
 measure of association, the spoken first idea is too erratic to be trust- 
 worthy, and measures too much else besides association. He offers 
 few definite suggestions as to methods of procedure. 
 
 38 Zeitschrift fiir Psychologie, 13, 1897. 
 
 39 Op. cit. 
 
6 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 In 1901 "Wissler, in working over the results of the Columbia 
 freshmen tests from the point of view of correlation, finds so little 
 that he concludes that they tell nothing as to the general intelligence 
 of individual college students or adults. If a functional relation- 
 ship exists it must be more complex than is usually supposed and it 
 needs further testing. He remarks that correlating successive trials 
 would help show the precision of a test. 
 
 Two years later appeared Binet's 40 account of careful and re- 
 peated tests, extending over several months, on his two little 
 daughters. Methods and results are given in detail and the con- 
 clusions drawn from them as to the characteristics of the two sub- 
 jects. Many of the twenty different tests were those already utilized 
 in work among school children, notably the written descriptions of 
 objects and pictures. His object was qualitative and descriptive 
 rather than normative, and in consequence the actual tests are 
 supplemented by long and careful questioning as regards imagery 
 and analysis of associations. 
 
 The same year, in the introduction to the first volume of the 
 "Beitrage zur Psychologie der Aussage," Stern again criticizes cur- 
 rent methods of investigation. He points out that by them either 
 time or numbers is sacrificed, whereas data from many people should 
 be amassed by trained observers, and similarly treated. Instead of 
 one experimenter using a few volunteer students as subjects, another 
 large or selected groups of school children, another his own patients, 
 another criminal cases, and still another results of a few experiments 
 on himself and treated by original methods — the general results 
 being confusion rather than cohesion — there should be an Institute 
 for Applied Psychology, to act as a centralizing and unifying agency, 
 a sort of clearing house, with the services of a trained statistician 
 always available! The tests used should represent actual life con- 
 ditions as nearly as possible and not be at all of the type of immedi- 
 ate memory for colors, tones, etc., which tell as much about the 
 memory as a microscopic study of the finger would tell of its func- 
 tion. How well he has succeeded in justifying his position may be 
 gathered from the successive volumes of the Beitrage and the Zeit- 
 schrift filr angewandte Psychologie. 
 
 The next year a distinct advance towards synthesis and standardi- 
 zation of tests was made in the carefully prepared work of Toulouse, 
 Vaschide, and Pieron. 41 Without quoting results to be expected or 
 norms to be employed, explicit directions are given for the adminis- 
 tration of nearly fifty tests, more than half of which are on memory. 
 
 40 "L 'etude experimentale de 1 'intelligence, ' ' 1903. 
 ""Technique de Psychologie Experimentale," 1904. 
 
HISTOBY OF INTEREST IN INDIVIDUAL DIFFERENCES 7 
 
 Ways of scoring are also illustrated at some length. The tests sug- 
 gested have been selected from a wide and lengthy laboratory and 
 clinical experience, and are, some of them, unduplicated in America, 
 so far as I know. A condensed list will be given later. The methods 
 of scoring too, do not seem so well known as Kraepelin's, for in- 
 stance, perhaps because England and America are more apt to bor- 
 row from German than from French sources.* 
 
 There have been since then two types of test series in use,~one of 
 a simple nature useful in determining differences of large classes of 
 people, the other of a more elaborate sort, applicable to a study of 
 individual differences within a group, or to stages of development, or 
 in some studies to the elucidation of the tests themselves. Thus epi- 
 leptics, feeble minded, backward and truant children are studied as 
 different from the normal type; twins, bright and dull children, 
 younger and older children are compared, and individual differences 
 in fatiguability by mental work, etc., investigated by the use of tests. 
 
 2. Representative Lists of Tests 
 
 By way of comparison some of the more representative lists are 
 here given. They are not all complete, since the purely anthro- 
 pometric tests have been omitted. It will be noted that a given test 
 such as cancellation or tapping may be differently classified by dif- 
 ferent investigators. 
 
 Cattell's list, for students at Pennsylvania includes — 
 
 Eate of movement — of hand and arm through 50 cm. 
 
 Least noticeable difference in weight — lifted pairs (similar to Galton's test). 
 
 * After the experiments to be reported in this study had been made, there 
 appeared Burt's article in the British Journal of Psychology, 1909, on "Experi- 
 mental Tests of General Intelligence' 7 and Whipple's "Manual of Mental and 
 Physical Tests." The former contains four new and interesting tests, and an 
 elaborate treatment by the method of correlation. The latter is exactly what its 
 title would indicate. Besides minute and explicit directions for administration 
 and statistical interpretation of the fifty-four tests described, the published 
 norms and extensive bibliographies are particularly helpful. The present study 
 is a more specific attempt to determine relative values in the case of certain of 
 the tests from which on the basis of general experience and a critical survey, 
 Professor Whipple has chosen his standard series. 
 
 Finally there are now being published reports of the Committee on Tests of 
 the American Psychological Association, which began its work in 1907. So far 
 three studies have been reported: "Methods for the Determination of the In- 
 tensity of Sound," by W. B. Pillsbury; "The Measurement of Pitch Discrim- 
 ination," by C. E. Seashore; "The Determination of Mental Imagery," by 
 J. E. Angell ; all in Monograph Supplement No. 53 of the Psychological Review, 
 December, 1910. 
 
STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 Reaction time for sound. 
 Time for naming colors — 
 Space judgment — 
 Time judgment — 
 Memory and attention — 
 
 ten colors. 
 
 bisection of a 50-mm. line, 
 equate an interval to a 10-sec. standard, 
 number of letters correctly repeated 
 after one auditory presentation. 
 
 Jastrow's list for students at Wisconsin includes — 
 
 Eate of movement — touching two reaction keys 38 inches apart 
 
 in natural time, 
 touching two keys 3 inches apart in quickest 
 time. 
 Sense judgment — estimate an ounce. 
 
 equate two weights, 
 estimate 1 inch on the skin, 
 estimate position in guided movements, 
 equate bilaterally symmetrical free move- 
 ments. 
 
 Jastrow's list for volunteer subjects at the World's Fair. 
 
 Sensibility, of touch — 
 
 of touch and sight- 
 
 of sight only- 
 
 Memory — 
 
 Reaction time. 
 
 distances in length. 
 
 kinds of surface. 
 
 weights. 
 
 bilateral symmetry. 
 
 lengths. 
 
 direction. 
 
 location. 
 
 aiming at a target. 
 
 lengths of lines. 
 
 bisection, trisection, etc., of lines. 
 
 number of letters, words, squares, colors, 
 
 etc., seen in an exposure of 1/20 sec. 
 visual immediate, 
 recognition method for colors and forms. 
 
 This description of the list follows Binet's analysis. 
 Gilbert's list for testing school children. 
 
 threshold for lifted weights. 
 
 Muscle sense — 
 Suggestibility — 
 Voluntary motor ability "1 
 Fatigue 
 Reaction time. 
 Discrimination reaction. 
 Memory of time. 
 
 Oehrn's list for 10 subjects. 
 Perception — 
 
 size weight illusion, 
 rate of tapping. 
 
 counting letters, 
 proof reading, 
 cancellation test. 
 
HISTOEY OF INTEREST IN INDIVIDUAL DIFFEEENCES 
 
 Memory — 
 
 Association — 
 Motor — 
 
 Binet and Henri's suggested 
 Memory — 
 
 Images — 
 Imagination — 
 
 Attention — 
 
 time to learn 12 nonsense syllables, 
 time to learn 12 numbers, 
 adding one place numbers, 
 speed of writing from dictation, 
 speed of reading. 
 
 list. 
 
 of a geometrical design. 
 
 of 60-word sentences. 
 
 of musical phrases. 
 
 of colors (recognition method). 
 
 number of repetitions needed to learn 12 
 numbers. 
 
 letter square. 
 
 questions as to tastes, etc. 
 
 ink blots. 
 
 suggestion from abstract words. 
 
 coordination of a theme. 
 
 completion of a drawing. 
 
 construction of many sentences with given 
 nouns or verbs. 
 
 a ten-minute theme on a given subject. 
 
 development of a musical theme. 
 
 regularity of reaction times. 
 
 reproduction several times of a line seen 
 once. 
 
 speed at which two metronomes at different 
 rates can be counted. 
 
 simultaneous reading and writing of dif- 
 ferent content. 
 
 understanding of simple puzzle mechanisms. 
 
 differentiation of synonyms. 
 
 criticism of absurdities, fallacies. ^ 
 
 an increase-in-length-of-line trap. 
 
 discrimination of odors (odorless flasks). 
 
 name and unannounced sensation from im- 
 posing-looking apparatus (none given). 
 
 apprehension at second, slow trial of algo- 
 meter. 
 
 involuntary movements. 
 
 constancy in selection of rectangles, colors, 
 etc. 
 
 series of musical phrases. 
 
 kind of reaction to one photograph of 
 brutal horrors included in a series of 
 neutral scenes. 
 
 behavior at a sudden loud noise. 
 
 dynamograph. 
 
 (vaguely indicated) some form of maze test. 
 
 throwing 10 balls at a target. 
 
 It will be noticed that the emphasis is on the qualitative rather 
 
 Comprehension — 
 Suggestibility — 
 
 ^Esthetic choice — 
 
 Moral feelings- 
 
 Force — 
 Motor skill- 
 
10 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 than the quantitative side, even in a series to be given at one sitting 
 only. Following these suggestions, but with repeated sittings there is 
 
 Sharp's list, used with seven subjects. 
 
 Memory — 
 
 Images — 
 Imagination — 
 
 immediate for 12 letters, visual, 
 immediate for 12 numbers, visual, 
 immediate for words, auditory, disconnected, 
 immediate for sentences, short and long, 
 
 auditory, 
 for sounds, by question method, 
 letter square test. 
 questions, 
 ink blots. 
 
 puzzle watch and box. 
 development of themes, 
 questions on suggestions from abstract 
 
 Attention — 
 
 Observation- 
 
 Tastes — 
 
 Stern 's suggested list. 
 Type of perception — 
 
 Memory- 
 
 Apperception type 
 
 cancellation (in four variations). 
 
 reading time of concrete and abstract ma- 
 terial. 
 
 simultaneous reading aloud and writing. 
 
 description of picture exposed for 2 minutes. 
 
 memory of colors exposed for 5 seconds. 
 
 comparison of synonyms. 
 
 range of information about pictures. 
 
 number of pieces of sculpture, artists, mu- 
 sical composers named in 5 minutes. 
 
 naming one production of each of 10 com- 
 posers. 
 
 naming an author from hearing a selection 
 read. 
 
 things highly colored named in 5 minutes 
 
 (written), 
 things of vivid sound named in 5 minutes 
 
 (written), 
 color recognition, after 10 minutes' interval, 
 pitch discrimination with several minutes * 
 
 interval, 
 kind of mistakes in letter square test, 
 reproduction of melodies and rhythms after 
 
 several days' interval, 
 estimate of location of a rotating hand on 
 
 a dial after a given interval, 
 time to learn lists. 
 
 time to re-learn next day, noting accuracy, 
 reproduction of an anecdote immediately, 
 
 next day, a week, a month later, 
 reproduction of a story, 
 description of a picture, object, etc. 
 
HISTORY OF INTEREST IN INDIVIDUAL DIFFERENCES 
 
 11 
 
 Attention — 
 
 Combination ( construction )- 
 Judgment — 
 Natural tempo — 
 
 distractibility during work from alteration 
 in light. 
 
 distractibility during work from interrupt- 
 ing sounds. 
 
 formation of as many words as possible out 
 of a given selection of letters. 
 
 suggestibility by weights, odors, changes in 
 pitch. 
 
 constancy in rate on different days of beat- 
 ing a three-fold rhythm. 
 
 Binet's list, used with his daughters. 
 
 Association and imagery- 
 
 Attention — 
 
 Memory- 
 
 Space and time perception- 
 
 writing a list of 20 words. 
 
 first idea on auditory presentation of a word 
 with many questions for introspection. 
 
 writing sentences (time before beginning 
 noted). 
 
 completing sentences. 
 
 developing a theme. 
 
 writing down events recalled. 
 
 description of objects. 
 
 description of occurrences (pictures). 
 
 cancellation test, varied. 
 
 immediate memory of numbers heard. 
 
 number of glances needed to copy figures 
 and lines of prose. 
 
 copying a drawing exposed .07 of a second, 
 number of exposures needed. 
 
 regularity and judgment of reaction time. 
 
 amount of poetry learned in 10 minutes re- 
 called immediately and 6 months later. 
 
 immediate memory for unrelated words, 
 auditory. 
 
 immediate memory and description of ob- 
 jects seen. 
 
 immediate memory for drawings of objects 
 seen for 20 sec. 
 
 immediate memory of hieroglyphs seen for 
 15 seconds. 
 
 reproduction in movement of a given length 
 of line. 
 
 equating an interval to varied standards. 
 
 Toulouse,Vaschide and Pieron list. 
 
 Memory- 
 
 visual, of colors, lines, angles, curves, loca- 
 tion of dots in a circle, rates of movement. 
 
 auditory, of tones, chords, arpeggio inter- 
 vals. 
 
 muscular, of lines, curves, positions. 
 
 verbal, of numbers, letters, words, phrases 
 (auditory). 
 
12 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 objects, pictures of. 
 
 positions, jointed model of a human figure. 
 
 sketches. 
 
 musical, phrases, rhythms. 
 
 logical, of a prose passage, auditory and 
 visual. 
 
 localization, grouped and serial order of 16 
 printed nouns. 
 (All the above to be studied by both reproduction and recognition methods.) 
 
 time to learn long lists of numbers and 
 letters, length of retention of lists. 
 
 recognition of words in lists too long to 
 have been learned. 
 
 lists of words with prefix or suffix in com- 
 mon. 
 
 cancellation test of letters, hieroglyphs. 
 
 reaction time with discrimination, and ir- 
 regular intervals. 
 
 algometer. 
 — rate of tapping. 
 
 reaction time to sight, sound, touch. 
 
 first idea, orally, from a starting word or 
 object drawn. 
 
 words with or without specified letters. 
 
 associate or dissociate of a verb. 
 
 free association, orally, for 30 seconds from 
 a word or object drawn. 
 
 spelling words backwards, visual or audi- 
 tory. 
 
 giving syllables backwards, auditory. 
 
 theme about a picture or drawing. 
 
 species-genus first idea. 
 
 detection of absurdities, fallacies, etc. (oral 
 presentation) and in drawings. 
 
 completion of syllogisms. 
 
 criticism of given syllogisms. 
 
 Attention- 
 
 Suggestibility — 
 
 Perception type (objectivation)- 
 
 Association and imagery— 
 
 Imagination- 
 
 Abstract synthesis — 
 Judgment and observation 
 
 Eeasoning — 
 
 Cattell's Columbia freshmen tests. (* = discontinued.) 
 
 Sense discrimination and per- 
 ception of space and time — 
 
 Memory — 
 
 Imagery- 
 Motor — 
 
 reproduction and bisection of a line. 
 
 pitch discrimination. 
 
 sesthesiometer. 
 
 reproduction of regular rhythm. 
 
 perception of weight (distance). 
 
 numerals heard, immediate. 
 
 numerals seen, immediate. 
 
 logical, of a prose passage read aloud to 
 
 them, 
 retrospective, of line drawn and bisected, 
 
 after 50 minutes' interval, 
 questions, 
 ergometer. 
 
} 
 
 HISTOEY OF INTEEEST IN INDIVIDUAL DIFFEEENCES 13 
 
 rate and accuracy of dotting. — ■- 
 
 reaction time to sound. 
 
 tremor in drawing a line.* 
 Perception — reaction with discrimination.* 
 
 cancellation test. 
 
 naming 100 colors. 
 Association — first idea, written. 
 
 (opposites, written). 
 ^Esthetic choice — color liked and disliked of models shown. 
 
 Attention 
 Apperception 
 Suggestibility 
 
 Whipple in his Manual 42 does not propose his list as one to be 
 used in its entirety as an inventory of an individual, but would prob- 
 ably claim, and with much justice, that an adequate inventory would 
 require his 54 tests or more and an expenditure of something like 
 an equal number of hours. His list is not quoted, though it is the 
 most important single contribution of the last decade to the topic, 
 because it is readily accessible. It should be carefully studied by 
 any one whose interests lead him to read the present report. 
 
 3. Aim of the Present Study 
 
 Without discussing the difference in aim revealed in the character 
 of these series nor the results obtainable by the different methods, 
 this study is concerned with only the usefulness of simple tests now 
 employed or of similar tests designed to supplement or replace them 
 because of greater significance or greater adaptability in content or 
 method. With the exception of one or two association tests all are 
 of the simplest type, and the question raised is, ' ' If this kind of test 
 is the sort frequently used, is it the best of its kind for the purpose ? ' ' 
 To answer this adequately would necessitate collecting every simple 
 test of intelligence known and experimenting with it from the points 
 of view of make-up of the test, method of administration, results, 
 change with practise, with maturity, with fatigue, etc. — too long and 
 complicated a task for this study. By limiting the field, however, is 
 caused the main defect of this work. If more of the time which has 
 been spent over the statistics resulting from the data gained had 
 been given in the first place to administering more tests of one func- 
 tion more carefully to more subjects there might be some definite 
 value. Nevertheless, for such as it is, this study is now presented. 
 
 My best thanks are due to a friend who assisted in standardizing 
 and correcting 360 pages of one of the cancellation tests, to the three 
 friends who cheerfully served as subjects for so many hours in the 
 
 42 Op. tit. 
 
14 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 hot summer days of 1907, to N. who also helped in many of the later 
 calculations, and lastly to Professor Thorndike for his ever ready 
 counsel and patient assistance in the revision of both data and 
 treatment. 
 
 In general this study is divided into two sections, one in which 
 about 45 different tests repeated on from three to seven subjects are 
 discussed from the point of view of correlation of the tests, change 
 with short practise and reliability of a single trial, the other in which 
 five very different tests practised with nine subjects are discussed 
 from the point of view of change in each, and similarity of changes. 
 
II 
 
 EXPERIMENTAL WORK WITH SEVERAL GROUPS OF 
 
 TESTS 
 
 Concerning certain of the tests supposed to inventory an indi- 
 vidual 's mental functions and measure his differences from the type 
 which are frequently given, as, for instance, the Columbia freshman 
 tests, 43 we are still undecided as to their exact value. We need to 
 know, (1) whether they test fundamental qualities slowly changing 
 by general mental growth and the effects of training in general, or 
 whether they measure degree of attainment in some specialized 
 ability. If large areas of the mind are reached, then much might be 
 predicted from them; if only narrow habits are tested, then little 
 could be predicted from them. One line of evidence is their sus- 
 ceptibility to practise ; for a test in which there is much change in a 
 short period of practise is evidently measuring something other than 
 a general function — it might be specialized ability, or the fact of be- 
 coming adjusted to test conditions, or the adoption of some device 
 with regard to certain material. 
 
 We need to know, (2) in case general qualities can be measured 
 by these tests, whether the test chosen is the best of its kind, the most 
 typical. One line of evidence here is the correlation of different tests 
 all supposed to measure the same thing. 
 
 We need to know, (3) how accurately the few trials made, often 
 only one, will measure the function directly tested, how far, for 
 instance, the result may be affected by the understanding of the 
 subject of what he is to do and how he is to do it. The reliability 
 of first trials can be worked out to give light here. 
 
 We need to know too, (4) how far results are influenced by dif- 
 ferences in the method of administration. Can differences in atti- 
 tude be made in the subject by varied direction of the attention? 
 Practically the question is — ■ ' How could the tests now in use be im- 
 proved in significance and accuracy ?" 
 
 The methods at present in use with the students from Columbia 
 and Barnard colleges must of necessity be more or less rough and 
 ready, since only from fifty to sixty minutes are occupied in giving 
 
 48 For a full list and descriptions of these, see Wissler, ' ■ The Correlation of 
 Mental and Physical Tests," Psych. Bev. Mon. Suppl, Vol. 3, No. 6, 1901. 
 
 15 
 
16 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 some twenty to twenty-four tests; and in successive years they are 
 given by different experimenters. Some of the subjects, particu- 
 larly the girls, are too nervous to do themselves justice at the be- 
 ginning of the hour, a fact which, as seniors, they frequently recall 
 with amusement or deprecation. Comparison in such cases between 
 performance as freshman and as senior will tend to overweight the 
 gain shown in the results of the seniors' tests, and the consequent 
 inferences as to the beneficial effect of college training. 
 The problems with which this first section deals are — 
 
 A. How far is each test susceptible to practise, especially to short 
 
 practise ? 
 
 B. What is the value of each test as a measure of the individual's 
 
 ability in some general function or group of functions, 
 such as memory, association or sensory discrimination ? 
 
 C. How can we get the best possible measure from a single trial ? 
 In general the procedure was as follows : 
 
 1. Three subjects, a highly selected group, made twenty trials 
 of each of certain selected tests during six weeks in the summer of 
 1907. Of these three, N. had had comparatively little linguistic 
 training, but, on the other hand, had exceptional preparation in 
 psychology, particularly in giving tests similar to these. She was 
 unusually quick in thinking and talking, also in writing and hand 
 movements. W. and F. both had a more inclusive linguistic train- 
 ing, F. particularly so. Both had done graduate work in psychology, 
 not including, however, much work of this nature. W. was somewhat 
 variable in speed, F. was rather slower on the whole, with two 
 notable exceptions, and was the least likely of the three to be put out 
 or upset nervously. Conditions were made as uniform as possible 
 during the tests, and record kept of the weather and temperature 
 conditions from day to day. The association, perception, and mem- 
 ory tests were practised by the three subjects in a group. The dis- 
 crimination and motor tests were practised by each separately, as 
 individual attention and timing were necessary. The group work 
 took about three quarters of an hour daily, the individual work from 
 20 to 30 minutes for each subject. The last two sets of trials were 
 made under rather forced circumstances, as it became necessary to 
 complete the twenty sets a little earlier than had been expected. 
 The general trend of the practise curve was not affected however. 
 
 2. From experience with this group, called the " long-term- 
 practise group" for convenience, certain of these tests, along with 
 others supposed to be of a similar nature or to test the same mental 
 process, were repeated in the spring of 1908 with a larger group of 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 17 
 
 subjects varying from six to eight members. These were junior or 
 senior women students in Teachers College, four rather young, three 
 rather more mature, and one man, some of whose records in the as- 
 sociation tests had to be omitted owing to some difficulty with the 
 English language. As much as possible was done with these sub- 
 jects working in a group, for which purpose they met once a week 
 for two hours for six weeks. They made from two to ten trials with 
 different tests. Later, each came alone for work with some of the 
 tests requiring special apparatus or individual attention. -These 
 subjects are referred to as the ''short-term-practise group." 
 
 3. Certain random groups of college students were used either 
 as opportunity offered or definitely in order to procure a larger num- 
 ber of control cases. One such group of nineteen summer session 
 students spent an hour in 1908 in taking various association and per- 
 ception tests ; another group of similar size in the winter term spent 
 half an hour on some of the tests. These have been called the "in- 
 structed group. ' ' Single tests are frequently given to large groups 
 for demonstration purposes, and where available, these records have 
 been utilized to get a standard average and deviation for maturer 
 students working in a group. These are referred to as "control 
 cases. ' ' 
 
 In discussing the work each test is taken separately and report 
 made, first of general experience with the test, including the fresh- 
 man results for men and women, then of the instructed group, men 
 and women separately where so distinguished, next of the short- 
 term-practise group, last of the long-term-practise group. Thus 
 there is quoted first the result as found by the present test and 
 method; next the results from more mature students, sometimes by 
 a slightly different method; then the change taking place in naive 
 mature subjects with only a few repetitions ; last, what change may 
 take place even in habituated, mature subjects with more extended 
 practise. 
 
 A test in which there is not much change will, other things being 
 equal, be the more reliable to use for a single trial with naive sub- 
 jects. The "other things" must of course include ease with which 
 directions are understood, simplicity of required reaction, and free- 
 dom from all pitfalls or traps for the well-intentioned but unwary 
 subject. 
 
 For each group of tests the questions of change by practise, in- 
 tercorrelation and precision are then taken up and recommendation 
 made of one or another of the tests tested. 
 
18 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 1. Tests on Association 
 
 A. Descriptive 
 
 The first group of tests to be reported on will be those on associa- 
 tion. 
 
 The Columbia freshmen are given one test only, the first idea, the 
 Barnard freshmen that and an opposite test. 
 First Idea. 
 
 This consists of the blank given below. 
 
 House 
 
 Tree 
 
 Child 
 
 Time 
 
 Art (N. B. This and many other blanks 
 
 London appear here in reduced size.) 
 
 Napoleon 
 
 Think 
 
 Bed 
 
 Enough 
 
 The test is explained to the students as one of rapidity in think- 
 ing rather than of quality. They are told to write as quickly as pos- 
 sible after each word the first idea — preferably one word — that oc- 
 curs to them. Practise is given orally with a sample word, then the 
 students are handed the blank. The time taken to finish the blank is 
 taken on a stop-watch, and the blank is filed. 
 
 One's common observation in giving this test to the freshmen is 
 that it is particularly hard to follow the directions, and to write 
 down actually the first idea that occurs on reading the word. Sub- 
 jects will sit blankly, stopped by a word, obviously choosing the 
 fittest of several ideas, however well it may have been explained to 
 them that it is primarily a test of the rate rather than of the quality 
 of thinking. The averages calculated from 250 Columbia and 100 
 Barnard freshmen show that the men take 55.4 seconds to write down 
 10 ideas, the girls 71.8 seconds. The P.E. for Columbia students is 
 22.9, quite the largest P.E. found for any of the freshman tests. To 
 make these figures easily comparable with those to be given for sub- 
 jects after short and long practise, they may be put thus : in 15 sec- 
 onds men, as tested in the regular manner, wrote 2.7 first ideas, girls 
 wrote 2.1 first ideas, or the average time to call up and write one 
 idea is 5.54 seconds for men and 7.18 seconds for women. In this test 
 then, the girls seem specially hampered ; for the results of other tests 
 of the rate of association, such as adding, and giving the opposites 
 of words show no such superior speed for males. 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 19 
 
 The method used in the present investigation was to explain very 
 carefully just what was wanted, giving oral practise with two sample 
 words. Subjects were told to begin at the signal "go" and get as 
 much as they could done till the signal "stop" was given. They 
 were warned that they would not have much time, though the actual 
 number of seconds was not told them in advance. (The three sub- 
 jects who took the long term of practise soon came to know the time 
 allowed for the different tests.) For the first idea test, the time- 
 limit was 15 seconds. The score was kept in number of words 
 written. Three letters counted as a word if the subject could ex- 
 plain that he had surely thought of something. 
 
 A single trial with 37 unpractised subjects, 19 men and 18 women, 
 with the time-limit of 15 seconds gave an average of 5.6 words 
 written, with an average deviation of 2.19 or an average of 2.68 sec- 
 onds to call up and write a word. The men and women had exactly 
 the same average, but the A.D. for the men was 2.58, for the women 
 1.78. Unless then, the apparent sex difference in the freshman re- 
 sults is due to difference in the relative immaturity of the subjects, it 
 may be produced by the method of giving the test. (For conveni- 
 ence, the method by which a subject is told to work as quickly as pos- 
 sible and the time taken to finish the test is noted will be called the 
 "amount-limit" and the method by which the subject starts and 
 stops at a given signal, and a certain time-limit unknown to the sub- 
 ject beforehand is allowed, will be referred to as the "time-limit" 
 method. The latter has obvious conveniences in testing groups of 
 subjects.) In each test where both methods were used, comparison 
 will be made of the results by each method, and a special section 
 devoted later to a summing up of these results. 
 
 By the amount-limit method 2.7 first ideas were written in 15 sec- 
 onds by the men, by the time-limit method 5.6; by the women the 
 averages are 2.1 and 5.6 respectively. These differences suggest first, 
 that the amount-limit method leaves the test ambiguous, the time 
 being a measure partly of slowness in associations and partly of as- 
 sociations called up and rejected; second, that a time-limit acts as a 
 spur, making subjects work more quickly than if simply directed to 
 write as quickly as possible, and making them less fastidious in 
 selection of associations when speed is so much emphasized. It is 
 known that "controlled association time" is often shorter than free 
 association time, the theory being that the setting of the attention 
 and judgment beforehand holds certain paths open for use more 
 readily than others; it may be then that attention is aided in a 
 somewhat analogous fashion by the incentive to do as much as pos- 
 
20 STUDY OF TESTS FOE INDIVIDUAL DIFFEEENCES 
 
 sible in a given time. The anticipation of the signal "stop" seems 
 
 to give a more definite aim than merely one's best effort after speed. 
 
 TABLE I 
 
 Words written in Seconds required 
 
 15 seconds per word 
 
 Men Women Men Women 
 
 Amount limit 2.7 2.1 5.54 7.18 
 
 Time limit 
 
 Instructed 5.6 5.6 2.68 2.68 
 
 reversed 4.6 3.26 
 
 f 1st 7.0 2.14 
 
 Short J Average 7.85 1.91 
 
 [4th 8.2 1.83 
 
 1st 7.0 2.14 
 
 Long -{ Average 7.83 1.92 
 
 20th 8.6 1.75 
 
 { 
 
 It was, however, suggested, that the list of words as printed lent 
 itself to higher scores by the time-limit method than by the amount- 
 limit, as the more concrete words come near the beginning, and the 
 most difficult are the three last. To test this point, the list was type- 
 written in reverse order and then used as a time-limit test with two 
 other groups of students, 29 rather young women, and 34 in a mixed 
 group of men and women somewhat older. The average number 
 written in 15 seconds was 4.6 words. Asked to repeat the test com- 
 mencing with the bottom word, the average in 15 seconds was 4.8 
 words. Thus the greater speed does not seem to be entirely due to 
 the kind of words encountered at the outset. 
 
 In the short term of practise, 4 trials on different days by 6 sub- 
 jects by the time-limit method, the average was 7.85 first ideas 
 written in 15 seconds, or 1.91 seconds per word. In the long term of 
 practise, 20 trials by 3 subjects, the average was 7.83, or 1.92 seconds 
 per word. The number written at the first trial by each group was 
 7.0. Taking all the trials of these two groups into account, 85 in all, 
 there were 14 occasions, or 16 per cent, of the total number, when 
 the test was completed in 15 seconds. The two lowest records, made 
 only once each, were 3 and 5 first ideas, both considerably higher 
 than the freshman results by the amount-limit method. 
 
 The difference appears even more striking when the fairly con- 
 stant factor of speed of writing is discounted. Three subjects were 
 given six trials each in writing ten words of some familiar sentence* 
 under each other in a vertical column. The average time for the 18 
 
 * Two clauses from the Lord's Prayer: (1) Our Father, etc.; (2) Lead 
 us, etc.; and (3) "Little Jack Horner sat in corner eating his Christmas pie. " 
 The number of letters were 40, 43, and 48. 
 
EXPERIMENTAL WORK WITH SEVEEAL GROUPS OF TESTS 21 
 
 trials was 13.38 seconds or 1.34 seconds a word. Thirty subjects, 
 naive except for an hour's work in other tests, were asked to write a 
 single word similarly with a time limit first of 10 seconds, then of 15 
 seconds. Half of them wrote the word " watch" in the 10-second 
 test, the word ' "father" in the 15-second test; the other half wrote 
 "father" in the first test, "watch" in the second. The results were 
 for the 10-second test 5.1 words, for the 15-second test 7.75 words, or 
 an average time of 1.95 a word or .355 second a letter. Thus the 
 average extra time needed for association over mere writing isj in the 
 case of the amount-limit method, about five seconds a word; in the 
 case of the time-limit method less than 1 second a word. 
 
 In absolutely free association — i. e., when a starting word only 
 was given and the subjects wrote down whatever series of things they 
 thought of, an average of 11.5 words was written in 15 seconds, or at 
 the rate of 1.31 seconds a word. (Incidentally it is interesting to 
 note that serial connections are more rapidly written than even the 
 same word in repetition, thus : 
 
 Familiar sentences, 3 subjects, 18 trials, 1.34 seconds per word, .307 per letter. 
 Free association, 6 subjects, 30 trials, 1.31 seconds per word, .240 per letter. 
 "Father" or " Watch," 30 subjects, 60 trials, 1.95 seconds per word, .355 per 
 letter. 
 
 though this difference is partly due to the fact that the 18 trials came 
 from 3 practised subjects on different days, the 30 trials from 6 sub- 
 jects after the short term of practise, the 60 trials from 30 subjects 
 after 1 hour's work with various tests.) 
 
 It seems certain then that the first idea test, as usually given, does 
 not measure the rate of association. Nor apparently can any test 
 involving the writing of words do so. For not only is the average 
 rate of mere writing no less per letter than the average rate of writ- 
 ing words under some associative requirement, but in certain cases 
 where the description of the association involves writing a phrase or 
 long word such as "eyes, nose and mouth," "kerosene oil" or 
 "pussy-willow," the writing time entirely obscures the association- 
 time. 
 
 Considering it from the point of view of practise, in the short 
 irregular practise with the average score of 7.85 the fourth trial 
 showed a gain of 1.17 or 17 per cent, over the first. With the three 
 subjects who repeated the test twenty times there was a practise gain 
 of 1.6 or 23 per cent. 
 
 In the five trials with the absolutely free association test there 
 was quite the reverse of practise effect. The starting words used at 
 
22 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 the five trials were, respectively, house, read, black, table, ball. The 
 average amount done in 15 seconds was 11.5 words, or one word in 
 1.31 seconds, the deviation of the first trial from this 11.5 being 
 + 1.8, of the fifth — .8. 
 
 The correlation of the first idea test with other association tests 
 will be taken up later. 
 
 Opposites Test. 
 
 In giving this test the usual experience is that some words are 
 uniformly hard, and that when once at a loss for the opposite to any 
 word that has presented difficulty, an enormous amount of time may 
 be spent. Some subjects will go on writing the easier ones, returning 
 afterwards to those that have proved puzzling. If these have been 
 retained subconsciously there is probably a saving of time. Usually 
 no hint is offered about "skipping" in this way to the freshmen, 
 though where this test has been used in group work with children and 
 others, with a time limit, usually no skipping is allowed. It then 
 becomes impossible to know how much of the time is spent over per- 
 haps one word in the list, so that the final record is very much af- 
 fected by the inherent difficulty of the test-words. The standard set 
 prepared by Woodworth and Wells* is not in common use yet, and 
 the Columbia set presents several difficulties. It is as follows: 
 
 Write as quickly as you can beside each word in the column a word which 
 means the opposite thing from it. 
 
 barbarous 
 
 simple 
 
 rude 
 
 obscure 
 
 gentle 
 
 to expand 
 
 elation 
 
 adroit 
 
 loquacious 
 
 to degrade 
 
 to hinder 
 
 precise 
 
 permanent 
 
 repulsion 
 
 to respect 
 
 genuine 
 
 separate 
 
 deceitful 
 
 grand 
 
 * To be reported as a publication of the American Psychological Association. 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 23 
 
 Other sets used in comparison were : 
 
 Opposites Tests 
 
 day- 
 
 I 
 
 vertical 
 
 right 
 
 good 
 
 asleep 
 
 to spend 
 
 love 
 
 outside 
 
 absent 
 
 to reveal 
 
 rude 
 
 quick 
 
 brother 
 
 level 
 
 just 
 
 tall 
 
 best 
 
 ignorant 
 
 lie 
 
 big 
 
 above 
 
 past 
 
 tidy 
 
 loud 
 
 big 
 
 part 
 
 cruel 
 
 white- 
 
 backwards 
 
 motion 
 
 run away 
 
 light 
 
 buy 
 
 to hold 
 
 best 
 
 happy 
 
 come 
 
 generous 
 
 quick 
 
 false 
 
 cheap 
 
 proud 
 
 remember 
 
 like 
 
 broad 
 
 diligent 
 
 dressed 
 
 rich 
 
 dead 
 
 stupid 
 
 to be hit 
 
 sick 
 
 land 
 
 serious 
 
 lose 
 
 glad 
 
 country 
 
 frequently 
 
 mend 
 
 thin 
 
 tall 
 
 weary 
 
 disobey 
 
 empty 
 
 son 
 
 wicked 
 
 clean 
 
 war 
 
 here 
 
 to create 
 
 noisy 
 
 many 
 
 less 
 
 to enrage 
 
 rough 
 
 above 
 
 mine 
 
 stormy 
 
 cross 
 
 friend 
 11 
 
 serious 
 
 high 
 
 great 
 
 vertical 
 
 grand 
 
 up 
 
 hot 
 
 ignorant 
 
 clumsy 
 
 wet 
 
 dirty 
 
 rude 
 
 to win 
 
 new 
 
 heavy 
 
 simple 
 
 to respect 
 
 soft 
 
 late 
 
 deceitful 
 
 frequently 
 
 wider 
 
 first 
 
 stingy 
 
 to lack 
 
 wrong 
 
 left 
 
 permanent 
 
 apart 
 
 yes 
 
 morning 
 
 over 
 
 stormy 
 
 young 
 
 much 
 
 to degrade 
 
 motion 
 
 laugh 
 
 near 
 
 weary 
 
 forcible 
 
 winter 
 
 north 
 
 to spend 
 
 to float 
 
 weak 
 
 open 
 
 to reveal 
 
 straight 
 
 forget 
 
 round 
 
 genuine 
 
 to hold 
 
 wild 
 
 sharp 
 
 level 
 
 after 
 
 beginning 
 
 east 
 
 broken 
 
 unless 
 
 straight 
 
 known 
 
 wild 
 
 rough 
 
 raise 
 
 something 
 
 part 
 
 to bless 
 
 rough 
 
 stay 
 
 past 
 
 to take 
 
 love 
 
 push 
 
 permit 
 
 exciting 
 
 noisy 
 
 nowhere 
 
 precise 
 
 In scoring these, a mark of 2 was given for the best choice, 1 for 
 a second best choice, and for a bad choice. The key used in scoring 
 will be found, alphabetically arranged, in the appendix. From the 
 very fact that so many words could be offered as opposites to certain 
 
24 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 given words, it will be seen how valuable a standardized set would 
 be. In the various tables that follow a score for accuracy is given in 
 terms of the per cent, which the score given to the individual in ques- 
 tion was of the score he would have received had every opposite 
 written by him been rated as worth 2 credits. Thus a record of five 
 opposites valued as 2, 0, 1, 1, 2 respectively is scored 6/10 or 60 
 per cent. 
 
 First, to compare the various blanks used. Columbia freshmen 
 have not been put through this test. Barnard freshmen have usually 
 taken the "barbarous" blank, though 14 were given "vertical I." 
 "Barbarous" took 166 seconds on the average or 8.74 seconds per 
 word compared with 105 seconds, or 5.25 seconds per word for "ver- 
 tical I"; the scores for accuracy were (average) 69 per cent, and 72 
 per cent, respectively. The short-term practise group who also 
 worked with each blank, and by the same method, took 141 seconds, 
 or 7.42 seconds per word for "barbarous," and 89 seconds, or 4.45 
 seconds per word for "vertical I." Their average scores were 69 
 per cent, and 71 per cent. Thus the difference in time taken shows 
 that the ' ' barbarous ' ' blank is more difficult than ' ' vertical I. ' ' The 
 average score for "barbarous" is also lower than that for any other 
 blank, as may be seen from Tables II and III. An easier blank, 
 such as "serious" or "day" would probably be more suitable for 
 this type of subjects. 
 
