Untitled-18 Small Group Interaction 157 157 Small-Group Interaction among Professional Librarians Kathleen R. Tower Libraries are using small groups to make decisions, and it is important that the decisions these groups make are effective. Small-group schol- ars argue that group processes and interactions play an important role and influence effective decision-making. Randy Y. Hirokawa developed a theory called Vigilant Interaction Theory which maintains that group interaction affects decision-making performance by directly shaping the quality of vigilance that leads to a final choice. Small groups consisting of professional librarians were used to test Hirokawa’s theory, and the results showed that group decision performance is directly related to the group’s efforts to perform critical vigilant decision-making functions. Specifically, groups that show a pattern generating more alternative solutions are more likely to develop effective decisions. Kathleen R. Tower is Head of Special Collections/Government Publications in the John U. Tomlinson Library at Mesa State College, Grand Junction, Colorado; e-mail: ktower@mesa7.mesa.colorado.edu. communicate efficiently and creatively with ever-increasing sophistication. Little research has been done on com- munication among librarians. Studies have explored interpersonal communi- cation between librarians and patrons during the reference interview,1 effects of nonverbal communication on the pa- trons� perceptions of the library and li- brarians,2 communication between li- brarians and faculty,3 communication be- tween public and school librarians,4 writ- ten communication between music li- brarians,5 and the type of communication activity in which librarians find them- selves involved, such as with whom they communicate, by what channels, and how frequently.6 Group processes and interactions play an important role and greatly influence effective decision-making and problem- solving. A large number of studies in ibrarians work in a world that is in constant change, accom- panied by increasing demands for accountability. Decisions made within the library are complex, costly, and visible to the public, making it crucial that good decisions are made. Along with other organizations, librar- ies are looking to small groups to solve problems. Library administrators are us- ing small-group techniques such as qual- ity circles, total quality management, nominal group techniques, and self-man- agement team techniques to help with decision-making. In this environment, it is imperative that library professionals develop their understanding of the pro- cess of communication and group deci- sion-making, and develop their ability to work effectively within new environ- ments. The need for effective decision- making requires library professionals to 158 College & Research Libraries March 1997 small-group communication have been concerned with group performance, group productivity, and group decision- making effectiv eness.7 Small-group scholars are suggesting that interactions that occur among group members as they work on a task or problem represent an important key to understanding why cer- tain groups are more successful than oth- ers in meeting their goals and objectives.8 �Organizations are only as good as the decisions they make.�9 Communication is one of the most abstract, pervasive, important, and com- plex cluster of behaviors which can be defined as a message that is transmitted from sender to receiver.10 The term small- group communication refers to the theory and practice relating to one or more meet- ings of a small group of people who com- municate face-to-face in order to fulfill common purposes and to achieve group goals.11 The 1980s saw the development of three distinct small-group communica- tion theories.12 One of these was devel- oped by Randy Y. Hirokawa, who con- ducted a number of studies in small- group communication. Hirokawa re- searched group decision-making effec- tiveness in an effort to identify the fac- tors that influence effective group perfor- mance and developed a theory he calls Vigilant Interaction Theory.13 His theory argues that the manner in which group members establish operating procedures, analyze a problem, establish criteria for evaluating it, and evaluate suggested al- ternatives affects the quality of the solu- tion presented by the group. This theory maintains that group interaction affects decision-making performance by directly shaping the quality of vigilance (or critical thinking) that leads to a final choice. In short, the theory argues that the manner in which group members talk about the problems, options, and consequences facing the group affects the way they think about those problems, options, and con- sequences, which in turn ultimately determines the quality of final choices