 
 
 
 TABLE 
 
 II 
 
 
 
 
 
 Speed and 
 
 Accuracy in 
 
 Writing 
 
 Opposites 
 
 
 
 "Barbarous" 
 
 "Vertical I' 
 
 ' 
 
 "Vertical II 
 
 
 •O u 
 0> O 
 
 ill 
 
 I5 H 
 
 111 
 
 02 CT 
 
 «1L 
 Mi 
 
 o o 
 
 .t5 tM OB 
 
 ELSE 
 
 ill 
 
 g EN 
 
 EH.S 
 
 b 
 S 
 
 Ph 
 
 II 
 
 o F 
 o 
 
 (D 
 
 02 
 
 a 
 
 o 
 u 
 
 0) 
 
 & 
 
 > 
 
 m 
 
 O) o>© 
 
 g coin 
 
 •1 
 
 £g 
 
 on n 
 
 •O 0) 
 
 a P< 
 
 0>rt 
 
 o S 
 8 u 
 
 *! 
 
 a 1 
 
 d 
 
 o 
 6 
 fa 
 9 
 
 P. 
 
 <1 
 
 n .... 166 
 
 8.74 
 
 69 
 
 105 
 
 5.25 
 
 72 
 
 
 
 
 93 
 
 4.89 
 
 69 
 
 
 
 
 96.2 
 
 4.81 
 
 71 
 
 Seniors 
 
 Short-term ... 141 7.42 69 89 4.45 71 
 
 So far as these blanks reveal differences in maturity, there is a 
 decided improvement in speed with more mature subjects; the fresh- 
 men take a longer time than the short-term group at their first trial 
 with both the difficult blanks, and considerably longer than the 
 seniors. The accuracy is practically the same for all these three 
 groups on the same blanks. Looking also at Table III, all the rec- 
 ords from the short-term group are poorer than even the first record 
 of the more mature long-term group for "vertical II" which is a 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 25 
 
 fairly difficult blank, though the easier blank "day" seems too easy 
 to show differences in the groups of subjects. In this table all the 
 records are reduced to the amount done in 30 seconds, and the ac- 
 curacy score to percentage, whether the test was by amount-limit or 
 time-limit method, and no matter what the blank. 
 
 To compare differences in method, a group of Barnard seniors 
 were given "vertical II" by the amount-limit method, and a group 
 of Teachers College women students the same blank by the time-limit 
 method, with scarcely any difference in the results, though: what 
 there was, was in favor of the time-limit method, as will be seen by 
 Table III. These two groups were of about the same maturity, but 
 again with the slight difference in favor of the Teachers College 
 students, so that either this factor, or that of difference in method 
 may be responsible for the very slight difference in the figures. 
 
 TABLE III 
 
 Speed and Accuracy in Writing the Opposites of Given Words 
 
 Speed is measured by the number of seconds required per word. Accuracy is 
 measured by the average per cent, of the maximum credit that was obtained. 
 
 * ' Barbarous " " Vertical I " " Vertical II " " Serious ' ' 
 
 Test Test Test Test "Day "Test 
 
 Accu- Accu- Accu- Accu- Accu- 
 
 Speed racy Speed racy Speed racy Speed racy Speed racy 
 Amount limit 
 
 Freshmen 8.74 69 5.25 72 
 
 Seniors 4.89 69 4.81 71 
 
 Short term .... 7.42 69 4.45 71 
 
 Time limit 
 
 Instructed 4.62 73 2.36 93 
 
 Short f 1st 4.48 70 2.21 91 
 
 term | last .... 4.55 75 2.03 94 
 
 r 1st 3.23 91 3.13 86 2.50 94 
 
 ° ng J Average 2.48 88 2.22 88 2.19 95 
 
 erm [ 10th trial 2.17 89 1.76 90 2.07 94 
 
 To test the effect of practise, the short-term group were given six 
 different tests, the "day" being repeated after six weeks, giving 7 
 trials in all with the time-limit of 30 seconds, also "vertical II" 
 once with a time-limit of 30 seconds. The Columbia blanks were 
 given on the fifth day by the amount-limit method, so that a total of 
 10 trials was made by this group of subjects. 
 
 Since the "day" test when repeated after practise with "good," 
 "great," "vertical," and "right" shows so little gain the practise 
 effect is very slight, and the test continues to be an association test 
 rather than a series of specially trained responses. 
 
 Even special practise with the same blank shows rather slow im- 
 
26 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 provement. The long-term group used three blanks only, "day," 
 "serious," and "vertical II." After the first two trials these were 
 used in rotation till it was evident that the easy "day" blank had 
 been memorized. The other two were used ten times each, on alter- 
 nate days, and beginning alternately at the top and the bottom of the 
 column. There was, of course, a gain in speed, the time per word 
 being reduced from 3.23 to 2.17 and from 3.13 to 1.76 in the 10 trials, 
 but the rate is still much above that for writing the numbers from 
 one to twenty or other familiar series. 
 
 Comparing this test with the first idea in rapidity, it will be seen 
 that this form of controlled association does take slightly longer 
 with subjects practised with both tests. 
 
 TABLE IV 
 
 Seconds Eequired per Word to Write (1) The First Idea Called up by a 
 Printed Word, (2) A Series of Words Started by a Printed Word, and 
 (3) The Opposites of the Words of the "Day" Blank 
 
 (1) (2) (3) 
 
 Time limit 
 
 Instructed group 2.68 1.31 2.36 
 
 Short-term group 1.91 2.11 
 
 Long-term group 1.92 2.19 
 
 . Other controlled-association tests used in comparison with this 
 were: for the "instructed" group, two in number, the preceding 
 letter, and complete the word; for the "long-term" group, six in 
 number, these two and also the subject predicate, difference between, 
 Ebbinghaus combination, and addition; for the "short-term" group, 
 the first five given above, a different set of addition and subtraction, 
 noun and adjective, nonsense words, and one or two nonsense sen- 
 tences, genus species, multiplication. They will be taken up in that 
 order. 
 
 Except where otherwise stated, these were always given by the 
 time-limit method. 
 
 Preceding Letter. 
 
 The series of stimulus letters is as follows : 
 f 
 k 
 s 
 
 P 
 w 
 
 1 
 
 e 
 
 r 
 a 
 
EXPERIMENTAL WORK WITH SEFEEAL GROUPS OF TESTS 27 
 
 o 
 v 
 
 J 
 n 
 t 
 h 
 
 The time-limit was 15 seconds. The subjects were told to "write be- 
 side each letter the letter which precedes it in the alphabet," oral 
 examples being given by two letters. With 197 subjects, one trial, the 
 average number written was 5.5 letters, a clear mode of 5, a range of 
 from to 12 and an average deviation from the mode of 1.6. One 
 letter thus required 2.73 seconds (Av.) or 3 seconds (Mode). Intro- 
 spective evidence shows that this is a peculiarly difficult test to start 
 right in spite of the preliminary oral practise. Old habit asserts 
 itself to such an extent that many subjects are unable to react at all 
 without mentally repeating the whole of the alphabet up to the test 
 letter. Others try to repeat it backwards; others to make use of 
 visual imagery. If this is the first test given in an hour's work on 
 various tests, it seems particularly bad. When it is the sixth or 
 seventh test given, the average on three different occasions with 
 small groups, making 36 subjects in all, was 6.1 letters in the 15 
 seconds, or 2.46 seconds per letter, with an A.D. of 1.2. 
 
 The short-term group used it three times with an average of 7.3, 
 the first day's average, 5.6, deviating by — 1.7, the last by +1.0, 
 showing a very decided practise effect for so few trials. The long- 
 term group made averages of 7.3 letters or 2.05 seconds per letter, 
 6.3, or 2.05 seconds per letter, 8.6, or 1.74 seconds per letter, and 9.3, 
 or 1.61 seconds per letter, in their first four trials. They were also 
 very variable throughout the entire 20 trials. This test then seems 
 to be a specially bad one. 
 
 Complete the Word. 
 
 The form of the test was as follows : 
 
 1. ri 11. med 
 
 2. bon 12. bus 
 
 3. mil 13. spo 
 
 4. la 14. gam 
 
 5. flo 15. an 
 
 6. chi 16. che 
 
 7. dr 17. chu 
 
 8. fas 18. we 
 
 9. sk 19. rec 
 10. bra 20. par 
 
 21. chap 
 
28 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 Fifteen seconds was allowed. Eight subjects used it three times, 
 and the three subjects ten times, beginning with the first or second 
 column or at the end, after which they made ten more trials with 
 fresh sets. 
 
 In a first trial it is very noticeable that a subject may think of 
 long words in the beginning, and continue to think of them even 
 when shorter words are completed in the spelling out of the word 
 actually written, as "ri" suggesting "ribbon" when "rib" would 
 suffice, or when cognates would be shorter, such as rite for ritual. At 
 the same time it is introspectively an easier test than the first idea, 
 because, in the first place, the subject seems to be less suspicious of 
 what may be demanded of him, and feels more free to write down 
 what he has actually thought of ; in the second place, parts of words 
 seem to be more suggestive of whole words than one word is of 
 another, perhaps for two reasons; first the conditions are more like 
 ordinary reading, second the motor or auditory imagery or perhaps 
 the incipient movements of the speech organs seem to perform the 
 task of completion automatically, while all the judgment has to do is 
 to acquiesce. With both this and the absolutely free association test, 
 the factor of long words may increase the time taken through the 
 mere mechanics of writing. The statistical results will favor those 
 who think of short words as well as the rapid thinkers. 
 
 For the "instructed" group of 37 subjects the average number 
 of words completed in 15 seconds was 8 (1.88 seconds per word), with 
 a range of from 3 to 15, and an A.D. of 2.8. 
 
 TABLE V 
 
 Number of Words Completed in 15 Seconds 
 
 No. ofsubj. No. of Av. No. written Sec. req. per word 
 
 Men Women trials Men Women A. D. Men Women 
 
 Instructed group 19 18 1 8.2 7.7 2.8 1.83 1.94 
 
 Short-term group (using the same blank) : 
 
 1st 9.5 1.58 
 
 average 7 3 9.1 2.0 1.65 
 
 last 11.4 1.31 
 
 Long-term group (using different blanks) : 
 
 1st 9.3 1.61 
 
 average 3 10 10.5 .8 1.43 
 
 last 11.8 1.27 
 
 The short-term practise group in three trials made an average 
 of 9.1 words completed or 1.65 seconds per word, with a range of 
 from 4 to 15 and an A.D. of 2. 
 
 The long-term practise group averaged 10.6 words in 15 seconds 
 or 1.42 seconds per word in their first trial. After 10 trials with the 
 
EXPEEIMENTAL WOEK WITH SEVEEAL GEOUPS OF TESTS 29 
 
 same blank, improvement being very rapid, 10 more trials were made, 
 with two or three from the original blank introduced into each set. 
 The average was then 10.5, ranging from 9.3 on the eleventh day to 
 11.8 on the twentieth, showing a slight practise effect. Had the 
 word beginnings been absolutely new, the practise effect would pre- 
 sumably have been still less. 
 
 Six of the short-term practise group later took this test orally by 
 the amount-limit method. Eight trials were made with different lists. 
 In this way it could be seen how a poor record is made by the influ- 
 ence of some one combination which halts a subject unduly long 
 rather than by slowness in general. One list seemed easy for all sub- 
 jects, but no one list was hard for all subjects; one or two excep- 
 tionally poor records occurred with every list. The combination 
 "urn" halted three subjects a comparatively long time. One subject 
 made the worst record 7 times out of the 8, though in the written 
 test by the time-limit method she had been one of the best subjects. 
 Introspectively, all preferred the oral method. Compared with 
 other tests, completing words is less disturbing than the first idea> 
 but less definite than the opposites. 
 Subject-predicate. 
 
 As a test this is not in common use, so that the blanks were pre- 
 pared in round handwriting, which may have retarded the speed 
 somewhat as compared with the first idea and opposites tests, which 
 were printed. Mimeographed sets were later used for the short-term 
 practise group. 
 
 Subject-predicate Lists 
 
 convenes 
 
 matriculates 
 
 stings 
 
 brays 
 
 confesses 
 
 butts 
 
 scratches 
 
 parries 
 
 steals 
 
 lubricates 
 
 explodes 
 
 earns 
 
 waxes 
 
 preaches 
 
 hatches 
 
 hops 
 
 bleats 
 
 prescribes 
 
 plays 
 
 disperses 
 
 sucks 
 
 illuminates 
 
 swims 
 
 arrests 
 
 reverberates 
 
 plants 
 
 paints 
 
 enlists 
 
 lectures 
 
 hoards 
 
 chases 
 
 flies 
 
 buys 
 
 flashes 
 
 smoulders 
 
 alleviates 
 
 experiments 
 
 quacks 
 
 rings 
 
 ordains 
 
 extinguishes 
 
 strikes 
 
 applauds 
 
 fights 
 
 nourishes 
 
 re-acts 
 
 reaps 
 
 sews 
 
 condemns 
 
 sneers 
 
 ebbs 
 
 cackles 
 
 navigates 
 
 graduates 
 
 performs 
 
 composes 
 
 inherits 
 
 freezes 
 
 burns 
 
 sells 
 
 shoots 
 
 learns 
 
 riots 
 
 drives 
 
 amputates 
 
 bites 
 
 blows 
 
 sues 
 
 cleanses 
 
 neighs 
 
 stitches 
 
 testifies 
 
 disbands 
 
 crows 
 
 rotates 
 
 trumps 
 
 owes 
 
 governs 
 
 calculates 
 
 fades 
 
 shines 
 
 adjourns 
 
 roars 
 
 haunts 
 
 bets 
 
 hammers 
 
 sings 
 
 occurs 
 
 melts 
 
 tolls 
 
 marries 
 
 sacrifices 
 
 raves 
 
 limps 
 
 foretells 
 
 trots 
 
 flows 
 
 surrenders 
 
 withers 
 
 barks 
 
30 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 Subjects were warned not to supply a subject by forming a noun 
 in "er" from the verb such as "singer" sings, nor by using indefinite 
 words as "man," "boy," but to supply the definite agent such as 
 "bird." Two or three examples were illustrated. One hundred 
 verbs were made up in ten sets of ten, each being used twice for the 
 long term of practise, and once each on typewritten sheets for the 
 short term of practise. Unfortunately for strict comparison they 
 were not given in the same order for the short practise as for the 
 long. The scoring for accuracy was done as for the oppo sites test, 
 giving 2 for the best choice, 1 for a poorer one, for a poor one. 
 
 N.= 
 
 Ace. = 
 
 Order given 1 
 Tests confesses 
 Subjects N. Ace. 
 
 Bu 10 75 
 
 Gr 2 25 
 
 J 4 63 
 
 L 5 70 
 
 M 5 30 
 
 Ba 10 65 
 
 Bf 
 
 Averages . 6 
 
 Medians .. 64 
 
 TABLE VI 
 
 number of subjects written to fit given predicates in 20 seconds. 
 = per cent, of maximum credits obtained. 
 
 2 
 
 ebbs 
 
 N. Ace. 
 
 9 100 
 6 100 
 6 67 
 6 100 
 6 100 
 9 44 
 
 100 
 
 cackles navigates 
 N. Ace. N. Ace. 
 
 89 
 71 
 71 
 79 
 60 
 
 7.3 
 
 6 92 
 2 100 
 
 6 83 
 
 71 
 33 
 
 9 33 8 75 
 
 5.3 
 
 brays convenes graduates 
 N. Ace. N. Ace. N. Ace. 
 
 71 
 
 79 
 
 8 100 
 5 100 
 
 8 88 
 5 100 
 5 100 
 
 9 89 
 8 100 
 6.8 
 
 100 
 
 8 99 
 
 5 90 
 
 6 50 
 4 63 
 
 4 100 
 
 7 86 
 
 5 80 
 5.5 
 
 86 
 
 6 92 
 
 6 100 
 
 7 86 
 5 80 
 3 100 
 
 8 63 
 7 100 
 6.0 
 
 92 
 
 8 9 10 
 
 performs stings matriculates 
 N. Ace. N. Ace. N. Ace. 
 
 4 75 10 80 
 
 9 100 
 7 64 
 4 100 
 
 7 64 
 
 8 88 
 
 7.5 
 
 75 10 70 
 
 30 
 
 6 92 
 5 50 
 
 6 
 
 71 
 93 
 50 
 
 10 55 10 30 10 70 
 8 100 8 69 8 100 
 
 8.4 
 
 88 
 
 69 
 
 71 
 
 TABLE VII 
 
 N. = number of subjects written in 20 seconds to fit given predicates. 
 
 Ace. = per cent, of maximum credits obtained. 
 
 First trials 
 1-10 
 N. Acc. 
 
 Av. Median 
 
 performs 5.6 100 
 
 stings 7.0 94 
 
 matriculates 6.6 93 
 
 ebbs 7.0 94 
 
 brays 8.3 95 
 
 cackles 7.6 94 
 
 convenes 5.6 100 
 
 navigates 7.0 81 
 
 graduates 6.3 93 
 
 confesses 8.8 100 
 
 Average 7.0 95 
 
 Second trials 
 
 11-20 
 
 N. Acc. 
 
 Av. Median 
 
 9.1 
 
 100 
 
 7.1 
 
 93 
 
 7.1 
 
 100 
 
 8.0 
 
 94 
 
 8.8 
 
 100 
 
 7.3 
 
 100 
 
 8.0 
 
 88 
 
 8.6 
 
 94 
 
 7.1 
 
 100 
 
 8.6 
 
 95 
 
 8.0 
 
 96 
 
 The results for the short-term group are shown in Table VI. The 
 practise effect is apparently very slight, the last five tests being only 
 a trifle better in speed or accuracy. Further tests are, however, 
 needed to separate the influence of differences of the tests in diffi- 
 
EXPEBIMENTAL WOBK WITH SEVEBAL GBOUPS OF TESTS 31 
 
 culty from that of practise, and from that of the chance variations 
 in the subjects. 
 
 The results for the long-term group are summarized in Table 
 VII. The practise effect of ten trials, including one of the same 
 blank, is in general to increase the speed only by a seventh, leaving 
 the accuracy uninfluenced. 
 
 The time required in these tests is about the same as that in the 
 difficult "vertical" opposite test. 
 
 The "Difference Between." 
 
 The form of the test used is as follows : 
 
 Answer these questions as quickly and as well as you can. 
 
 1. What is the difference between grab and take? 
 
 2. What is the difference between eat and devour? 
 
 3. What is the difference between a stream and a river? 
 
 4. What is the difference between a wagon and a cart? 
 
 5. What is the difference between sorry and sad? 
 
 6. What is the difference between naughty and bad? 
 
 7. What is the difference between homely and ugly? 
 
 8. What is the difference between right and correct? 
 
 Other lists used were : 
 II 
 confess, reveal 
 confine, limit 
 colleague, partner 
 bend, curve 
 # resistance, opposition 
 deceive, mislead 
 adrift, afloat 
 extend, increase 
 
 IV 
 show, indicate 
 watch, observe 
 trial, test 
 contract, bargain 
 peace, repose 
 clear, obvious 
 cleanse, purify 
 classify, arrange 
 
 VI 
 
 chuckle, giggle 
 honest, honorable 
 procure, obtain 
 haste, hurry 
 crayon, chalk 
 antagonist, opponent 
 puff, swell 
 abrupt, blunt 
 
 III 
 above, over 
 demonstrate, illustrate 
 deluge, flood 
 guardian, keeper 
 merry, gay 
 bring, fetch 
 heavy, weighty 
 innocent, harmless 
 
 V 
 get, provide 
 win, gain 
 pair, two 
 parcel, bundle 
 womanish, feminine 
 put, place 
 boat, ship 
 clever, talented 
 
 VII 
 walk, march 
 ignore, overlook 
 corpse, carcass 
 early, soon 
 allude, refer 
 drag, pull 
 
32 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 VIII 
 
 walk, march 
 deceive, mislead 
 corpse, carcass 
 colleague, partner 
 drag, pull 
 adrift, afloat 
 try, test 
 extend, increase 
 
 The subjects were told that the quickest way to answer was either 
 
 to explain one word in terms of the other, or to write 1 = 2 =*= 
 
 , not wasting time by repetition. Notwithstanding this, many to 
 
 whom it was given used an unnecessary number of words in expla- 
 nation, thus taking longer to write. From the point of view of time 
 consumed, then, it is not a useful nor a satisfactory test whether 
 given by the time-limit or by the amount-limit method. Not only as- 
 sociation and speed of writing enter in, but the ability to profit by 
 the advice in the instructions, and ability to condense — also, of 
 course, linguistic discrimination. This test is, besides, not very easy 
 to score, as the answers may vary considerably. 
 
 Blank I was kindly filled in at leisure by one of the professors 
 in the English department. Answers were then compared with these 
 standard answers and each of the eight scored 2, 1 or 0, as in the case 
 of the opposites and subject-predicate tests. For the remaining 
 blanks, dictionaries and books of synonyms were resorted to for 
 standard answers, or, failing anything sufficiently discriminating 
 there, the experimenter's own judgment of the best answer in the 
 group was followed. 
 
 An "instructed" group of about 200 were tested with Blank I, 
 time-limit of 120 seconds. In 49 of these chosen at random the aver- 
 age number of answers written was 4.4, with an A.D. of 1.08 and a 
 range of 2 to 8. The average score for accuracy was 89 per cent, 
 (reliability 1). 
 
 The short-term practise group took this test only twice, using 
 Blanks I and VIII. The reason more time was not spent with them 
 on the various blanks was that previous experience with the long- 
 term practise group seemed to indicate that the test was not a valu- 
 able one. For the same reason and also because the 49 control cases 
 from the "instructed" group were in terms of time-limit, this group 
 were tested by the amount-limit method. Their record for Blank I 
 was: average time taken 217 seconds, score for accuracy 73 per cent. ; 
 for Blank II, 233 seconds, score for accuracy 63 per cent. ; for both 
 blanks together, average time taken, 225 seconds, A.D. 25.5, average 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 33 
 
 score 68 per cent. For them, then, Blank I was easier since they 
 made a better showing with it, although it was the first one given. 
 
 An "instructed" group of 49, tested with Blank I, with a time- 
 limit of 120 seconds, averaged 4.4 answers written, A.D., 1.08. The 
 average accuracy was 89 per cent. 
 
 The long-term practise group used seven different blanks alto- 
 gether, each one three times except the last, beginning with the 1st, 
 3d, or last of the 8 pairs of terms. A time-limit of 60 seconds was 
 allowed. Their average for Blank I was 4.6, score of 66 per cent. 
 The average number written for all 20 trials was 3.2, the first day's 
 average deviating by + 1.4, the last by + .4. The average score for 
 accuracy was 70 per cent., the first day's average deviating by + 6 
 per cent., the last by + 3 per cent. Thus the difference in the diffi- 
 culty of the blanks again disguises any practise effect. If the records 
 of the first three trials which were made with Blank I are omitted, 
 the average number written is 2.7, the fourth day's average deviating 
 by — .7, the last by + .9, so that there seems a slight gain in speed. 
 The average score for accuracy is then 77 per cent., the fourth day's 
 average deviating by — 2 per cent., the last by — 4 per cent. 
 
 Nothing can be surely inferred from these records save that for 
 them less than 20 seconds sufficed to think of and write out a differ- 
 ence (only 13.1 seconds for Blank I). A much longer time limit 
 should have been given. 
 
 On the whole, as will appear when the facts concerning correla- 
 tions and reliabilities are given, this test, if useful at all, is useful 
 only as a specialized measure of linguistic knowledge and facility in 
 expression. The times 27.3 seconds per difference for 49 subjects 
 using Blank I, 27.1 seconds per difference for 6 subjects using Blanks 
 I and VIII, and 18.8 seconds for 3 subjects using Blanks I-VII, 
 show that an elaborate process of selective thinking is involved. 
 Ebbinghaus Combination Test. 
 
 This test was as follows. For the short-term group certain para- 
 graphs of convenient length, averaging 100 words, were chosen from 
 such varied materials as newspaper reports, scientific articles, essays, 
 novels, narrative poems. These were typewritten, with 10 to 16 
 words, according to the length of the paragraph, omitted in various 
 places, blank spaces being left in their stead. One such paragraph 
 was placed before the subject, who was instructed to write down an 
 appropriate word for each space. The time taken was noted, and a 
 score was made of the suitability of the words supplied in terms of 
 per cent, of a perfect record. Five of the short-term practise group 
 
34 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 took ten such tests, repeating the first paragraph used at the 10th 
 trial three weeks later. 
 
 In general, subjects will either skim two thirds to the whole of 
 the paragraph at the outset, going back to fill in the spaces, or they 
 will rush at the first phrase, fill in the first thing that occurs, and 
 get tangled up before the end of the first sentence unless the subject 
 matter is very easy. From one or two such experiences the subject is 
 generally led to adopt the other method. 
 
 The short-term group took an average of 103 seconds to complete 
 a paragraph, with an A.D. of 32. Comparing their two trials (three 
 weeks apart) with the same paragraph there was an improvement in 
 average speed from 173 seconds to 71 seconds, the A.D.'s 33 and 6 
 respectively. Their accuracy rose from 70 per cent. t6 80 per cent, or, 
 omitting one subject who seemed very much upset at the first trial, 
 it was 80 per cent, on both occasions. 
 
 The long-term group was tested with 20 paragraphs averaging 
 92 words long, each with ten words omitted ; they averaged 80.2 sec- 
 onds, A.D. 18 seconds. Variations of 10 per cent, or less in the length 
 of the passage caused no appreciable differences in the time required. 
 Variations in the content are very influential. The poetry was diffi- 
 cult for these subjects, the average time for that being 108 seconds. 
 Newspaper reports were easy, the average time for them being only 
 54.4 seconds. Picking the first trial of each kind of material, and 
 comparing it with the last of each, there was an improvement in 
 speed from an average of 104 seconds to 89 seconds. These figures 
 do not measure practise with surety, owing to possible variations in 
 the difficulty of even the same kind of material. The average accu- 
 racy was 87 per cent, with no discoverable practise effect. The para- 
 graphs they used are given in the appendix. 
 
 In general it appears that adaptation to the form of problem set 
 by the Ebbinghaus test is likely to count considerably, especially with 
 untrained subjects. 
 Addition. — The blank used was as follows: 
 
 
 
 Addition Examples 
 
 
 
 17 
 
 26 
 
 27 
 
 72 
 
 23 
 
 42 
 
 51 
 
 24 
 
 14 
 
 47 
 
 38 
 
 47 
 
 83 
 
 39 
 
 86 
 
 91 
 
 82 
 
 19 
 
 81 
 
 54 
 
 54 
 
 63 
 
 45 
 
 26 
 
 36 
 
 17 
 
 42 
 
 38 
 
 91 
 
 36 
 
 26 
 
 51 
 
 47 
 
 82 
 
 26 
 
 27 
 
 24 
 
 83 
 
 19 
 
 45 
 
 72 
 
 14 
 
 39 
 
 62 
 
 63 
 
 23 
 
 47 
 
 86 
 
 54 
 
 54 
 
EXPEBIMENTAL WORK WITH SEVEBAL GBOUPS OF TESTS 35 
 
 41 
 
 53 
 
 67 
 
 78 
 
 86 
 
 52 
 
 67 
 
 86 
 
 37 
 
 32 
 
 86 
 
 34 
 
 23 
 
 96 
 
 44 
 
 23 
 
 78 
 
 45 
 
 72 
 
 36 
 
 35 
 
 19 
 
 67 
 
 23 
 
 68 
 
 45 
 
 52 
 
 19 
 
 45 
 
 23 
 
 13 
 
 86 
 
 78 
 
 67 
 
 72 
 
 68 
 
 23 
 
 67 
 
 78 
 
 36 
 
 77 
 
 35 
 
 23 
 
 37 
 
 68 
 
 86 
 
 67 
 
 86 
 
 96 
 
 39 
 
 A score of 1 for each column added was given and 0.5 deducted 
 for each wrong figure in an answer. The time limit was 60 seconds. 
 The results as to rate will be discussed in connection with those of 
 the next test. Since these experiments were made, it has been shown 
 by Wells and Thorndike that even so familiar a process is, under test 
 conditions, subject to adaptation and practise effects. In these sub- 
 jects these effects were shown chiefly or wholly in the speed of the 
 process. The short-term group averaged 16, 19, and 18 columns, and 
 .5, .67, and 1.33 errors in three trials on February 15, March 7, and 
 March 7. The long-term group gained in twenty trials about 20 per 
 cent, in speed but lost somewhat in accuracy, so that their net im- 
 provement was 17 per cent. 
 
 Addition and Subtraction. 
 
 The short-term group used a blank, given on the next page, from 
 the collection prepared by Woodworth and Wells. 
 
 The test consists of adding a certain number to each figure in 
 succession in the column, or subtracting it, as directed, and writing 
 down the result. One column was counted as a test, making 25 times 
 that a given number was added or subtracted and the result written. 
 Twelve such tests were made, six times with a time-limit of 40 sec- 
 onds, six times with a time-limit of 30 seconds. In cases where a sub- 
 ject completed the series in less than the allotted time her time was 
 recorded. The key numbers were 3, 4, 5, 6, 7, 8, each added in one 
 test, subtracted in another. Four tests were made in succession, the 
 order in which they were given being as follows : 
 
 I. 
 
 7 added 
 
 3 subtracted 
 
 4 added 
 
 5 subtracted 
 
 40 sec. 
 
 II. 
 30 sec. 
 
 5 added 
 
 \ 40 sec. 
 7 subtracted J 
 
 3 added 
 
 \ 30 sec. 
 4 subtracted > 
 
 III. 
 
 6 added I 4Q gec 
 
 8 subtracted > 
 
 6 subtracted ) OA _ 
 >• 30 sec. 
 8 added i 
 
36 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 64 
 
 72 
 
 47 
 
 30 
 
 49 
 
 35 
 
 43 
 
 56 
 
 62 
 
 51 
 
 35 
 
 44 
 
 57 
 
 30 
 
 64 
 
 31 
 
 68 
 
 56 
 
 49 
 
 37 
 
 74 
 
 44 
 
 67 
 
 60 
 
 53 
 
 36 
 
 28 
 
 71 
 
 67 
 
 73 
 
 46 
 
 48 
 
 25 
 
 63 
 
 55 
 
 53 
 
 40 
 
 47 
 
 65 
 
 61 
 
 61 
 
 43 
 
 70 
 
 36 
 
 71 
 
 66 
 
 41 
 
 42 
 
 33 
 
 69 
 
 62 
 
 34 
 
 38 
 
 37 
 
 25 
 
 39 
 
 28 
 
 39 
 
 40 
 
 33 
 
 65 
 
 32 
 
 57 
 
 73 
 
 41 
 
 59 
 
 26 
 
 38 
 
 50 
 
 31 
 
 68 
 
 63 
 
 42 
 
 60 
 
 66 
 
 58 
 
 58 
 
 48 
 
 27 
 
 32 
 
 52 
 
 54 
 
 51 
 
 59 
 
 70 
 
 46 
 
 69 
 
 52 
 
 26 
 
 55 
 
 29 
 
 45 
 
 34 
 
 27 
 
 74 
 
 72 
 
 45 
 
 29 
 
 50 
 
 54 
 
 As we now know through the work of Browne, 44 Stone, 45 and 
 others, the adding and subtracting abilities are two very different 
 things ; also some figures are easier to handle than others, a combina- 
 tion such as 9 + 2 being different from and easier than 2 + 9. These 
 facts complicate the issue. 
 
 However, it seems clear that adaptation to the test does bring 
 about a practise effect in the first few trials. The speed with + 8 
 in the last of the twelve tests is for every subject save Ji. greater than 
 for + 7 in the first of the twelve. 
 
 By any rational estimate also the second day's records are above 
 the first in general, and in the case of all but one of the subjects 
 measured. They were so probably for Bu. also. 
 
 Using the easiest set of these additions of a 1 place to a 2 place 
 number (+3), we find the time per operation to be Bu., .76 second ; 
 Gr., .96 second; Ji., 1.04 seconds; Le., 1.43 seconds, and Mo., 1.43 
 
 444 'The Psych, of the Simpler Arithmetical Processes," Am. J. of Psych., 
 17, 1906. 
 
 ""Arithmetical Abilities . . .," Col. Contr. to Educ, 19, 1908. 
 
EXPERIMENTAL WOBK WITH SEVEBAL GBOUPS OF TESTS 37 
 
 TABLE VIII 
 
 Eesults in the Add and Subtract Columns Test from the Short-term 
 
 Practise Group 
 
 A = amount done in time limit. 
 
 E = errors. T = seconds actually taken. 
 
 Column 
 
 Operation 
 
 Time limit in 
 
 seconds 
 
 I 
 
 +7 
 
 40 
 
 2 
 -3 
 
 40 
 
 3 
 
 +4 
 
 30 
 
 4 
 —5 
 
 30 
 
 5 
 40 
 
 6 
 
 —7 
 
 40 
 
 7 
 +3 
 
 30 
 
 8 
 —4 
 
 30 
 
 9 
 
 +6 
 
 40 
 
 10 
 -8 
 
 40 
 
 11 
 —6 
 
 30 
 
 12 
 
 +8 
 
 30 
 
 Bu. 
 
 A 
 E 
 
 f 
 
 ! 
 
 ? 
 
 ? 
 
 25 
 
 25 
 
 25 
 
 25 
 
 25 
 
 25 
 
 25 
 
 25 
 
 
 T 
 
 
 
 
 
 23 
 
 27 
 
 19 
 
 24 
 
 22 
 
 34 
 
 24 
 
 25 
 
 Gr. 
 
 A 
 
 E 
 
 21 
 1 
 
 ? 
 
 22 
 1 
 
 20 
 
 25 
 
 24 
 
 25 
 
 17 
 
 25 
 
 1 
 
 25 
 
 25 
 
 25 
 
 
 T 
 
 
 
 
 
 34 
 
 
 24 
 
 
 34 
 
 36 
 
 24 
 
 25 
 
 St. 
 
 A 
 
 E 
 T 
 
 13 
 
 22 
 3 
 
 11 
 
 12 
 
 1 
 
 25 
 
 38 
 
 21 
 
 25 
 
 26 
 
 21 
 
 25 
 34 
 
 21 
 
 20 
 1 
 
 21 
 
 Ji. 
 
 A 
 E 
 T 
 
 20 
 
 ? 
 
 22 
 
 14 
 
 16 
 
 11 
 
 21 
 
 14 
 
 21 
 1 
 
 11 
 
 13 
 
 9 
 
 L. 
 
 A 
 
 E 
 T 
 
 9 
 
 21 
 
 16 
 
 11 
 
 18 
 
 17 
 
 21 
 
 16 
 
 24 
 1 
 
 20 
 
 16 
 
 20 
 
 Mo. 
 
 A 
 
 E 
 T 
 
 18 
 
 18 
 
 18 
 
 13 
 
 
 
 
 13 
 
 25 
 38 
 
 14 
 1 
 
 12 
 1 
 
 17 
 
 Ba. 
 
 A 
 
 E 
 T 
 
 19 
 
 12 
 
 1 
 
 19 
 
 15 
 
 
 
 
 
 
 
 
 
 Bf. 
 
 A 
 E 
 T 
 
 17 
 
 21 
 
 25 
 1 
 
 13 
 
 
 
 
 
 
 
 
 
 seconds; a median of 1.04 and an average of 1.12 seconds. On 
 March 15 the short-term group was tested with 100 mixed examples, 
 such as 9 + 7, 8 — 3, 6 — 2, 5 + 8, etc., 70 seconds time being given. 
 The results were Bu., 100 ; Gr., 100 ; Ji., 69 ; Le., 63 ; Mo., 67 ; Ba., 64 ; 
 Bf., 63. Le. made 1 and Ba. 2 errors. The median time per opera- 
 tion was thus 1.04 seconds, as for the easiest addition to a 2-place 
 number. The average time was probably .9 second. In adding in 
 columns with 5 two-place numbers, for example, in which about three 
 fourths of the additions are to a two-place number, and in which the 
 number added is more often harder than easier than 3, the results 
 were, after the first trial, an average of .67 second per operation 
 (median .87 second). Although the average especially is perhaps 
 too low because the number of actual conscious operations was prob- 
 ably reduced by grouping in the case of the more rapid workers, the 
 fact remains that the mere writing time for a two-place number may, 
 
38 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 especially with slow writers, be greater than the time required to add 
 a one to a two place number without writing. One has only a choice 
 of evils. Column addition permits grouping and so mixes the rate 
 of association with the power to associate three numbers with their 
 sum in one connection. A test in writing additions and subtractions 
 with two place answers measures the rate of mere writing in very 
 rapid computers or very slow writers. 
 
 Noun and Adjective. 
 
 Two blanks with 20 adjectives on each were arranged as follows : 
 
 Complete the following sentences, 
 after the model of the first one, that 
 is, by adding to each a noun at the 
 beginning, and a second adjective at 
 the end — the whole to make sense: 
 
 The hill 
 
 high and wooded. 
 
 soft 
 
 cold 
 
 new 
 
 smooth 
 
 red 
 
 round 
 windy 
 
 clean 
 
 bent 
 wooden 
 
 deep 
 empty 
 narrow 
 
 loose 
 
 bitter 
 
 level 
 
 stale 
 
 oily 
 
 heavy 
 woolen 
 
 II 
 
 Complete the following sentences, 
 by adding a subject and an addi- 
 tional adjective, as m the first sen- 
 tence : 
 
 Her taste 
 
 refined and delicate. 
 
 portable 
 unexpected 
 
 ridiculous 
 interesting 
 
 imported 
 
 probable 
 
 tapering 
 dangerous 
 
 complete 
 
 unusual 
 
 metallic 
 
 spacious 
 
 painless 
 
 excessive 
 seasonable 
 
 desolate 
 
 frequent 
 
 distinct 
 
 select 
 
 temporary 
 
 A score of 1 was given for each appropriate word written, ma- 
 king 40 the maximum score for a test. Sometimes an indeterminate 
 adjective such as "nice" or "long" would be written several times 
 in succession, and the possibility of this detracts from the value of 
 the test. One subject wrote the pronoun "it" instead of a noun, as 
 directed, and so made a low scoring; otherwise this seems an easy 
 test, for the average accuracy score was 38, or 95 per cent. 
 
 The short-term group took this test four times only, the first time 
 with a time-limit of 120 seconds, the other three times by the amount- 
 
EXPEBIMENTAL WOBK WITH SEVEBAL GBOUPS OF TESTS 39 
 
 limit method. The average time taken to finish was 135 seconds, A.D. 
 27, or an average speed per word written of 3.37 seconds. There was 
 a slight practise effect in speed even with so few tests, but none in 
 the accuracy. It was written more slowly than the opposite and 
 subject-predicate tests, but this may be due to the arrangement of 
 the blank, and the need of an additional movement of the hand. 
 Blank I. is, so far as the records from six subjects go, much easier 
 than Blank II., taking only about three fourths as long with equal 
 precision. 
 
 English and Nonsense. 
 
 The following blank was used three times, a time-limit of 60 sec- 
 onds being given for each section with 3 minutes interval between the 
 sections. 
 
 A. Mark the (familiar) English words among the following groups of 
 letters : 
 
 nop 
 
 yas 
 
 jeb 
 
 cug 
 
 pin 
 
 warn 
 
 hay 
 
 bot 
 
 hub 
 
 kib 
 
 max 
 
 dug 
 
 faw 
 
 rab 
 
 sid 
 
 ven 
 
 mar 
 
 pid 
 
 baw 
 
 moy 
 
 mud 
 
 yim 
 
 nam 
 
 Ian 
 
 ram 
 
 l rox 
 
 fub 
 
 hor 
 
 tey 
 
 deb 
 
 pow 
 
 was 
 
 jig 
 
 ges 
 
 lud 
 
 wid 
 
 jom 
 
 kus 
 
 dix 
 
 bag 
 
 cay 
 
 yut 
 
 dam 
 
 lax 
 
 sor 
 
 not 
 
 har 
 
 vim 
 
 pab 
 
 fon 
 
 tus 
 
 rit 
 
 kay 
 
 bir 
 
 wep 
 
 bow 
 
 lix 
 
 mur 
 
 seg 
 
 voy 
 
 sir 
 
 pex 
 
 heg 
 
 rum 
 
 gid 
 
 neg 
 
 fim 
 
 tip 
 
 loy 
 
 dut 
 
 wut 
 
 tox 
 
 gem 
 
 ruy 
 
 gor 
 
 vig 
 
 jad 
 
 kow 
 
 ton 
 
 sut 
 
 tir 
 
 hig 
 
 med 
 
 fox 
 
 bep 
 
 nis 
 
 vun 
 
 dow 
 
 gax 
 
 can 
 
 jup 
 
 nun 
 
 yow 
 
 mig 
 
 dat 
 
 tar 
 
 soy 
 
 few 
 
 lun 
 
 taw 
 
 B. 
 
 Mark all groups of letters in 
 
 the following list that i 
 
 are not 
 
 (familia 
 
 Qglish 
 
 words : 
 
 
 
 
 
 
 
 
 
 men 
 
 sar 
 
 bet 
 
 won 
 
 pox 
 
 hus 
 
 nib 
 
 ket 
 
 sum 
 
 hip 
 
 tug 
 
 mop 
 
 jaw 
 
 bux 
 
 cub 
 
 gas 
 
 pay 
 
 rib 
 
 her 
 
 num 
 
 vat 
 
 nay 
 
 gup 
 
 bun 
 
 fit 
 
 keg 
 
 sop 
 
 yes 
 
 com 
 
 fur 
 
 pum 
 
 web 
 
 ten 
 
 wox 
 
 dip 
 
 jug 
 
 sew 
 
 jis 
 
 toy 
 
 gig 
 
 lip 
 
 tar 
 
 jet 
 
 pus 
 
 rob 
 
 feg 
 
 coy 
 
 win 
 
 kid 
 
 gum 
 
 pew 
 
 mix 
 
 lep 
 
 sar 
 
 job 
 
 vap 
 
 bid 
 
 yeb 
 
 den 
 
 low 
 
 sap 
 
 ren 
 
 fow 
 
 new 
 
 red 
 
 lug 
 
 hod 
 
 kin 
 
 dot 
 
 ses 
 
 bip 
 
 led 
 
 war 
 
 his 
 
 tid 
 
 buy 
 
 sex 
 
 did 
 
 rag 
 
 hop 
 
 yew 
 
 mub 
 
 got 
 
 tax 
 
 put 
 
 hen 
 
 vot 
 
 jar 
 
 key 
 
 him 
 
 fad 
 
 tub 
 
 nor 
 
 fix 
 
 pern 
 
 vow 
 
 doy 
 
 let 
 
 nex 
 
 lay 
 
 Introspectively it was difficult to take B so soon after A, so that 
 the blank might be cut in two instead of being used as it is. Another 
 difficulty was found in the arrangement of the syllables. There was 
 a tendency to work by vertical columns rather than across the sheet, 
 and section B was confusing for the eye. Either explicit directions 
 should be included, or the syllables printed in even columns. 
 
40 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 No one made a perfect record in the time given, but in about all 
 of the "Mark English words" tests, and in some of the "Mark non- 
 sense words" tests the entire blank was gone over within the time, 
 the rest of the time being spent in looking back for omissions. Since, 
 moreover, there were many of both omissions and errors, the meas- 
 urement of the time of the process is not feasible. 
 
 The second test is much harder. The requirement in it of equa- 
 ting time, errors and omissions in the case of almost every subject is 
 troublesome. This difficulty exists to a less degree with the "Mark 
 English words" test. 
 
 The amount of improvement due to familiarization with the plan 
 of the test would not apparently be so great as to be very trouble- 
 some. When the same blank was used twice, as here, the change of 
 the third over the first trial was for the marking nonsense words 
 about 25 per cent, more words correctly marked, and about 30 per 
 cent, fewer words wrongly marked, with a slight increase in 
 omissions. 
 
 The remaining three tests were not given each sufficiently often 
 to allow discussion of any practise effect. They were included for 
 purposes of comparison and correlation when taking one or two 
 trials; so that the "short-term group" becomes, to all intents and 
 purposes, nothing more than an "instructed" group in those tests, 
 except for their general experience of test conditions. 
 
 B. Relative Value of these Tests 
 
 The question of the variability and correlation of these association 
 tests will now be taken up. 
 
 The resemblance between an individual's average ability in the 
 first idea, day opposite, vertical opposite, preceding letter and com- 
 plete the word tests combined, and his ability in each of these tests 
 separately, was calculated in order to discover the extent to which 
 each single test is significant of the more general ability. This re- 
 semblance was calculated both from the percentage of unl ike-signed 
 pairs, and also by the Pearson coefficient of correlation. 
 
 In the case of these and all correlations to follow, the reader will 
 understand that I am not measuring the correlations between the 
 true abilities which would be found from an infinite number of 
 trials with each test, but only the correlations between the measures 
 got from 1, 2, 3, or 4 trials, as the case may be. The question is not 
 of the significance of certain traits in human nature, but only of 
 certain previously defined tests of those traits. 
 
 It will be understood also that other results, mostly from only 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 41 
 
 10 and in some cases only 6 individuals, are very unreliable. They 
 are however much more reliable than mere opinions. 
 
 The performances of the 36 individuals in the "instructed" 
 group were thus correlated with the following results: 
 
 TABLE IX 
 
 Average of 
 these five 
 tests and 
 
 COS7T&T 
 
 First idea 749 
 
 Day opposite 844 
 
 Vertical opposite 509 
 
 Preceding letter 368 
 
 ^ Complete the word , .425 
 
 r 
 
 (Closest correla- 
 tion =J1) 
 
 .623 
 
 2 
 
 .671 
 
 1 
 
 .615 
 
 3 
 
 .484 
 
 5 
 
 .607 
 
 4 
 
 Thus by both methods the easy opposites seems to be the best test 
 so far as it measures the element common to all these tests on asso- 
 ciation. By both methods also the preceding letter seems the poorest. 
 
 Next were used the results (in the first two trials) of the ten 
 individuals in both the long-term group and the short-term group in 
 the following tests: first idea, vertical opposite, day opposite, pre- 
 ceding letter, complete the word, free association, subject-predicate, 
 difference between, addition, Mbbinghaus combination. 
 
 Again each test was correlated with the average for all, with the 
 following results. 
 
 TABLE X 
 
 COSttCT 
 
 First idea 22 
 
 "Vertical" 92 
 
 "Day" 79 
 
 Preceding letter 81 
 
 Complete word 37 
 
 Free association > — .13 
 
 Subject -predicate 37 
 
 Difference between 64 
 
 Ebbinghaus combination •. .66 
 
 Addition 79 
 
 The two methods do not agree so well this time, but again the 
 easy list of opposites correlates high. The preceding letter correlates 
 rather low by the Pearson coefficient method, high by the percentage 
 of like-signed pairs. As this latter method takes account only of 
 number of cases of difference whereas r is affected as well by the 
 amounts of difference, it is obvious that a few cases of wide diver- 
 gence from the average, or in other words a subject making an 
 unusually low record in a certain test, will bring about the dis- 
 crepancy between the two methods. On examining the original data 
 this is precisely what is found: one subject usually far below the 
 
 r 
 
 (Closest = 1) 
 
 .39 
 
 8 
 
 .48 
 
 3 
 
 .71 
 
 1 
 
 .42 
 
 4 
 
 .09 
 
 9 
 
 .11 
 
 10 
 
 .47 
 
 6-7 
 
 .23 
 
 6-7 
 
 .67 
 
 2 
 
 .39 
 
 5 
 
42 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 average made a very good record at the second trial, and one of the 
 very best subjects made the lowest record of anybody at this pre- 
 ceding letter test. The Pearson coefficient is greatly affected by these 
 records, and is correspondingly low ; by the percentage method their 
 influence is only slightly felt. 
 
 Complete-the-ivord, which was low for the instructed group is 
 also low for these two groups, extremely so by the Pearson coeffi- 
 cient. The other test with very low correlation, the free association 
 has inverse relationship by the percentage of like-signed pairs. 
 This means that although the majority of subjects reacted differently 
 in this test from their average reaction in association tests, yet their 
 individual records differ only slightly from each other — the A.D. 
 for this test being very low. 
 
 The Ebbinghaus Combination test correlates fairly closely by 
 both methods. 
 
 The Free association test correlates so slightly probably because, 
 as was shown, it is largely a test of the rate of writing for many 
 subjects. 
 
 The value of each test of association has been discussed from two 
 standpoints thus far, that of significance measured by highest corre- 
 lation with the average of all tests in the series and that of least 
 disturbance by practise. A third standard would be that of ascer- 
 taining for each test the unreliability of any given number of trials. 
 Where possible this has been measured in the case of: (1) the first 
 four or five records of each member of the short-term practise group, 
 and (2) the first five and sometimes the last five records of each 
 member of the long-term practise group. The average results of (1) 
 and of (2) are presented in the following table in percentage state- 
 ments. The higher the figure the greater the unreliability of a 
 single trial and vice versa. To this table is added a column to give 
 the number of trials that would be needed to reduce the unreliability 
 to 1 per cent., and a column to give the consequent time it would 
 take to get such reliable information about a person's ability in that 
 test, using as a basis for this calculation the average time taken in an 
 amount-limit test, the time allowed in a time-limit test. 
 
 Such determinations are difficult because of the practise effect, 
 and the difference in difficulty of different blanks of the same series. 
 From the gross differences found in an individual's trials, one must, 
 in order to get an approximate measure of how much difference is 
 due to chance variations in the individual, eliminate these two added 
 causes of difference. This can be done only approximately and by 
 more or less arbitrary criteria. 
 
 In tests involving differences in quality as well as rate of achieve- 
 
EXPERIMENTAL WORK WITH SEVERAL GBOUPS OF TESTS 43 
 
 ment there is the further difficulty that one performance may differ 
 from another in quality and in speed or vice versa. The reliability 
 of the test as a whole as a measure of efficiency in the function in 
 question can then be determined only after the combination_of the 
 measures for quality and speed into a single measure. 
 
 The method taken may be shown best by an example. The records 
 of the three long-term subjects in the " day " opposite test were: 
 
 
 
 
 TABLE 
 
 XI 
 
 
 
 H 
 Amount 
 13 
 
 Quality 
 25 
 
 W 
 
 Amount 
 
 11 
 
 Quality 
 21 
 
 F 
 
 Amount 
 
 12 
 
 Quality 
 22 
 
 Av. 
 Amount Quality 
 
 12 22.6 
 
 15 
 
 29 
 
 12.5 
 
 25 
 
 13 
 
 26 
 
 13.5 26.6 
 
 15 
 
 29 
 
 13 
 
 25 
 
 13 
 
 26 
 
 13.6 26.6 
 
 17 
 
 31 
 
 14 
 
 26 
 
 14 
 
 26 
 
 15 27.6 
 
 15.5 
 
 29 
 
 14 
 
 26 
 
 14 
 
 27 
 
 14.5 27.3 
 
 Since the quality was substantially equal throughout for each 
 individual, the reliability may be measured from the differences in 
 the amount score alone. Since, as will be shown in a later section, 
 individuals cluster around a central tendency in respect to changes 
 in the rate of improvement, the general practise effect shown in the 
 average column may be applied to each individual. That general 
 effect smoothed may be taken as 12.5, 13.5, 14, 14.5, 15. So it may be 
 assumed without great inaccuracy that, apart from the chance varia- 
 tions of the subject, the records would have been approximately — 
 
 n. w. F. 
 
 13.5 11.5 11.5 
 
 14.5 12.5 12.5 
 
 15 13 13 
 15.5 13.5 13.5 
 
 16 14 14 
 
 The deviation of the single trials due to the person's varying 
 condition are then for 
 
 N. 
 .5 
 
 w. 
 .5 
 
 F. 
 .5 
 
 .5 
 
 
 
 .5 
 
 
 
 
 
 
 
 1.5 
 
 .5 
 
 .5 
 
 .5 
 [6 
 
 
 .2 
 
 
 
 A.D. 
 In per cent, of Av. Amt. 4.0 1.5 2.3 
 
 So far as these three subjects go, the probable average divergence of 
 the result obtained from a single trial with the ' ' day ' ' test from the 
 probable true result is then 2.9 per cent, of the former's amount. 
 
44 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 To show the reliability of these estimates of reliability themselves, 
 the results from all the short-term and from the long-term subjects 
 are given separately. 
 
 TABLE XII 
 
 Eelative Precision of Association Tests 
 
 Approxi- 
 mate No. of 
 Probable Average Divergence of the Trials Nec " 
 Result Obtained from 1 Trial from the ,? ssary to 
 Probable True Result, in Per Cents. M ^ u ^ a A ^^^ 
 of the Former Person Approxi- 
 
 No.of of tne former w ith an mate Time 
 
 Seconds Short Lon £ Term Data Com- Average Di- of Tests so 
 for 1 Term Early Late bined Es- vergence of to Measure 
 Test Trial. Data Trials Trials timate 1 Per Cent, a Person 
 
 49 121 min. 
 
 Easy opposites [day, 
 
 good, great, high] 30 6.9 2.9 5 25 12* " 
 
 Hard opposites [ver- 
 tical, serious] ... . 30 7.4 7.5 56 28 " 
 
 Addition [of 5 two 
 
 place numbers] .. 60 6.0 6.5 5.1 6 36 36 " 
 
 Preceding letter 15 10.0 12.4 18.1 13 169 42 " 
 
 Complete the word . 15 12.6 8.8 11.2 11 121 30 " 
 
 The facts in the case of the subject-predicate, add and subtract 
 columns, mark nonsense and English words are too intricate to allow 
 even an approximate estimate. So also with difference between, 
 Ebbinghaus combination, noun and adjective, and free association 
 starting from one given word, though these four are all apparently 
 very much more unreliable than those listed. It appears then that for 
 freedom from ambiguity, significance as a symptom of the condition 
 of the association processes in general, freedom from disturbance by 
 adaptation to the test shown in great early practise effect, and reli- 
 ability, the best single written test of these is one in giving easily 
 thought of opposites. In administering it, skipping should be 
 allowed. 
 
 2. Tests on Memory 
 
 A. Descriptive 
 
 Along with these tests on association another group of tests on 
 memory was given. Four memory tests are given to the freshmen, 
 the auditory figures, visual figures, logical memory and retrospective 
 memory. The method of giving them is as follows. For the auditory 
 figures, each series of 8 numerals is read aloud at a rate of about 2 
 per second, after which the subject writes them down " in the order 
 given." In visual figures, corresponding sets of 8 numerals are 
 shown one at a time at the same rate. These numerals (Willson's 
 black gummed) are mounted on cards, held in the hand and exposed 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 45 
 
 by turning them singly to face the subject. In logical memory, a 
 passage — to be quoted later — is read to the subjects who then write 
 as much of it as they can. Attempt is made to give the thought com- 
 pletely, and the words where possible. In retrospective memory^ the 
 subjects are asked to reproduce a line 5 cm. long which they drew 
 as a perception-of-size test at the beginning of the hour, also to ' ' do 
 with it as they did before. ' ' 
 
 Other visual and auditory tests were used with the practise 
 groups; a few other paragraphs were used though no other change 
 made in the logical memory test ; but no other ' ' retrospective ' ' 
 memory test at all similar to this was devised. 
 
 The classification into ' ' auditory, visual ' ' and the like may well 
 seem misleading, as it by no means implies that auditory stimuli 
 are remembered in auditory terms, nor, more usually, that visual 
 stimuli will not be translated by the subject into auditory terms. 
 No warning is given to the freshmen with regard to this, and observa- 
 tion shows that the great majority of them do repeat orally the 
 numerals presented visually. Any comparison of tests, then, does 
 not signify a comparison of kinds of memory, but of varied stimuli 
 or material, and varied ways of presenting material. On the report 
 sheet sent to the freshmen care is taken to say ' ' numerals heard, ' ' 
 and ' ' numerals seen ' ' ; but here, for brevity 's sake, the more usual 
 designation of auditory, visual, etc., will be adhered to, with the 
 understanding that the words refer to stimuli, not to memory terms. 
 For convenience sake also, the tests with auditory stimuli are dis- 
 cussed first, those with visual stimuli later, though the related words 
 might possibly be classified as a logical memory test. 
 
 Auditory Figures. — Experience with this familiar test as given to 
 the freshmen shows that most of them group the 8 numerals in two 
 groups of four. Enquiry reveals that many depend upon a memory 
 after-image for the last four, and memorize the first group only. 
 The average number correctly remembered is 7.6 for the men, 6.7 for 
 the women. This test is thus too easy, many of the individuals 
 obtaining perfect scores. 
 
 The chief difficulty in comparing people's work on memory lies 
 in the variable methods of scoring, especially with regard to trans- 
 positions. If the order is 76431528, and a subject writes 7463 . . ., 
 some experimenters call it two errors because both the 4 and the 6 
 are in the wrong places ; other experimenters call it one error because 
 by making one change — by ' ' lifting ' ' the 6 over the 4, it is corrected. 
 The latter method seems preferable. Supposing a subject were to 
 write 87643152, eight errors would be scored by the first method 
 since each numeral is misplaced; by the latter method only 
 
46 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 one error is scored, since one change would set all right. Also, a 
 misplacement error would be rated more nearly as an omission. A 
 subject writing 76-31528 would be scored one error for omitting the 
 4, but two if he places it before the 6, by the first method ; in either 
 case he is scored just one error by the latter method, putting mis- 
 placements and omissions on an equal basis. 
 
 In the work to be reported on therefore, the second method was 
 used, only that a positive score was used instead of counting the 
 errors. Thus each numeral given correctly was scored 1/2, and if 
 it was in the right place — interpreting this as relative place not abso- 
 lute place — it was scored 1/2 more. This modification has the 
 advantage of being rapid to use in determining the score, especially 
 of the different kinds of material used in the tests. It is also much 
 easier and can be used more rapidly than the Spearman " foot-rule " 
 method, or the modification recommended by "Whipple (" Manual," 
 p. 266). If it is too cumbersome when it comes to calculating corre- 
 lations, the figures can be very quickly read off as numbers of errors. 
 
 According to this method the average freshmen scores would be, 
 as before, 7.6 for the men, 6.7 for the women. 
 
 To the " instructed " group of eighteen subjects, two sets of ten 
 numerals were given, with an average score of 7.2 figures remembered 
 for the men, A.D. .75 ; and 6.1 for the women, A.D. .85. This agrees 
 with the superiority shown by the men over the women in the fresh- 
 men results, though showing lower scores. 
 
 The short-term group made six trials with ten numerals at a time, 
 with an average score of 8.8 numerals remembered, A.D. .7. The 
 series of 10 was long enough to measure all in this group. No 
 practise effect was observable. 
 
 The long-term group made twenty trials with ten numerals at a 
 time. One subject made only four errors in the whole series, her 
 memory span for this being evidently greater than ten; in conse- 
 quence her records were not used in estimating practise. For the 
 other two subjects the average score was 9.55, the first day's average 
 deviating by — .55, the last by — .5, or taking the first two and the 
 last two trials, the deviation at first was — .45, and at last + .2. 
 
 For these two subjects also the list of 10 was not long enough to 
 measure the practise effect accurately, there being numerous perfect 
 scores. Their records were, in order (in errors) : 
 
 N. 12101 21001 00110 11002 
 F. 3 112 3 01002 10010 22000 
 
 Two other auditory tests were used, (1) series of fifteen related 
 words, and (2) mixed series of unrelated units, including besides 
 
EXPEBIMENTAL WOBK WITH SEVEBAL GBOUPS OF TESTS 47 
 
 
 Lists of 
 
 Belated Words 
 
 
 I 
 
 II 
 
 III 
 
 IV 
 
 College 
 
 See 
 
 Book 
 
 Holiday 
 
 course 
 
 sensation 
 
 author 
 
 excursion 
 
 grade 
 
 perception 
 
 style 
 
 boat 
 
 graduate 
 
 interpret 
 
 classic 
 
 train 
 
 senior 
 
 illusion 
 
 literature 
 
 ticket 
 
 dues 
 
 cortex 
 
 essay 
 
 early 
 
 money 
 
 hemisphere 
 
 poem 
 
 seat 
 
 purse 
 
 ganglion 
 
 rhyme 
 
 hot 
 
 lost 
 
 dendrite 
 
 meter 
 
 window 
 
 advertise 
 
 branch 
 
 scan 
 
 draught 
 
 reward 
 
 conduct 
 
 quantity 
 
 cold 
 
 deceive 
 
 intercept 
 
 Latin 
 
 bronchitis 
 
 angry 
 
 numb 
 
 translate 
 
 doctor 
 
 threaten 
 
 injury 
 
 language 
 
 medicine 
 
 blows 
 
 paralyze 
 
 accent 
 
 cure 
 
 V 
 
 VI 
 
 VII 
 
 VIII 
 
 Noise 
 
 Sunset 
 
 Time 
 
 Black 
 
 cat 
 
 dusk 
 
 test 
 
 negro 
 
 baby 
 
 lamp 
 
 , write 
 
 Africa 
 
 child 
 
 table 
 
 quickly 
 
 Congo 
 
 kindergarten 
 
 play 
 
 maze 
 
 Leopold 
 
 child-study 
 
 deal 
 
 difference 
 
 rubber 
 
 psychology 
 
 lead 
 
 sorting 
 
 cruel 
 
 Thorndike 
 
 queen 
 
 color 
 
 atrocity 
 
 chickens 
 
 trump 
 
 forms 
 
 remonstrate 
 
 monkeys 
 
 short 
 
 remember 
 
 America 
 
 bananas 
 
 partner 
 
 auditory 
 
 Eockefeller 
 
 fruit 
 
 trick 
 
 score 
 
 millions 
 
 skin 
 
 point 
 
 improve 
 
 oil 
 
 slice 
 
 rubber 
 
 average 
 
 monopoly 
 
 supper 
 
 stop 
 
 twenty 
 
 trusts 
 
 IX 
 
 X 
 
 XI 
 
 XII 
 
 Picture 
 
 Child 
 
 Sunday 
 
 Finance 
 
 photograph 
 
 teacher 
 
 rest 
 
 stocks 
 
 pose 
 
 rude 
 
 church 
 
 rise 
 
 recognize 
 
 naughty 
 
 sing 
 
 fortune 
 
 because 
 
 punish 
 
 choir 
 
 invest 
 
 older 
 
 sorry 
 
 organist 
 
 dividends 
 
 friend 
 
 forgive 
 
 training 
 
 railroad 
 
 together 
 
 better 
 
 abroad 
 
 anthracite 
 
 travel 
 
 promise 
 
 Germany 
 
 Phoebe 
 
 foreign 
 
 broken 
 
 Berlin 
 
 advertisement 
 
 steamer 
 
 hardened 
 
 university 
 
 magazine 
 
 seasick 
 
 discourage 
 
 philosophy 
 
 story 
 
 improve 
 
 report 
 
 research 
 
 read 
 
 turbine 
 
 trouble 
 
 valuable 
 
 hammock 
 
 Cunard 
 
 consult 
 
 publish 
 
 trees 
 
48 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 XIII 
 
 XIV 
 
 XV 
 
 XVI 
 
 Dog 
 
 Sky 
 
 Paper 
 
 Teach 
 
 kind 
 
 cloud 
 
 envelope 
 
 physics 
 
 terrier 
 
 raining 
 
 write 
 
 experiment 
 
 rats 
 
 wet 
 
 letter 
 
 light 
 
 hunt 
 
 spoilt 
 
 parents 
 
 refraction 
 
 catch 
 
 new 
 
 away 
 
 angle 
 
 trap 
 
 expensive 
 
 seaside 
 
 measure 
 
 poison 
 
 money 
 
 sands 
 
 survey 
 
 antidote 
 
 draw 
 
 bathe 
 
 instrument 
 
 doctor 
 
 bank 
 
 swim 
 
 careful 
 
 ambulance 
 
 cashier 
 
 deep 
 
 understand 
 
 policeman 
 
 dishonest 
 
 cramp 
 
 accurate 
 
 Irish 
 
 abscond 
 
 drowning 
 
 rely 
 
 Murphy 
 
 scandal 
 
 revive 
 
 promote 
 
 milk 
 
 newspaper 
 
 thankful 
 
 successful 
 
 words, numerals, letters of the alphabet and sounds such as clapping 
 the hands, tapping, ringing a bell, shuffling the feet, whistling, etc., 
 the necessary movements being out of sight of the subjects. 
 
 The short-term group made five trials using series I., II., III., IV., 
 and VI. Besides scoring in the manner described, note was kept of 
 whether the errors were those of omission or misplacement, or 
 whether extra words were put in. At first sight it would seem best 
 to handle this score by keeping it in terms of errors made ; but as the 
 score is given for the right words in the right order, additional words 
 practically counted as errors. From the point of view of interest in 
 individual differences, however, it was felt worth while to keep track 
 of the number and occasion of additional words ; also to note whether 
 any one list seemed more tempting to the imagination than others. 
 In a total of 30 records, eight of them had extra words, one subject 
 supplying them three times. She remembered the greatest number 
 of words correctly. The subject with the lowest score put in extra 
 words twice. Every subject misplaced some words, the one with the 
 best score doing so most often. The average score was 8.9 words 
 A.D. 2.5. There was no practise effect discernible. 
 
 The long-term group in a total of twenty trials made an average 
 score of 12.6 words, A.D. 1.5. The first two trials deviated by — .75, 
 the last two by — .15, but there seemed no certainty of practise effect. 
 The lists of 15 words were just long enough to measure the most 
 capable of these subjects; toward the end of practise a list of 16 
 would be better for regular use. No particular list seemed specially 
 liable to error. The subject with the highest and least variable 
 record wrote the fewest extra words, and made six perf ft ^t records. 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 49 
 
 Both of the other subjects showed considerable variation, one having 
 five perfect records, but misplacements 50 per cent, of the time, the 
 other having no perfect records, and only three free from extra 
 words or misplacements. The one with the greatest number of mis- 
 placements also wrote the greatest number of extra words. The sub- 
 ject who had so good a record with the auditory numerals was not 
 the best in this test. 
 
 Auditory Mixed. — The object in giving this test was to present 
 material absolutely disconnected, yet with each of the units in the list 
 having its own meaning. Even with nonsense syllables some fanci- 
 ful connections are usually made, so that it was not supposed that 
 artificial associations could be entirely avoided; nevertheless by in- 
 trospection there seemed to be very few of them in this case. There 
 is some difficulty in presenting nonsense syllables orally, but with 
 this incongruous yet senseful material there is less danger of errors 
 in hearing on the part of the subjects. The tendency to groupings 
 of four was broken up somewhat by the introduction of the various 
 sounds or noises (shown in the list by italics). By introspection this 
 test proved difficult and irritating to those accustomed to the other 
 material. 
 
 The lists used were as follows : 
 
 (1) 
 
 
 (2) 
 
 (3) 
 
 Carriage 
 
 
 Distance 
 
 Oo 
 
 F 
 
 
 as 
 
 but 
 
 adversary- 
 
 
 whistle 
 
 16 
 
 preach 
 
 
 flag 
 
 resting 
 
 stamp with 
 
 foot 
 
 require 
 
 clucking noise 
 
 lamp 
 
 
 38 
 
 organ 
 
 never 
 
 
 other 
 
 3 
 
 ring a bell 
 
 
 harper 
 
 spring 
 
 K 
 
 
 clap hands 
 
 W 
 
 green 
 
 
 H 
 
 matches 
 
 (4) 
 
 
 (5) 
 
 (6) 
 
 And 
 
 
 Monstrous 
 
 99 
 
 20 
 
 
 (jingle keys) 
 
 monotone 
 
 ring a ~bell 
 
 
 X 
 
 scrape with foot 
 
 wall paper 
 
 
 Symphony 
 
 alphabet 
 
 stampede 
 
 
 tap with pencil 
 
 tomahawk 
 
 tap with finger 
 
 she 
 
 jingle keys 
 
 M 
 
 
 whistle 
 
 asleep 
 
 symmetry- 
 
 
 bugle 
 
 purple 
 
 stamp with 
 
 foot 
 
 typewriter 
 
 tap, or clap 
 
 56 
 
 
 ice-cream 
 
 because 
 
 The short-term group made only 2 trials, with an average score 
 
50 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 of 8.15, A.D. .45. The long-term group made 20 trials, with an aver- 
 age score of 9.2 of the ten remembered, A.D. .35. The detailed re- 
 sults were, in order (in terms of errors) — 
 
 N. 35211 22123 21121 21102 
 W. 02222 11121 31110 31112 
 F. 22223 11031 21201 32112 
 
 There was no practise effect discoverable. The subject who was 
 so very competent with the auditory figures was also the best in this 
 test. The misplacements were unfortunately not noted, so that no 
 comparison can be made in this respect with the related words. 
 
 Visual Figures 
 
 Three sets of eight numerals are shown serially to the freshmen. 
 No apparatus is used, and some little practise is required on the part 
 of the experimenter to expose the cards regularly and at a convenient 
 angle. As said before, no warning is given about not repeating to 
 one's self orally what is shown. The men remember 6.9 correctly on 
 the average, the women 5.7. 
 
 Two of these sets were used with the "instructed" group. The 
 men made an average score of 5.85, the women of 5.15, again agree- 
 ing with the freshmen results in the superiority of the men's record 
 over the women's, though showing lower scoring for both men and 
 women than in the case of the freshmen. The percentages would be 
 73 and 64. 
 
 The short-term group made 5 trials with sets of 8 numerals ; their 
 average score was 7.5, A.D. 0.5. Series of 8 are thus too short for an 
 adequate measure of visual as well as auditory memory. 
 
 The long-term group made 20 trials with sets of 10 numerals. 
 For the first four trials cards were used as for the freshmen. After 
 this, as a screen with a slit was in use for other visual material it was 
 used for the numerals also. This screen was a very simple affair of 
 pasteboard with a 2-inch square opening in the middle. The visual 
 stimuli were written or drawn with charcoal on a long strip of card- 
 board which was pushed along behind the screen, allowing one sec- 
 ond for the exposure of each unit in the series. By reversing the 
 strip, one series could be used as two different tests on different days. 
 Sixteen trials were made with this, making twenty in all. Even 
 series of 10 numerals are too short for adequate measurement of these 
 subjects, perfect records being made frequently after the first three 
 trials. 
 
 Their average score was 9.4, A.D. .5, the range from 8 to 10. 
 The first day's average deviated by — .1, the last by + .8. 
 
EXPERIMENTAL WOEK WITH SEVEEAL GEOUPS OF TESTS 51 
 
 Other visual tests were : grouped forms, serial forms, grouped ob- 
 jects, serial objects, forms recognized. 
 
 Grouped Forms.' — Five different sets were used, one of which was 
 as follows : 
 
 -^ 
 
 /N cz5i 
 
 These forms were drawn roughly with crayon on a small black- 
 board which could be turned and exposed to view for 10 seconds, then 
 turned away again. The short-term group made only two trials with 
 sets 2 and 4. Their average score was 5.4 forms, A.D. .9. The long- 
 term group made 10 trials, average score 8.15 forms, A.D. .1.0. The 
 first day's trial deviated by — 1.35, the last by + .35. It had been 
 intended to make 20 trials with this as with the others ; but very soon 
 the question arose whether it was not much easier to look at a group 
 of 10 for 10 seconds than to see 10 units one at a time for one second 
 each, in the same way that the numerals are shown, with no chance 
 of looking twice at any one of them. It was decided to compare 
 the grouped with the serial method, both for forms and objects, 
 though cutting down the number of trials to 10 each, for this group 
 of subjects. 
 
 Serial Forms. — The cardboard screen and strip, as described be- 
 fore, were used in this test. The sets of forms were similar to those 
 used in the grouped forms test. Two of them are here reproduced. 
 
 oc-- r i d -'rix'ijj 
 
 > -riliOlziz 
 
52 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 The short-term group made 4 trials, of which the average score was 
 6.5, A.D. .75. The averages of the successive trials were 5.66, 5.83, 
 7.07, 7.43 showing a greater gain for them in this test than in the 
 other immediate memory tests. Probably this is due to the initial 
 comparative unfamiliarity of the material used. 
 
 The long-term group made 11 trials, average score 7.95, A.D. .95. 
 The first day's average deviated by — 1.95, the last by + 1.7, show- 
 ing a very great practise effect. 
 
 Grouped Objects 
 Ten familiar objects chosen from about 25 in daily use, such as 
 a watch, box of matches, bunch of keys, spool, envelope, pack of 
 cards, books, scissors, fish-hook, soap, were arranged in the same 
 groupings as that used for the grouped forms, a row of three, a row 
 of four, a row of three, thus, — 
 
 XXX 
 
 X X X X 
 
 XXX 
 
 on a small table behind a screen. At the signal the screen was raised 
 for 10 seconds. The subjects then wrote down the names of the 
 things seen, grouping the names as the objects had been grouped. 
 
 Only the long-term group practised with this test, their average 
 score in ten trials being 8.85. The first day's trial deviated by 
 — 1.25, the last by — .1. On the fifth and eighth trials, perfect 
 scores were made, however, by all three subjects. 
 
 Serial Objects 
 
 In this test, the same sort of objects were picked up one at a time 
 and shown for one second each above the screen. 
 
 The long-term group in ten trials made an average score of 
 9.3, the first day's average deviating by — .3, the last by + .1. 
 
 So far then as serial grouped method is concerned there seems, 
 by examination of the accompanying table, 
 
 Serial Grouped 
 
 ( 6.85 (4 trials) 5.4 (2 trials) ^) 
 
 Short-term | g Q (firgt g ^.^ I Formg 
 
 _ . < 7.95 8.15 
 
 Long-term | 83 g 75 J 0bjects 
 
 to be a slight balance in favor of the serial method, probably because 
 this is the familiar method used for numerals, and in auditory 
 stimuli. Introspectively, the long-term group found the grouped 
 
EXPEEIMENTAL WOBK WITH SEFEBAL GBOUPS OF TESTS 53 
 
 forms easier than the serial forms. The reason is, perhaps, that with 
 the latter method the second of exposure is not always sufficient for 
 the recognition of some of the forms, whereas when grouped, the total 
 10 seconds can be distributed in the most economical manner, the 
 eyes pausing longer, or returning to those forms not so readily 
 apperceived. In the case of objects shown, this factor of appercep- 
 tion scarcely entered in, as each object was readily recognized, and 
 mentally named in its one-second exposure. A slightly higher 
 score was made on the average for objects shown serially than shown 
 grouped. 
 
 Forms Recognized 
 The blanks used in this test are reproduced on this and the three 
 following pages. 
 
 *^<5># 
 
 G V 
 
 oasoo 
 
 AA 
 
 □ o$^ 
 
 □ 
 
 1 (a) 
 
 The subject is given the small sheet with instructions to study it in 
 any way preferred till at the end of 60 seconds he is given another 
 sheet on which he is to mark as quickly as possible all the forms he 
 remembers having seen on the first sheet. It will be noticed that on 
 (1) 24 can be marked, on (2) only 18. 
 
 The time taken to mark the second sheet is noted, also the total 
 number marked, and the number correctly marked. 
 
54 
 
 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 Set (1) was given to the Barnard freshmen of the class of 1912. 
 The average time taken by 49 of them was 66 seconds, A.D. 16.2, with 
 15.6 correctly marked, A.D. 2.3, and 5 wrongly marked. 
 
 Six members of the short-term group and the most rapid worker 
 in the long-term group made one trial with this set. Their average 
 time was 81 seconds, or, not counting N., 88 seconds, A.D. 22.5 with 15 
 correctly marked and 2 wrongly marked. These subjects made trial 
 
 □0#OiOOV#C 
 
 ^AOdOOaO 
 
 ^7 
 
 ^AO^OA^ 
 
 □ 
 
 OA 
 
 ®A$ 
 
 1 (6) 
 
 ^7 
 
 also with (2), where their average time was 115 seconds, A.D. 33, 
 with 9.5 correctly marked, A.D. 1.3, and 3.5 wrongly marked. It is 
 much more difficult than set (1). 
 
 The attempt thus to measure memory by a combination of the 
 amount recalled, the quickness with which it is recalled, and the 
 errors made, should be carried on with better material. The results 
 obtained here are of value only for measurements of the significance 
 of this particular test by its correlations. 
 
 Two other memory tests were given, the logical memory and the 
 retrospective memory. 
 
 Logical Memory. A paragraph is read aloud to the subjects who 
 then write out as much as they remember of it, stress being laid upon 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 55 
 
 the matter rather than on the words remembered. To the freshmen 
 the following paragraph (I.) is read: 
 
 Tests such as we are now making are of value both for the advancement of 
 science and for the information of the student who is being tested. IT is of 
 importance for science to learn how people differ and on what factors these 
 differences depend. If we can disentangle the complex influences of heredity 
 and environment we may be able to apply our knowledge to guide human devel- 
 opment. Then it is well for each of us to know in what way he differs from 
 others. We may thus in some cases correct defects and develop aptitudes which 
 we might otherwise neglect. 
 
 C 
 
 v. u m 
 
 M 
 
 2 (a) 
 
 The men remember 44.5 per cent, of the ideas contained in it on 
 the average, the women 51.2 per cent. 
 
 The short-term group made four trials, once with this paragraph, 
 and once with each of three others II., III., and IV. 
 
 Other Passages Used 
 II 
 
 Could the young but realize how soon they will become mere walking bundles 
 of habits, they would give more heed to their conduct while in the plastic state. 
 We are spinning our own fates, good or evil, and never to be undone. Every 
 smallest stroke of virtue or of vice leaves its never so little scar. 
 
 Ill 
 
 Measures of the variability of the individual measures are of two sorts: 
 measures of the averaging type and measures of the percentile type. 
 
56 
 
 STUDY OF TESTS FOE INDIVIDUAL DIFFEEENCES 
 
 The mean square deviation equals the square root of the average of the 
 squares of the deviations of the individual measures from their average, median, 
 or mode. 
 
 IV 
 
 The abstract scheme of successive predications, extended indefinitely, with 
 all the possibilities of substitution which it involves, is thus an immutable system 
 of truth which flows from the very structure and form of our thinking. If any 
 real terms ever do fit into such a scheme they will obey its laws. 
 
 c m 
 
 ^ 
 
 n 
 
 n 
 
 2 (6) 
 
 The average percentage remembered was 39.1 ; for paragraph I. 
 alone it was 49 per cent., slightly lower than was the case with the 
 
EXPEBIMENTAL WOBE WITH SEVEBAL GBOUPS OF TESTS 57 
 
 Barnard freshmen. These tests were given primarily as a means of 
 estimating the significance of so called M logical memory," and no 
 data on the effect of practise were secured. 
 
 
 TABLE 
 
 XIII 
 
 
 
 
 Individual Credits 
 
 for Memory Passages 
 
 
 
 Graded 
 
 Oil 
 
 a 
 
 scale of ten 
 
 
 
 Bu. ... 
 
 I 
 7.0 
 
 
 
 II 
 7.0 
 
 5.5 
 
 1.0 
 6.5 
 5.5 
 
 III 
 5.5 
 
 2.0 
 
 3.5 
 1.5 
 1.5 
 
 IV 
 6.0 
 
 Gr. ... 
 
 3.5 
 
 3.0 
 
 St. ... 
 