they make as a group.14 Hirokawa�s theory grew out of a need to explain inconsistent findings in re- search relating communication to group performance.15 He set out to identify qualities of groups that are most associ- ated with effective decision-making.16 His theory takes a comprehensive look at the variety of error sources in group deci- sion-making with an eye toward identi- fying the kinds of things groups must address to become more effective. The se- quence of activity identified by Hirokawa mirrors John Dewey�s problem-solving sequence, which, since the publication of How We Think in 1910, has greatly influ- enced twentieth-century pragmatic thought.17 Hirokawa�s research was conducted to learn about the differences in group pro- cedures when members make �good� and �bad� decisions, thus studying �best� and �worst� groups.18 He con- cluded that members of the effective groups examined the validity of one another �s opinions and assumptions; people in the more successful groups compared alternatives to a set of criteria they believed a good decision should meet; and both groups based their choices on premises. The premises of the successful groups followed directly from facts that the members possessed about each option and that the premises of the ineffective groups did not have a good basis in fact.19 According to Hirokawa�s theory, the most important characteristics of effec- tive groups are: 1. Groups assess and understand the problem. 2. Groups identify alternatives. 3. Groups assess positive and negative alternatives vigorously Small Group Interaction 159 sional librarians. Like other organiza- tions, libraries are populated by staff with varied interests and attitudes.23 How do librarians as professionals communicate among themselves in small groups? Hirokawa indicates that studies simi- lar to his are needed to provide internally valid data establishing the importance of vigilant interaction for organization group decision-making. The questions explored in the study presented here were: (1) Do relationships exist between the frequency of certain communicative behaviors and group decision-making effectiveness among professional librar- ians? (2) Are the communication patterns of interaction within effective decision- making groups different from those within ineffective decision-making groups made up of professional librar- ians? For this study, a small group is defined as three or more people interacting with one another face-to-face in such a man- ner that each person influences, and is influenced by, each other working to- gether to reach a common goal. An ef- fective small group is one whose activ- ity leads to a positive, workable decision and whose activity produces a desired effect. Methodology This study employed an experimental de- sign where small leaderless groups of volunteer subjects participated in thirty- minute discussions that were videotaped and tape-recorded. Each group consist- ing of four professional librarians worked on the same problem-solving task and was asked to develop a solu- tion that was a consensus of the group. The librarians were recruited by contact- ing library directors at various institu- evaluating the validity of each other �s opinions and assumptions. 4. Groups select the alternative (makes a decision).20 Hirokawa�s theory says that effective groups clarify, modify, and test opinions of their members and that ineffective groups gloss over the opinions of their members. Members of effective groups carefully test their decision alternatives and match them against the group�s pre- established decision criteria, and criti- cally evaluate each alternative in the same manner that they evaluate the opinions and assumptions of their fellow mem- bers, forcing themselves to consider all aspects of an alternative. They ask whether the alternative is fair, warranted, appropriate, or reasonable to ensure that they have examined it adequately. Hirokawa and his associates found that high-quality groups use facts and infer- ences that are accurate and reasonable.21 Hirokawa identified factors that con- tribute to a faulty decision-making pro- cess. The first is improper assessment of the problem. This involves inadequate or inaccurate analysis of the problem or situation in which the decision-making occurs. A second source of error in deci- sion-making is inappropriate goals and objectives. The group may neglect impor- tant objectives that ought to be achieved, or it may work toward unnecessary ob- jectives. A third problem is improper as- sessment of positive and negative quali- ties. The group does not see the advan- tages and/or disadvantages of various proposals, or it may overestimate the positive or negative outcomes. Fourth, the group may develop an inadequate information base, and finally, the group may be guilty of faulty reasoning.22 There appears to be a relationship be- tween interaction behavior and group performance outcomes. The purpose of the study the author presents here was to investigate if this would be the case with groups whose members are profes- There appears to be a relationship between interaction behavior and group performance outcomes. 