 3.0 
 
 
 J 
 
 
 1.0 
 
 L 
 
 5.0 
 
 3.0 
 
 M 
 
 6.0 
 
 3.5 
 
 Ba. ... 
 
 
 
 
 6.0 
 
 1.5 
 
 3.5 
 
 Retrospective Memory. — Instead of the test given the freshmen, 
 which consists of reproducing a line the same length as one seen and 
 reproduced an hour previously, the long-term group made ten trials 
 in eight of which they were asked to reproduce the list of 15 related 
 words given as an auditory test on the previous day. On that occa- 
 sion the list had of course been read, written more or less correctly 
 and then re-read for the subjects' satisfaction in their performance, 
 so that there had been three repetitions of the list, two of them 
 correctly, followed by an interval of about twenty-four hours. At 
 the third and seventh trials other material was used. Once they were 
 asked to reproduce a paragraph used the day before ina" complete 
 the paragraph test," and once to give the ten kinds of objects used 
 in a "naming 100 objects" test — yet to be described. It would be 
 interesting to prolong and vary this test indefinitely, as individuals 
 differ so much in their ability to recall different kinds of things after 
 different intervals, and so many human interests depend upon the 
 accuracy and length of retention j but as the object here was merely 
 to discover any tendency to practise effect in such mature subjects, 
 and as time and opportunity were lacking for more prolonged series, 
 only these ten trials were made. The score was 9.9 on the average, 
 with no practise effect discernible. 
 
 B. Relative Value of these Tests on Memory 
 
 On the whole, there is no evidence that in any of these tests of 
 immediate memory, a first trial measures a markedly different process 
 from later trials after the subject is adapted to the form of the test. 
 No great difference can exist, or it would show itself in the work of 
 the short-term group. With the possible exception of serial forms, 
 
58 STUDY OF TESTS FOR INDIVIDUAL DIFFERENCES 
 
 there is no test in which the second trial shows any greater propor- 
 tionate improvement over the first than the fourth or fifth shows 
 over the third or fourth. Indeed, in almost every case it is among the 
 records of the long-term group that evidence of the existence of any 
 practise effect must be sought. 
 
 The tests rank in respect to susceptibility to practise as follows : 
 
 Very slight, not discernible in these cases Auditory mixed. 
 
 Serial objects. 
 
 Retrospective. 
 Slight (less than 10 per cent, in 20 trials) Auditory figures. 
 
 Auditory words. 
 
 Visual figures. 
 Considerable Grouped objects. 
 
 Grouped forms. 
 Most Serial forms. 
 
 Certain correlations of these various tests on memory have been 
 computed. 
 
 First of all, taking the short-term and long-term groups together, 
 the average of the first three records of each subject in the following 
 tests were compared, each test with the average for all six tests: 
 auditory figures, related words, auditory mixed, visual figures, 
 grouped forms, serial forms. In calculating this set of correlations 
 the deviations of each subject in the short-term group from the 
 average of her own group were taken, not from the average for the 
 ten subjects treated as one group. 
 
 Next, the records of the " instructed group " with auditory 
 figures and visual figures — 18 cases, two trials for each — were corre- 
 lated; also the same tests for the short- and long-term group, as 
 above. Similarly nine subjects' records with auditory figures and 
 related words, and five subjects' records with related words and 
 logical memory. 
 
 Third, all auditory tests, viz., auditory figures, related words, and 
 mixed series, were averaged, and each test correlated with the average 
 of all, using again the average of the first three records of both short- 
 and long-term groups. 
 
 Fourth, using 10 subjects as above, the correlation of grouped and 
 serial forms was computed. 
 
 Last, visual figures was compared with forms recognized using the 
 records of the 49 freshmen, and also those of the short-term group. 
 The latter test was also compared with grouped forms, a supposedly 
 similar test. 
 
 All these results are presented in the following table, where in 
 addition to the Pearson coefficient, the rougher correlation by the 
 
1. Average of these 
 six tests and 
 
 EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 59 
 
 method of unlike-signed pairs is given wherever justified by the 
 number of cases available. 
 
 It will be understood that these correlations are to measure the 
 significance of three (or two, as noted) trials of a given test, not the 
 true relation between an individual's total ability in one trait and 
 his ability in another. The reader is again reminded that the results 
 commonly from only ten subjects are only very coarse approxima- 
 tions, but are nevertheless by so much better than nothing. 
 
 TABLE XIV 
 
 cos Ti-U 
 Auditory figures 31 
 
 Belated words 93 
 
 Mixed series 31 
 
 Visual figures ? 
 
 Grouped forms 95 
 
 Serial forms 31 
 
 2. Auditory figures and Visual figures 
 
 Auditory figures and Visual figures 
 
 Auditory figures and Eelated words 
 
 Logical memory and Eelated words 
 
 3. Average of these f ^T "T" ft 
 
 ., , J Eelated words 93 
 
 three tests and \ __. . . __ 
 
 I Mixed series 93 
 
 4. Grouped forms and Serial forms 81 
 
 5. Forms recognized, and Visual figures 03 
 
 Forms recognized, and Visual figures 
 
 Forms recognized, and Grouped forms 
 
 In the first set of correlations, with varied material and including 
 auditory and visual tests it would be surprising to find high correla- 
 tions. Grouped forms stands out conspicuously therefore as a typi- 
 cal test — in so far as it measures whatever element may be common 
 to all these six tests. Related words comes next by both methods of 
 correlation, while visual figures is actually an inverse relationship. 
 
 In the second set it is seen that auditory and visual figures have 
 a very low correlation, none by the percentage of unlike-signed pairs. 
 Clark Wissler, who differentiates between numerals correctly given 
 and those correctly placed, found correlations of .29 and .39 re- 
 spectively. 
 
 The correlation of auditory figures and related words is, however, 
 still lower, though too much can not be argued from the records of 
 only 9 subjects. The very few records for related words and logical 
 memory similarly cautious against too great emphasis on the higher 
 correlation found there, though this is certainly more what might be 
 
 r 
 .51 
 
 No. of 
 
 Cases 
 
 10 
 
 .64 
 
 10 
 
 .05 
 
 10 
 
 ? 
 
 10 
 
 .91 
 
 10 
 
 .45 
 
 10 
 
 .21 
 
 18 
 
 .17 
 
 10 
 
 .12 
 
 9 
 
 .55 
 
 5 
 
 .69 
 
 10 
 
 .58 
 
 9 
 
 .64 
 
 10 
 
 .76 
 
 10 
 
 .37 
 
 49 
 
 .13 
 
 6 
 
 .26 
 
 6 
 
60 STUDY OF TESTS FOE INDIVIDUAL DIFFEEENCES 
 
 expected. The unreliability of these two Pearson coefficients is (P.E. 
 r true — r obtained) .021 and .184 respectively. 
 
 In the third set, it is interesting to see that all the correlations of 
 the auditory group are fairly high, and that auditory figures come 
 out better than related words reckoning the Pearson coefficient only, 
 though in the first set this was not the case. Even the mixed series 
 correlates well with the average of the group, and the coefficient is 
 higher than that of logical memory and related words (in the second 
 set), rather unexpectedly. 
 
 Summing up this work on memory from the point of view of in- 
 tercorrelations, auditory figures and related words seem tests fairly 
 typical of any presented to the ear. Grouped forms seems distinctly 
 typical as, taken all through, its correlations are high. 
 
 As to the question of the relative precision of the different tests 
 of memory, making a reasonable allowance for practise effect, where 
 such exists, the unreliability of single trials with the tests described 
 are as shown in Table XV. The unreliability of a test with visual 
 figures can not be properly estimated. The series of eight were, as has 
 been stated, too short, and the series of ten was for the long-term 
 group too short toward the end of practise. From the early trials 
 of these three subjects the average divergence of the result from a 
 single trial from the true result may be estimated as from 5 to 7 
 per cent, according to how the probable course of practise is esti- 
 mated. 
 
 TABLE XV 
 
 Eelative Precision of Memory Tests 
 
 Most Probable Average Divergence of the Approximate 
 Result Obtained from 1 Trial from the Prob- No. of Trials 
 able True Result, in per cents, of the Former Necessary to 
 
 Measure a Per. 
 Long Term Data son with an 
 
 Short Com- Average Diver- 
 
 Term Early Late bined genceof 
 
 Test Data Trials Trials Records 1 per cent. 
 
 Auditory words . . .' 18.0 ' 13.1 12.8 14.6 213 
 
 Auditory mixed 4.3 3.5 3.9 15 
 
 Visual grouped forms 14.6 12.1 13.3 177 
 
 Visual serial forms 9.7 9.9 13.6 11.1 123 
 
 Visual grouped objects 5.4 10.8 8.1 65 
 
 Visual serial objects 3.1 4.0 3.5 12 
 
 Visual figures 6.7 (45) 
 
 So far as the data go, Auditory mixed series, Visual serial objects 
 and probably Visual figures (with a long enough series) have de- 
 cided advantages from the point of view of precision over the other 
 tests. Auditory figures was, as given, too easy a test to measure the 
 subjects and therefore could not be included in this list. 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 61 
 
 If a choice of tests were to be made therefore, a good test, corre- 
 lating with other auditory tests and not much subject to practise 
 with mature subjects, and requiring few trials for a fair degree of 
 precision is Auditory figures. Belated words is good except for the 
 lack of precision accentuated by the fact that any selected Jist of 
 words with its varied appeal to different types of subjects would be 
 less simple than numerals with their greater similarity of associa- 
 tions. 
 
 In spite of its susceptibility to practise and the greater number 
 of trials required to give a fair degree of precision Grouped 
 forms is suggested as the best visual test for three reasons: 
 (1) it is significant of memory in general; (2) subjects have slight 
 tendency to repeat the name of the form, so that it appeals merely 
 to the eye better than do numerals or objects; (3) it is equally easy 
 if not easier to give than Visual figures, requiring less dexterity in 
 manipulation. Standard groups could easily be drawn or printed on 
 cardboard, say two feet six inches square, and thus used for small 
 groups as well as for individual work. 
 
 These tests complement one the other and would together make 
 an easily given, easily scored and fairly significant and precise test. 
 
 3. Tests on Perception 
 A. Descriptive 
 
 The A Test. — The following blank, here reduced in size, is used 
 with the freshmen. 
 
 OYKFIUDBHTAGDAACDIXAMRPAGQZTAACVAOWLYX 
 
 WABBTHJJANEEFAAMEAACBSVSKALLPHANRNPKAZF 
 
 YRQAQEAXJUDFOIMWZSAUCGVAOABMAYDYAAZJDAL 
 
 JACINEVBGAOFHARPVEJCTQZAPJLEIQWNAHRBUIAS 
 
 SNZMWAAAWHACAXHXQAXTDPUTYGSKGRKVLGKIM 
 
 FUOFAAKYFGTMBLYZIJAAVAUAACXDTVDACJSITJFMO 
 
 TXWAMQEAKHAOPXZWCAIEBRZNSOQAQLMDGUSGB 
 
 AKNAAPLPAAAHYOAEKLNVFARJAEHNPWIBAYAQEK 
 
 UPDSHAAQGGHTAMZAQGMTPNTJRQNXIJEOWYCREJD 
 
 UOLJCCAKSZAUAFERFAWAFZAWXBAAAVHAMBATAD 
 
 KVSTVNAPLILAOXYSJUOVYIVPAAPSDNLKRQAAOJLE 
 
 GAAQYEMPAZNTIBXGAIMRUSAWZAZWXAMXBDXAJZ 
 
 ECNABAHGDVSVFTCLAYKUKCWAFRWHTQYAFAAAOH 
 
 There are 100 A's on it, and the directions are to mark as quickly 
 as possible all the A's. Since several A's occur together more than 
 once it might be better to tell them to mark each A. 
 
 The men take 100 seconds on the average, the women 87.3 seconds, 
 agreeing with the general conclusion that women are quicker with 
 this sort of test — noticing details — than are men. The general ex- 
 
62 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 perience is that all the A's are not marked by either the men or the 
 women, so that when using these figures comparatively, i. e., when 
 60 A's are scored in 60 seconds for the men, and 68.7 for the women, 
 it must be understood that they are only approximately correct, are 
 in fact a little too high. 
 
 In testing this test, the following blanks were used. No. 2 has 
 also 100 A's; No. 3, 50 of each of the letters A, B, K, S. 
 
 Set No. 2 
 GAAQYEMPAZNTIBXGAIMRUSAWZAZWXAMXBDXAJZ 
 ECNABAHGDVSVFTCLAYKUKCWAFRWHTQYAFAAAOH 
 UOLJCCAKSZAUAFERFAWAFZAWXBAAAVHAMBATAD 
 KVSTVNAPLILAOXYSJUOVYIVPAAPSDNLKRQAAOJLE 
 AKNAAPLPAAAHYOAEKLNVFARJAEHNPWIBAYAQRK 
 UPDSHAAQGGHTAMZAQGMTPNURQNXIJEOWYCREJ.D 
 TXWAMQEAKHAOPXZWCAIRBRZNSOQAQLMDGUSGB 
 FUOFAAKYFGTMBLYZIJAAVAUAACXDTVDACJSIUFMO 
 SNZMWAAAWHACAXHXQAXTDPUTYGSKGRKVLGKIM 
 JACINEVBGAOFHARPVEJCTQZAPJLEIQWNAHRBUIAS 
 YRQAQEAXJUDFOIMWZSAUCGVAOABMAYDYAAZJDAL 
 OYKFIUDBHTAGDAACDIXAMRPAGQZTAACVAOWLYX 
 WABBTHJJANEEFAAMEAACBSVSKALLPHANRNPKAZF 
 
 No. 3 
 GWBTBVKIKSCSAUEBCIWVABZSMDUBKLWHKHYCGYGK 
 NANNCBVBSAKOIUPEKCXVGSTVRIWYBYGKHAZLPBYO 
 XAPYEXXHUFSBVDYDIAZLRSATZAZVFCOFSAIPTDOK 
 BBISKAKHXDYIUZRHVRZYSCIGECPOFKBICBMGFSDC 
 YHSRMVBLYICKZBMXFVBBIKUCBZLOGLVKGFMOATUN 
 SHOFHXIMKUXLDZKMRYRLVUWWKYEUVECSOUWBADEX 
 ALUAKRMSFTGXWLVGAOWBTPODXBNSFSFSWSDRSMPO 
 KBRIGAXZBZACKFBBEVWCGSWBMFEMXXOKRDIWGGBL 
 BTPNSKBAGVTCSSRKUBURUDMZEWIZFESTMZEBWAFI 
 BKSGYHSLSFABTLTIUDXGAKROZYKOBHEAALPMLLKC 
 GVCWKKPTUYUGSTSSDWNKSIEICSNBTVADKANTKKPB 
 UXGTSOSUZPNBKRBAFDYFOVYBMPSOMBUOPMEGKKTA 
 COWVFXATSVAPAKYVAHNFXSBDAZYDCFDPPKNPHAMM 
 XUNKDXSRAAMDVOPECXRKTLHAXVKSHYWEWMMNNHBR 
 SLSOZFBZGRRIIHKRLEKHEZRGSCYKUIPSLECKYNDA 
 UGKLLEMAXFYERKWZYSNTTUAVSNAAMNWSAODFWAEH 
 WBNSPAKBBAOAHPHBHNRDELDLMPWZTAIORTSKLBAZ 
 HNBKXPSNXAZHNIPHFGTE 
 
 The disturbing effect of adaptation and practise with this test 
 is very slight. The short-term group using blank 2 required .783 
 second per A marked in their first trial of 45 seconds and .869 sec- 
 ond per A marked in a second trial of 60 seconds. The long-term 
 group using blank 1 required .643 second per A marked in the first, 
 and .636 second per A in a second trial, each of 60 seconds. 
 
EXPEEIMENTAL WOBK WITH SEVERAL GBOUPS OF TESTS 63 
 
 An "instructed" group of eleven subjects who marked A, B and 
 K in order in three successive trials with blank 3, took only nine 
 tenths as long per K as per A ; but the same proportionate time was 
 taken when K was given, as the first to be marked, to one group of 
 18 and A to another group. The difference was therefore probably 
 largely due to the greater ease of marking K. 
 
 To determine the relative difficulty of finding A, K, B and S on 
 No. 3 blank, four similar groups of 19 subjects were tested, each 
 group marking a different letter. A time limit of 105 seconds (1} 
 minutes) was allowed to mark the 50 letters. The results were as 
 follows : 
 
 TABLE XVI 
 
 Blank Letter Time Av. Marked A.D. No. of Cases 
 
 No. 3 A 105 41.3 5.1 19 
 
 No. 3 B 105 40.0 5.2 19 
 
 No. 3 K 105 37.5 3.9 19 
 
 No. 3 S 105 44.6 5.1 19 
 
 The time was possibly too long to measure all adequately in the 
 case of the letter S. 
 
 The short-term group gave the following results which, in view 
 of the probability that practise effect is very slight, may be used to 
 estimate the relative difficulty. 
 
 TABLE XVII 
 
 Time Av. Sec. 
 
 Letter Method in Sec. Marked A.D. per Letter 
 
 S Time limit 40 26.0 8.0 1.54 
 
 S Time limit 30 18.0 5.0 1.67 
 
 (Three other trials intervening) 
 
 B Amount limit Av. 117 47 17 sec. 2.49 
 
 K Amount limit Av. 112 43.5 12.3 see. 2.61 
 
 A Time limit (not reached) 90 50 f 1.80 
 
 A Time limit 60 31 5.3 1.94 
 
 K is a little harder than B as before, and S is easier than A by 
 about the same proportion as before. A and S can not properly be 
 compared with B and K since the announcement of a time-limit seems 
 to have a stimulating effect. 
 
 An "instructed" group of eleven subjects in a 60 second test 
 with the order ABKS gave averages marked of 30.1, 32.7, 27.0, and 
 37.1 respectively, or 2.0, 1.83, 2.22, and 1.62 seconds per letter 
 marked. These figures where the practise effect for A in comparison 
 with S is reversed confirm the others. 
 
 Concerning the influence of the time-limit versus amount-limit 
 method the following records show that the former does seem to act 
 
64 
 
 STUDY OF TESTS FOE INDIVIDUAL DIFFEEENCES 
 
 as a suggestion to greater efficiency. Those subjects who with 
 amount-limit required more than 105 seconds, often completed the 
 blank with that time limit, making as high scores for accuracy as 
 with the longer time. The facts are: 
 
 Time Limit 105 
 
 Fifth Test 
 Letters Marked 
 
 Gr 40 
 
 L 45 
 
 M 47 
 
 Ba 48 
 
 Sixth Test 
 
 Gr 39 
 
 J 49 
 
 M 42 
 
 Ba 37 
 
 TABLE XVIII 
 
 Amount Limit 
 Eighth Test 
 Time Letters marked 
 
 47 
 
 149 
 125 
 117 
 127 
 
 46 
 46 
 47 
 
 Ninth Test 
 111 41 
 
 110 50 
 
 134 45 
 
 127 38 
 
 Time Limit 105 
 Thirteenth Test 
 Letters Marked 
 
 48 
 
 Marking 
 B 
 
 Marking 
 K 
 
 a — t Test. — The blank is as follows : parts A and B are generally 
 used for separate tests. 
 
 (A) 
 
 Dire tengo antipatia senores; esto seria necedad, porque hombre vale siempre 
 tanto como otro hombre. Todas elases hombres merito; resumidas cuentas, sulpa 
 suya vizxonde; pero dire sobrina puede contar dote viente einco duros menos, 
 tengo apartado; pardiez tamado trabajo atesorar-los para enriquecer estrano. 
 Vizconde rico. Mios, quiero ganado sudor f rente saiga familia; suyo, pertenence, 
 tendran. Conozeo marido pueda convenirle Isabel; Carlos, sobrino. Donde 
 muchacho honrado, mejor indole, juicioso, valiente? Quieres sobrino. Esposo 
 parece natural, pero. Pero, pero, diablos, objeciones hacer. Posible quedandonow 
 solos siempre hacer oposicion. Solo delante hentes eres ministerial. Pues, sidens 
 siempre plan, dicho antes, porque hace tiempo notade cose aflige cierto. Sabes 
 cuante quiero Carlos; consuelo apoyo; despues persona quiero mundo. Como 
 eres buene amable, quieres porque, darme gusto, pero quisiera. Palabra cuesta 
 trabajo; parece sino teines miedo agasajarle, manifestarle carino. Veces tratas 
 cumplimiento veces senor. Probare; ejemplo pudiendo abandonar case negocios, 
 deseaba hubiese acompanado viaje; preferiste sola sobrina doncella. Quise con- 
 tradecir, pero para sentimento, para tambien. Voto gasta palabra, dice frases, 
 dice; pero alia adentros quiere. Mientras estado malo, puesto dirigir casa; 
 pardiez aunque carrera, hacia mejor; cabo tiene sobre ventaja poca edad, activa- 
 dad zelo, pues para contigo digo. Siempre ordenes; dejaria matar alcanzarte 
 billete para opera para baile. Necsitamos para felices; algo estrano, desconocido. 
 Esta resuelto; supuesto hemos hablado esto, mismo, preciso empieces darle con- 
 ocer nuestros planes. Quien mejor. Opone nunca deseos, sera facil nadie per- 
 suadirle. Probare menos, preciso sino creere tienes interes decidido proteger 
 vizconde. Pudieras creer siempre inclinado senores cabra tira monte. Pero 
 tengo nada ellos esposo tienes siempre pensativo -siempre trists. Diablos tiene 
 Carlos acercate tiene hablarte. Holo parece sacado letargo tengo algunas instruc- 
 ciones cajero marcha dentro poco. Para empresa piensa usted establecer Habana. 
 
EXPERIMENTAL WOEE WITH SEVEBAL GROUPS OF TESTS 65 
 
 Precisamente bonita especulacion bien manejada sobre todo. Espero poro tengo 
 entre manos etro proyecto interesa aqui estabamos ocupando pienso. Eres porque 
 
 (B) 
 B. 
 quieres porque e tragas defensa peligro lugar huir mujer, harto debil duda pero 
 algun desgracia tuviese luchar sentimientos seme j antes tuyos, lejos ceder ellos 
 cobardemente moriria pero triunfaria. Tendras menos valor tendre darte lec- 
 ciones valor energia. Vamos, Carlos, amigo creeme sentimiento, profundo razon 
 pueda subyugar, desgracia grande pueda soportar veneer nuestro corazon. Ofrezeo 
 apoyo eres creo sequiras consejos. Bied, hable usted. Quiere casarte Isabel. Isabel, 
 prima imposible; quiere otro, vizconde amigo. Preciso persuadirselo hare otros 
 partidos habra jamas para jurado nada espero pero conservare siempre entero este 
 amor ella ignora unos juramentes recibido. Enhorabuena otro medio asequarara 
 tranquilidad, uya destino ofrecido aleja Madrid, preciso aceptarle. Privarme pre- 
 sencia felicidad hecho usted para consejo especie embargo preciso seguirle solo 
 puedes conservar amistad elige. Jamas caballero crei usted digno consejos dejo 
 usted abandouado mismo nada tango decirle Carlos aleja, echa mirade salir Dona 
 mira; suspira sale. Porque inquieta partida desterremos para siempre memoria 
 quiero puedo presente temo; ausente, echo menos, verle sonrojo, nombre hace 
 temblar. Embargo nunca dicho debiera ignorario Dios Dame f uerzas para resistir. 
 
 Subjects are told to mark every word that contains both an a and 
 a t. If they look doubtful, examples are given of words such as cat 
 which should be marked, and paper which should not. Even so, ex- 
 perience shows that further directions are often necessary even for 
 educated adults. Some subjects mark the letters a and t in the word 
 rather than the word; others do not mark a word unless the a pre- 
 cedes the t, others unless the a and t are together. A sample line 
 with a judicious mixture of words correctly marked might be printed 
 on the blank, and subjects told to look at it for a minute before the 
 signal to begin is given. Those subjects who hit soon upon the de- 
 vice of looking for the rarer and projecting letter t first and then 
 to see if there is an a as well, make better scores than the others. 
 This method might be more easily suggested if the directions said 
 "both a t and an a." Other letter combinations might be better. 
 
 Two ' * instructed ' ' groups using the first part with a time-limit of 
 45 seconds marked, one an average of 11 words correctly, A.D. 2.5, 
 the other an average of 10.2 words, A.D. 1.7. There was an average 
 of 1.4 omissions for the second group, the greatest number being 
 made by those below the average score. 
 
 The short-term group improved from 9.3 to 13.3 words correctly 
 marked in their second test with the first division of the blank and 
 from 7.5 to 10.7 words marked in the second test with the second 
 division. Thus even over an interval of one or more weeks the ac- 
 quaintance with the form of the test or the special blank or both has 
 an effect of over 40 per cent. gain. The long-term group taking the 
 
66 STUDT OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 two divisions alternately gained in days 3 and 4 9.5 per cent, over 
 days 1 and 2. In 20 days they improved from 15.6 and 9.6 words 
 marked for the two divisions to 20.0 and 15. Apparently much of 
 the improvement of the short-term group was due to familiarity with 
 the form of the test rather than with the special blank. 
 Misspelling. — The blanks used are as follows : 
 
 (A) 
 Mark Every Word that is not Spelled Correctly 
 
 1. On the 3d of September, 1832, inteligence was broght to the collecter of 
 Tinnevelly that som wildd eliphants had appeared in the neighborhod. A hunt- 
 ing party was imediately formed, and a large number of nattive hunters were en- 
 gaged. We left the tents, on horsback, at half -past sevin o 'clock in the morrning 
 and rode thre miles to an open spote, flanked on one sid bye Rice-fields, and on 
 the other by a jungle. 
 
 2. After waiting som time, Captain B and myself walked acros the 
 
 rice fields to the shad of a tree. There we herd the trumpett of an elephant; we 
 reshed acros the rice-fields up to our knes in mud, but all in vaiu, thogh we came 
 upon the trak of one of the animels, and then ran five or six hundredd yards iutoo 
 the jungle. 
 
 3. After varius false allarms, aud vane endevors to discuvor the obgects of 
 
 our chace, the colector went into the jungle, and Captin B and myself into 
 
 bed of the stream ' where we had sen the traks ; and here it was evedent the ela- 
 phents had passed to and fro. Disapointed and impasient, we allmost determened 
 to giv up the chace and go home; but shots fird just before us reanimated us, 
 aud we proceded, and found the collecter had just firred twicce. 
 
 4. Of we went throuh forest, over ravin, and through strems, till att last, at 
 the top of the ravine, the elephants were seen. This was a momant of excitment ! 
 
 We wer all scatered. The collector had taken the midle path; Captain B , 
 
 some huntsmen, and myself took to the f ef t ; and the other hunters scrabled down 
 that to the rite. At this momunt I did not see enything but after advanceing a 
 few yards, the hugh hed ef an elephunt shaking abuve the jungle, withen ten 
 yards of us, burst sudenly upon my view. 
 
 5. Captain B ande a hunter justt befor me; we al fired at the same 
 
 moment, and in so dirrect a line that the percussion-cap of my gun hitt the hunt- 
 er, whome I thougt at first I had shoot. This acident, thogh it prouved slight, 
 troubled me a litle. The grate excitement ocasioned by seeing, for the first tim, 
 a wild best at liberty and in a state of natur, product a sensation of hop and fear 
 that was intens. 
 
 (B) 
 Mark Every Misspelled Word 
 
 I percieved, about four years ago, a large spiider in one korner of my room, 
 makeing its web; and through the maid frequentely leveled her fatale brom 
 against the lobors of the little anemal, I had the good fortoone then to prevente 
 its distrucsion, and, I may say, it mor than paid me by the intertainement it 
 aforded. 
 
 In thre days the weeb was, with encredable diligence, compleeted; nor could 
 I avod thinkeing that the insect seemed to exult in its new abode. It often trev- 
 ersd it round, and exsamined the strenth of every part of it, retierd into its whole, 
 and came out very ferquently. The first inemy, however, it had to inconter was 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 67 
 
 another and much larger spidur, which, having no web of its owne, and hareing 
 probibly hexausted all its stock in former labors of this kind, came to invaide the 
 prouperty of its nieghbore. 
 
 Soon a terreble encounter ensooed, in which the invader seemed to have the 
 victorie, and the laborius spider was obleeged to take ref ug in its hole. Upon this 
 I perceived the victer useing every art to draw the enemey from his strongholde. 
 He seemed to go of, but quicklie returned, and, when he found all arts vane, be- 
 gan to dimoilish the new web withoute mercy. This broght on another battle, 
 and contary to my expextations, the laborious spider became conckeror, and fairly 
 killed his antagonist. 
 
 Nou in pieceable possession of what was justely its own, it awated three days 
 with the uttmoste impatients, repairing the breeches of its web, and taking no 
 sustenance that I could perceive. Ate last, houever, a large blue fly fell into the 
 snaire, and strugled hard to get lose. The spider gave it leeve to intangle itself 
 as much as possible, but it seemed to be to strong for the cobwebe. 
 
 I must own I was grately serprised when I saw the spider imediately sally 
 out, and in lese than a minite wheave a new nett around its capthive, by wich the 
 moshun of its wings was stoped, and, when it was f airely hampered in this maner, 
 it was siezed and druged into the houle. 
 
 In this manner it lif ed, in a precarious staite, and Natcher seemed to have fited 
 it for such a life, for upon a singl fly it subsested for a weak. I put a waspe into 
 the neat, but the spider sit it free. 
 
 To a class of 183 members blank B was given. In 30 seconds the 
 average number marked was 18.3 at the first trial, A.D. 4.5, and 18.2 
 at the second trial, A.D. 3.4, when beginning at the third paragraph. 
 There was a total of 34 errors in the first trial, 63 in the second. 
 There were also 156 omissions in the first trial, 160 in the second, the 
 mode being 1 both for errors and omissions, the average omission 2.8. 
 
 The short-term group made four trials with each blank beginning 
 with the first and third paragraphs alternately, 8 tests in all. Their 
 average on the A blank in a time limit of 30 seconds was 18.2; for 
 the B blank, 18.8, or 19.6 for the first paragraph, 18.0 for the third. 
 
 The effect of practise and adaptation was as follows: the record 
 with the two divisions of blank A in the first two sets was 13.1 words 
 marked, 3.1 omissions for Al and 18.8 words, 4 omissions for A2. In 
 the seventh and eighth tests it was 17.7 words, 5.1 omissions, and 
 23.4 words, 6.1 omissions. If one word is deducted for each omission 
 the individual scores become : 
 
 TABLE XIX 
 
 First and Second Trial : Repeated after Four Other Tests: 
 
 Blank Al Blank A2 Blank Al Blank A2 
 
 Bu 7 19 15 16 
 
 Or 12 20 6 15 
 
 Ji 8 14 6 4 
 
 Le 5 9 15 17 
 
 Mo 8 17 5 17 
 
 Ba 16 26 23 27 
 
 Bf 12 17 18 24 
 
 Average 9.7 17.4 11.1 17.1 
 
68 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 The long-term group made 20 trials all with B blank, beginning 
 at different trials with the first, second, third or fourth paragraphs. 
 In a time-limit of 30 seconds their average was 28.4 correctly marked. 
 For the first paragraph it was 30.5, for the third 23.9, with a very 
 slight practise discernible which is here probably traceable to ac- 
 quaintance with the blank. From the first four trials to the last 
 four the change was only from 26.5 words to 28.8 and from 2.2 to 
 1.8 omissions. These blanks should be revised to make each of even 
 difficulty throughout, and to make sure that the A and B blanks are 
 of equal difficulty. The following table shows their present defects 
 and also gives an approximate idea of the time required to find and 
 mark a misspelled word such as these. 
 
 TABLE XX 
 
 B Blank Seconds 
 
 A Blank First Third per Word 
 
 Class of 183 18.3 18.2 correctly marked 1.64 
 
 Instructed 16.0 correctly marked 1.87 
 
 Short-term 18.2 19.6 18.0 correctly marked 1.61 
 
 Long-term (first) 29.3 22.6 correctly marked 1.16 
 
 Long-term (average) . 30.5 23.9 Correctly marked 1.11 
 
 At the end of the 20 trials, each of the three subjects completed 
 the blank, i. e., the amount-limit method was used. Two subjects 
 were slower by this method, the third quicker than she was on the 
 average by the time-limit method. This one subject, who was the 
 most rapid in this test, did not with the amount-limit method exceed 
 her maximum speed with the time-limit method. The following 
 table will make this clear. 
 
 TABLE XXI 
 
 Misspelling Test 
 
 Subject Time Right Wrong Omitted R—{W+0) 
 
 Eecord in last four tests, i N. 120 108 1 6 101 
 
 Blank B, beginning at | W. 120 111 10 101 
 
 If 1, 2, 3, 4, 30 sec. each I F. 120 124 6 118 
 
 Eecord in amount-limit r N. 118 92 1 7 84 
 
 test | W. j 130 94 1 5 88 
 
 I F. 93 98 1 97 
 
 N. lost approximately 15 per cent. 
 
 W. lost approximately 13 per cent. 
 
 F. gained approximately 6 per cent. 
 
 Approximate average loss by amount-limit 7 per cent. 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 69 
 Perception of Forms. — The two blanks used were as follows : 
 
 ;:;S|Si}:;[1'Bi«p:|8!8 
 
 1111111 
 
 :::::::::::::::::si:: 
 
70 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 No. 2 is very convenient as it has eight different geometrical 
 forms of which there are 50 each on the sheet : it is thus to some de- 
 
 □ 
 
 OrrDoUhs^OcNOi 
 
 b 
 
 oa 
 
 o^g^nDgaoDb^A 
 <RoR^o<JoDnG 
 
 =fau 
 Joo 
 
 on 
 
 D 
 
 §D<M3 
 
 QgggSQBg 
 
 at 
 ooDdDPd a<oa>- 
 
 gree comparable with "A" blank No. 3. The square and rectangle 
 may, however, be easily confused, and for that reason were not used. 
 No. 1 has four forms of which there are but 50 each ; but in the first 
 place this blank is exceedingly trying for the eyes, and in the second 
 
EXPERIMENTAL WORK WITH SEVEBAL GROUPS OF TESTS 71 
 
 place forms No. 1 and 3 are not easily and rapidly distinguishable 
 from other forms that appear fairly often. The long-term group 
 had to use this, however, as at the time of their practise the other 
 blank had not been prepared. 
 
 Blank No. 2 was given to the "instructed" group with directions 
 to mark every triangle. The time limit was 60 seconds. The average 
 number marked was 35.2, A.D. 5.8, or 1.71 seconds per triangle. 
 
 The short-term group made two trials marking the trapezoid in 
 each case. The time limit was 70 seconds. The average number 
 marked was 39.3 (A.D. 3.1) in the first and 41.4 (A.D. 3.9) in the 
 second trial. 
 
 Tests were made also with five other forms, but as the subjects 
 after completing all the lines looked back to seek omissions, instead 
 of reporting themselves as having finished, the records are not usable 
 to estimate either practise effect or the difference in difficulty of the 
 forms. The circle and semi-circle are proved to be much easier than 
 the trapezoid, since within 60 seconds the blank was completed by all 
 for the circle (Av. No. marked 48.3) and by three out of seven for 
 the semicircle (Av. No. marked 42.4, Median 41). The last measure 
 is valid, so that we may assume the trapezoid to be approximately a 
 sixth harder to locate than the semicircle on this blank. 
 
 This group made also two trials with blank I. They were told to 
 study the selected pattern at the bottom of the sheet on the word 
 "go," till the signal "now," when they were to mark as rapidly as 
 possible every one exactly like it till the signal "stop." Five sec- 
 onds was allowed for the study, 55 seconds for the marking. With 
 form 1, their average was 13 marked, with form 2 it was 10.6. 
 
 The long-term group made 20 trials with blank I following 
 the directions given above. As they took the different forms in ro- 
 tation they had only five trials with each form. The average for 
 any form was 19.4, the first four trials' average deviating by — 3.2, 
 the last four by + 2.8. 
 
 This and the a — t test gain from repetition with the same blank 
 far more than do the A test and misspelled word tests. The gain 
 would appear therefore to be due more to becoming accustomed to a 
 novel problem in identification rather than to partial memorizing 
 of the positions on the blank. The latter should have been most in- 
 fluential in the A test when repeated 20 times with just, the same 
 arrangement of objects to be marked. 
 
 On examining the records to see if one form benefited more than 
 another, it was seen undoubtedly that form 2, subjectively the 
 easiest, benefited most, and form 4 next. The average number 
 marked in the five trials with each was respectively, 24.6, 22.2, 18.4, 
 
72 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 12.3. Thus No. 4 proved the most difficult. Errors and omissions 
 were not counted on this blank, as it was judged that its difficulty 
 put it on altogether a different plane from the A, a — t, and mis- 
 spelled words tests. 
 
 At this point some note may be taken of the speed attained in 
 these tests. The process required is so similar in all of them — to 
 look for some special thing, and mark it when seen, that more uni- 
 formity in speed might be expected than was found among the as- 
 sociation tests. One test classified under association requires this 
 same process of checking rather than writing words or parts of 
 words, and the consideration of speed in that was deferred for com- 
 parison with these tests. It was the marking of nonsense syllables 
 and English words out of a mixed list. For purposes of comparison, 
 all are reduced to the time required to find and mark one object of 
 the specified sort. The conditions of the surroundings of the object 
 must be kept in mind in considering these figures. 
 
 TABLE XXII 
 
 Scores in Early Trials 
 
 Sec per Unit Found and Marked 
 Short Long Various 
 
 Term Term Instructed 
 
 100 A's amongst 400 other letters 83 .64 1.06 .94* 
 
 50 A 's amongst 650 other letters 1.87 2.00 
 
 50 B 's amongst 650 other letters 1.83 
 
 50 K 's amongst 650 other letters 2.22 
 
 50 S's amongst 650 other letters 1.61 1.62 
 
 50 triangles amongst 350 other forms 1.71 
 
 50 trapezoids amongst 350 other forms 1.74 
 
 50 semicircles amongst 350 other forms .... 1.44 
 
 Misspelled words amongst 300 other words . . 1.61 1.16 1.87 
 
 25 nonsense syllables amongst 75 confusion 
 
 words 3.10 
 
 * Columbia and Barnard students. 
 
 From the difference found in marking A's, it is evident that the 
 arrangement of the blank itself and the possible number of units to 
 be examined is one of the largest factors in the rate of marking. 
 
 Another test commonly classified under perception tests, though 
 totally different from all so far described, is that known as " per- 
 ception of size." The freshmen are given a sheet of paper bearing 
 a 5-cm. line, which is placed to their left, also a blank sheet of paper. 
 They are asked to draw a line the same length as the standard with- 
 out moving the papers or measuring in any way, then to bisect the 
 line drawn, then to erect a perpendicular the length of the line. 
 Columbia freshmen are also asked to bisect the right-hand angle. 
 
EXPERIMENTAL WOBK WITH SEVERAL GROUPS OF TESTS 73 
 
 The men make an average error of 2.4 mm. in drawing the first 
 line, the women 3.7 mm. 
 
 The records of three graduate men students who made 50 trials 
 in five sets of ten each of drawing a line equal to a standard line, 
 were examined. These three were chosen at random from a class of 
 eleven. The average errors in 50 trials were respectively 2.cT mm., 
 3.7 mm., and 1.8 mm. A changed from 1.5 for the first group to 4.5 
 in the last, mainly on account of developing a positive constant error. 
 B changed from 1.6 to 5.5, also because of a large positive constant 
 error. C changed from .7 to 1.0, his larger average for the total 
 series being influenced by a negative constant error in the fourth 
 group. 
 
 The short-term practise group made ten trials of each of the four 
 
 processes required of the freshmen, after taking the test as a whole 
 
 once. Unlike the method in other tests, they made all ten trials of 
 
 one process at one sitting, as the three subjects A, B, and C had done. 
 
 The results were, in terms of error : 
 
 Av. a.d. 
 
 Line 3.4 mm. 1.8 
 
 Vertical 5.7 mm. 3.9 
 
 Bisect line 1.5 mm. * 1.0 
 
 Bisect angle 3.2° 1.7 
 
 As might be expected from the illusion involved in erecting the 
 perpendicular, the largest error is found there, and is a negative 
 constant error. The average for drawing the line equal to the stan- 
 dard is very near that of the Barnard freshmen. No subject did 
 equally well in all four processes ; in fact the one with the least error 
 in drawing the line made the greatest in bisecting the line, and 
 another who made the least error in bisection of the line made the 
 greatest in erecting the perpendicular. 
 
 No practise effect was discernible in the ten trials, and since 
 the tendency of a rather longer practise is to confirm a constant error, 
 the earlier trials may perhaps give more accurate results, though 
 they may not reveal individual differences in habituation. 
 
 B. Relative Value of these Tests on Perception 
 There can be no question that in freedom from ambiguity due to 
 measuring, in early trials, a combination of ability to perceive ob- 
 jects and ability to get used to the form of a test the A test and 
 geometrical forms test are markedly superior to the a — t and the 
 hieroglyph tests. There is some uncertainty with respect to the 
 misspelled words test, but it is at least probable that the first trial 
 with it is largely influenced by a person's ability to set his mind to 
 
74 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 the novel task. It is unnecessary to repeat details here as it will 
 appear that for other reasons the misspelled words is an undesirable 
 test. 
 
 The question of the significance of these tests of perception as 
 shown by their correlations was next studied. 
 
 First of all the performances of the eighteen instructed subjects 
 were compared in the four tests, A, a< — t, triangle or perception of 
 forms and misspelled words. Each test was compared with the aver- 
 age for all four. 
 
 The coefficients are: 
 
 TABLE XXIII (a) 
 Tests Cos ttU 
 
 Perception of geometrical forms .90 
 
 A 34 
 
 a — t 81 
 
 Misspelled words 64 
 
 Average of 
 these four < 
 tests and 
 
 r 
 
 Av. 
 
 Order 
 
 .65 
 
 .78 
 
 1 
 
 .82 
 
 .58 
 
 4 
 
 .49 
 
 .65 
 
 3 
 
 .85 
 
 .75 
 
 2 
 
 r 
 
 Av. 
 
 Order 
 
 .83 
 
 .87 
 
 1 
 
 .16 
 
 .32 
 
 
 .65 
 
 .63 
 
 3 
 
 .72 
 
 .81 
 
 2 
 
 .35 
 
 — .18 
 
 4 
 
 .57 
 
 .74 
 
 2 
 
 .54 
 
 .38 
 
 
 Next the first two trials of the short-term practise group were 
 compared in seven tests — a — t, e — r, A, misspelling, perception of 
 forms (2 blanks), perception of size, each with the average of all. 
 The results are : 
 
 TABLE XXIII (6) 
 Tests cos tj-U 
 
 Perception of geometrical forms .90 
 Forms 1 and 2 (hieroglyphs) . .48 
 
 Average of A 61 
 
 these seven - a — t 90 
 
 tests and Misspelled words 
 
 e — r 90 
 
 Perception of size .22 
 
 Next, the performances of the long-term group were compared 
 in the four perception tests with which they practised. For this all 
 the 20 records for each subject were averaged. As there were four 
 forms in the perception of forms, and two parts to the a — t blank 
 it was all the more advisable to avoid making any selection from the 
 total number of trials. It should be noted that this group used dif- 
 ferent blanks in the case of the A test, and perception of forms from 
 those used by the other two groups, also that in the A test these sub- 
 jects reached something presumably near the physiological limit. 
 
 The correlations were: 
 
 Order 
 
 {Forms 1, 2, 3, 4 (hieroglyphs) ... r = .87 3 
 
 A r = .88 2 
 
 a — t r=.98 1 
 
 Misspelled words r = .79 4 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 75 
 
 It appears that even so few as two tests of approximately a 
 minute with the A, a — t or geometrical forms tests are significant 
 of an individual 's ability in visual perception. Amongst these three 
 tests there is little choice. The geometrical forms test is perhaps the 
 most typical of the general function in question, but both the A or 
 the a — t are satisfactory in this respect. 
 
 The precision of the otherwise desirable tests of perception was 
 measured, as for the association and memory tests, in terms of the 
 average divergence of the result obtained from a single trial from 
 the individual's true total ability, and the amount is expressed, as 
 before, in per cent, of the former. 
 
 TABLE XXIV 
 
 Kelative Precision of Perception Tests 
 
 Probable Average Divergence of 
 
 the Result Obtained from 1 Trial 
 
 from the Probable True Result, 
 
 in Per Cent, of the Former 
 
 Time in Short Long Term, 
 
 Test Seconds Term Early 
 
 A (Blanks 1 and 2) 60 5.4 2.8 
 
 S on blank 3 35 5 
 
 a—t 45 7 4.6 
 
 Misspelled words 30 10 5.4 
 
 Forms (trapezoid) 70 4 
 
 Here again, marking letters, marking words containing certain 
 letters, and marking geometrical forms are all fairly satisfactory 
 with little to choose among them. On the whole perhaps the A test 
 and geometrical forms used together would be the best. The latter 
 has the advantage of being uninfluenced by habituation to any one 
 visual alphabet, and is therefore adaptable to more kinds of people, 
 e. g., young children or members of different racial groups. 
 
 4. Tests on Discrimination 
 A. Descriptive 
 
 Another test given the freshmen is that of naming 100 colors as 
 quickly as possible. 100 1 cm. squares of 10 different colors are ar- 
 ranged in chance order on a white ground. Care is taken that the 
 students have a ready name for each color there before beginning 
 the test; then they are asked to read off — or name — all the colors 
 there as rapidly as possible, while the time taken is noted. A name 
 like "old rose," preferred by some students to "pink," makes an 
 appreciable delay, so that it might be better to have 10 indisputable 
 shades, or even briefer names assigned in print to a sample row. 
 
76 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 The men take 85 seconds on the average (P.E. 14) to read the 
 100 colors, and the women 67.2 seconds. Here, as in the marking 
 100 A's, the women are quicker than the men. 
 
 The short-term group made 6 trials with this test individually. 
 Their average time on the first trial was 56 seconds; for the total 
 series it was 53.1 seconds, with A.D. 9.9. In half the cases there 
 was a slight practise effect discernible. The A.D. of the successive 
 averages was only 1.2. The successive averages were 56, 54, 51.5, 
 51.7, 51.8, and 53. 
 
 The long-term group made, as usual, 20 trials, using a rather 
 smaller piece of apparatus. Their average time was 46.7 seconds, 
 the first trial's average deviating by + 16, the last by — 4. The 
 greatest gain was made from the first to the second trial. The 
 first six averages were 62.7, 49.6, 50.8, 48.1, 50.9, and 46.6. It was 
 interesting to note that the most rapid talker was considerably the 
 slowest at the beginning of this test, though by the twentieth trial 
 she had caught up with the second quickest. The one who did the 
 best seemed to acquire her speed principally by careful economy of 
 breath. On three occasions she read the 100 colors in 36 seconds. 
 
 At the end of the 20 trials each was asked to read off 100 color 
 names without discrimination; that is, to move eyes and hand in 
 pointing as before but to use the same word 100 times. The respec- 
 tive times taken for this were 37.5, 33, and 31 seconds, as compared 
 with 44, 44, and 40 seconds at the 20th trial. The average extra 
 time needed for discrimination beyond the mechanics of the test was 
 therefore at the end 8.2 seconds. 
 
 Naming Forms. 
 
 Along with this test it was thought that comparison of forms and 
 objects might be made, as similar material was being used in the 
 memory and perception tests. Accordingly 100 squares were filled 
 with 10 each of 10 different forms in chance order. These forms 
 were star, cross, square, oblong, spiral, circle, "dots" (three dots 
 spaced to form an equilateral triangle), oval, line, and triangle, and 
 were drawn in ink or stamped from rubber type in black on a white 
 ground. The whole resulting square was only four inches. Only 
 the long-term group practised with this test. In 20 trials the aver- 
 age time taken was 53.3 seconds, the first day's average deviating' by 
 + 16.7, the last by — 5.3. Again the greatest gain was made from 
 the first to the second trial. The first six averages were 70.0, 58.5, 
 59.2, 58.0, 57.6, 54.8. More errors in naming were made with this 
 than with naming colors, though very few all told, a total of 9 for one 
 subject, 6 for another, 4 for the other. Introspectively, these errors 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 77 
 
 are not due to faulty recognition but to difficulty in saying the right 
 word; in the rapid enunciation the speech channel got blocked, or 
 the "tongue twisted" as we say commonly, so that a circle would be 
 called spiral, the subject being conscious of the error at the time of 
 making it. Just here a question arises : the freshmen make slips in 
 naming the colors too, and the directions should include advice 
 about going on in spite of mistakes recognized as soon as made, or 
 going back to correct them. Otherwise a considerable difference 
 occurs in the time taken. The Barnard freshmen are told to go on 
 usually, but in spite of this some conscientious students go back. 
 Individual differences come out rather well on this point but escape 
 the measuring rod of the statistician. 
 
 To return to the long-term group — the same subject was quickest 
 in these two tests, but the other two changed rank. In neither of 
 these two tests could there presumably have been any memory aid, 
 as on successive trials the apparatus was turned round and the 
 reading begun from a different corner. 
 
 Naming Objects. 
 
 A third test was devised, that of naming 100 objects. Owing to 
 the trouble involved in collecting these and setting them out on a 
 small table, four readings were made on the same day by each sub- 
 ject for five separate days, instead of one a day. They began at a 
 different corner for each reading, however. The objects included 
 keys, spoons, nails, screws, corks, pencils, books, tumblers, hairpins, 
 spools, paper, matches, candles, checkers, picture-hooks ("hang- 
 ers"), boxes, bottles, flowers, leaves, berries — all small but familiar 
 objects, arranged again in chance order in 10 rows of 10. Intro- 
 spectively this was a harder test, the space taken up in three dimen- 
 sions seeming to confuse the subjects. The average time taken was 
 56.2 seconds, the first trial's average deviating by -f- 8.4, the last by 
 — 1.3. The greatest gain was made from the first four readings to 
 the next four, not from the first, to the second, nor was there any 
 marked improvement from the first to the second reading on any 
 one day. The first eight averages were — 64.6, 61.3, 65.1, 59.9, 54.3, 
 53.9, 53.1, and 52.3. It may be therefore that the particular com- 
 bination and arrangement of the objects on the first day was more 
 difficult to read off than on any other day; or else that the new, 
 strange feeling persisted through all four readings on the first day, 
 but disappeared on the second occasion when four readings were 
 to be made. 
 
78 
 
 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 B. Relative Value of these Tests on Discrimination 
 
 First the correlation of these tests was examined. 
 Again all 20 records for each subject were utilized, as any selec- 
 tion of records seemed to measure the effect of practise at different 
 
 The results were : 
 
 TABLE XXV 
 
 {Naming colors r = .67 
 Naming forms r = .99 
 Naming objects r = .96 
 
 Naming colors and objects r = .45 
 
 Naming forms and objects r = .93 
 
 Naming colors and forms r = .73 
 
 From this it would seem that naming colors is unlike the other 
 two tests devised, as it does not correlate so closely with the average 
 for the three as do the other two, nor are its intercorrelations close. 
 Naming forms seems more a typical test in so far as it measures an 
 ability common to these three tests. These relationships persist 
 through " trial correlations" of selected records. 
 
 Unfortunately there were no records available from the "in- 
 structed" group to give greater weight to these correlations. 
 
 All three of these tests are of the same general degree of pre- 
 cision, color naming being somewhat the best. It is noteworthy that 
 the individual variation of daily trials is so great in so simple a per- 
 formance. The facts follow in Table XXVI. 
 
 Test 
 Name colors . . . 
 Name forms . . . 
 Name objects . . 
 
 TABLE XXVI 
 
 Average Divergence of the Rate Found 
 
 in One Trial from the Individual's 
 
 True Rate. In Per Cent, of 
 
 the Former 
 
 Short- Long-term Group 
 
 term Early Late 
 
 Group Trials Trials 
 
 . 3.8 6.6 5.0 
 
 6.6 
 6.8 
 4.6 
 
 5.1 
 
 8.3 
 
 
 Probable 
 
 
 Number of 
 
 
 Trials Re- 
 
 
 quired to 
 
 Time Per 
 
 Reduce the 
 
 Trial in 
 
 Unreliability 
 
 Seconds 
 
 to 1 Per Cent. 
 
 50 
 
 26 
 
 53 
 
 35 
 
 56 
 
 42 
 
 Introspectively, naming objects is most unlike the other two 
 tests ; it is certainly the most awkward to use. In the memory tests, 
 objects seemed to have the advantage over forms, but there, of 
 course, there was no question of speed in making the test, and as 
 mental speech was a distinct help in remembering, objects stood a 
 better chance with their definite names than did unnamed forms. It 
 could be wished that perception of colors had also been used, to make 
 comparison possible between colors and forms in the two processes 
 
EXPEEIMENTAL WORK WITH SEVEBAL GBOUPS OF TESTS 79 
 
 of checking and naming, though the supposition would be that un- 
 less the colors were unequivocably distinguished some students might 
 suspect it as a test of artistic taste or ability to match shades. 
 
 From experience with these tests it is suggested that names of 
 forms would be less indefinite to read off than are those of colors; 
 and as colors are apt to fade, the forms test has a slight advantage. 
 The forms test is as easy to administer, is almost or quite as desir- 
 able from the point of view of susceptibility to practise and unre- 
 liability, and is perhaps more significant of the process of naming 
 in general. 
 
 5. Discrimination and Motor Tests 
 A. Descriptive 
 
 Another allied series of discrimination tests was practised by the 
 long-term group, but they are discussed separately as they involved 
 a different motor reaction. The series included sorting ordinary 
 playing cards by suit, similar sized cards by number, and small 
 objects by size, color, or shape, making five tests in all. Similar 
 tests have been devised before and used in such studies as Berg- 
 strom 's. 46 
 
 Sorting Cards. — An ordinary pack of cards was well shuffled, 
 and then, held face up, dealt out into four piles according to suit, 
 the subjects choosing their own positions for the piles. Before 
 making the first trial, each subject dealt a pack into four piles with- 
 out discrimination of suit, as one deals when playing a game; the 
 respective times taken in this preliminary trial were 17 seconds, 17.2, 
 and 19, as against 26.4, 39.2, and 28.2 for the first trial with dis- 
 crimination. Thus, the average extra time needed for the discrimi- 
 nation process was 13.5 seconds. The average time taken through 
 the 20 trials was 26.5 seconds, the first day's average deviating by 
 + 4.8 seconds, the last by — 2.7. Near the beginning there was no 
 marked improvement; the greatest change occurred between the 
 eighth and ninth trials. The slowest subject made a total of eleven 
 errors, the quickest two, the other one none. On four days two 
 trials were made in succession, and of the twelve records, there were 
 five where the second trial took less time than the first. 
 
 Sorting by Number. — Compared with this was a test in which 
 60 cards — 10 each of 6 different numerals, were to be sorted into 6 
 piles. These sets were selected from the complete pack of 150 used 
 in playing "Flinch," care being taken not to confuse the eye by in- 
 cluding 5's, 3's, and 8's in the same set of 60. Different sets were 
 
 46 Am. J. Psy., 6, 24. 
 
80 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 used on different days. On ten occasions the subjects knew before- 
 hand what numbers to expect ; on ten, they had to find out as they 
 dealt. As before, they were at liberty to place their piles as they 
 wished, but in this test the cards were held face down. 
 
 The average time for the 20 trials was 58.4 seconds, the first day's 
 average deviating by -j- 7.4, the last by — 4.6. The greatest im- 
 provement occurred near the beginning, between the second and 
 third trials. Comparing the ten trials when the numbers were 
 known beforehand with those when they were not, there was an 
 average difference of 2 seconds in favor of knowing them. 
 
 At the end of the 20 trials each subject dealt the 60 cards into 
 6 piles without discrimination. The times taken were respectively 
 24, 26, 25 seconds, as compared with 55, 55 and 51 at the 20th trial. 
 The average extra time needed for discrimination was then 28.8 
 seconds. 
 
 Comparing the two tests — with the more familiar material, an 
 easier manipulation and a narrower choice, a card was handled in 
 .51 of a second on the average. With numbers, an additional move- 
 ment, and six instead of four, possibilities, in .97 of a second. Elim- 
 inating the discrimination, before practise the playing cards were 
 handled at the rate of one in .34 of a second; with the additional 
 movement and after practise, the numbered cards at the rate of one 
 in .42 of a second. This extra time is probably taken up by the 
 turning of the cards. Unfortunately, trials by both methods with 
 each kind of material were not made to make this point decisive. 
 There is also the possibility that the pack of " Flinch'' cards was less 
 easy to handle than any of the three ordinary packs of cards. 
 
 The subjects held the same relative rank for speed in these two 
 
 For the other three tests small objects such as pieces of thick 
 cardboard, checkers, buttons, marbles, kindergarten beads, chess 
 pawns, ' ' halma ' ' men, ping-pong balls, candle-ends, small spools and 
 children's alphabet blocks were used. Three sets of 60 objects each 
 were made up from this assortment, one to be sorted by size, another 
 by color, the third by shape. In sorting by size, the objects were all 
 discs, but varied in color as well as in thickness and diameter. In 
 sorting by color, all sizes and shapes were included, and in sorting 
 by shape, all sizes and colors. 
 
 The 60 objects were contained in a cardboard box ; from this 
 they were to be sorted into six smaller cardboard (shoe) boxes placed 
 in a row. The subjects were at liberty as in the card sorting test to 
 distribute as they wished rather than to memorize the experimenter's 
 choice of the position of the different kinds of material. Usually the 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 81 
 
 three tests were taken one after the other with about two minutes' 
 interval. The order was varied from day to day to equalize the 
 interference effect. On the first day, each subject had the benefit of 
 watching the other two do two of the tests, herself going through 
 the third test in their presence before they did it. Otherwise these 
 trials were made alone. 
 
 The general experience with these tests was that the subjects did 
 not take any object that was nearest and then place it in the right 
 box, but tried to get all 10 of one kind of object before beginning on 
 another kind. This was not invariable however, as there was also a 
 tendency to handle the largest objects first whatever they might be. 
 No restrictions were put upon the subjects except that the objects 
 were to be handled one at a time. This ruled out an ingenious de- 
 vice of one subject, of leaving the thinnest and flattest till the last 
 and then pouring out all 10 at once straight from one box into the 
 other. Careful observation showed that the training of the left 
 hand played no small part in the gain in speed. 
 
 Sorting by Size. — The average time taken was 31.5 seconds, the 
 first day's average deviating by + 4.3, the last by + 1.7. The best 
 record was made on the 18th trial. In all 60 cases there were but 
 five errors. 
 
 Sorting by Color. — The colors were black, white, red, blue, green, 
 and yellow. The average time taken was 33.5 seconds, the first day's 
 average deviating by + 7.0, the last by -}- 2.0. The greatest im- 
 provement came between the second and third trials. The best score 
 was at the 16th trial. 
 
 The most rapid worker made eight errors, the other two five 
 each. Thus there was greater inaccuracy with the color discrimina- 
 tion than with the size. 
 
 Sorting by Shape. — The shapes were — cube, sphere, cylinder, 
 disc, flat-square, and halma man (resembling a chess-pawn, but only 
 three fourths inch high). The average time taken was 47.5 seconds, 
 the first day's average deviating by + 10.4, the last by — 6.7. For 
 the first nine trials the improvement was very irregular (av. 51.4, 
 A.D. 3.7), but from the tenth trial on it was much more regular 
 (av. 44.4, A.D. 2.1). The best score was the 20th. The most rapid 
 worker made 14 errors, the next 12, the slowest 8. 
 
 Sorting by Size was least influenced by adaptation and practise, 
 sorting by color next, while sortiyxg by shape, though irregular in its 
 course, showed a gain of from 25 to 30 per cent, in twenty trials. 
 
 This and also the time per unit of the process is shown by 
 Table XXVII. 
 
82 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 TABLE XXVII 
 
 Average Time of Three Subjects in Successive Daily Trials 
 with the Sorting Test 
 
 Time Required Per Unit Sorted, in Seconds 
 
 Cards with Large Numbers 
 Held Face Down, Into 6 
 
 Plaving Cards 
 
 Held Face Up, 
 
 Into 4 Piles, 
 
 by Suit 
 
 .60 
 
 Piles, by Varying Number 
 
 Number Number 
 
 Known Unknown 
 
 Beforehand Beforehand 
 
 1.10 
 
 By Size 
 
 Into 
 6 Boxes 
 
 .60 
 
 Sorting 60 Objects 
 By Color 
 
 Into 
 6 Boxes 
 
 .68 
 
 By Shape 
 
 Into 
 6 Boxes 
 
 .98 
 
 .60 
 
 
 1.09 
 
 .57 
 
 .64 
 
 .87 
 
 .58 
 
 1.00 
 
 
 .53 
 
 .56 
 
 .98 
 
 .62 
 
 
 1.02 
 
 .55 
 
 .54 
 
 .88 
 
 .58 
 
 
 .98 
 
 .52 
 
 .52 
 
 .74 
 
 .56 
 
 
 .93 
 
 .52 
 
 .58 
 
 .85 
 
 .59 
 
 
 1.07 
 
 .54 
 
 .57 
 
 .82 
 
 .53 
 
 
 .96 
 
 .55 
 
 .55 
 
 .76 
 
 .47 
 
 
 .99 
 
 .55 
 
 .54 
 
 .84 
 
 .48 
 
 
 1.03 
 
 .45 
 
 .53 
 
 .73 
 
 .44 
 
 1.01 
 
 
 .47 
 
 .51 
 
 .74 
 
 .43 
 
 
 .96 
 
 .51 
 
 .52 
 
 .72 
 
 .49 
 
 .94 
 
 
 .49 
 
 .55 
 
 .80 
 
 .48 
 
 .97 
 
 
 .54 
 
 .58 
 
 .81 
 
 .46 
 
 
 .93 
 
 .50 
 
 .53 
 
 .78 
 
 .45 
 
 .96 
 
 
 .54 
 
 .52 
 
 .77 
 
 .46 
 
 .92 
 
 
 .51 
 
 .58 
 
 .72 
 
 .47 
 
 .89 
 
 
 .49 
 
 .56 
 
 .72 
 
 .43 
 
 .93 
 
 
 .55 
 
 .58 
 
 .68 
 
 .46 
 
 .90 
 
 
 .55 
 
 .60 
 
 .68 
 
 Comparing all three tests, the same subject was quickest in all 
 of them, and was also the second quickest in the two card sorting 
 tests. Neither of the other two kept the same rank throughout. In 
 the average time taken, it would have been expected that sorting by 
 size might be different from the others, as there was not quite the 
 same variety in the material, and the objects were slightly more 
 tiresome to handle. However, the average times for size and color 
 are about the same, 32 and 34 seconds, while that of shape was con- 
 siderably longer, 47 seconds. Introspectively, sorting by shape was 
 the most difficult, perhaps the least familiar way of regarding things. 
 
 B. Relative Value of these Discrimination-motor Tests 
 
 These various "discrimination-motor" tests were correlated, 
 using as before all available records from the three subjects of the 
 long-term group. The results were as follows: 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 83 
 
 TABLE XXVIII 
 
 {Sorting objects by shape r = .68 
 
 Sorting objects by color r = .9S 
 
 Sorting objects by size r = .99 
 
 By shape and by color r = .54 
 
 By shape and by size r = .55 
 
 By size and by color r = .98 — _ 
 
 Sorting cards by number and by suit r = .96 
 
 From this it appears that sorting by shape is most unlike the 
 other tests, agreeing with the introspective evidence and the observ- 
 er's notes at the time; otherwise, all the correlations are close. If 
 however we include the two tests with cards and correlate each of the 
 five with the average of all five sorting tests, sorting by shape is 
 found to be the best representative. One individual who was the 
 slowest in sorting objects by size and color and in the second place 
 in sorting objects by shape was the most rapid in both tests with 
 cards and the correlations became : 
 
 Average of these 
 five tests and 
 
 TABLE XXIX 
 
 Sorting objects by shape 99 
 
 Sorting objects by color 52 
 
 Sorting objects by size 61 
 
 Sorting cards by suits 63 
 
 Sorting cards by 6 numbers 43 
 
 The measurements of relative precision on the basis of early and 
 late trials of the three subjects show, as with the naming 100 colors, 
 shapes, and objects, a large variation due to accidental causes in- 
 cluding those which differentiate one day's condition from another. 
 Even so simple a process repeated 60 times needs apparently from 
 10 to 50 trials, or from 8 to 30 minutes to measure a person within 
 1 per cent. Sorting by size is especially variable, and sorting by 
 mnriber least so. The facts are as given in Table XXX. 
 
 Test 
 A By size (60 objects) . 
 
 B By shape (60 objects) 
 
 C By color (60 objects) 
 
 D By suit (52 cards) . . 
 
 E By number (60 cards) 
 
 TABLE XXX 
 
 Precision of Sorting Tests 
 
 Probable Average Divergence of the Result 
 Obtained from One Trial from the Prob- 
 able True Ability. (3 Individuals) 
 
 First Five Trials 
 As Per 
 Cent, of the 
 Time Re- 
 quired by 
 Individual 
 8.6 
 
 In 
 Seconds 
 . 3.0 
 
 Last Five Trials 
 As Per 
 Cent, of the 
 Time Re- 
 quired by 
 Individual 
 10.3 
 
 In 
 
 Seconds 
 2.9 
 
 4.4 
 2.0 
 2.1 
 2.0 
 
 8.3 
 5.5 
 6.0 
 3.3 
 
 1.4 
 2.7 
 1.5 
 1.5 
 
 3.3 
 8.3 
 6.6 
 
 2.8 
 
 Approx- 
 imate 
 Average 
 Time Nec- 
 essary to 
 Sort the 60 
 
 (52 in 
 
 Case of D) 
 
 31 
 
 47 
 
 33 
 
 26 
 
 58 
 
 Approx- 
 imate 
 Number 
 of Trials 
 Needed to 
 Reduce 
 the Aver- 
 age Diver- 
 gence to 1 
 Per Cent. 
 88 
 34 
 48 
 40 
 9 
 
 Approx- 
 imate Time 
 in Minutes 
 Necessary to 
 Reduce the 
 
 Average 
 Divergence 
 
 to 1 Per 
 Cent. 
 45.5 
 
 26.5 
 
 26.5 
 
 17.3 
 
 7.7 
 
84 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 From these facts, and from experience with the tests it is sug- 
 gested that sorting small objects by color is a good test. It is less 
 confusing than sorting by shape, yet can be varied more than sorting 
 by size. In sorting cards one is confronted with the very unequal 
 abilities people possess in their manual dexterity owing to previous 
 experience; in using objects, the extra trouble in providing them is 
 offset by the greater equality in experience of subjects at the start. 
 Otherwise, pictures, words, figures, geometrical forms, material in 
 great variety can be prepared on cards. 
 
 6. Tests for Speed and Accuracy of Movements 
 A. Descriptive 
 
 To the freshmen is given the following blank with directions, for 
 the first half, to place a dot in each square as rapidly as possible. 
 
 The average time taken by the men is 34 seconds, P.E. 4 ; by the 
 women 30.8 seconds. 
 
 In the second half of the test the subjects are required to strike 
 each dot. The average times taken are 49 seconds by the men, 45.5 
 by the women. The average error in accuracy has been measured 
 only for the men ; with them it is .8 mm. 
 
 Trials of this by the short-term group were not sufficiently 
 numerous to develop a practise effect, but only to give a basis for 
 correlation with other tests. Their average speed in the first half 
 was the same as the freshmen's, though given by the time-limit 
 method. This might suggest that an easy test such as this, where 
 speed is the only thing emphasized, could be given by either method 
 without suffering in rate. In the second part of the test, the short- 
 term group worked proportionately slower than the freshmen, ma- 
 king an average of 59 hits in 30 seconds (or needing 50 seconds to 
 complete the test). Three fifths of these were not separated from 
 the dot to be struck so that their average deviation from the mark 
 might be called the radius of the pencil mark plus the radius of the 
 printed dot (the latter is about .25 mm.). But the dot is often a 
 very short dash and its radius or width varies so that such measure- 
 ments are hardly of value. Wissler, who computed the average error 
 of .8 mm. for the freshmen does not state how he computed it. 
 
 More attention was given by the short-term group to the various 
 forms of maze tests that have been prepared. Of these the following 
 five were used, known respectively as the curved, straight, combined, 
 black, and spiral. The instructed and long-term groups used only 
 the curved. The directions in each case were to draw a line between 
 the two lines without touching either, working as quickly as pos- 
 

 '} 
 
86 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 sible. Care was taken also to see that the blank was placed always 
 in the same position before the subject, and that it was not moved 
 during the tracing. In general, most subjects in a single test pay 
 more attention to the accuracy than to the speed; with repeated 
 tests, however, the emphasis tends to shift, with the result that in a 
 long period of practise the accuracy decreases for a while and the 
 speed increases very considerably. Once conscious of this, the sub- 
 jects will redirect their chief attention to the accuracy so that after 
 20 to 24 days' practise the speed may have increased but slightly, 
 while the accuracy may have improved a great deal. Having real- 
 ized this, with both the instructed and the short-term-practise group 
 — who, it will be remembered, were tested some months after the 
 long-term group, although their results have here been noted first — 
 the emphasis was chiefly and continuously laid on the accuracy, in 
 the hope of getting the practise effects shown in speed, with errors 
 constantly at zero, or sufficiently near it to be almost negligible. A 
 more rapid improvement might thus be looked for, with unwavering 
 attention to one factor, and also the scoring would be much 
 simplified. 
 
 Curved Maze. 
 
 The instructed group used this as a time-limit test. In 60 sec- 
 onds they traced (omitting one subject who completed the blank, 
 but with 26 touches) 41.4 per cent, on the average, with 2.9 touches. 
 The short-term group made three trials with this. The first two were 
 amount-limit tests, with an average time taken of 169.5 seconds. 
 The third trial was meant as a time-limit test and so announced, but 
 all the subjects except one finished before the 165 seconds limit set. 
 As in the cancellation test then and in the first-idea test, the an- 
 nouncement of time limit spurred on most of the subjects to work 
 faster. Taking the three tests together, the average number of 
 touches were 1, 3 and 1. 
 
 The long-term group made 20 trials with this as a time-limit test, 
 using 60 seconds. The average amount traced was 76 per cent., the 
 first day's average deviating by — 7, the last by + 1.6. The average 
 number of touches was 11.3. In these subjects no steady improve- 
 ment was noticed. N in the first five trials paid most attention to 
 speed, with an average of 16 touches. In the next four trials, with 
 more attention to accuracy the average number of touches dropped 
 to 8, while the speed very slightly decreased. After this, her records 
 were not so markedly irregular. W was most ambitious to complete 
 the maze within the 60 seconds at least once. For this reason she 
 began on the ninth day to spurt, succeeding on the thirteenth day 
 
EXPEEIMENTAL WOEK WITH SEVEEAL GEOUPS OF TESTS 87 
 
 Curved 
 
 jlJiUryiJiUrlJn! 
 
 jvynyrynuryi 
 jiinyryiyiJriJi) 
 jijiinyryiunyn] 
 
 C omi/'rtcd 
 
 Blaxk 
 
 Spiral 
 
 N. B. These are reduced to f actual size. 
 
88 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 in finishing. During this spurt her number of touches rose from an 
 average of 12 to an average of 19, after which they dropped back 
 again to 12. The third subject was slower and steadier than the other 
 two. Finding, however, by the fifth day that she did not get so far 
 as the others, she attempted for two days to put on speed with the re- 
 sult that her average number of touches rose from 6.5 to 15.5. There- 
 after she paid most attention to accuracy and kept the number of 
 her touches down. As these spurts by the three subjects did not 
 occur simultaneously, the resulting average curve scarcely reveals 
 the real conditions. On the whole there was a gain of 10 or 15 per 
 cent, in the 20 days. 
 
 It appears then that if subjects work with the curved maze at a 
 very high speed they gain perhaps one half of one per cent, a day. 
 If they work with care so as to have only one or two touches they can 
 increase their speed much more than that per day. 
 
 From these observations therefore, in practise with the other 
 maze tests with the short-term group, accuracy was strongly and con- 
 tinuously emphasized, to see (1) if when errors were kept at zero 
 there would be a practise effect in speed, and also (2) if there was an 
 optimum time discoverable which could be used as a standard when- 
 ever such maze was to be used with large groups of subjects as a 
 time-limit test. 
 
 Straight Maze. 
 
 This maze has two advantages — that of permitting a regular 
 familiar movement, and that of presenting units easily measurable. 
 Each blank can be used as the basis of five separate trials, and was 
 twice so used by the short-term group. For the first five, time limits 
 of 60, 50, 40, 30 and 30 seconds were set. At the beginning the 
 subjects were told that they would have plenty of time to finish 
 without touching, later on that they would have a little less. The 
 first trial, of eight subjects two did not finish and two made touches 
 (2 and 1). The second trial, one did not finish, and one made one 
 touch. The third trial, three did not finish and one made a touch. 
 The fourth time, six did not finish, two made touches (1 and 1). 
 The last time, three did not finish, two made touches (2 and 1). 
 Thus no gain in accuracy was made by the increase from 30 to 60 
 seconds, though most of the extra time was used. 
 
 The next time the blank was used it was given as an amount- 
 limit test, or rather as five such tests, as each line was taken as a 
 unit. In the five trials the average times taken were 29.3, 27.3, 27.9, 
 24.1, 23.5 seconds; the average numbers of touches were .4, .9, .1, 
 .3, and .7. 
 
EXPEBIMENTAL WOBK WITH SEVEBAL GBOUPS OF TESTS 89 
 
 The combined, maze and black maze were used each only once 
 with the short-term group by the amount-limit method. The average 
 time taken for the combined maze was 294 seconds, A.D. 13; the 
 touches were 2, 3, 5, 6, 12, and 13. The average time taken for the 
 black maze was 202 seconds; the touches were 0, 0, 0, 1, 2, 2,-3.- 
 
 The spiral maze was designed to provide another regular move- 
 ment and one more natural perhaps, than the straight. 
 
 Endeavors were made to practise this keeping the touches at 
 zero, and it was also hoped to practise with and without turning the 
 paper, with wrist and with free-arm movements, beginning from 
 the outside and from the center ; but after a few trials this hope was 
 given up, as all the subjects complained so much of eye-strain in- 
 volved, and the unpleasant after images. 
 
 The average times taken in successive trials were 360, 360, 298, 
 and 316 seconds. The average number of touches was in the first 
 trial 2.3; in the second 2.8; in the third 2.4; in the last 2.0. The 
 time taken would alone show how tiring to the eyes this might be, 
 staring at a heavy black spiral for over five minutes, and following 
 the pencil point round dizzyingly. The number of touches was very 
 low all through with one glaring exception when one subject de- 
 creased her time from 475 to 288 seconds and increased her touches 
 from 2 to 13. In 27 records there were 6 of zero touches, 5 of 
 1, and 6 of 2. 
 
 Of the tests tried none are injuriously susceptible to adapta- 
 tion to the task and practise. The straight maze is the easiest to 
 score. The spiral is too much a test of ability to stand eye-strain. 
 It would also be the easiest to use if the rate of the subjects was to 
 be controlled so as to compare individuals in accuracy alone. 
 
 B. Relative Value of these Motor Tests 
 
 The data serviceable for correlation are given in Table XXXI. 
 Having two records for each test, one of amount done, the other of 
 number of touches in the case of a time-limit test — one of time taken, 
 
 
 
 
 TABLE 
 
 XXXI 
 
 
 
 
 
 Subject 
 Bu 
 
 Curved Maze 
 Av. of 3 Trials 
 Time Touches 
 ... 142 1.3 
 
 Straight 
 
 5 Lines 
 Time Touches 
 145 3 
 
 Black 
 
 1 Trial 
 Time Touches 
 207 2 
 
 Spiral 
 Av. of 4 Trials 
 Time Touches 
 341 3.0 
 
 Gr 
 
 ... 136 
 
 3.7 
 
 147 
 
 4 
 
 224 
 
 
 
 315 
 
 3.5 
 
 J 
 
 ... 177 
 
 3.7 
 
 146 
 
 1 
 
 227 
 
 1 
 
 310 
 
 2.0 
 
 L 
 
 ... 182 
 
 1.0 
 
 112 
 
 
 
 225 
 
 
 
 359 
 
 .5 
 
 M 
 
 ... 147 
 
 .7 
 
 128 
 
 2 
 
 195 
 
 
 
 324 
 
 1.0 
 
 Ba 
 
 ... 126 
 
 
 
 125 
 
 4 
 
 154 
 
 3 
 
 397 
 
 2.3 
 
 Bf 
 
 ... 128 
 
 
 
 119 
 
 2 
 
 175 
 
 
 
 302 
 
 1.3 
 
90 STUDY OF TESTS FOE INDIVIDUAL DIFFEEENCES 
 
 one of number of touches in the case of completing the maze — the 
 resulting score must be arbitrarily determined, if a single measure 
 for efficiency is to be used for correlations. 
 
 As a fairly just method 5 seconds per touch has been added. 
 
 The Pearson coefficients are then, 
 
 TABLE XXXII 
 
 Average of all four tests ap- 
 proximately equal weight 
 in determining the average 
 being given to each. 
 
 Curved maze 60 
 
 Straight maze 49 
 
 Black maze 76 
 
 Spiral maze 29 
 
 The tests of rate of putting dots in the squares and of hitting 
 the dots showed little or no correlation with each other or with these 
 maze tests. 
 
 In estimating the relative precision of these tests of motor con- 
 trol two methods have been used. First, each individual's several 
 trials have been expressed as deviations from the probable result, in 
 view of the practise effect which he would have shown apart from 
 other variations than those due to the general tendency to improve 
 with practise. This is the result hitherto employed. Second, each 
 individual 's several trials have been expressed as deviations from the 
 average score of all the group on that day, and then the average 
 deviation of these deviations has been computed. 
 
 The following will illustrate the second method. The five suc- 
 cessive trials with the straight maze, gave, as average times for the 
 seven subjects, 29.3, 27.3, 27.9, 24.1, and 23.5. L, whose times were 
 30, 22, 25, 18, and 17 deviated by + .7, — 5.3, — 2.9, — 9.9, and 
 — 7.1. The deviations of these latter from their central tendency 
 ( — 4.9) were 5.6, .4, 2.0, 5.0, and 2.2, averaging over three seconds, 
 or 13 per cent, of L's average time. 
 
 With the first method in the case of the short-term group addi- 
 tions were made to the time to compensate for the touches. With the 
 second, no account was kept of touches. The results are given in per 
 cents of the time taken. The probable average divergences of the 
 score in one record from the individual's true ability are for the 
 curved, spiral, and straight mazes in order 10, 6, and 6 per cent, by 
 the first method, and 7, 9, and 9 by the second. Early trials of the 
 curved maze with the three long-term subjects showed by the first 
 method a corresponding figure of 7.3. Remembering the relative 
 lengths of the time required it will be seen that the straight maze has 
 a great advantage over the curved maze and a still greater advantage 
 over the spiral. 
 