160 College & Research Libraries March 1997 tions. No attempt was made to manipu- late the makeup of the groups. Group membership was developed at the indi- vidual libraries by the library director or one of his or her representatives. Each group was presented with the group task. Each received an identical presentation of background information and was told that its members would be given thirty minutes in which to discuss the problem and come up with specific recommendations. The groups were in- formed that the entire discussion would be videotaped and tape-recorded. One member of the group was asked to be re- sponsible for submitting written recom- mendations reflecting the views of the entire group. After addressing any pro- cedural questions, the researcher left the room and allowed the group to discuss the problem for the entire thirty-minute period. After thirty minutes, the re- searcher reentered the room and collected the recommendations proposed by the group, at which point subjects were in- formed of the true purpose of the re- search. The entire data-gathering process was repeated until all ten videotaped and tape-recorded discussions and group proposals had been obtained. The task selected for discussion was a problem that librarians would encoun- ter in a working situation. The problem was one that could be discussed and a decision reached in the time allotted. For this study, each group was to develop a program for training library staff in new technologies and for keeping them up- to-date. Library staff included those who worked in the library�librarians, sup- port staff, and student workers. The pro- gram was not to include training or teaching patrons. The technologies could be new CD-ROM products, changes in the CD-ROM products already in the li- brary, changes on the online catalog, the use of Internet, and the use of locally cre- ated databases. Groups were not to limit their plan. Group members were to as- sume that the library director was com- mitted to the project and had guaranteed the funds and staff time for implementa- tion of the program that was developed. Once all the group discussions had been videotaped and tape-recorded, and their respective group proposals col- lected, steps were taken to determine the quality of the training program devel- oped by each group to ascertain effective and ineffective groups. Three judges, who possessed the knowledge and ex- pertise necessary to evaluate the propos- als properly and fairly, were asked to evaluate the training programs. The judges were given identical instructions. Each judge worked independently and gave each program ratings as to appro- priateness, workability, completeness, and timeliness. Inappropriate 1 2 3 4 5 6 7 Appropriate Workable 7 6 5 4 3 2 1 Unworkable Incomplete 1 2 3 4 5 6 7 Complete Timely 7 6 5 4 3 2 1 Not timely In awarding scores, the judges were told to be as consistent as possible and to keep in mind that the groups had only thirty minutes to discuss the problem. The groups were not told in advance what topic they would be discussing, nor were they given any time to research the problem before the discussion. A summation of the four scores for each of the training programs proposed by a group thus served as the measure of group decision-making effectiveness. In order to have adequate degrees of free- dom for the statistical tests in this study, the groups were classified in two sub- sets��high effective� and �low effec- tive.� Content analysis was used to analyze the group interactions. Content analysis Group members were to assume that the library director was com- mitted to the project Small Group Interaction 161 2. Ask for opinions 3. Ask for consent/approval 4. Ask for clarification 5. Ask for summary/synthesis (Hirokawa, 49�51) In order to organize the five general functional categories and sixteen specific behavior categories into a workable sys- tem for analyzing group interaction, Hirokawa developed a two-level analy- sis system (Hirokawa, 51). The first level consists of the five general task-achieve- ment functions. A sixth �miscellaneous� category was created to accommodate any verbal behaviors that performed a function other than the five being exam- ined. The second level consists of sixteen interaction categories that represent spe- cific types of verbal behaviors which can perform any of the five task-achievement functions. The eighty-one-behavioral-cat- egory system was assumed to contain be- haviors most likely to be related to group decision-making or problem-solving ef- fectiveness (Hirokawa, 51�52). It assumes that the basic unit of analysis is the func- tional utterance that is defined as �an un- interrupted utterance of a single group member which is perceived to perform a specific function (or action) within the group interaction process� (Hirokawa, 63). Once the ten-group proposals had been evaluated and the corresponding group discussions had been coded, the data were analyzed using three different statistical models: Chi-square analysis, correlation analysis, and multiple dis- criminate analysis. Results An objective of the study was to assess the relative effectiveness of the solutions developed by the ten groups. The evalu- ation process was accomplished with the assistance of three judges who possessed the knowledge and expertise necessary to evaluate the proposals properly and fairly. The sum of the individual scores for each of the recommendations pro- was used to facilitate the objective analy- sis of the discussions and to eliminate bias in the investigative process and to decrease subjectivity. Without knowledge of the results of the three judges� scor- ing, utterances were coded using the Function-Oriented Interaction Analysis System (FOIAS) designed by Hirokawa.24 Hirokawa developed the system to or- ganize group interaction process in terms of those communication behaviors that performed functions essential to success- ful task performance. He identified these functions as follows: 1. Establishment of operating procedures; 2. Analysis of the problem fac- ing the group; 3. Establishment of criteria for evaluating alternative solutions; 4. Generation of alternative so- lutions to solving the problem; 5. Evaluation of each suggested al- ternative with the eventual selection of the best available alternative(s). (Hirokawa, 49) Hirokawa�s FOIAS identified sixteen interactive categories. Eleven are orga- nized under the heading of �assertions,� which are comments that present infor- mation, and five under the heading of �requests,� which are comments that ask for information (Hirokawa, 49). I. Assertions 1. Introduction of a fact 2. Introduction of an opinion 3. Restatement of a fact 4. Restatement of an opinion 5. Development of a fact 6. Development of an opinion 7. Substantiation of a fact 8. Substantiation of an opinion 9. Stated agreement 10. Stated disagreement 11. Summary/synthesis II. Requests 1. Ask for facts 162 College & Research Libraries March 1997 Judge 3; and the forty scores from Judge 2 were compared with the forty scores from Judge 3. The resulting Pearson prod- uct-moment r�s of .2646 (p < .099), .6709 (p < .001), and .2949 (p < .065) at the .10 level indicated that the three judges basi- cally agreed with one another regarding their evaluations of the group proposals. Table 2 summarizes the results of the cor- relation. Having established each group�s rank and the judges� reliability, the groups� evaluation scores and coded discussions were utilized in a series of nonparamet- ric and parametric statistical models. The first was Chi-square analysis (x2). The purpose of using this model was to de- termine whether a relationship existed between the performance of the five gen- eral discussion functions and group de- cision-making effectiveness. The ten groups were divided into two subsets on the basis of their effectiveness scores. Those with scores above the mean (20�22.6667) were classified as �high ef- fective� and those with scores below the mean (17.3�19) were classified as �low ef- posed by the groups served as quantita- tive measures of the groups� decision- making effectiveness. Table 1 summa- rizes the results of the judges� evaluation. The mean scores ranged from a low of 17.0000 (group 7) to a high of 22.6667 (group 6), with a standard deviation of 1.86. The distribution of scores approxi- mated the �normal� distribution (kurotosis = -.74) being a little more peaked than normal and somewhat skewed to the right (skewness = .50). After obtaining a quantitative index of the decision-making effectiveness of the groups, an attempt was made to de- termine the reliability of the judges. The three judges independently rated each of the ten-group solutions such that each solution received separate scores for ap- propriateness, workability, timeliness, and completeness. Estimates of interrater reliability was established using Pearson product-moment correlation statistics (r). The forty scores from Judge 1 were com- pared with the forty scores from Judge 2; the forty scores from Judge 1 were compared with the forty scores from TABLE 1 Group Decision-Making Effectiveness Scores Judge 1 Judge 2 Judge 3 Total Mean Rank Group 1 20 23 22 65 21.6667 2 Group 2 21 16 20 57 19.0000 5 Group 3 22 22 19 63 21.0000 3 Group 4 27 12 21 60 20.0000 4 