 Comparing all five maze tests as to the time taken to complete 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 91 
 
 with no touches, it is found that the curved and the straight take 
 about equal time, 156 and 155 seconds respectively, the black takes 
 somewhat longer — 199 seconds, the combined 327 or more, and the 
 spiral longest of all, 364 seconds. From the point of view of discom- 
 fort the spiral and the black are hardest on the eyes, and even -the 
 combined becomes somewhat dazzling when over five minutes is spent 
 following its windings. For a short, convenient test either the 
 curved then, or the straight maze might be used. This last has, as 
 before mentioned, advantages of regularity of movement and ease of 
 measurement, but to offset this, it may be suggestive of jerky, dis- 
 crete movements by its very angularity ; also the units are very small. 
 From all these indications the choice would lie between the 
 straight for its convenience and precision, the black and the curved 
 for their higher correlation. Of these two the first has also some dis- 
 advantages, already mentioned, which the others have not, and since 
 the black is somewhat trying to the eyes and takes longer, the choice 
 would rest upon the curved maze as a suitable and convenient second 
 motor test. It would probably keep its present advantages and gain 
 others if arranged in a series of straight lines each repeating some 
 simple series of curves. The spiral maze has no merits. 
 
 7. Miscellaneous Tests 
 A. Descriptive 
 
 Six of the short-term group spent some time practising seven 
 other tests that are usually given the freshmen, viz: perception of 
 force of movement, with the monochord, the aesthesiometer and the 
 algometer, all of which test perception in some form; each also 
 practised 40 to 80 times with reaction time, 10 to 15 times with the 
 dynamometer and 5 times with the spring ergometer, all three tests 
 of movement in various ways. This work was done not so much to 
 find out anything about each test when practised as to get a basis for 
 intercorrelations when there was more than one trial of each — which 
 is all the freshmen take — and to get a basis of comparison with some 
 of the other tests already described. 
 
 With some few tests records of long practise were also available 
 from two subjects who were making some cross-education experi- 
 ments. 
 
 Perception of Force of Movement. — This is as often considered 
 a test for perception of weight, or perception of distance. As de- 
 scribed by Wissler 46a the test is as follows : ' ' the lift is vertical and 
 the dynamometer gives a pressure of 1 kg. to 10 cm. A mechanical 
 
 46a Psy. Rev. Mon. Suppl., No. 16, 1901. 
 
92 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 stop is provided at a pressure of 1 kg. to give the student his standard. 
 In making the test he is told to lift the handle to the stop three times 
 and then make ten (more recently five) attempts to lift it to the same 
 height after the operator has removed the stop. Each lift is to be 
 made in about 2 sec, with equal pauses between. A graphic record 
 of the lifts is taken on a kymograph." The errors are afterwards 
 recorded in cm. The men make an average error of 1.44 cm., the 
 women of 1.8 cm. 
 
 The apparatus has been criticised on the ground that it is sure to 
 induce a positive constant error because of the impact necessary in 
 the first three trials while getting the standard. Even with directions 
 to the Barnard students to be very careful in the first three trials, 
 this positive error persists; and after even 75 trials with some of 
 the short term group it was not overcome, though the subjects had 
 the benefit of seeing their records after every 15 trials. 
 
 In tabulating the results only the average error was considered. 
 Six of the short-term group and one member of the original long- 
 term group made from 9 to 15 groups of 5 trials, and the two other 
 extra subjects made 36 such groups of trials each. 
 
 TABLE XXXIII 
 Errors in cm. Made in Perception op Force op Movement 
 
 Av. Error No of Groups 
 
 Subject First Total Last of Trials 
 
 Ba 1.06 1.70 .88 13 
 
 Bf 1.52 .85 1.22 13 
 
 Bu 2.12 1.29 .52 12 
 
 J 1.74 .74 .22 15 
 
 L 32 .97 .44 9 
 
 M 80 .64 .20 10 
 
 N 74 .34 .46 10 
 
 E 1.54 .65 .40 36 
 
 Wy 42 .67 .68 36 
 
 From the above table it will be seen that there is a certain amount 
 of practise since the error is reduced in all cases except two. That 
 improvement with practise is slow and irregular may be seen from 
 the single records and even from the averages of the seven subjects 
 for each successive group of five trials, up to ten groups, which were : 
 
 123456789 10 
 
 1.21 1.06 .93 .92 1.28 .73 1.24 1.04 .98 .76 
 
 The record is better than the freshmen records. 
 
 It might be better to require the subject to make a given number 
 of movements of approximately the force shown him with the stop, 
 each as nearly as possible equal in force to the one just made> and 
 
EXPEBIMENTAL WORK WITH SEVERAL GEOUPS OF TESTS 93 
 
 to use the successive differences as the measure of his efficiency in the 
 test. 
 
 With the monocliord, the freshmen are tested for perception of 
 pitch as follows : The instrument is tuned so that F below middle C 
 is given when the bridge is at 75 cm. The tone F is given twice~at-an 
 interval of about 2 seconds while the subject's back is turned. The 
 bridge is then shifted and the subject told to find the tone given. 
 The position is recorded. Then the original tone is given as before, 
 and the bridge shifted to the place where it was left by the subject 
 in his first trial ; he is told this, and again required to find the tone. 
 The position is recorded. Also, before the test is begun, the subject is 
 shown how to use the instrument. 
 
 In general, if a subject is diffident, or slow in moving the bridge, 
 or by chance tries at first tones a long way from the standard, he 
 rapidly gets confused and forgets the original tone. On the other 
 hand, a very good record at the first trial is followed frequently by 
 a very poor one at the second, showing that in addition to memory 
 and celerity in moving the bridge, something is due, with poor sub- 
 jects, to chance. This seems to be a test of memory of pitch and of 
 general intelligence in using the instrument as much as of perception 
 of pitch. 
 
 Among the men 10 per cent, make an error of less than one tenth 
 of a tone, 53 per cent, of one tenth to one tone, and 37 per cent. 
 an error of more than one tone. For the women the corresponding 
 percentages are 17 per cent., 63 per cent., 20 per cent. 
 
 TABLE XXXIV 
 
 Accuracy in Placing a Bridge on the Monochord so as to Produce a Tone 
 of the Same Pitch as a Eemembered Tone; in Millimeters 
 
 Av. Error Av. Error on 75 
 
 Subject in mm. A.D. Position 
 
 Ba 37.2 26.0 24.6 
 
 Bf 10.7 6.0 7.8 
 
 Bu 7.2 5.0 4.2 
 
 J 31.8 29.7 47.5 
 
 L 9.1 5.0 10.5 
 
 M 24.4 17.0 36.8 
 
 Average 20.1 
 
 Average of successive records on 75 cm. 12 20.8 21 36 31 15 
 
 With this group of six subjects, after the preliminary trials, 
 eighteen to twenty further trials were given on different days, using 
 ten other standards ranging from 58 cm. to 93 cm. and also the 
 original standard 75 on four more occasions. At their last trial 
 they were asked to move the bridge till the tones on each side of it 
 were of the same pitch, thus eliminating the memory factor. This 
 
94 STUDY OF TESTS FOE INDIVIDUAL DIFFEEENCES 
 
 was of course done without looking at the instrument, though even 
 so, only two subjects realized that the bridge would have to be in 
 the exact middle. In this last trial the greatest error made by any 
 one was a difference of 3 mm., whereas, as is seen in the table above, 
 only one subject was distinctly good at the test given in the usual 
 way. 
 
 The variability from one trial to the next, particularly in the 
 ease of those with poor records, completely disguises any practise 
 effect, and emphasizes the need of more than one trial at the orig- 
 inal test. 
 
 For sensation areas, "the points of the cesthesiometer are 2 cm. 
 apart and the instrument is applied longitudinally to the back of the 
 left hand between the bones of the second and third fingers. Five 
 tests are made, the student being touched with one or two points in 
 the order, two, two, one, one, two, and being required to decide in 
 each case whether he was touched with one or with two points. ' ' Of 
 the men, 63 per cent, are correct four or five times, of the women 
 52 per cent. 
 
 With six subjects the right and left hands were used alternately 
 with the above series of touches twice each day for three days, 
 twelve tests in all. The total average error for the E. hand was 
 40.5 per cent., for the L. hand 40.6 per cent., or practically no dif- 
 ference. As this means that they were correct only three times out 
 of five on the average with either hand, they were rather below the 
 Barnard standard. There was no discernible improvement with 
 practise. 
 
 The algometer used has a pressing surface 1 cm. in diameter 
 which is made of rubber. It is applied with gradually increasing 
 pressure till the student signals that it is felt as disagreeable. 
 Usually there is some little difficulty in making students understand 
 just what is wanted. Some are nervous and afraid of receiving 
 electric shocks, others consider it a test of endurance, particularly 
 if it is given later in the series than the ergometer. With suggest- 
 ible subjects too the judgment is apt to be based on the rate at which 
 increasing pressure is applied. At the second trial with either hand 
 when an equivalent time has passed the student will frequently 
 signal "stop" though the pressure is only from a half to two thirds 
 of what it was at the first trial. 
 
 The averages for the men are : E. hand 5.9 kg. ; L. hand 5.6 kg. ; 
 for the women, 3.8 kg. and 4.3 kg. respectively. 
 
 The short-term group made eight trials with each hand on dif- 
 ferent days. Two subjects showed considerable difference from the 
 first to the last trials, one changing from 7.25 kg. to 3.5 kg., the 
 
EXPERIMENTAL WORK WITH SEVERAL GROUPS OF TESTS 95 
 
 other from 4.7 kg. to 2.5 kg. With the other four there was an 
 average reduction of only .5 kg. The averages for the whole series 
 of trials were : R. hand, 3.7 kg., L. hand 3.4 kg. The averages for 
 the first four successive trials (both hands together) were 4.7, 3.9, 
 4.6, 3.7. There would thus be no very great advantage in making 
 a first trial merely for adaptation to the test and using the second: and 
 later trials as the record. The test doubtless measures an individ- 
 ual's notion of the meaning of "painful" as well as his threshold for 
 pain as he defines it. Even so it is a significant test ; the correlation 
 between the first eight and the last eight trials of the same individual 
 is close. 
 
 In reaction-time the freshmen are tested five times in succession, 
 with the Hipp chronoscope. The average of the five tests for the 
 men is .159 second, for the women, .186 second. 
 
 The short-term group and the two extra subjects made from 40 
 to 75 trials each. Up to 30 trials, the average from each group of 
 five was recorded, as well as each separate trial, after that the aver- 
 age from each group of three trials only. There is apparently a 
 considerable effect from adaptation to the form of the test. The 
 average times for the eight subjects in the first six successive 5-trial 
 groups run 155, 158, 139, 133, 129, 130.5. This is also disturbing 
 since the relative rates assigned to individuals from the first ten 
 trials do not correspond at all perfectly to those assigned from say 
 the next twenty trials. In these eight subjects the deviations were 
 as follows: 
 
 TABLE XXXV 
 
 Deviation of the Individual's Average Keaction-time prom the Average of 
 the Group's in Thousandths of a Second 
 
 Subject First 10 Trials Next 20 Trials 
 
 Ba +46.5 +20 
 
 Bf + 5.5 +10 
 
 Bu + 5 — 0.5 
 
 J —12.5 —11 
 
 L —16.5 —11 
 
 M — 1 — 6.5 
 
 R —10 + 2 
 
 Wy —17 —12 
 
 These give a correlation of less than .09. The records of the first 
 reactions correlate with those of the twenty from the 11th to the 
 30th by less than .07. It would seem worth while to take 15 re- 
 actions, discarding the first five. 
 
 With the oval dynamometer the freshmen make two trials with 
 each hand in the order R. L. ; L. R. The average strength of grip 
 
A.D. 
 3 
 
 Av. 
 
 19.8 
 
 19.6 
 
 L. 
 
 A.D. 
 1.8 
 
 2.2 
 
 14.8 
 
 
 3.8 
 
 96 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 found is for men, R. hand 36.3 kg. ; L. hand 33.5 kg. ; for the women, 
 R. hand 25.8 kg. ; L. hand 23.6 kg. 
 
 The short-term group made, on different days, from nine to six- 
 teen trials, but this series also was not long enough to develop notice- 
 able practise, with one possible exception. Their averages were as 
 follows : 
 
 Av. 
 First 21.8 
 
 Average 21.5 
 
 Last 22.4 
 
 In this test a good deal of interest has attached to the question of 
 whether the maximum strength is attained at the first or at the 
 second trial, it being claimed that since a larger percentage of 
 women reach their maximum at first than do men, and that the left 
 or weaker hand in men is more apt to reach its maximum first than 
 the stronger hand, that therefore to do so is a sign of weakness. 
 However this condition goes with all degrees of strength of grip 
 among the freshmen; and experience with repeated sets of trials 
 with even this small group indicates that an individual may vary 
 very much in the relationship of the first two trials. The following 
 table illustrates this: 
 
 Equal 
 R. L. 
 
 2 
 
 1 2 
 
 2 
 
 1 
 
 1 1 
 1 
 4 7 
 
 Too much must not then be argued from the comparison of only 
 one set of trials. According to these records a single trial is subject 
 to an average divergence from an individual's true ability of 9.5 
 per cent. The difference between two single trials would then be 
 subject to an average divergence from the true difference of 
 \/9.5 2 + 9.5* or 13.4 per cent. 
 
 Cattell's spring ergometer is used for a test of fatigue with the 
 freshmen. The student is shown how to work the instrument with 
 particular attention to the use of only the end of the first finger on 
 the top of the piston. He is instructed to press the piston down as 
 far as possible fifty times without stopping. A rhythm of about 
 
 "Rfl 
 
 
 Gre 
 the 
 R. 
 . . 2 
 
 TABLE 
 
 ater 
 first 
 
 L. 
 
 2 
 
 2 
 2 
 3 
 2 
 2 
 13 
 
 XXXVI 
 
 Greater 
 
 the second 
 
 R. L. 
 
 1 3 
 
 Bf 
 
 
 .. 4 
 
 1 
 
 Bn 
 
 
 . . 4 
 
 1 2 
 
 J. . 
 
 T, 
 
 
 .. 4 
 .. 2 
 
 1 2 
 
 2 2 
 
 M. 
 
 Total . . 
 
 .. 3 
 .. 19 
 
 1 1 
 6 11 
 
EXPEBIMENTAL WOBK WITH SEVEBAL GBOUPS OF TESTS 97 
 
 one a second is set by counting aloud at the outset. The reading on 
 the dial for each ten pressures is recorded. 
 
 The men's average for the total amount of work done in the 50 
 pressures is 284.3 kg., the women's 172.9 kg.; the degrees of fatigue 
 are 65 per cent, and 63 per cent, respectively. 
 
 The short-term group made five trials with this on different days. 
 Their average amount of work was 267 kg., considerably nearer the 
 men's than the women's average among the freshmen. There was 
 the reverse of a practise effect from trial to trial, the average of the 
 last was 254 kg. The percentage of fatigue likewise increased. 
 With extended practise by the two extra subjects there was a similar 
 falling off for the first eight days; then one of them reached and 
 maintained her original level, and the other reached it and during 
 the last seven days of the twenty- two days' practise, went far beyond 
 it. As the average amount of work done for the first 10 pressures 
 of the series varied scarcely at all, however, what practise effect was 
 present was due to the increased power of endurance. The data 
 for the comparison of these tests were scarcely reliable enough to 
 warrant computing correlations by the Pearson coefficient. In gen- 
 eral there seemed to be correlation between reaction time and speed 
 of perception, and to be a slightly closer relation in speed in all the 
 tests than in accuracy. 
 
 A summary of the results found in Section II. will be deferred 
 till the end of the study. 
 
Ill 
 
 CHANGES WITH PRACTISE 
 1. Methods of Measuring such Changes 
 
 Before taking up the work of individual differences and the 
 practise curve, it would be well to take up some of the difficulties of 
 interpretation due to the method of constructing such curves. Dif- 
 ferent units may be taken as the basis, the starting-point may be ob- 
 scured by the use of percentile values only, and units may be dif- 
 ferently equated, perhaps distorted, in different parts of the curve. 
 
 First as to the kind of units used. 
 
 Curves may be constructed in terms of decrease in error (a time 
 or amount-limit test), decrease in time (amount-limit test), or in- 
 crease in amount (time-limit test). Or, whether time-limit or 
 amount-limit test, the scores may be reduced to the hundredths of a 
 second required to perform a definite minimum of work such as 
 adding two figures, cancelling one letter, etc. Bair, in his "Practise 
 Curve," 47 used units both of errors made after a given number of 
 practises, and of number of trials necessary to eliminate all errors. 
 His curves then slope down from left to right. Bryan and Harter 48 
 in their study of the acquisition of telegraphy used the number of 
 letters tapped per minute. Swift 49 in his experiments with the 
 typewriter used the number of words written during an hour, 
 smoothing the curve by averaging each successive three scores. In 
 later similar work undertaken with Schuyler, 50 two units were used, 
 one of strokes made on the typewriter, one of errors made. His 
 curves then — for no tables are given — show one a rise, the other a 
 slight drop. Coover and Angell 51 in making tests on the vexed ques- 
 tion of the general practise effect of special exercise, used variously 
 the number of right judgments before and after training, the de- 
 crease in time in 100 reactions, and the similar decrease in errors. 
 Where practise has meant a long period of exercise taken regularly 
 on successive days, the unit may be the average deviation of each 
 
 47 Hon. Suppl. to Psych. Bev., 1902. 
 
 48 Psych. Bev., 4, 1897, and 6, 1899. 
 
 49 Psych. Bull., 1, 1904. 
 
 50 Psych. Bull., 4, 1907. 
 
 11 Am. J. Psych., 18, 1907. 
 
 98 
 
CHANGES WITH PRACTISE 99 
 
 day's performances, giving a downward sloping curve for any one 
 individual. 
 
 So long as only one individual's curve is being considered, or 
 only the mean curve, the use of such varied units presents little 
 difficulty; but when comparisons are to be made of the curves of 
 learning whether of different subjects in the same test, or those of 
 the same subject in different tests, it becomes important to know 
 whether a different choice of units may show the same performance 
 in two different ways, and whether the units are alike all through 
 the curve. Otherwise, the questions "Does practise increase or 
 decrease differences?" and "Who profit most by practise, those 
 whose initial record is best or poorest V may receive quite differ- 
 ent answers according to the varied statistical treatment of identical 
 facts. 
 
 There is considerable divergence of custom. One method has been 
 to keep all scores in gross amounts, basing conclusions directly on 
 them. Examples of this would be Swift's and Schuyler's work 
 already referred to, and Smythe Johnson's experiments on motor 
 education. 52 Let us call this the gross method. 
 
 Another method is to turn each score into percentile values of 
 the initial record, or perhaps of the maximum reached before 
 fatigue sets in. Examples of this are Gilbert's work on develop- 
 ment of school-children, 53 Oehrn's on the work-curve of 10 sub- 
 jects, 54 Coover and Angell as already referred to, and Wells in 
 reports before the New York Branch of the American Psychological 
 Association. Let us call this the percentile method. 
 
 Another way of expressing percentile values used by Smythe 
 Johnson, 55 and modified by him from Amberg 56 is as follows: The 
 difference between the first and second scores, first and third, and 
 so on, is taken, and the sum of gains so found averaged and ex- 
 pressed in percentage of the first score. This process is repeated 
 with the second score used as basis, again with the third, and so on 
 through the series. Finally, all percentages are averaged. He says 
 "The significance of such percentages is that they give us a true 
 standard for the comparative influence of practise on different indi- 
 viduals" (page 61). That part of Amberg 's method which was 
 modified was, instead of averaging the ft — 1 different percentile 
 values, to weight each one, multiplying the first by n — 1, the second 
 by n — 2, etc., adding the products and dividing by (ft — 1) + 
 
 62 Yale Studies, 6, 1898. 
 
 63 Yale Studies, % 1894. 
 "Psych. Arbeiten, 1, 1896. 
 65 Yale Studies, 6, 1898. 
 
 69 Psych. Art., 1, 1896. 
 
100 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 (n — 2) + (n — 3) ■•••!, According to Amberg the resulting figure 
 "giebt mithin in moglichst einwandfreier Weise" the average per-' 
 centile increase by practise for the whole test. 
 
 Just to illustrate to what various conclusions one may be led 
 solely from differences in methods of portraying practise data, the 
 following tables and figures were made from five supposititious cases. 
 
 In 15 seconds, using as a score units of gross amount, suppose 
 that in seven trials, five subjects scored as follows : 
 
 TABLE XXXVII 
 Gross Amounts in Successive Trials 
 
 vidual 
 
 A 
 
 .. 5 
 
 6 
 
 7 
 
 8 
 
 9 
 
 10 
 
 Tc 
 
 10 
 
 >tal Increase 
 Units 
 
 5 
 
 B 
 
 .. 9 
 
 12 
 
 16 
 
 16 
 
 17 
 
 17 
 
 18 
 
 9 
 
 C 
 
 . . 10 
 
 10 
 
 10 
 
 12 
 
 13 
 
 14 
 
 15 
 
 5 
 
 D 
 
 .. 6 
 
 9 
 
 11 
 
 12 
 
 12 
 
 15 
 
 18 
 
 12 
 
 E 
 
 .. 5 
 
 7 
 
 9 
 
 10 
 
 12 
 
 14 
 
 15 
 
 10 
 
 Average . 
 A.D 
 
 .. 7.0 
 .. 2 
 
 8.8 
 
 10.7 
 
 11.6 
 
 12.6 
 
 14.0 
 
 15.2 
 2.25 
 
 8.2 
 
 It might be stated then that D improves most, and A and C 
 improve least. 
 
 This same table turned into units of time required to do one unit 
 of work, using hundredths of a second as the basis becomes : 
 
 TABLE XXXVIII 
 
 Gross Time for Work Unit in Successive Trials 
 
 Individual Hundredths of a Second Total Decrease 
 
 A 300 250 214 187 166 150 150 150 
 
 B 166 125 93 93 88 88 83 83 
 
 C 150 150 150 125 115 107 100 50 
 
 D 250 166 136 125 125 100 83 167 
 
 E 300 214 166 150 125 107 100 200 
 
 Average . . 233 181 155 136 124 110 103 130 
 
 A.D 60 19 
 
 It might be stated now that E improves most and C improves 
 least. 
 
 The two sets of curves as plotted* are not strictly comparable, 
 except that the same individuals are alike at the starting point in 
 each, and at the end. Otherwise, in answering the question whether 
 differences are increased or diminished by practise, the curves show 
 graphically that in the first case they apparently are increased, in 
 the second considerably decreased. The tables show the same thing, 
 
 * See Fig. 1. 
 
CHANGES WITH PBACTISE 
 
 tt>f 
 
 if the A.D. for the first trial is compared with the A.D. for the last, 
 in each table. In the first case there is a slightly greater difference 
 at the end, in the second, there is less. 
 
 The inference is then, that the change from the use of one kind 
 of unit to another in expression of one and the same performance 
 makes an appreciable change in its interpretation. 
 
 Fiftf. 
 
 Percentile Amoutvt (a) 
 
 Suppose however, as is sometimes the case, it were desirable to 
 compare one individual quantitatively with another, it could be 
 said from the first form of presentation that A and C improve 
 equally, and half as much as does E ; and that B improves three 
 quarters as much as D. In the second case it might be said that no 
 
102 STUDY OF TESTS FOB INDIVIDUAL DIFFEEENCES 
 
 two subjects improve equally though A and D are nearly equal; 
 that A improves three times as much as C, and three quarters as 
 much as E. 
 
 Evidently the value of such statements would be conditioned by 
 the nature of the test, for units near the physiological limit would 
 not be equal to those in the lower ranges. In a test such as mental 
 multiplication, the gain of the last few units may be far more diffi- 
 cult than that of the first many. In a cancellation test, the units may 
 possibly be of rather more equal difficulty, conditioned as they are 
 by factors of amount of eye movement necessary, and rejection of 
 wrong stimuli. In a feat such as juggling with balls, the first three 
 or four units may be harder to gain than fifteen such units later. 
 In other words, sharp slants or a plateau may be produced by the 
 nature of the variations in the real value of the units scored as equi- 
 valent, so that a "typical" curve for certain work may really exist. 
 
 If, as is more customary when individuals are to be compared, 
 the method of percentile values is used, the above table of gross 
 scores becomes: 
 
 TABLE XXXIX 
 Percentile Amounts Done 
 
 
 
 
 
 
 
 
 
 Total Gain 
 
 A .... 
 
 . .. 100 
 
 120 
 
 140 
 
 160 
 
 180 
 
 200 
 
 200 
 
 100 
 
 B .... 
 
 . .. 100 
 
 133 
 
 177 
 
 177 
 
 188 
 
 188 
 
 200 
 
 100' 
 
 C .... 
 
 . .. 100 
 
 100 
 
 100 
 
 120 
 
 130 
 
 140 
 
 150 
 
 50 
 
 D .... 
 
 . .. 100 
 
 150 
 
 183 
 
 200 
 
 200 
 
 250 
 
 300 
 
 200 
 
 E .... 
 
 ... 100 
 
 140 
 
 180 
 
 200 
 
 240 
 
 380 
 
 300 
 
 200 
 
 Av. ... 
 
 ... 100 
 
 129 
 
 156 
 
 171 
 
 188 
 
 212 
 
 230 
 
 130 
 
 A.D. . 
 
 
 
 15 
 
 
 
 
 
 56 
 
 
 From this it could be said that D and E improve most and C 
 least. 
 
 Again turning this table into units of time taken and expressed 
 in percentile values of the starting point it becomes: 
 
 TABLE XL 
 
 Percentile Decrease in Time Taken 
 
 Total Improve- 
 ment Per Cent. 
 
 A 100 83 71 62 55 50 50 50 
 
 B 100 76 56 56 53 53 50 50 
 
 C 100 100 100 83 76 71 66.6 33.3 
 
 D 100 66 54 50 50 40 33.3 66.6 
 
 E 100 71 55 50 42 36 33.3 66.6 
 
 Average 100 79 67 60 55 50 46 
 
 A.D 9.8 9.8 10.8 
 
CHANGES WITH PBACTISE 
 
 103 
 
 As from the preceding table, the conclusion would be that A and 
 B make equal gain, that so do D and E, and that C gains least; but 
 whereas before C's gain was half A's and B's, and one fourth D's 
 and E's, now it looks like one half that of D and E. Again, in each 
 table of percentile values the A.D. tends to increase, and evidently, 
 since in the curves the starting point is a common zero, they in- 
 evitably diverge later, and might be interpreted to mean that differ- 
 ences increase by practise. 
 
 In general then, this particular use of the method of percentiles 
 must confuse the issue unless each individual's starting point is 
 given, i. e., unless some statement of gross scores is also made. 
 
 Working over the original scores given above by both Smythe 
 Johnson's and Amberg's methods, the percentile increase is as 
 follows : 
 
 ABODE 
 
 Smythe Johnson 23 19 15 38 40 
 
 Amberg 32 29 19 53 56 
 
 Here the subjects keep the same relative position, though the 
 statements of how much more one improved than the other would 
 not be alike in the two cases. E improves most and C least is all 
 that can be said. 
 
 Just to put these varying interpretations into strong contrast 
 the following table has been prepared, giving for six ways of ex- 
 pressing the facts very varying answers to the question of relative 
 improvement. 
 
 TABLE XLI 
 
 Improvement of Seventh over First Practise Period in 
 
 Gross Gross Time Percentile Percentile Time 
 
 Amount per Amount per By Smythe 
 
 Individual Work Units Work Unit Work Units Work Unit Johnson By Amberg 
 
 A 5 150 100 50 23 32 
 
 B 9 83 100 50 19 29 
 
 C 5 50 50 33.3 15 19 
 
 D 12 167 200 66.6 38 53 
 
 E 10 200 200 66.6 40 56 
 
 Av 8.2 130 130 53.3 27 37.8 
 
 Gained most D E DE DE E E 
 
 Gained equally AC None D and E D and E None None 
 
 A and B A and B 
 
 Gained least AC C C C C C 
 
 Other statements E gains E gains E gains E gains E gains E gains 
 
 twice as four times twice as twice as between nearly 
 
 much as as much much as A much as C two and three 
 
 C or A as C and four three times 
 
 times as times as as much 
 
 much as C much as C as C 
 
104 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 One more such ease will be considered, but, for brevity, instead of 
 the similar four first tables and curves only the first and last scores 
 in gross amount of work done by four subjects in 10 units of time 
 is given, and a set of comparisons worked out as in the table just 
 preceding. 
 
 TAELE XLII 
 
 Improvement of Last over First Period of Practise in 
 
 Gross Gross Time Percentile Percentile By 
 
 Score Indi- Amount per Amount Time per Smythe 
 
 First Last vidual Work Units Work Unit Work Qnits Work Unit Johnson By Amberg 
 
 20 30 W 10 17 50 66 14.7 17.3 
 
 12 20 X 8 33 66 40 19.2 19.6 
 
 15 25 Y 10 26 66 40 23 27.7 
 
 16 24 Z 8 21 50 33 14.5 17.0 
 Average 9 24.2 58 45 17.8 20.4 
 
 Most gain WY X XY W Y Y 
 
 Equal gain W and Y None X and Y X and Y None None 
 
 XandZ W and Z (WandZ) (WandZ) 
 
 Least gain XZ WWZ Z Z Z 
 
 Other statements W gains W gains W gains W gains W gains W gains 
 
 more less equally twice slightly slightly 
 
 than Z than Z with Z as much more more 
 
 as Z than Z than Z 
 
 The conclusion would be that if one wishes to compare one indi- 
 vidual with another in rate of improvement, or one individual's per- 
 formances in two different kinds of tests, any statement based upon 
 a comparison of difference between the last score and the first score 
 will be seriously affected by the kind of units chosen, and may be 
 the more misleading the more definitely comparative they are made. 
 All of these methods alike ignore the actual starting and finishing 
 points which might be useful objective data, and may outrage the 
 sense of fairness by equating units taken from different points of the 
 scale. Thus it seems absurd to call A and C equal because each 
 gains 5 units, since they start and finish at such different points. 
 But to imagine that expressing A's performance as 100 per cent, 
 gain, C's as only 50 per cent, and therefore conclude that A does 
 twice as well as C, may be equally absurd, since it may be no nearer 
 the truth than was the first statement. There is no magic in per- 
 centile statements, except it be in blinding people to the actual 
 efficiency of a performance. 
 
 Then too, useful information may be obscured by stating merely 
 the amount of gain or loss whether in gross or percentile statements, 
 information which the full tables would have given and which is of 
 interest; such as, in the first example, that at the start C is much 
 better than E, but after seven periods of practise their performances 
 
CHANGES WITH PRACTISE 105 
 
 are equal, and that A after practise reaches only the point where C 
 started. Also from the second example, W who was best at the start 
 maintains his lead and is best at the finish ; X who was poorest at the 
 start was also poorest at the finish. Facts such as these are not 
 brought out by a mere statement of gain, nor by the percentile 
 tables and curves, though they would be by the gross amount tables 
 and curves; yet they are of value in application to everyday tasks 
 where objective norms must hold in speed and accuracy. 
 
 At this point examples may well be given of the treatment actu- 
 ally given to practise records — or fatigue. Gilbert 57 argues in favor 
 of the percentile measures thus: "To have expressed the fatigue 
 merely by the difference between the two rates of tapping would not 
 have expressed the truth : e. g., one child who tapped 19 and 15 for 
 the respective periods of 5 seconds lost a great deal more than 
 another who tapped 38 and 34 respectively : each lost 4 taps but the 
 first lost 21 per cent., the second only 11 per cent." His curve shows 
 the average per cent, of loss for each age, which means for eleven- 
 year olds, that children whose records were 30 to 24, 35 to 28, and 
 25 to 20 were considered equal. Later he says, "The average boy 
 . . . taps 29.4 times in Hye seconds, the average girl taps 26.9 times, 
 thus tapping 8.5 per cent, slower than boys. The average boy . . . 
 loses 18.1 per cent, by fatigue, the average girl loses 16.6. In other 
 words the boys lose 1.5 per cent, more by fatigue and yet tap 8.5 
 per cent, faster. This leaves the balance greatly in favor of boys." 
 Elsewhere, however, he does give a table of gross averages. 
 
 "Wells, in a report read before the New York Branch of the Amer- 
 ican Psychological Association in 1910 quoted some practise results 
 in two different tests without giving starting points, concluding that 
 as there was 71 per cent, gain in one test and 94 per cent, in the 
 other, there was greater gain in this than in the first. In a published 
 article on practise in free association 58 the curves in that test are 
 plotted on the gross decrease in units of time ; but when comparison 
 is made of susceptibility to practise in this test and in two other 
 tests, no gross figures for the others are given at all but only the 
 ratio of the mean of the nineteenth and twentieth days to the mean 
 of the first and second days practise, and the conclusions based on 
 those ratios. 
 
 Davis in his studies of cross-education 59 gives no gross gains, 
 only the percentage. The ratio is taken on the basis of the first trial 
 which is called 1 ; then the result is stated that the left hand gained 
 
 57 Yale Studies, 2, 1804. 
 
 58 Am. Jour, of Psych., 22, 1911. 
 69 Yale Studies, 8, 1900. 
 
106 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 more than the right. In earlier work 60 on the same problem he 
 quotes initial and final scores, gross and relative gains, and plots his 
 curves in gross errors. 
 
 Woodworth and Thorndike 61 carefully point out that one's in- 
 terpretation of what equal improvement or indeed proportionate 
 improvement means depends upon what is taken to be the starting 
 point, and they recommend the use of at least two measures of ac- 
 curacy. They use the gross error, also the ratio of errors after 
 practise to errors before practise, so that improving from 166 to 130 
 errors or 78 per cent., is considered about equal to improving from 
 302 to 232 errors, or 77 per cent. Later a statement occurs, "the 
 improvement in — is not equalled in the other functions." Seven 
 years later Thorndike gives this warning : 62 ' ' In estimating individ- 
 ual differences in amount of improvement . . . the ratios listed 
 must not be taken thoughtlessly at their face value. For a person 
 to change from 400 seconds per example to 200 is not necessarily the 
 same amount of improvement as for him or another to change from 
 200 seconds to 100 seconds. The second is probably an improvement 
 which f ewer individuals would be capable of, which the same individ- 
 ual would take longer to attain. ... To call the two equal as frac- 
 tions must not lead one to infer any thorough-going equality in 
 the facts which the fractions only partially represent. ... In fact 
 every measure of improvement by a gross difference or by a ratio 
 must be accompanied by a statement of the initial or final gross 
 actual ability. ' ' Such statements are given both in this and in later 
 work, 63 where no conclusions are drawn as to whether one individual 
 improved more or less, especially by how much more or less than 
 another. In presenting a curve which might be representative of 
 the general law of change, whether from the beginning of the test 
 to the end, or between two arbitrarily chosen points each within 
 every individual's compass, it is plotted according to the central 
 tendency of a series of points determined for each individual by the 
 formula 
 
 first score — score in question 
 first score — last score 
 
 But this average or mean curve is characterized as mongrel since 
 changes in the rate of improvement are due "to the action of radi- 
 cally different laws acting on different individuals according to the 
 
 80 Yale Studies, 6, 1898. 
 
 61 Psych. Bev., 8, 1901. 
 
 62 Am. Journ. of Psych., 19, 1908. 
 
 63 Am. Journ. of Psych., 21, 1910. 
 
CHANGES WITH PEACTISE 107 
 
 different physiological changes in them to which the improvement 
 is due." 
 
 It would seem then that the answer to the question "How much 
 relative improvement is there, or how much more does one individual 
 improve than another ?" can be given only for some arbitrarily 
 chosen definitions of "how much" and "how much more." The 
 nature of the work, the inevitable relativity of the starting points 
 and of the units, and one's preferred method of interpreting sta- 
 tistics will all modify such answer. What must be done is to keep 
 the first factor in mind, to present the second fully, and in more 
 than one way, to be wary and undogmatic as to the third, allowing 
 others to be the same. 
 
 There are other questions commonly asked, however, and 
 answered simply from examination of curves plotted according to 
 gross amounts, or somewhat variously by the use of certain formulae. 
 
 For example, it is of great importance in relation to measure- 
 ments of the relative parts played by heredity and environment in 
 producing the differences between individuals to determine whether, 
 and how far, different amounts of training account for individual 
 differences. The most usual and convenient measurement is of 
 whether and how far equal amounts of practise will reduce individ- 
 ual differences. To make this measurement one might : 
 
 1. Examine the average deviations from the average at the first 
 trial, and also after practise, and compare them directly. Then ac- 
 cording as one's units of measurement increase in amount or de- 
 crease in time or error, so will the deviations in all probability. 
 
 A.D. 
 
 2. Use the formula -r — ■ for both beginning and end, and make 
 
 comparisons. 
 
 A.D. 
 
 3. Use the preferred formula '__! and compare. 
 
 yAv. 
 
 4. Study the ratio of the range at both the beginning and at the 
 end, by finding in each case the ratio of best to worst, second best to 
 second worst and so on, and comparing each such ratio with the 
 corresponding ratio at the end. 
 
 Moreover any of these four methods could be applied not only 
 to the first and last scores, but to averages of the first few and the 
 last few, or the middle, or to each if necessary. Using all four meth- 
 ods on the two examples given, the figures would stand: 
 
108 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 
 
 
 TABLE XLIII 
 
 
 
 
 
 From Example 1 
 
 
 
 Gross Amount 
 First Last 
 
 Gross Time 
 First Last 
 
 Per Cent. Amount 
 First Second Last 
 
 Per Cent. Time 
 First Second Last 
 
 Average , 
 
 ,. 7 
 
 15.2 
 
 233 
 
 103 
 
 100 230 
 
 100 46.6 
 
 Gross A. D. 
 
 .. 2 
 
 2.24 
 
 60 
 
 18.6 
 
 15 56 
 
 9.8 10.5 
 
 A.D./Av 
 
 . . 29* 
 
 14* 
 
 25* 
 
 18* 
 
 11* 24* 
 
 12* 23* 
 
 A.D./V5v\ ... 
 
 . . 75* 
 
 57* 
 
 393* 
 
 183* 
 
 132* 369* 
 
 110* 154* 
 
 Worst and Best 
 
 . 2.00 
 
 1.80 
 
 .50 
 
 .55 
 
 1.50 2.00 
 
 .66 .50 
 
 Next Worst \ 
 Next Best J 
 
 1.80 
 
 1.20 
 
 .55 
 
 .83 
 
 1.16 1.50 
 
 .85 .66 
 
 
 or from twice 
 
 or from half 
 
 or from 1.50 
 
 or from .66 times 
 
 
 as good 
 
 to 1.80 
 
 as good to 
 
 as good at the 
 
 as good at the 
 
 
 times as good 
 
 .55 as 
 
 good 
 
 second trial to 
 
 second trial to 
 
 
 
 
 
 
 twice as good 
 
 only half as good 
 
 TABLE XLIV 
 From Example 2 
 
 Gross Amount Gross Time 
 
 First Last First Last 
 
 Average 15.7 24.7 65 41 
 
 A. D 2.2 2.7 9.2 4.5 
 
 A. D./Av 14* 10* 14* 11* 
 
 A.D./VSv 57* 54* 114* 70* 
 
 Worst and Best . . 1.66 1.50 .60 .6Q 
 Next Worst 1 
 
 NertBest }" 1M 1M M - 95 
 
 or from 1.66 from .60 as 
 
 times as good good to .66 
 
 to 1.50 times as good 
 as good 
 
 Per Cent. Amount 
 
 Per Cent 
 
 ;. Time 
 
 First Last 
 
 First 
 
 Last 
 
 100 139 
 
 100 
 
 63.3 
 
 1.9 
 
 
 
 3.3 
 
 13* 
 
 
 
 5* 
 
 161* 
 
 
 
 42* 
 
 1.16 
 
 
 
 .91 
 
 1.16 
 
 .91 
 
 From the tables in gross amounts it would be concluded that 
 individual differences tend to increase with practise; but the terms 
 in which the score is kept, and the method of comparing variations 
 make a great difference in the apparent amount or ratio of that de- 
 crease. The last method illustrated needs perhaps a word of caution. 
 In the second column — although the figures increase from .50 to .55 
 and .60 to .66, this means a decrease in differences of range, as the 
 interpretative readings added for both the first and second columns 
 show. Obviously, in the next two columns by the percentage in- 
 crease or decrease scoring, individual differences must be shown to 
 increase by practise, since all are made to start equal. The answers 
 to the questions obtained by such methods are then necessarily 
 absurd. 
 