Group 5 25 10 20 55 18.3333 8 Group 6 27 15 26 68 22.6667 1 Group 7 19 13 19 51 17.0000 10 Group 8 20 11 21 52 17.3333 9 Group 9 22 17 17 56 18.6667 6 Group 10 21 15 20 56 18.6667 7 Total 583 194.34 Mean 58.3 19.43 Standard Deviation (SD) 5.579 1.86 Mean (Judge 1) = 22.40 S.D. (overall) = 1.86 Mean (Judge 2) = 15.40 Kurtosis (overall) = -.74 Mean (Judge 3) = 20.50 Skewness (overall) = .50 Mean (overall) = 19.43 Small Group Interaction 163 TABLE 2 Judges Scores for Appropriateness, Workability, Completeness, and Timeliness Appropriateness Group Judge 1 Judge 2 Judge 3 1 6 6 6 2 6 6 6 3 7 6 6 4 7 4 6 5 7 4 6 6 7 6 7 7 6 4 5 8 6 4 6 9 6 4 5 10 6 4 5 Workability Group Judge 1 Judge 2 Judge 3 1 6 6 6 2 6 5 6 3 7 6 5 4 7 4 6 5 6 4 5 6 7 6 6 7 6 4 5 8 6 4 6 9 6 6 5 10 6 3 6 Completeness Group Judge 1 Judge 2 Judge 3 1 4 6 4 2 5 3 4 3 4 4 4 4 7 3 5 5 6 1 4 6 7 2 7 7 5 3 5 8 6 2 5 9 4 5 3 10 5 4 4 Timeliness Group Judge 1 Judge 2 Judge 3 1 4 5 6 2 4 2 4 3 4 6 4 4 6 1 4 5 6 1 5 6 6 1 6 7 2 2 4 8 2 1 4 9 2 2 4 10 6 4 5 Judge 1 Judge 2 Judge 3 Judge 1 1.0000 .2646 .2709 (40) (40) (40) p = * p = .099 p = .000 Judge 2 .2636 1.0000 .2949 (40) (40) (40) p = .099 p = * p = .065 Judge 3 .6709 .2949 1.0000 (40) (40) (40) p = .000 p = .065 p = * *Value could not be computed. utilize all the observable behaviors. The observed frequencies for each of the twelve cells were compared to the theo- retical frequencies that would be ex- pected if no relationship existed between fective.� The eighty-one behavioral cat- egories within FOIAS were collapsed into the six major functions, and a 2 x 6 con- tingency table was created. A �miscella- neous� function was included so as to 164 College & Research Libraries March 1997 the two variables. The resultant x2 was found to be significant (x2 = 41.814; df = 5; p <.001). This indicated that a relation- ship does exist between the performance of the general discussion functions and group decision-making effectiveness. A test of the strength of the relationship using Cramer�s V (V = 0.145) indicated that the relationship was a weak one. Table 3 summarizes the results of the Chi- square analysis. Establishing that a relationship exists between the general discussion functions and group decision-making effective- ness, the data were further analyzed to determine the nature of that relationship. An attempt was made to determine the extent to which the general discussion functions were related to group decision- making effectiveness by employing the statistical model of correlation analysis. To compensate for the varying length of the ten-group discussions, the observed frequencies for each of the six general functions were converted into propor- tions. The discussions ranged from 156 to 247 units of behavior. The �miscella- neous� function was again included in the analysis. Taking one function at a time, the observed frequencies for that function were correlated with the group effectiveness scores using the Pearson product-moment correlation statistic (r). The resulting six correlation coefficients were tested for statistical significance to identify the strongest relationships. A sig- nificance level of .10 was established for all tests. Table 4 summarizes the observed frequencies for each of the six general functions, the corresponding effective- ness scores, and the resulting Pearson product-moment coefficients. The results indicate that a statistically significant negative relationship exists between �Es- tablishment of operation procedures� (Function 1) and group decision-making effectiveness. This suggests that the more group members attempted to establish operational procedures, the more likely they were to come up with an ineffective decision. An attempt was then made to deter- mine which specific communicative be- haviors within those functions accounted for the overall relationships observed. The observed frequencies were again converted to proportions. The observed frequencies for each of the eighty-one be- havioral categories in FOIAS were cor- TABLE 3 Contingency Table for Chi-Square Analysis Functions Row 1 2 3 4 5 6 Totals High Effect. Observed freq. 19 138 14 491 117 9 788 Theoretical freq. 19.68 177.52 8.27 433.75 132.65 16.14 Low Effect. Observed freq. 31 313 7 611 220 32 1,214 Theoretical freq. 30.32 273 12.73 668.25 204.35 24.86 Column totals 50 451 21 1,102 337 41 2,002 x2 = 41.814 v = 0.145 df = 5 p <.001 There appears to be a fairly strong negative relationship between Function 1 (�Establishment of operating procedures�) and group effectiveness within the �low� effective groups. Small Group Interaction 165 related with the group effectiveness scores using the Pearson product-mo- ment correlation statistic (r). Table 5 sum- marizes the results of the correlation analysis. A .15 level of significance was established for all tests. Only those cor- relation coefficients that were significant, or close to it, are reported in the table. An attempt was made to determine whether those relationships were consis- tent enough to allow the functions to serve as �predictors� of deci- sion-making effective- ness. To make this de- termination, the data were analyzed using the statistical model of multiple discriminant analysis. The six major discussion functions were the discriminating variables and were en- tered in the analysis in a stepwise fashion based on the discrimi- nating power of the variable. Wilk�s lambda was used as the TABLE 4 Correlations between Major Functions and Group Effectiveness Scores Group Score F1 F2 F3 F4 F5 F6 6 68 .00 .17 .00 .64 .08 .04 1 65 .03 .11 .05 .71 .10 .00 3 63 .01 .30 .02 .43 .23 .01 4 60 .00 .13 .00 .70 .17 .00 2 57 .00 .41 .00 .40 .19 .00 5 55 .00 .17 .00 .64 .18 .00 9 56 .05 .16 .00 .51 .22 .05 10 56 .02 .33 .01 .55 .06 .04 8 52 .06 .25 .02 .49 .17 .02 7 51 .06 .21 .00 .41 .30 .02 r(F1/Score) = -.5715 (p<.084) r(F4/Score) = .4120 (p<.237) r(F2/Score) = -.2701 (p<.450) r(F5/Score) = -.5231 (p<.121) r(F3/Score) = .3415 (p<.334) r(F6/Score) = -.1648 (p<.649) stepwise criterion. The two subsets (�high� and �low� ef- fective groups) served as the �groups� variable in the analysis. Table 6 summarizes the results of the multiple dis- criminant analysis. The re- sults reveal one discriminant function with a canonical cor- relation of .8599 (p < .3471). Based on the calculations of Wilk�s lambda, the best dis- criminating variable appears to be Function 4 (�Generation of alternative solutions�) (Wilk�s lambda = .260573; p < .3471). The discriminant function was found to suc- cessfully classify 90 percent of the known cases. It appears that �high� and �low� effec- tive groups can be discrimi- nated by a linear combination of Function 4. Having discovered that Function 4 can be used to predict �high� and �low� ef- fective groups, multiple discriminant analysis was again used to attempt to de- termine whether the relationship was consistent enough to allow Function 4 to serve as a �predictor � of decision-mak- ing effectiveness. This time, the discrimi- nant function correctly classed 80 percent TABLE 5 Expanded Correlation Analysis between Behavioral Categories and Group Effectiveness Behavioral Major Code Description Function r p 101 Introduction of fact 1 -.6674 .035 112 Ask for facts 1 -.6369 .048 114 Ask for consent/approval 1 .7848 .007 208 Substantiate an opinion 2 -.5065 .135 209 Stated agreement 2 -.7514 .012 408 Substantiate an opinion 4 .4870 .153 411 Summary/synthesis 4 .7862 .007 511 Summary/synthesis 5 -.5166 .126 516 Ask for summary/synthesis 5 -.6835 .029 166 College & Research Libraries March 1997 of the known classes. The results indi- cate that Group 3 should fall in the �low� effective subset, and Group 5 should fall in the �high� effective subset. Table 7 sum- marizes the results of the multiple dis- criminant analysis. Knowing that Function 4 (�Generation of alternative solutions�) can be used to discriminate between �high� and �low� effective groups, an attempt was made to discover why 20 percent of the known cases were incorrectly classified by the discriminate function. Using the fre- quency proportions, each major discus- sion function was correlated with the group effectiveness scores within each of the two subsets. The resulting twelve Pearson product-moment correlation co- efficients are summarized in table 8. There appears to be a fairly strong negative rela- tionship between Function 1 (�Establish- ment of operating procedures�) and group effectiveness within the �low� effective groups. This was the only relationship found. It appears that the discriminant function incorrectly classified 20 percent of the cases because the direction and strength of the relationships between the TABLE 6 Summary of Multiple Discriminant Analysis Step Variable Wilk’s Number Entered Lambda Probability 1 F4 .72563 .1202 2 F1 .65458 .2269 3 F3 .50824 .2252 4 F2 .31490 .1510 5 F6 .26058 .2236 6 F5 .26057 .4169 Number Canonical Wilk’s Removed Eigenvalue Correlation Lambda x2 df p 0 2.8377 .8599 .260573 6.724 6 .3471 Standard Discriminant Unstand. Discriminant Variable Function Coefficient Function Coefficient F1 2.54245 105.0800612 F2 2.26023 24.1930890 F3 -1.59288 -100.1188734 F4 1.67319 15.6536852 F5 -.04120 -.5517650 F6 -.70543 -34.0934141 Constant -14.6372858 Actual Groups Predicted Group Membership No. of Subset Code Cases Subset 1 Subset 2 High 1 4 4 0 100.0% 0.0% Low 2 6 1 5 16.7% 83.3% 90 percent of the known cases were correctly classified. Small Group Interaction 167 TABLE 7 Summary of Multiple Discriminant Analysis Step Variable Wilk’s Number Entered Lambda Probability 1 F4 .72563 .1209 Number Canonical Wilk’s Removed Eigenvalue Correlation Lambda x2 df p 