 Therefore in using any of these four methods to examine the 
 variability one should again: (1) beware of being misled by the kind 
 
CHANGES WITH PBACTISE 109 
 
 of units used, both at the chosen starting point and at any point 
 in the practise series: (2) prefer gross to percentile measures of the 
 ability in question: (3) remember that only general tendencies are 
 given, not specific comparisons. 
 
 Even the fourth method would not make comparisons- atways 
 between the same pairs of individuals unless they happened to retain 
 their relative position all through the series, since it is engaged in 
 studying the range whoever may be at or near the extremes. But 
 this very point of individual comparisons is also of interest — whether 
 the one who is best at the start is also best after practise even though 
 the curve may have a less sudden slant than that of the worst at the 
 start, and whether those who start with a poor record will still be 
 poor, or the poorest at the end. The fourth method could be modi- 
 fied to answer that, but there are at least two common procedures. 
 One is to compare the position at the start with the total gross gain 
 or percentile gain or both; the other is to rank all individuals at 
 their first trial and at their last trial and compare the rankings. 
 
 By the former method, applied to example 1, between ability at 
 the start and gross gain there is correlation of — .32 ; between ability 
 at the start and percentile gain a correlation of — .55, from which 
 the inference would be that those who start well gain less than those 
 who start poorly. 
 
 By the latter method (used by Wimms 64 in his work with school- 
 boys in various mental tests) correlating by the "foot-rule" method, 
 R^.75. 
 
 Even this ranking method has been variously applied. Wimms, 
 for instance, also tabulates the percentage increase of each of his 
 subjects from the first to the last series of tests and ranks his sub- 
 jects accordingly. He then finds that the two ways of ranking, this, 
 and by numerical difference of absolute achievement in the last 
 series, do not agree. 
 
 Oehrn, 65 whom Wimms quotes, after stating that practise has 
 two effects, that of shortening the time for successive groups of trials, 
 and that of reducing each subject's variability in series of such 
 groups, ranked his subjects first in decrease in gross time taken, also 
 in percentage of reduction of variability, and found that the two 
 ways of ranking were not proportional. His correlations are based 
 on the ranking for the time taken. In his work too he introduces 
 another point as the basis of reckoning for the " work-curve, ' ' 
 namely the maximum performance of any individual, which he says 
 is a better standard than the starting-point because more constant 
 
 64 Brit. Journ. of Psych., 2, 1907. 
 
 65 Psych. Aroeiten, 1, 1896. 
 
HO STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 for each individual. This is rather a novel procedure, which though 
 it may have suited his conditions — continuous mental work for two 
 hours measured every quarter of an hour — would not suit work like 
 Bair's or Bryan and Harter's where the maximum performance was 
 emphatically not a constant. 
 
 In general, this ranking method tells precisely what a direct in- 
 spection of individual curves would do ; but since with large groups 
 it would be inconvenient and confusing to plot all the curves, tables 
 of ranks would be likely to give direct information about relative 
 improvement. If the question were "Are those who are best at the 
 start also best at the finish V 9 then ranks in initial and final tests 
 would be needed. If the question were "Do those who are best at 
 first improve most or those who are poorest?" then ranks by the 
 initial record and total increase would be needed. The absolute^gain 
 would be the more objective record perhaps, but here, at least, so 
 long as gross measures are available, a percentile or proportional 
 gain would not be misleading, and would often give just the practi- 
 cal information required. 
 
 Now this tedious elaboration has been based on simple and sup- 
 posititious records, solely to bring out possible discrepancies in 
 results and conclusions according to the use of one method rather 
 than another. Actual published results could be worked out in the 
 same way and contrasts drawn. That would, however, be beyond 
 the scope of the present investigation. 
 
 That the practise or rather the "work-curve" may be compli- 
 cated beyond easy and rapid inspection, Kraepelin has endeavored 
 to show 66 when he takes the record of one subject in continuous work 
 for two hours and at great length analyzes and plots curves for at 
 least seven factors: practise, fatigue, adaptation (or warming-up 
 period), inclination (or attitude towards work), initial and final 
 spurts, the desire to improve, and recovery by rest. He points out, 
 too, the difference between morning and evening workers, and the 
 effects of a recent meal or period of sleep. 
 
 Who would study individual differences as revealed in or af- 
 fected by practise has no easy task. 
 
 2. Results from a Special Series of Tests 
 So far in this study, the statistics of practise with the short or 
 long term groups have been confined to the starting point, average 
 and finishing point in gross amount for each group, with no com- 
 parison of individuals. Too few subjects made up the long-term 
 group to make any extended comparisons worth while, and the larger 
 
 60 Phil. Studien, 19, 1902. 
 
CHANGES WITE PBACTISE HI 
 
 group made too few trials with most tests to do more than indicate 
 the trend of individual curves at the beginning of practise. 
 
 Also, the results have been stated as if a typical curve for a test 
 or a group of tests could be determined. But it is a question whether 
 individuals will not differ so much in their improvement with- any 
 test as to make the average or mean curve unreliable, or rather 
 representative of nothing. It is also a question whether an individ- 
 ual's improvement in one test will not so parallel his improvement 
 in another as to make his curve typical of him rather than of the 
 kind of work. Or again, a "motor minded" individual might show 
 a different rate of practise in a motor test from one who is an abstract 
 thinker, and different also from his own improvement in another 
 field. In other words is "the practise curve" that of (1) the kind 
 of Work, or (2) of the general abilities of an individual, or (3) of 
 •special abilities of individuals? 
 
 In the hope of getting a little light on this problem, a further set 
 of tests was undertaken with a larger group of subjects, a long 
 period of practise, and with five tests of presumably very different 
 functions. 
 
 Supposing tests could be selected with which the subjects had 
 had no previous experience, then if all show slants and plateaus at 
 about the same level of practise judged by time or amount, the 
 curve would be typical of the kind of work. If there is greater re- 
 semblance between all curves from one individual than between one 
 individual and other, then the curve is typical of the kind of per- 
 son rather than of the kind of work. If any one subject's curves in, 
 say two motor, or two mental tests, resembled each other and were 
 unlike the mean curve, but in tests of some other function were like 
 some other individual's curves, then the curve is typical of special- 
 ized abilities in individuals. Lastly, if the mean curve for one in- 
 dividual in several tests is indistinguishable from the mean curve of 
 several individuals in one test there would be no evidence one way 
 or the other except that practise must produce the same results in 
 people whatever the work, and so must reduce differences between 
 people. 
 
 In order to discover which of the above conditions would prevail, 
 a group of subjects was put through a period of practise for twenty 
 days, excluding Sundays, in November and December of 1909. 
 
 The subjects, nine in number (the tenth did not continue suffi- 
 ciently long for any use to be made of her records) were all women 
 selected from among Teachers College students on the basis of their 
 needing financial help in working through college and so responding 
 to an appeal for subjects. From the group those were used who 
 
112 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 could give from one and one half hours a day at the beginning to 
 whatever time the tests took at the end of the period of practise, al- 
 ways at the same time of day. Four distinctly different nationalities 
 were represented, and five different departments in the college. 
 One was constitutionally delicate, two others showed signs of strain 
 and worry, the other six were in good health. One was over forty, 
 one over thirty, the others under twenty-five. Their college stand- 
 ing for the year 1909-10 was also examined, and they themselves 
 were carefully observed for general temperament as revealed dur- 
 ing the practise of one test. These facts are tabulated below : 
 
 Subject Nationality Department Health Relative College 
 
 Age Standing 
 
 C American Mathematics Delicate Young Good 
 
 E American Eng. & Dom. Sci. Tired Over 30 Very good 
 
 Go Eussian Jewess (German) Good Young Variable 
 
 H American English Good Young Poor 
 
 Jb German Domestic Art Good Young Fair 
 
 Nb American English Good Young Good 
 
 P American English Good Young Fair 
 
 Sch German German Good Over 40 Good 
 
 Sa Jewess Physical education Strained Young Fair to good 
 
 The tests selected were five in number: one for accuracy and 
 speed in movement, one for sensory discrimination, one for discrimi- 
 nation plus movements, one cancellation or perception test, one 
 purely mental test. The tests were explained orally to the sub- 
 jects and demonstrated, after which a manuscript book was given 
 to each with the directions for each test written out, and spaces pre- 
 pared for the required entries. The subjects were asked to select 
 whatever time of day was most convenient for them, and to work 
 always at that time through the whole number of days that the tests 
 lasted. Four of the tests were thus practised independently and 
 always in the same order; but for the discrimination of lifted 
 weights, which test needs of course an observer, each subject came 
 at an appointed hour. 
 
 For the first test the curved maze already described (see page 87) 
 was used. The directions were as follows: 
 
 "1. Place the maze so that the words begin here are at the left- 
 hand bottom corner. Do not turn the paper about during the test. 
 See that you have a sharp pencil. 
 
 "2. Note the time when you begin: (wait until the second hand 
 of your watch is at 60). 
 
 "3. Draw a line between the two lines of the maze without touch- 
 ing either, working as fast as you can. 
 
CHANGES WITH PRACTISE 113 
 
 1 ' 4. Note the exact time at which you finish, entering both times 
 in the proper columns opposite. 
 
 "5. Write your name on the blank, also the number of the ex- 
 periment. ' ' 
 
 The spaces ruled for entry were headed: 
 
 Date Time of Day Physical Condition Time at Start Time at Finish 
 
 In this third column they were directed to grade their felt con- 
 dition from A, excellent, to D, miserable. Thus a check of health 
 and weather could be applied to each subject's performances. 
 
 The " purely mental" test consisted of three sums in mental 
 multiplication of a three-place number by a three-place number. 
 The directions were: 
 
 "1. Beginning at the middle of this book you will find, under 
 day 1, 2, etc., three sums to be multiplied, each 3 figures by 3 figures. 
 
 ' ' 2. Cover up all but the one to be worked ; take note of the time. 
 
 1 ' 3. Multiply it mentally. Do not write anything at all till you 
 get the final answer, then write that down. 
 
 "4. Record for each sum in the appropriate column the time at 
 the beginning and the time at the end. Do not rest more than three 
 minutes between examples." 
 
 This wording might have been still more explicit, but the sub- 
 jects understood that "take note of the time" meant to write it 
 down, and also that the recording was to be for each sum, not after 
 all three were finished. The spaces for entry were headed: 
 
 Day 1. 
 
 First Sum 
 Time at 
 Start Finish 
 
 Second Sum 
 Time at 
 Start Finish 
 
 Third Sum 
 Time at 
 Start Finish 
 
 Day 2. 
 
 
 
 
 Etc. 
 
 
 
 
 For the sorting test, Dennison's colored cardboard counters 1% 
 inches in diameter, % of an inch thick were used, and for the "box," 
 the 5-cent size ice-cream carton. The directions were: 
 
 "1. In the little bag are 50 counters all of one color; in the box 
 are 50 counters of five different colors. Empty the varied ones into 
 some convenient place, and empty the bagful into the box. 
 
 ' ' 2. Distribute the 50 from the box at random into five piles. In 
 doing this use one hand only, and pick up only one at a time. Work 
 as rapidly as possible. Do this twice, just for practise in manipu- 
 lating the counters. Return them to the bag. 
 
 "3. Shuffle the 50 mixed colors well, and put them into the box. 
 Time yourself as in the other tests, and sort the 50 into five heaps 
 according to color, using the same care in handling as before. Re- 
 cord the time at the finish. 
 
114 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 "4. On the 1st, 10th, and 20th days, record also the time before 
 and after one distribution of the 50 all of one color. ' ' 
 
 Spaces were prepared for the entries of time at start and finish 
 each day as before, also for the three additional entries. 
 
 For the cancellation test, two copies of each of two back num- 
 bers of the Journal of Philosophy, Psychology, and Scientific Meth- 
 ods were provided for each subject. From these certain pages were 
 selected which were fairly evenly filled with print, in the hope of 
 getting about the same number of a's for each experiment, also about 
 the same number of lines for the eye to traverse. Previous work 
 with this test had shown how soon a blank is memorized, so that it 
 seemed advisable to use more ordinarily available reading matter. 
 Pages of a foreign text would have been still preferable. 
 
 The directions were: 
 
 1 ■ 1. Find the pages for the day : be ready to turn over quickly. 
 Note the time. 
 
 "2. Mark, on the pages designated every small print a you see, 
 going line by line over the two pages. To underline is the quickest 
 method. 
 
 "3. Note the time at start and finish as before." 
 
 The spaces for entry were headed as before, besides indicating for 
 each day exactly which pages were to be used. A second trial with 
 the same page was made only four times, and then it came at least 
 ten days later than the first trial, so that there was practically no 
 memory of the location of the a's. The average total number of a's 
 for the daily task was found to be 338, but unfortunately with a 
 large range of from 268 to 410, which complicated the latter calcu- 
 lations very much. 
 
 For the lifted weights test thirty weights ranging from 40 to 130 
 grams were prepared. These were unpainted wooden cylindrical 
 boxes containing lead or small shot to make up the required weight. 
 Six of these were used as standards of comparison, a 40, 55, 75, 90, 
 110, and 130 box, so labelled, and kept apart by themselves to the 
 side of the twenty-four test boxes. Of these, there were nineteen 
 different weights ranging by differences of 5 grams from 40 to 130 
 grams, and also six duplicates, one each of the 45, 60, 75, 90, 105, 
 and 120 gram weights. It will be noticed that of these duplicates 
 two are identical with two of the standards. By using six standards 
 scattered through the range, and by using steps of five grams it was 
 hoped to make the test easier and therefore likely to be completed 
 more rapidly than if merely one of the extremes had been used as 
 
CHANGES WITH PBACTISE 115 
 
 the sole standard or if very fine discriminations had been necessary 
 (see Thompson's work 67 ). 
 
 The twenty-four test weights were arranged in three rows of 
 eight, and daily rearranged in a different order with care to avoid 
 strong contrast effects and consequent probable illusions. Secret 
 marks on the side nearest the observer permitted immediate and 
 rapid checking up of the judgments made. For the first two days 
 preliminary experience was allowed in hefting the six standard 
 weights and one or two test weights. Thereafter the subjects began 
 immediately upon the test. 
 
 The first box in the nearest row was hefted with the fingers of 
 the right hand, then one of the standards, whichever would be 
 selected as probably the nearest, then the judgment was generally 
 made in terms of grams. However the subjects were free to try 
 another standard if the first was presumably not near the testbox 
 in weight and then to heft the testbox again. In this way emphasis 
 and help were given to making correct judgments. No fixed speed 
 was insisted on, but a check was kept on the total time taken daily 
 for the whole set of twenty-four judgments. Only on three occasions 
 were subjects hurried up, and then when they had exceeded 25 
 seconds in arriving at a judgment. Otherwise the aim was to leave 
 the subjects as free as possible. 
 
 Each subject came 16 times for this test, though as all did not 
 begin on the same day, any particular arrangement of the boxes 
 would not fall on say the fifth trial for everybody. After a certain 
 date too, each subject after having made a judgment was told what 
 the real weight was, in the hope of facilitating practise by this 
 means. Again, this additional means of training did not begin at 
 the same point in the series of 16 tests for each subject. In the 
 curves this point is indicated for each individual by a small cross. 
 
 In working up the results, judgments for weights below 60 g. 
 and over 105 g. were not used, in order to avoid the influence of the 
 * ' end error. ' ' The curves then are plotted from the average error in 
 14 judgments of 10 different weights from the middle of the series, 
 4 of which were duplicates and 2 of those duplicates identical with 
 2 of the standards. This leaves a total of 2,016 judgments instead 
 of 3,024. 
 
 The method of scoring was to enter immediately the errors in 
 grams, plus or minus. After the date on which the subjects were 
 told the real weights, the last 12 judgments of the 24 were recorded 
 in ink instead of pencil. In this way could be found (1) the average 
 error with each weight for each subject, (2) the constant error for 
 
 67 < « The Mental Traits of Sex. ' ' 
 
116 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 O ^ N IO O) O O M l> OO M W S H 
 
 o m rl 15 w w i> a! tjI 10 t> 
 
 w 
 
 
 N N W H oi CO CO H 
 r-( r-l ri ri 
 
 co h ^ oq « 10 eo mi a h iq h t» ^ iq 
 
 © N ri n o w i> w to h h ri © ' h w 
 
 I + I + + + + + I + + + | I + 
 
 {CO tH CO in CO rJJ OS r-J t-H lO OS <* CO CO CO tJH 
 ao'oiNoidooNNsri©'*^'*^ 
 cocMOs<McqcoT^co # ^Haqosb^aqcoinoq 
 in c> ri ri ri h ©' ri to s 06 w w ri n ri 
 ++++++++++++++++ 
 
 MSSHCOSOOOIMS^NIOSSH 
 
 fodinNoidto'oJoio'ffiio'sinin'td 
 
 i-i r-H iH 
 
 j « N rl lO Ifl M »J N H ft H M N S «5 O 
 
 id id cm* ri cm" " ri ih ri c4 tjh ri ' ■*' in 
 
 L + + + + I I + I I +++++++ 
 
 (D H N q © M l> UJ q h q w a to H M 
 
 roi h d in 06 h in n in i> in in i> d t> ^' 
 
 i-( r-i r-i t-I 
 
 inN©inscqcqi<icqi>oqoq©0)HT)i 
 <m" ri ri ri id os ni ®ci cm* <m" m* c6 cm" «h 
 
 ++++++++++++++++ 
 
 ^ 00 q w n s n i> h «) iq 01 01 o h m t> 
 f©dcoo6ddt>©ooso6i>ri©inin 
 
 |> 00 . J r* rt 
 
 v-"* o °]*HiqwHOsoq!qo5Wiqq co co «-j 
 
 fcxj 9 tj* r-i cm' * cm* i>" ri ri ©j" ' t> in o in in d 
 
 I L l + I i ++++++++ +++ 
 
 3 * 
 
 PQ 2 OJMTljHM©M©Oj©HNinoOinH 
 
 <3 5 f CX) d © © o> d Tji oi 06 ■* © 00 N © N* 06 
 
 O ^JcqcOOjOOHOiHrjMj»aHM©HM 
 
 O ri d N H w ri ©i W to TH N S <** r-i i-i 
 
 S L I +++++++++++++++ 
 
 (D H M OO 00 ffl S P5 &5 S r| N r-J 00 iH CO 
 OS* CM O* <M* r-i T** id "*' C* O* r-i 00* (O t> CD O 
 
 < 
 
 I 6 ,' 
 
 • r*j < CO iqcq©©NNNr||CO(M00l>(MH 
 
 ri O N IO 00 ^' O O K5 ri ri CO* r-i ' CO* r-i 
 
 L 7 1 1 ++++++ 1 + + + + + 
 
 N a m « s ^ q © <>j iq i| n s s q q 
 
 r ri r-i co* t^ id cd in "*" to" t^ cd od id id id id 
 ,0 J ■- H 
 
 ►^-(inr-IOSOOO CO rH OS OOSO r-i OOOOCOOO 
 
 CM* r-i CO* id CM* Tti (M* CO* 'H* CO* id CO CM* CM* "* CM* 
 
 + + + + + + + + I + + + + + + + 
 
 q^sctoocqoMqncMHN© 
 
 "^ f o co' id ri e<3* id -^i ri in id tj* ri* ri co* id ■* 
 
 *a \ r-i 
 « j 
 
 .£ ") »: co # rij >s in r}j in in co r-j cq oq os co <m 
 
 ■2 I id O * r-i ^} (M* rH* <M* (M* CO* (M* CO* CO* CO* Tti ri 
 
 »^+ I I + + + + + + + + + + + + 
 
 « HNCOMIintONOOOOHOlCO^lOtO 
 
 H r-l i-H rH r-l rH tH rH 
 
CHANGES WITH PBACTISE 117 
 
 each weight, (3) the average error daily, (4) the constant error 
 daily, (5) the improvement daily during the test. 
 
 Below is given the average error for each weight through the 
 whole period of practise: 
 
 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 "120 125 
 2.1 4.4 5.7 6.1 6.3 7.8 7.2 7.1 9.1 7.8 7.6 8.3 8.2 7.9 7.3 5.5 5.3 3.9 
 
 From this the influence of the "end error" is clearly visible, 
 though not so far into the series as it had been expected. The 
 weights of 75 and 90 grams do seem to show the benefit of both their 
 identity with the standards and the double practise they received; 
 the 60 grams perhaps shows the double practice benefit, but the same 
 can not be said of the other weight, the 105. 
 
 Table XLV gives for each subject for each day the average error 
 and the constant error for the set of 14 judgments. The scores in 
 italics show the first day on which additional help was given by being 
 told the real weight. 
 
 In general this shows a slow reduction in the average error for 
 each subject, a tendency to a positive constant error, a disturbance 
 in the constant error on the day of the change in method, and that 
 the greatest fluctuations occurred between the first and second trials. 
 
 Eight other individuals also took this test in this form once each. 
 For these control cases are here given also the constant error and the 
 average error for the set of 14 judgments. 
 
 
 
 TABLE XLVI 
 
 
 
 Ind. 
 
 Const. Error 
 
 Av. Error 
 
 Rank for 
 Accuracy. 
 1 = Least Error 1 = 
 
 Rank for 
 
 Time. 
 
 = Least Time 
 
 1 
 
 +4.3 
 
 6.4 
 
 4 
 
 2 
 
 2 
 
 + .7 
 
 17.2 
 
 8 
 
 8 
 
 3 
 
 +2.1 
 
 5.0 
 
 1 
 
 6 
 
 4 
 
 —1.4 
 
 10.7 
 
 7 
 
 1 
 
 5 
 
 — .3 
 
 9.7 
 
 6 
 
 5 
 
 6 
 
 —1.0 
 
 6.1 
 
 3 
 
 7 
 
 7 
 
 —1.8 
 
 5.4 
 
 2 
 
 4 
 
 8 
 
 +9.6 
 Average 
 
 9.6 
 8.8 
 
 5 
 
 3 
 
 Compared with the first trial by the practising group, 11.1, the 
 average record for these is somewhat better. The curves as plotted 
 for each of the group of nine subjects from their daily average error 
 are shown in Fig. 2. The dotted line shows the most probable 
 "smoothed" curve. The two individuals most unlike are Go. and Sa. 
 The latter had the benefit of knowledge of the correct weight longer 
 than did the others; she was also the slowest of the nine. Go. gave 
 
118 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 
 o 
 
 
 u 
 
 g 29 £ 2 
 
 00 vO 
 
CHANGES WITH PRACTISE 119 
 
 the impression of being very careless and indifferent, she took about 
 half the time that Sa. did. Nb., who took about three sevenths the 
 time Sa. did and was the quickest, has a curve more like Sa.'s than 
 has any one else. Taking the average of the first two trials and of 
 the last two (gross score), the gross gain, percentile gain, the- time 
 taken on the average, and ranking the nine subjects by each of these 
 scores, we get : 
 
 
 
 TABLE XLYII 
 
 
 
 E. ... 
 
 Rank at 
 
 Start. 
 
 (1 = Least 
 
 Error) 
 
 .... 3.5 
 
 Rank at 
 
 Finish. 
 
 (1 = Least 
 
 Error) 
 
 8 
 
 Rank for Av. 
 
 of Total Series 
 
 (1 = Least 
 
 Error) 
 
 8 
 
 Gross 
 
 Gain. 
 
 (1 = Most 
 
 Gain) 
 
 7 
 
 Percentile 
 Gain. 
 
 (1 = Most 
 Gain) 
 
 9 
 
 Time 
 
 Taken. 
 
 (1 = Least 
 
 Time) 
 
 7 
 
 H. .. 
 
 .... 5 
 
 1.5 
 
 6 
 
 3 
 
 2 
 
 4.5 
 
 Jb. .. 
 
 .... 7 
 
 7 
 
 3.5 
 
 4.5 
 
 6 
 
 2.5 
 
 Sch. . 
 
 .... 6 
 
 6 
 
 7 
 
 4.5 
 
 5 
 
 8 
 
 C. ... 
 
 .... 2 
 
 5 
 
 5 
 
 9 
 
 8 
 
 6 
 
 P. ... 
 
 .... 3.5 
 
 4 
 
 2 
 
 6 
 
 4 
 
 4.5 
 
 Go. .. 
 
 .... 9 
 
 9 
 
 9 
 
 1 
 
 3 
 
 2.5 
 
 Nb. .. 
 
 .... 8 
 
 1.5 
 
 3.5 
 
 2 
 
 1 
 
 1 
 
 Sa. .. 
 
 .... 1 
 
 3 
 
 1 
 
 8 
 
 7 
 
 9 
 
 from which the correlations by the method of rank differences are as 
 follows : 
 
 Position at start and at finish B = +.27 
 
 Position at start and average in the whole series +.45 
 
 Position at start and gross gain — .98 
 
 Position at start and percentile gain — .65 
 
 Average in whole series and time taken — .04 
 
 This means that, with these subjects at least, their performance at 
 the first two days' trial was relatively more like their average per- 
 formance than it was like their performance during the last two 
 days. Those who were poorer at the start made a greater relative 
 gain and a much greater gross gain than those who were better at the 
 start. Within the range of accuracy attained there was practically 
 no relationship to the speed of judgments. 
 
 For the control cases also the correlation of accuracy and speed 
 in this was — .07, very near the figure for the practising group, and 
 meaning again practically no relationship. 
 
 To notice the improvement if any during the daily test the aver- 
 age errors of the first twelve and the last twelve judgments of each 
 subject were compared. The twelve were of course carefully dis- 
 tributed over the whole range of weights. The errors are as follows : 
 
120 STUDY OF TESTS FOB INDIVIDUAL DIFFEBENCES 
 
 First 12 Last 12 
 
 E 6.7 6.8 
 
 H 5.1 5.5 
 
 Jb 5.4 5.9 
 
 Sch 5.5 5.1 
 
 C 5.1 5.3 
 
 P 5.1 5.4 
 
 Go 8.2 6.5 
 
 Nb 5.4 5.7 
 
 Sa 4.1 5.0 
 
 Average 5.6 5.7 
 
 There is no "warming up" effect discoverable from the first half 
 to the second half of the test daily. On the whole there is scarcely 
 any difference, though for some subjects there is a decided increase 
 in error, which in Sa. 's case may be due to fatigue, since she was the 
 slowest. 
 
 The scoring of the a's test was not so easy, because of unequal 
 numbers of a's in the daily tasks of two pages each. Instead of re- 
 taining the gross time taken to cover two pages, it seemed fairer to 
 make the following reduction: find the time that would have been 
 required (proportionately) to cover a space including 250 a's with 
 the same accuracy as was actually shown for the whole two pages, 
 i. e., with the same proportion of errors and omissions. This reduc- 
 tion is accomplished by use of the formula, 
 
 time taken 
 number marked 
 
 Thus, the score for a subject who in 420 seconds marked 286 a's is 
 
 4-90 
 
 2^X250 = 367. 
 
 This score is, essentially, the time for covering a given space, 
 and therefore grows smaller with increase in efficiency. 
 
 In the following table are given the daily scores for each indi- 
 vidual, and also the total number of a's in the day's task. The 
 curves as plotted from these scores are shown in Fig. 3. The great- 
 est difference is from the first to the second day's trial. 
 
 There are several curves fairly similar, Jb.'s and P.'s, for in- 
 stance, also H.'s and Go.'s, perhaps E.'s and Nb.'s. When smoothed 
 out, there are seven very similar, namely those of all except H. and 
 Go. The two most unlike are Go.'s and C.'s, the former irregular, 
 showing a poor record at the start and a rapid improvement, the 
 latter very smooth, with a good record at the start and gradual but 
 steady improvement. In percentile improvement the two were 
 nearly equal. 
 
CHANGES WITH PRACTISE 
 
 121 
 
 
 CJ 
 
 8 
 
 o "o o o o o 
 o o o o o o 
 
 ts ^ in 7 W (^ 
 
 <0 
 
 / 
 
 OOOOOOQO 
 
122 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 TABLE XLVIII 
 
 Score in A's Test 
 
 a's E. H. Jb. Sen. C. P. Go. Nb. Sa. Av. 
 Day Possible 
 
 1 299 642 882 550 298 463 542 933 611 508 603 
 
 2 314 461 892 596 374 367 480 558 517 498 527 
 
 3 299 506 737 514 330 371 398 457 372 428 457 
 
 4 285 447 508 581 392 341 418 425 476 426 446 
 
 5 345 402 502 497 413 314 366 390 418 414 413 
 
 6 374 595 520 418 352 326 391 312 416 298 403 
 
 7 355 395 496 408 442 287 257 273 348 519 381 
 
 8 368 388 458 347 282 296 302 219 299 362 328 
 
 9 365 307 421 381 293 258 293 273 342 386 328 
 
 10 318 365 517 360 373 285 341 334 310 417 367 
 
 11 409 270 409 314 323 285 329 232 324 382 318 
 
 12 333 331 484 347 209 280 383 281 335 420 341 
 
 13 327 334 481 340 210 266 343 349 285 412 335 
 
 14 268 345 611 365 251 278 390 413 242 356 361 
 
 15 315 306 520 340 277 252 343 350 315 365 341 
 
 16 360 299 453 315 262 229 312 303 275 327 308 
 
 17 334 304 465 325 170 267 297 326 294 360 312 
 
 18 313 349 493 413 209 248 334 402 ! 336 348 
 
 19 410 279 431 347 246 179 262 271 279 327 ,291 
 
 20 409 274 400 319 146 173 304 307 236 333 277 
 
 Average 338 380 534 354 293 288 354 370 352 394 
 
 As before, comparing the average of the first two days with the 
 average of the last two days, taking also the average for the whole 20 
 days, the gross gain, the percentile gain, and ranking the nine sub- 
 jects for each of these and also for speed and for accuracy, we get : 
 
 TABLE XLIX 
 
 Position At Average 
 
 at Start Finish Position Gross 
 
 l=Short- l=Short- l=Short- Gain 
 
 est Time est Time est Time l=Most 
 
 E 5 4 7 4 
 
 H 9 9 9 1 
 
 Jb 7 8 4.5 5 
 
 Sch 1 2 2 9 
 
 C 2 1 1 6 
 
 P 4 5 4.5 7 
 
 Go 8 6 6 2 
 
 Nb 6 3 3 3 
 
 Sa 3 7 8 8 
 
 The correlations by the method of rank differences are : 
 
 Position at start and at finish B = .72 
 
 Position at start and average position 58 
 
 Position at start and gross gain — .90 
 
 Position at start and percentile gain — .38 
 
 Speed and accuracy — .37 
 
 Per Cent. 
 
 Gain 
 
 l=Most 
 
 5 
 
 Speed 
 
 l=Least 
 
 Time 
 
 1 
 
 Accuracy 
 
 l=Fewest 
 
 Errors 
 
 7 
 
 4 
 
 9 
 
 4 
 
 7.5 
 
 7 
 
 9 
 
 7.5 
 
 3 
 
 6 
 
 2 
 
 2 
 
 5 
 
 6 
 
 6 
 
 1 
 
 1 
 
 4 
 
 8 
 
 3 
 
 5 
 
 3 
 
 9 
 
 8 
 
 2 
 
CHANGES WITH PEACTISE 
 
 123 
 
 Here the subjects kept their relative positions through the test fairly- 
 well. Those who were poorest at the start made a greater relative 
 gain than those who were better, and had almost a guarantee that 
 they would make a greater gross gain. The quicker ones are rather 
 less accurate than the slower ones. 
 
 In the sorting test one subject so misunderstood directions that 
 all her records had to be discarded; another so confused her first 
 nine entries that they too could not be used; a third showed a care- 
 lessness in entering whole minutes rather than seconds. Only seven 
 and a half complete records were therefore available of which one 
 is not so reliable as are the others. 
 
 The following table gives the daily scores, from which the curves 
 
 as plotted are shown in Fig. 4. The scores on the three occasions on 
 
 which the " control time for movements only" was recorded are 
 
 indicated on each curve by a cross. The missing curves are suggested 
 
 by a dotted line. 
 
 TABLE L 
 
 Seconds Kequired to Sort 60 Counters 
 
 E. Jb. Sch. C. P. Go. Nb. Sa. Average 
 
 Day 
 
 1 240 90 150 120 240 150 145 162 
 
 2 120 90 60 60 180 90 85 98 
 
 3 60 100 75 89 180 60 90 93 
 
 4 60 85 55 90 240 60 80 96 
 
 5 60 170 56 90 240 90 72 111 
 
 6 90 90 65 95 240 90 68 105 
 
 7 90 80 55 65 120 90 114 88 
 
 8 90 70 50 70 120 72 88 80 
 
 9 60 110 50 50 120 72 75 77 
 
 10 120 105 120 50 55 150 54 77 91 
 
 11 60 80 60 56 60 120 66 90 74 
 
 12 60 60 100 53 68 120 66 95 78 
 
 13 90 60 85 50 60 60 60 69 67 
 
 14 60 110 60 40 50 60 8-5 10O 71 
 
 15 45 50 90 45 60 60 60 69 80 
 
 16 60 50 115 45 55 60 50 60 62 
 
 17 60 108 60 45 60 60 60 62 64 
 
 18 60 60 90 45 61 60 60 68 63 
 
 19 60 60 60 50 60 60 65 66 60 
 
 20 60 60 90 43 50 60 50 65 60 
 
 Average ... 80 84 85 57 68 127 72.5 82 
 
 The curves most alike are those of P. and Nb., though when 
 smoothed out those of Sa. and C. are also similar. Those most unlike 
 are Go. — irregular and rapidly improving — and C, very regular 
 with almost all the improvement at the beginning. Since Go. 's scores 
 were so poorly kept, a better instance of dissimilarity might be C.'s 
 
124 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 4 
 
 iC QD X 
 
 / 
 
 8 I 
 
 MS 
 
 '* is c 
 
 c6 
 
CHANGES WITH PEACTISE 125 
 
 curve and Jb.'s, the latter showing great irregularity from day to 
 day and the reverse of improvement near the beginning. 
 
 Below are the rankings of the subjects according to position at 
 the start (average of two days), position at the finish, average posi- 
 tion, gross gain and percentile gain. 
 
 TABLE LI 
 
 E. ... 
 
 Position 
 at Start 
 l=Least Time 
 6 
 
 At Finish 
 
 l=Least Time 
 
 4.5 
 
 Average 
 
 l=Least Time 
 
 4 
 
 Gross Gain 
 
 l=Most Gain 
 
 2 
 
 Per Cent. Gain 
 
 l=Most Gain 
 
 2 
 
 Jb. . . 
 
 1.5 
 
 4.5 
 8 
 
 6 
 (7) 
 
 7 
 
 7 
 
 Sch. .. 
 
 
 ? 
 
 C. ... 
 
 3 
 
 1 
 
 1 
 
 4 
 
 3 
 
 P. 
 
 1.5 
 
 2 
 
 4.5 
 
 4.5 
 
 2 
 
 8 
 3 
 
 6 
 
 1 
 3 
 
 6 
 
 Go. .. 
 
 7 
 
 1 
 
 Nb. .. 
 
 5 
 
 4 
 
 Sa. .. 
 
 4 
 
 7 
 
 5 
 
 5 
 
 5 
 
 From these the correlations by the method of rank differences are : 
 
 Position at start and at finish B = .56 
 
 Position at start and average position 58 
 
 Position at start and gross gain — .92 
 
 Position at start and percentile gain — .86 
 
 Here there was more change in the relative position through the 
 test than in the marking a's. It should be noted, however, that the 
 "positions" were very close together at the end since nearly all got 
 down to about 60 seconds or slightly less in handling the 60 counters. 
 
 Again therefore, since all finish nearly alike, those who were 
 poorest at the beginning made the greatest relative and gross gain. 
 
 In the mental multiplication tests only digits from 3 to 8 were 
 used in the multiplicand, and from 2 to 7 in the multiplier. In ar- 
 ranging examples care was taken to have no two consecutive figures 
 alike in both multiplicand and multiplier — to minimize unnecessary 
 confusion. The subjects all dreaded this test at the outset, but after 
 two days' work with it they gained confidence in their ability. No 
 suggestion was given any of them as to using or discarding visual 
 or auditory imagery, nor as to devices for lessening the number of 
 figures to be remembered. But they were asked to note any change 
 in attitude or method that helped or hindered them. The following 
 notes are interesting. 
 
 E. after the second day decided that a pause between examples 
 was not worth while. For a time she visualized a series of dots as a 
 help in placing partial products. 
 
 H. found it better to do her adding as she went along rather than 
 to keep one partial product in mind while getting another. 
 
126 STUDY OF TESTS FOE INDIVIDUAL DIFFERENCES 
 
 Jb. discarded visualizing as it was a hindrance. She tried saying 
 the partial products aloud for awhile, finally took to adding two 
 partial products before finding the third. 
 
 P. also hit upon this method as early as the third day and kept 
 to it thereafter. 
 
 C. occasionally adopted a device, such as, with a multiplier like 
 625, dividing by 4 instead of multiplying by 25. This was seldom 
 possible however. Occasionally she noted that the answer seemed to 
 come automatically, in one process without consciously thinking 
 through the steps. " It opened out before me. " C. was specializing 
 in the mathematics department, so was probably better prepared 
 with devices and automatic calculations than any of the others. 
 
 In scoring this test, errors were penalized by adding on .2 of the 
 time taken for 1 error, .3 for 2 or 3 errors, .4 for 4 or 5 errors, and 
 .5 for 6 errors in the final answer. As it happens, subjects who are 
 usually accurate seem doubly penalized by this, since with them the 
 consciousness or suspicion of error lengthens their time in any case, 
 whereas with the habitually inaccurate an error more or less made no 
 appreciable difference in the time taken. 
 
 Records for each of the 60 examples were kept to see if any par- 
 ticular one was much more difficult or easy than the rest; but both 
 good and bad scores were made with almost every example, and none 
 could be singled out as specially difficult or easy. 
 
 In the table that follows the daily average score for each subject 
 is given, that is, the average score on three examples for 20 days. 
 The curves as plotted from them are in Fig. 5. 
 
 As each point on the curve represents an average rather than a 
 single trial, the curves may be considered partly smoothed already. 
 The two most regular and most alike are those of C. and Sa. ; the 
 most irregular is that of Sch. ; the most unlike any other is that of 
 H., though after the sixth day when her scores are within the range 
 of those of the other subjects, her curve is more regular, and not 
 unlike E.'s or Jb.'s. 
 
 The curves representing separately the factors of speed and ac- 
 curacy are shown by a continuous and a dashed line respectively in 
 Fig. 6. 
 
 From this it will be seen that there is very little if any improve- 
 ment in accuracy, but a good deal in speed. Also, of the most ac- 
 curate subjects, H. is the slowest, Nb., C, and Sa. are the quickest. 
 Also that there is more individual difference revealed in speed than 
 in accuracy, judging by the amount and regularity of improvement 
 in each. 
 