0 .3781 .5238 .72562 2.405 1 .1209 Standard Discriminant Unstand. Discriminant Variable Function Coefficient Function Coefficient F4 1.0000 9.3556054 Constant -5.1268718 Actual Groups Predicted Group Membership No. of Subset Code Cases Subset 1 Subset 2 High 1 4 3 1 75.0% 25.0% Low 2 6 1 5 16.7% 83.3% 80 percent of the known cases were correctly classified. major discussion functions and group de- cision-making effectiveness are not con- sistent across the two subsets. The results of the investigation indi- cate a rejection of the null hypotheses. Systematic relationships do exist be- tween the frequencies of certain com- munication behaviors and group de- cision-making among professional li- brarians, and communication patterns of interaction within effective decision- making groups are different from those within ineffective decision-making groups made up of professional librar- ians. In particular, a negative relation- ship was discovered between group d e c i s i o n - m a k i n g e f f e c t i v e n e s s and communication behaviors that estab- lished operating procedures (Function 1). A positive relationship was discovered between group decision-making gener- ating alternative solutions (Function 4). Though these two relationships were found to vary in both strength and di- TABLE 8 Correlations between Major Functions and Group Effectiveness Scores within Subsets Subsets Functions High Low 1 r = .1400 r = -.7347 p = .860 p = .096 2 r = -.0348 r = .4078 p = .965 p = .422 3 r = .1257 r = -.3444 p = .874 p = .422 4 r = .0304 r = .2290 p = .292 p = .663 5 r = .7077 r = -.5180 p = .292 p = .293 6 r = .7846 r = -.1329 p = .215 p = .802 rection across the �high� and �low� ef- fective groups, they were consistent 168 College & Research Libraries March 1997 enough to allow the two functions to serve as fairly accurate predictors of group decision-making effectiveness. Discussion The purpose of the present study was to examine the relationships between cer- tain communicative behaviors and group decision-making among professional li- brarians, and communication interaction patterns within effective decision-mak- ing groups and ineffective decision-mak- ing groups consisting of professional li- brarians. Using content analysis of actual group discussions, the study explored the distribution of interactions found be- tween effective and ineffective decision- making groups. One finding of the study is that a nega- tive relationship existed between behav- iors that function to �establish operating procedures� and group decision-making effectiveness. Specifically, the more group members introduced facts regarding pro- cedural directions and asked for facts re- garding procedural directions, the less likely they were to reach an effective de- cision. The results suggest that although structuring and organizing comments are important, an excessive amount of pro- cedural comments may be counterpro- ductive to effective group decision-mak- ing (Hirokawa, 119). This may be an ex- planation for the observed negative rela- tionship between procedural comments and group decision-making effective- ness. Once the group has discussed and decided upon an appropriate way to ap- proach the task facing it, additional pro- cedural comments may be counterpro- ductive to effective group decision-mak- ing (Hirokawa, 119). This is likely the ex- planation for groups consisting of pro- fessional librarians. The study found little evidence to sup- port the relationship between group de- cision-making effectiveness and commu- nication behavior that functions to ana- lyze the problem facing the group. The finding appears to contradict the views of many decision-making scholars. John Dewey, James H. McBurney, Kenneth G. Hance, and Robert F. Bales all suggest that groups need to assess the problem to be effective.25 The results suggest that the lack of re- lationship between analysis of the prob- lem and effective decision-making does not disprove the importance of group members� attempts to analyze and un- derstand the problem facing them. Schol- ars offer no evidence to support the as- sumption that the more group members attempt to analyze and understand the problem under consideration, the more effective the groups� decision is likely to be.26 Because the task presented for the groups to discuss in the present study was one that would typically occur in the work setting, the librarians were aware of the problem. Although none of the groups had developed training programs prior to the discussion, they were aware of the need to develop such a program. It is understandable, then, that the present study found no consistent rela- tionship between the frequency of com- municative behavior that functions to analyze the problem under consideration and group decision-making effective- ness. This study