CHANGES WITH PEACTISE 
 
 127 
 
 td 
 
 a 3 s s s s a * c 
 
 ^j 
 
 si 
 o 
 
 <0 
 
 •21 
 
 o 
 
 £ § 3 § « c ?§H^o 
 
128 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 U 
 
 < 
 ) 
 
 I § § r § § s?s> 
 
 CO 
 
 8888880 
 
 g g co so ^f ra 
 
CHANGES WITH PRACTISE 
 
 129 
 
 TABLE LII 
 
 Scores in Multiplication Test 
 
 E. H. Jb. Sch. C. P. Go. Nb. Sa. Average 
 Day 
 
 1 400 685 408 492 360 344 284 303 189 380 
 
 2 348 790 382 196 262 419 218 310 119 338 
 
 3 344 642 251 234 198 315 332 308 97 302 
 
 4 255 510 242 395 207 307 338 145 108 282 
 
 5 316 628 240 303 189 193 380 175 99 280 
 
 6 320 389 278 328 187 160 356 142 203 262 
 
 7 340 220 235 289 191 240 164 150 165 221 
 
 8 294 209 240 117 180 174 182 166 85 187 
 
 9 204 227 260 214 184 165 214 146 88 189 
 
 10 276 176 168 166 168 160 210 182 81 176 
 
 11 196 184 180 275 145 174 156 113 93 169 
 
 12 212 170 154 116 145 192 156 164 69 152 
 
 13 175 145 155 193 121 172 208 110 84 151 
 
 14 176 120 121 147 216 176 234 95 98 153 
 
 15 175 145 135 239 109 261 240 130 94 170 
 
 16 166 219 141 278 96 172 249 111 83 168 
 
 17 140 229 116 153 76 113 228 110 108 142 
 
 18 216 211 147 217 96 126 234 127 82 161 
 
 19 225 145 233 148 94 188 228 132 81 164 
 
 20 242 131 225 195 96 133 237 110 97 163 
 
 Average . . 251 307 216 235 166 212 242 163 106 
 
 Below are the rankings of the nine given as for the other three 
 tests considered so far. Jb. and E. are perhaps penalized here as 
 their last few records were worse than say the fourteenth and fif- 
 teenth. Otherwise the correlations would all be closer. It must be 
 remembered too that the steps in the speed ranking are much more 
 unequal than in some of the other tests. 
 
 E. . 
 H. . 
 Jb. 
 Sch. 
 C. . 
 P. . 
 Go. 
 Nb. 
 
 Position 
 at Start 
 
 (1 = Least 
 Time) 
 
 ... 6 
 
 At Finish 
 
 (1 = Least 
 
 Time) 
 
 7.5 
 
 4 
 
 9 
 
 6 
 
 2 
 
 5 
 
 7.5 
 
 3 
 
 1 
 
 TABLE LIII 
 
 Av. 
 
 Position Speed Accuracy 
 
 (1 = Least (1 = Least (1 = Fewest 
 
 Time) Time) Errors) 
 
 Gross Per cent. 
 Gain Gain 
 
 (l=Most) (l = Most) 
 7 
 
 The correlations are : 
 
 Position at start and at finish B = .44 
 
 Position at start and average position 58 
 
 Position at start and gross gain — .63 
 
 Position at start and per cent, gain — .52 
 
 Speed and accuracy 10 
 
130 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 The same general conclusions would be drawn as for the other 
 tests, except that there is a slight positive relationship between speed 
 and accuracy. Possibly the quasi-automatism in the familiar arith- 
 metic processes noticed by C. may account for this. 
 
 In the maze test the scoring was done — as with other subjects — 
 by adding .1 to the time taken for 1 or 2 touches, .2 for 3 or 4 touches, 
 .3 for 5 or 6 touches and so on. The daily scores resulting are given 
 below and the curves plotted from them in Fig. 7. 
 
 
 
 
 
 TABLE 
 
 ! LIV 
 
 
 
 
 
 E. 
 
 H. 
 
 Jb. 
 
 Sch. 
 
 C. 
 
 P. 
 
 Go. 
 
 Nb. 
 
 Sa. 
 
 Average 
 
 216 
 
 180 
 
 174 
 
 118 
 
 165 
 
 170 
 
 288 
 
 165 
 
 172 
 
 194 
 
 204 
 
 195 
 
 264 
 
 121 
 
 198 
 
 154 
 
 240 
 
 174 
 
 333 
 
 209 
 
 180 
 
 180 
 
 143 
 
 117 
 
 187 
 
 165 
 
 90 
 
 180 
 
 280 
 
 169 
 
 216 
 
 180 
 
 159 
 
 150 
 
 198 
 
 130 
 
 306 
 
 180 
 
 195 
 
 190 
 
 192 
 
 165 
 
 121 
 
 135 
 
 181 
 
 148 
 
 96 
 
 180 
 
 273 
 
 166 
 
 168 
 
 165 
 
 154 
 
 153 
 
 160 
 
 143 
 
 192 
 
 132 
 
 290 
 
 173 
 
 216 
 
 135 
 
 176 
 
 89 
 
 181 
 
 165 
 
 228 
 
 148 
 
 259 
 
 177 
 
 216 
 
 150 
 
 224 
 
 117 
 
 190 
 
 135 
 
 228 
 
 161 
 
 281 
 
 189 
 
 195 
 
 150 
 
 120 
 
 132 
 
 160 
 
 182 
 
 264 
 
 144 
 
 215 
 
 173 
 
 168 
 
 150 
 
 221 
 
 117 
 
 209 
 
 152 
 
 108 
 
 168 
 
 244 
 
 171 
 
 132 
 
 181 
 
 108 
 
 84 
 
 176 
 
 140 
 
 144 
 
 158 
 
 316 
 
 159 
 
 168 
 
 198 
 
 187 
 
 154 
 
 140 
 
 128 
 
 108 
 
 156 
 
 247 
 
 165 
 
 144 
 
 181 
 
 120 
 
 100 
 
 160 
 
 130 
 
 132 
 
 132 
 
 203 
 
 145 
 
 144 
 
 180 
 
 100 
 
 55 
 
 190 
 
 132 
 
 108 
 
 135 
 
 210 
 
 139 
 
 126 
 
 180 
 
 142 
 
 156 
 
 130 
 
 139 
 
 114 
 
 110 
 
 212 
 
 145 
 
 168 
 
 165 
 
 88 
 
 121 
 
 140 
 
 182 
 
 108 
 
 88 
 
 210 
 
 141 
 
 132 
 
 198 
 
 120 
 
 89 
 
 120 
 
 115 
 
 108 
 
 108 
 
 190 
 
 131 
 
 144 
 
 150 
 
 168 
 
 96 
 
 149 
 
 135 
 
 102 
 
 117 
 
 231 
 
 143 
 
 144 
 
 148 
 
 99 
 
 96 
 
 143 
 
 165 
 
 120 
 
 135 
 
 231 
 
 142 
 
 144 
 
 120 
 
 90 
 
 144 
 
 165 
 
 144 
 
 102 
 
 120 
 
 82 
 
 123 
 
 Av. 171 
 
 162 
 
 148 
 
 117 
 
 162 
 
 147 
 
 159 
 
 144 
 
 238 
 
 
 It must be remembered that these are only single trials; also, 
 from experience with other subjects, notably the long-term group 
 and R. and Wy., that a conscious attention to speed is accompanied 
 by decreased accuracy. No track was kept by these nine subjects 
 as to whether they attended more to speed or to accuracy. The oral 
 directions emphasized the latter, but the general conditions of the 
 test — timing themselves and having to enter the time — would prob- 
 ably emphasize the former. From these facts then very irregular 
 curves would be expected, which is exactly what is shown. 
 
 Go.'s apparent regularity in the second half is due partly to her 
 careless entries of whole minutes, partly to her consistently high 
 number of touches. H.'s comparative smoothness is due to her al- 
 most perfect record for accuracy. When these curves are smoothed 
 out C. and P. are most alike, Sch. and Sa. most unlike. 
 
CHANGES WITH PRACTISE 
 
 131 
 
 9<> n 
 
 z: 
 
132 STUDY OF TESTS FOE INDIVIDUAL DIFFEEENCES 
 
 The rankings are given below as for the other tests, and also the 
 correlations worked out from them. 
 
 
 
 
 TABLE 
 
 LV 
 
 
 
 
 E. .. 
 
 At Start 
 
 (l=Least 
 
 Time) 
 
 .... 5 
 
 At Finish 
 
 (l=Least 
 
 Time) 
 
 6 
 
 Average 
 
 Position 
 
 (l=Least 
 
 Time) 
 
 8 
 
 Speed 
 
 (l=Least 
 
 Time) 
 
 4 
 
 Accuracy 
 (1= Fewest 
 Touches) 
 
 8 
 
 Gross 
 
 Gain 
 
 (l=Most) 
 
 5 
 
 Per cent. 
 Gain 
 (l=Most) 
 
 5 
 
 H. .. 
 
 ....4 
 
 5 
 
 6.5 
 
 8 
 
 1 
 
 6 
 
 6 
 
 Jb. 
 
 .... 6 
 
 1 
 
 4 
 
 5 
 
 5 
 
 3 
 
 2 
 
 Sch. 
 
 .... 7 
 
 3 
 
 1 
 
 2 
 
 7 
 
 4 
 
 4 
 
 C. .. 
 
 .... 3 
 
 7.5 
 
 6.5 
 
 7 
 
 3 
 
 8 
 
 8 
 
 P. .. 
 
 1 
 
 7.5 
 
 3 
 
 6 
 
 4 
 
 9 
 
 9 
 
 Go. . 
 
 .... 8 
 
 2 
 
 5 
 
 1 
 
 9 
 
 1 
 
 1 
 
 Nb. . 
 
 .... 2 
 
 4 
 
 2 
 
 3 
 
 6 
 
 7 
 
 7 
 
 Sa. . 
 
 .... 9 
 
 9 
 
 9 
 
 9 
 
 2 
 
 2 
 
 3 
 
 The corrections are: 
 
 Position at start and at finish JB = — .21 
 
 Position at start and average position 33 
 
 Position at start and gross gain — .95 
 
 Position at start and per cent, gain — .90 
 
 Speed and accuracy — .93 
 
 In this test the subjects do not keep their relative positions 
 through the series; and, as might be expected, speed and accuracy 
 are almost completely inversely correlated. 
 
 Now to examine the data for answers to the questions raised: 
 first, is a mean curve for a test representative of the test or do indi- 
 vidual curves differ too much from it and each other to make it re- 
 liable? After all, since any average tells little unless accompanied 
 by a statement of the variability, and since a curve of practise is 
 nothing but a series of such non-significant averages, one would not 
 expect a mean curve to be representative of anything beyond the fact 
 of change. Still, the changes in rate of improvement as shown by 
 the mean curve may be different with different functions, or there 
 may be one typical curve of practise to which all functions approxi- 
 mate. In Fig. 8 are shown five mean curves, one for each test. 
 That for the maze is accompanied by a scattering of dots to show 
 the distribution of the nine around each average point; that for 
 mental multiplication is accompanied by the two most distinctly dif- 
 ferent curves, those of H. and Sa. to show the range. Without these 
 representations of variability there is nothing to distinguish one 
 curve from the others. All alike show greater improvement near the 
 beginning and only slight irregularity after about the seventh day. 
 
CHANGES WITH PBACTISE 
 
 133 
 
 
 I 
 
 I 
 
 s 
 
 o 
 
 ? 
 
 I 
 
 3 
 
 DD 
 
 i 
 
 8 8 8 8 8 8c 
 
 vo «o V V) & - * 
 
134 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 All the functions do seem to approximate one typical law for changes 
 in the rate of improvement. 
 
 The second question, are the changes in the rate of improvement 
 different with different individuals or is there one typical curve of 
 practise to which all individuals approximate is answered so far as 
 these data go by Fig. 9. In this are shown the nine sets of smoothed 
 curves, one for each individual. Those for C. and P. are different 
 from those of H. and Go., the former being level and smooth, the 
 latter with a sharp slant near the beginning. Sa. also belongs to the 
 former group, only her relative position in the various tests is very 
 different. Jb. and Nb. show a moderate slant in practically all; E. 
 and Sch. have a mixture of types. This may mean that practise does 
 disclose easily recognizable individual differences, that some people 
 improve rapidly at first, others at about the same rate all the time. 
 Or it may mean only that giving a few trials shows at the beginning 
 a great range of abilities and that the range is lessened with prac- 
 tise. Those who are poor in ability have the greatest leeway to make 
 up and so improve rapidly, while every one improves rather slowly 
 once a certain degree of ability is reached. Thus if comparison is 
 made after the sharp initial slant is over, individual curves will re- 
 semble each other in form very closely. In general it seems most 
 probable that if all individuals could start with absolutely zero 
 practise and their changes in rate of improvement up to the limit of 
 improvement be measured, that their curves would resemble each 
 other very closely. The apparent differences as found are so largely 
 caused by the very different levels at which they start, as well as to 
 chance variations in their daily performances. 
 
 Individual differences do however occur in the consistency of 
 performance shown by the relative freedom from irregularity in 
 the slope of the curve. If the irregularities of C. and Sa. on the 
 one hand and of E., Go., and Sch., on the other, were computed, the 
 general tendency of the three last to more irregular progress than 
 that shown by the former two would be found much greater than 
 would be expected by chance. This difference is, however, simply 
 one form of the general differences in variability of performance, 
 not anything peculiar to the learning process by itself. 
 
 However, since all C.'s curves are not alike, nor all Go.'s, it may 
 be that there is some truth in the third condition suggested, namely, 
 that a curve reveals special not general ability in an individual. 
 That is, that in some kinds of work an individual who is good in any 
 case when compared with others will make steady though slight im- 
 provement, while one who is relatively poor will either improve 
 rapidly at first and irregularly for a considerable period, as Sch. in 
 
CHANGES WITH PBACTISE 
 
 135 
 
 
 Si 
 
136 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 mental multiplication, or he will improve very little if at all, per- 
 haps regularly as E. in the maze but more likely irregularly, as E. 
 in weights and sorting, and Sa. in the maze. 
 
 In other kinds of work the individual's initial ability may be 
 relatively very different but his tendency toward great irregularity 
 in practise or the reverse may persist. Even in a test such as judg- 
 ments of lifted weights where all nine curves are more or less irreg- 
 ular, those from C. and Sa. and perhaps Nb. who were notably reg- 
 ular in the other tests are less irregular than those from E., Go., and 
 Jb., who were irregular in other tests as well. 
 
 Finally, if irregularity is disregarded and all curves smoothed 
 out, only those facts conforming to the "law of the practise curve" 
 are represented, namely, that a person improves in any work most 
 rapidly at first and makes little and slow improvement after reach- 
 ing a certain degree of ability. From this point of view, since 
 smoothed mean curves resemble each other no matter whence their 
 derivation, practise must tend to make people more alike. 
 
IV 
 CONCLUSIONS 
 
 Reviewing this experimental study as a whole, it may be said to 
 offer evidence in reply to certain criticisms of the method of mental 
 tests. 
 
 1. In the first place the kind of tests given are said to be of little 
 significance, that knowing how many A's an individual can cancel 
 in a given time, or how many objects he can sort or how many oppo- 
 sites he can name tells us very little about him. This is probably 
 true to a certain extent, since the simpler the performance the more 
 alike individuals will probably be. Complex processes from real 
 life may often be more significant but are necessarily less precise, 
 less convenient, less well recorded and scored, and may therefore be 
 limited to the descriptive stage of investigation. Making more pre- 
 cise measurements need not exclude descriptive work, however, for, 
 in individual tests at least, details of temperament, speed in respond- 
 ing, intelligence in understanding and following directions can be 
 noted, while in addition there will be the objective record to serve 
 as basis of comparison. Then too, with careful experimentation, the 
 tests proven most typical or significant can be selected and admin- 
 istered in the best way. For instance, the easy opposite test given 
 by the time-limit method seems to be a truer measure of the speed 
 of association than the first-idea test by the amount-limit method. 
 The straight maze if improved with respect to length and continuity 
 of movement would probably be more significant and precise as a 
 measure of speed and accuracy of movement than is the hitting 
 100 dots. 
 
 2. In the second place, the criticism that a single trial is unre- 
 liable is true but need not be exaggerated since other facts such as 
 state of fatigue, time of day, temporary embarrassment, inclination 
 for work and familiarity with the environment and the kind of 
 material used also enter in to make trials unreliable. To overcome 
 this in part, at least two trials should be made of any test, preferably 
 in addition to a few minutes fore-exercise in similar work. Fewer 
 tests each administered oftener would give a truer estimate of an 
 individual and a better basis for comparison and correlations. It 
 might be advisable to allow sufficient time for each test to get the 
 
 137 
 
138 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 average divergence of the obtained result for an individual from the 
 true result down to some standard of reliability agreed upon by 
 various investigators. 
 
 3. In the third place, the criticism that giving only a few trials 
 measures not the mental process supposedly tested but merely adapt- 
 ability to strange conditions such as apparatus, instructions, work- 
 ing for speed, and the particular requirements of the test is seldom 
 of weight. Early improvement due to this alone is rare, and even 
 so could be checked by proportionate fore-exercise and the choice of 
 a proper test. 
 
 4. In the fourth place, the criticism that tests measure the degree 
 or amount of previous similar experience rather than actual capacity 
 is true not only of such tests but of any form of mental measurement. 
 It should operate only against expecting too much from the tests, not 
 against their use, but rather, in fact, in favor of repeating them at 
 stated intervals. The only alternative — testing subjects with no simi- 
 lar previous experience or else those whose training had brought 
 them to the physiological limit — would be impracticable, and out of 
 the question. In general, tests of a novel, little-trained function such 
 as grouped objects or the a — t test show greater susceptibility to 
 practise than those of a frequently used, much trained function 
 such as addition. 
 
 5. In the fifth place, in estimating the nature and degree of 
 improvement in a function with repeated trials the nature of the 
 units used to express such improvement must be taken into consider- 
 ation, and misleading statements based upon one form of measure- 
 ment only must be guarded against. Moreover, when comparisons 
 of changes are to be made, whether between different processes in an 
 individual or a group, or between different individuals in one process, 
 it becomes still more important to use more than one way of treating 
 measurements. 
 
 6. In the sixth place, the criticism that practise may influence 
 individuals each by a law of his own and processes each by a law of 
 its own does not seem to hold so far as the general law of improve- 
 ment goes. On the whole, higher mental functions are sooner sus- 
 ceptible to practise than are sensory functions, the more so again if 
 they are novel. Individuals with low standing can and do improve 
 the most, judging objectively, though even so they may not, in con- 
 veniently measurable periods of time, overtake those whose standing 
 was high at the beginning. Characteristic variability or consistency 
 of performance may be disclosed whatever the process and whatever 
 the change in improvement. 
 
APPENDIX 
 
 Key fob Correction of Opposites 
 Eight, scored 2. (Second choice, scored 1.) Wrong scored 
 
 Above Below, beneath, under, down 
 
 Absent Present (here) 
 
 Adroit Awkward, clumsy (unskilful, unskilled) 
 
 After Before (ahead) 
 
 Apart Together (with, near) 
 
 Asleep Awake 
 
 Backwards Forwards (frontwards) 
 
 Barbarous Civilized (humane), tame, cultivated 
 
 Best 
 
 Big 
 
 Bless 
 
 Broad 
 
 Broken 
 
 Brother 
 
 Buy 
 
 Cheap 
 
 Clumsy 
 
 Come 
 
 Country 
 
 Create 
 
 Day 
 
 Dead 
 
 Deceitful 
 
 Degrade 
 
 Diligent 
 
 Elation 
 
 Enrage 
 
 Exciting 
 
 Expand 
 
 to Float 
 
 Forcible 
 
 Worst 
 
 Little (small) 
 
 Curse 
 
 Narrow (thin) 
 
 Whole (mended, unbroken, intact) 
 
 Sister 
 
 Sell 
 
 Dear, expensive 
 
 Adroit, deft, skilful, neat (adept, agile, graceful), clever 
 
 Go 
 
 City, town 
 
 Destroy, annihilate, tear down (abolish, spoil) 
 
 Night 
 
 Alive, living 
 
 Sincere, straightforward (truthful, honest, frank, candid, honor- 
 able), open, true, ingenious, upright 
 
 Elevate (exalt, uplift, raise, ennoble, promote, advance, restore, 
 honor) 
 
 Lazy, indolent 
 
 Depression, dejection (despondency, low-spiritedness) 
 
 Pacify (subdue, appease, calm), quiet 
 
 Depressing, quieting, soothing (calm, restful) 
 
 Contract, condense (decrease, narrow), enclose 
 
 Sink (anchor) 
 
 Weak (gentle), gently 
 Frequently Seldom, rarely (not often, occasionally) 
 Generous Stingy, parsimonious (miserly, greedy, mean), avaricious 
 
 Eough (rude, harsh) 
 
 False, spurious (counterfeit, sham, insincere, artificial, unreal, 
 imitation, fictitious), fake, bogus, adulterated, spurious 
 
 Simple, trivial (poor, petty, modest, ordinary, humble, mean, 
 ignoble, plain, commonplace, insignificant), tawdry, mediocre, 
 lowly 
 
 There 
 
 Help, aid, further (promote, advance, assist, hasten, quicken) 
 
 139 
 
 Gentle 
 Genuine 
 
 Grand 
 
 Here 
 Hinder 
 
140 
 
 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 Level 
 
 Hold Let go, release, drop (lose, give up, loosen), give, loose 
 
 If Unless (although, certainly) 
 
 Ignorant Wise (informed, learned, knowing, educated, intelligent) 
 to Lack Have, possess, abound (have in abundance, gain), attain 
 Land Water (sea) 
 
 More 
 
 Uneven, slanting, sloping, inclined (rugged, hilly, mountainous, 
 irregular, undulating), jagged, rough, 'bumpy, broken 
 Loquacious Taciturn, silent (quiet, reticent, reserved) 
 Mine Yours (his, theirs), your 
 
 Motion Eest (still, standstill, stillness, quiet) 
 
 Obscure Clear, lucid (plain, evident, light, bright), open, significant 
 Over Under (below, beneath) 
 
 Part Whole, meet (totality, entirely) 
 
 Past Future (present) 
 
 Permanent Temporary (transitory, transient, fleeting), ephemeral, evanescent, 
 
 unstable, changing 
 Permit Forbid, deny, prohibit (prevent, refuse), hinder 
 
 Precise Inexact (careless, slovenly, disorderly, lax, indefinite, vague, inac- 
 
 curate), irregular, loose 
 Proud Eumble, cosmopolitan, democratic 
 
 Eepulsion Attraction, liking, encouragement, acceptance 
 
 to Respect Despise (look down on, disregard, insult), abhor, scorn, loathe, 
 dislike 
 
 Conceal, keep secret (hide, obscure, cover up, keep back) 
 
 Smooth, gentle (calm, tender), easy 
 
 Polite, civil, courteous (cultured, sophisticated, obliging, gentle), 
 refined, fine, polished 
 
 Together, combined, meet, join, connect (collective, united, con- 
 tinuous) 
 
 Frivolous, gay (merry, laughing, joking, jocular, mirthful, lively), 
 jocose, funny, silly, cheerful 
 
 Complex (hard, wise, clever, complicated, difficult, intricate, pro- 
 found, elaborate) 
 
 Daughter (father) 
 
 Save (keep, hoard), hold 
 
 Calm (clear, quiet, fine, peaceful, smooth, tranquil), fair, mild 
 
 Crooked (curved) 
 
 Sensible, bright, clever (smart), wise, alert 
 
 Give (leave, let alone) 
 
 Short 
 
 If (in spite of, though), because 
 
 Horizontal (slanting), crooked, perpendicular 
 
 Fresh (refreshed, rested, brisk, lively), energetic 
 
 Eighteous, good (holy) 
 
 Tame, cultivated (civilized) 
 
 Lose 
 
 to Eeveal 
 
 Eough 
 
 Eude 
 
 Separate 
 
 Serious 
 
 Simple 
 
 Son 
 
 Spend 
 
 Stormy 
 
 Straight 
 
 Stupid 
 
 Take 
 
 Tall 
 
 Unless 
 
 Vertical 
 
 Weary 
 
 Wicked 
 
 Wild 
 
 Win 
 
 PAEAGEAPHS USED IN THE EBBINGHAUS COMBINATION TEST 
 
 I-XX were specially prepared for the long-term group. The remaining 
 paragraphs, prepared by other investigators, were used with the short-term 
 group. 
 
APPENDIX 141 
 
 I. 
 
 The argument amounts . . this, that like consequents must like ante- 
 cedents. But it is impossible for the antecedents to be alike, in that 
 
 the thoughts and feelings give rise to my movements are immediately 
 
 given, while which give rise to people 's movements are __. . . given. 
 
 The question presents , whether this essential in the mode of 
 
 existence . . the antecedents does not wreck the analogy. 
 
 II. 
 
 From the facts thus . . . presented, it would be natural to infer 
 
 mind and body are, in respect of action, on a footing . . equality. The inter- 
 
 actionist, at this point, might be tempted to set up the that every fact 
 
 showing the influence of .... upon mind can be matched with a . . . . showing 
 of ... . upon . . . . , and that by as much as the former demon- 
 strates the mind 'a dependence, the demonstrates its power. 
 
 III. 
 
 In every actual case of perception, the entire fact is not the presence 
 
 of a physical to consciousness, but at the same , and as a condition 
 
 of that presence, the existence of a train of and effects connecting the 
 
 object the percipient 's If I a table, this involves the pres- 
 ence in the world, along with the table, of light-rays passing from the 
 
 to the eye, and passing from the eye to the brain. 
 
 IV. 
 
 Parliament had hitherto very little attention on our Eastern 
 
 possessions. Since the death of George II., a rapid of weak admin- 
 istrations each of was in turn flattered and betrayed by the Court, had 
 
 held the of power. Intrigues in the palace, riots in the capital, and 
 
 insurrectionary in the American colonies had left the advisers of 
 
 the Crown little time to study Indian politics. When they did interfere their 
 
 interference was and irresolute. Lord Chatham had a bold 
 
 attack on the Company, but his plans were rendered by the strange 
 
 malady which about that began to overcloud his splendid genius. At 
 
 length it was generally felt that Parliament could no longer the affairs 
 
 of India. 
 
 V. 
 
 Very similar to this was the state of India sixty years .... Of the existing 
 
 governments not a single one could lay to legitimacy. There was 
 
 scarcely a province in which the real sovereignty and the sovereignty 
 
 were not disjoined. Titles and forms were still which implied that the 
 
 heir of Jamerlane was absolute when in reality he was a captive. The 
 
 Nabobs were, in some independent princes; in others, they had, their 
 
 master, become phantoms and the Company was supreme. Among the 
 
 Mahrattas the heir still the title of Kajah; but he was a prisoner, and his 
 
 prime minister had the chief of the state. 
 
 VI. 
 
 In a rude state of society men are children with a greater variety of ideas. 
 It is in such a state of society that we may to find the poetical tempera- 
 ment in its perfection. In an enlightened . . . there will be much 
 
142 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 intelligence, much , much philosophy, abundance of just classification and 
 
 subtle , abundance of wit and eloquence, abundance of verses and even 
 
 of .... ones ; but little Men will talk about the old poets and com- 
 ment on them, and to a certain extent them, but they will scarcely be 
 
 able to the effect which poetry produced upon their ruder , 
 
 the ecstasy, the plenitude of belief. 
 
 VII. 
 
 One of his gifts was a voice habitually deep and sonorous yet capable of 
 very low and gentle at the moment. About his ordinary bear- 
 ing was a certain fling, a fearless expectation of success, a confidence in 
 
 his own and integrity much fortified by contempt for .... obstacles or 
 
 seductions of he had had . . experience. Mr. B. perhaps liked him the 
 
 for the difference between . . . . , and certainly for being a stranger. 
 
 One can begin so many things with a . . . person ! 
 
 VIII. 
 
 He had never put any question concerning the nature of his illness, nor had 
 
 he betrayed any as to how far it might be likely to cut his labors 
 
 or his life. On this point, as on all others he from pity ; and if the 
 
 suspicion of being pitied for anything surmised or known in of himself 
 
 was embittering, the idea of calling a show of compassion by frankly 
 
 an alarm was intolerable. Every proud mind knows something of this 
 
 and perhaps it is only to be by a sense of fellowship deep 
 
 to make all efforts at isolation seem mean and petty of exalting. 
 
 IX. 
 
 Her belief that Eosamond could manage her papa was well founded. Mr. 
 
 Vincy had as of his own way as if he had been a prime minister : the 
 
 force of was easily too for him as it is for most pleasure- 
 loving, florid ; and Eosamond was forcible by means of that 
 
 mild persistence which enables a soft living substance to make its ... in spite of 
 opposing rock. Papa was no rock. He had no fixity but that of alternating 
 
 impulses sometimes habit, and was altogether unfavorable to his 
 
 taking a decisive line of in relation to his engagement. 
 
 X. 
 
 Soldier wake, the ... is peeping 
 Honor ne 'er was ... in sleeping, 
 Never .... the sunbeams still 
 Lay unreflected on the . . . . : 
 'Tis when they are glinted .... 
 From axe and armor, spear and jack, 
 
 That they promise story 
 
 Many a page of deathless 
 
 Shields that are the foeman's terror 
 Ever . . . the morning 's mirror. 
 
 Soldier, , thy harvest, fame; 
 
 Thy study, conquest ; war, thy 
 
APPENDIX 143 
 
 XI. 
 
 XII. 
 
 And is she happy? Does she see unmoved 
 
 The in which she have lived and loved 
 
 Slip without bliss slowly away, 
 
 One after one, like to-day? ~ — _ 
 
 Joy has . . . found her yet, nor ever will, 
 
 Is it this which makes her mien so still 
 
 Her features . . fatigued, her eyes, tho * sweet, 
 
 So sunk, so rarely save to meet 
 
 Her children's? She moves slow; her voice alone 
 Hath yet an infantine and silver tone, 
 
 But that comes languidly : in truth 
 
 She one dying in a mask of youth. 
 
 Move eastward, happy earth, and leave 
 
 Yon orange waning slow ; 
 
 From fringes of the eve 
 
 O, happy planet, go ; 
 
 Till over thy dark shoulder glow 
 
 . . . silver sister , and rise 
 
 To glass herself in dewy eyes 
 
 That me from the glen below. 
 
 Ah, bear me with . . . . , lightly borne, 
 
 Dip forward under light 
 
 And move me to my marriage .... 
 And round to happy night. 
 
 XIII. 
 
 Professor Crocker presented his trained animals yesterday afternoon and 
 
 and was greeted . . large houses on both The production is 
 
 unique and an interesting lesson in education, some . . the tricks 
 
 by the four-footed actors being really His troup con- 
 sists of 25 animals, and has a role to 
 
 XIV. 
 
 Weather that was pleasant only at times, and at times threatening or 
 
 rainy made unpleasant conditions . . . yesterday 's observance of 
 
 Dominion day, and .... a damper on many festivities. The morning dawned 
 
 bright and and scores of parties left the city on excursions. 
 
 Towards noon it became cloudy and there were some Again it 
 
 cleared up, only to be later by heavy thunder, lightning and rain, 
 
 though the in the city was light to what it was in the 
 
 surburbs. 
 
 XV. 
 
 The longshoremen of the Cunard pier who struck yesterday 
 
 the steamship Umbria arrived to the company to pay them sixty 
 
 instead of fifty-five cents an hour for Sunday , returned to work 
 
 to-day. Their demand was not The chairman of the said to-day 
 
 that he was at a loss to the reason for the action of the 
 
 men. He said the union did not the strike. 
 
144 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 XVI. 
 
 The magnetic dip needle is made in the form of a lozenge, to the 
 
 horizontal needle, but it is poised or by of a shaft running 
 
 through the center of the lozenge at right to it, and is held in 
 
 by agate bearings as in figure 20. In some types the cradle the 
 
 horizontal shaft is poised on a steel needle. The needle is thus to take up 
 
 a position and south and to incline on its 
 
 XVII. 
 
 It is natural to believe in great men. Nature seems to for the 
 
 excellent. The world is upheld by the veracity of .... men ; they make the 
 earth wholesome. They who lived .... them found life glad and nutritious. 
 Life is sweet and tolerable only in our belief in .... society ; and actually, or 
 ideally, we manage to .... with our superiors. We call our children and our 
 
 lands by their Their names are into the verbs of language, their 
 
 works and effigies are in our , and every circumstance of the . . . recalls 
 
 an anecdote of them. 
 
 XVIII. 
 
 If he had been an English nobleman on a pleasure tour, or a newspaper 
 
 courier, he could not have more quickly. The post boys wondered at 
 
 the fees he amongst them. How happy and green the country 
 
 as the chaise whirled from milestone to milestone, through neat country 
 
 towns where landlords out to welcome him with and bows ; by 
 
 pretty roadside inns where the signs .... on the elms, and horses and men were 
 
 drinking under the checkered of the trees ; rustic hamlets 
 
 round ancient grey churches, and through the friendly English landscape. To a 
 traveller returning it looks so kind. 
 
 XIX. 
 
 Nay, ye should not weep, my children! 
 Leave it to the faint and weak; 
 Sobs are ... a woman 's weapon 
 
 Tears befit a maiden 's 
 
 Weep not, of MacDonald ! 
 
 not thou, his orphan heir. 
 
 Not in shame, but honor 
 
 Lies thy slaughtered there. 
 
 Weep not, but when years are over 
 And thine arm is and sure, 
 
 Let thy heart be as iron 
 
 And thy wrath as fierce . . fire, 
 
 Till the hour when cometh 
 
 For the race that slew thy sire! 
 
 XX. 
 
 An electrical storm of severity passed over this district last night, 
 
 which burned barns, killed cows in the field, put telephones and lines 
 
 out of commission, knocked trees, and did a great deal of gener- 
 ally. The flag staff was struck and splintered and the slates were .... off the 
 
APPENDIX 145 
 
 roof. A barn was burned with a large of hay, and a driving shed 
 
 was destroyed. Crops in all were almost pounded into the 
 
 XXI. 
 
 We confess to something of sympathy ... the correspondent . . . hinted 
 
 yesterday that . . . children are ... over and killed by automobiles, the 
 
 is not always that . . the automobilist, . . . sometimes rests in some measure 
 
 on those who do not their children to avoid unnecessary It is 
 
 a plain . . . . , of course, that public highways are . . . the use of the whole 
 
 population, . . . that the automobilist is every obligation . . keep the 
 
 limitations of his rights and privileges . . mind as he goes along, but the road 
 is his . . well as other peoples. 
 
 XXII. 
 
 If we are well, thoroughly sound, we not be depressed. 
 
 The perfectly healthy animal ... no worries. The remedy has already 
 
 indicated. Eegretfully it is . . simple .... very few people take the trouble 
 
 to it it is clearly and widely recognized that is stupid, 
 
 that its .... is simple where is no organic trouble, worry will 
 
 Worry is simply a .... of, what, ... the sake of a nice large word, is called 
 
 1 ' neurasthenia, ' ' nerve-depletion plenty of recreation, plenty of fresh 
 
 air, and the man will not worry. 
 
 XXIII. 
 
 Park Hill on the Hudson offers you a solution . . the home problem to-day. 
 No home seeker . . investor . . . afford to ignore its claims. Escape the wear 
 and tear .. the city's noise ... rush .. this open air paradise, just .. the 
 city 's edge, . . all respects an ideal home location . . . yourself and family. 
 
 are cottages containing every improvement waiting . . . you to step . . 
 
 and make yourself comfortable. It not .... commands the most beautiful view 
 
 around New York ... is protected for all time intrusion. Choice lots 
 
 now on very easy terms. 
 
 XXIV. 
 
 A law . . defence of property rights in the broadest sense . . observed 
 
 almost abolish international conflicts. Gentlemen . . not fight with fists 
 
 . . money differences ... do they refer them . . courts of honor. Civil courts 
 
 are for that and are as useful for nations as for men. The sanction of 
 
 international law must . . merely moral, for a long time . . least. But in 
 
 that there should be . . . moral sanction there must . . a moral code. The prin- 
 ciples of .... a code are deducible .... treaties to which nations have set 
 their hands . . . seals. 
 
 XXV. 
 
 I asked the slovenly, . . . cheerful female . . . answered the bell . . . the 
 landlady, wondering the while .... I should say when I was asked . . . refer- 
 ences. The merriment had not been called forth . . anything amusing . . my 
 appearance, . . my vanity had feared, ... by a story which a man sitting . . 
 . . . head of the table was just finishing. The only vacant chair . . the room 
 
 was beside him, and, rather awkwardly, ... I felt that they were my 
 
 measure, I made my . . . toward it. As I ... down he greeted . . with a 
 polite bow. 
 
146 STUDY OF TESTS FOB INDIVIDUAL DIFFERENCES 
 
 XXVI. 
 
 The occult in everyday affairs is the of this new book . . Robert 
 
 Chalmers one of the thrilling stories of the volume is composed 
 
 . . the tale of some awful mysterious happening, some supernatural 
 
 beyond the of material reasoning of mortal man . . explain, which comes 
 
 the life of some ordinary, everyday man. The opening tells of a 
 
 dinner to a man deeply versed in occultism . . his American friends. 
 
 To these he gives many hints . . . suggestions of momentous things which he 
 . . . plainly see for them . . the future. 
 
 XXVII. 
 
 We believe we can prove . . you that this investment is . . secure . . . the 
 dividends so sure, that it justifies you . . withdrawing money .... the Savings 
 
 Banks, it is earning 3£# and putting it . . our business where it will 
 
 earn 1$. We are a New England enterprise, managed . . New England men, 
 
 and we have behind . . a record . . fourteen years of unbroken success 
 
 you have much or little you can not to let slip this opportunity of 
 
 doubling the from your savings. Prompt action in this matter will 
 
 you well. 
 
 XXVIII. 
 
 On the , it didn 't cost me a dollar. In fact, though at I 
 
 have found myself of considerable sums of ready money, I have 
 
 never a man of property . . the strict sense of the word. I abandoned 
 
 my , the law, . . I did not its practice so lucrative . . I had 
 
 hoped. For some years thereafter I traveled largely . . the Mississippi Eiver. 
 It . . . the decline in steamboating . . . the adoption . . less leisurely methods 
 
 of travel cut into my income and forced . . to come North and 
 
 in trade. 
 
VITA 
 
 Mary Theodora Whitley, the author of this dissertation, was born 
 October 4, 1878, in London, England. She was educated at home 
 and in private schools, taking second class honors in the Senior 
 Cambridge Local Examinations at Eastbourne, in 1895. After three 
 years of travel and three of private teaching, she entered Teachers 
 College, Columbia University, taking the B.S. degree in 1905 with 
 diploma in English, and the A.M. degree in 1906, specializing in 
 psychology under Professors Thorndike, Woodworth and Cattell. 
 From 1905 to 1907 she held the position of assistant in the depart- 
 ment of psychology, Teachers College; 1907-08, lecturer, 1908-10, 
 tutor, 1910- instructor in psychology in the same institution. 
 
 147 
 
THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 AN INITIAL FINE OF 25 CENTS 
 
 WILL BE ASSESSED FOR FAILURE TO RETURN 
 THIS BOOK ON THE DATE DUE. THE PENALTY 
 WILL INCREASE TO 50 CENTS ON THE FOURTH 
 DAY AND TO $1.00 ON THE SEVENTH DAY 
 OVERDUE. 
 
 NOV 1A IQ^fi 
 
 
 nFUV ID 15*00 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 « 
 
 LD 21-100m-8,'34 
 
it u / ot»b