found that group decision- making effectiveness was unrelated to the attempts of group members to establish criteria for evaluating the alternative so- lutions. Hirokawa suggests that the pro- cess of establishing criteria for evaluat- ing alternative solutions may not involve the direct discussion of such criteria (Hirokawa, 125). In the present study it The study found little evidence to support the relationship between group decision-making effective- ness and communication behavior that functions to analyze the problem facing the group. Small Group Interaction 169 is not unreasonable to speculate that the group members generally understood what criteria were appropriate for the evaluations of alternative solutions to problems and discussion of those crite- ria was not necessary for effective deci- sion-making. Also, it would not be un- reasonable to speculate that the groups do not need such criteria to be ranked as effective. The professional librarians were aware of the needs of their particular li- brary and staff, and therefore understood solutions presented during the discus- sion. This study found a relationship be- tween group decision-making effective- ness and certain communicative behav- iors that function to help a group gener- ate alternative solutions and suggestions. The results of the discriminant analysis showed that Function 4, which is the gen- eration of alternative solutions, could serve as a predictor of decision-making effectiveness 80 percent of the time. Spe- cifically, the more group members sub- stantiated an opinion and �summa- rized,� the more likely they were to reach an effective decision. The relationship between group deci- sion-making effectiveness and group members� attempts to generate alterna- tive solutions and suggestions, as indi- cated with the discriminate analysis, is generally consistent with small- group literature (Hirokawa, 128). The function also seems to confirm that the more group members consider alternative so- lutions, the more likely they are to select the �best� alternative for their final deci- sion. The study supports advocates of �brainstorming� sessions that utilize an idea-generating process which encour- ages any and all alternatives (Hirokawa, 128). Although there was an indication that the generation of alternative solutions could be used to predict effective deci- sions, there is a negative relationship (though not statistically significant) be- tween group decision-making effective- ness and communicative behaviors that function to evaluate alternative solutions. The possibility exists that a lot of evalua- tion comments may be counterproduc- tive to effective group decision-making. In this study, there was no significant difference between the two groups. This could indicate that all the participating librarians were aware of the needs of their libraries and staffs. The purpose of the research was to in- vestigate whether groups consisting of professional librarians would demon- strate the certain communication behav- iors and patterns of behavior as set forth in Hirokawa�s Vigilante Interaction Theory. The results provide support for Hirokawa�s findings that the group de- cision performance of groups made up of professional librarians is directly re- lated to a group�s efforts to perform criti- cal vigilant decision-making functions as investigated in the first question. The findings of the study indicate that com- munication interaction patterns are uniquely different between �high� and �low� effective groups as investigated in the second question. Specifically, groups that spend more time on attempting to establish operational procedures are less likely to develop effective decisions, and if groups show a pattern generating more alternative solutions, they are more likely to develop effective decisions. The present study was an exploratory study designed to examine the possible relationship between certain communi- cation behaviors and patterns of behav- ior and group decision-making effective- ness of small groups consisting of pro- fessional librarians. The objective of the investigation was to provide information that would test Hirokawa�s Vigilant In- teraction Theory. The study examined small groups in the field as suggested by Hirokawa and others. A logical extension of this study would be to examine a larger sample of groups consisting of professional librarians to further substan- tiate that group decision performance is 170 College & Research Libraries March 1997 directly related to a group�s efforts to perform critical decision-making func- tions. Another extension of this study could be to examine several studies with ten groups each doing a meta-analysis. Another extension would be the com- parison of this study with a study of small groups consisting of professionals from different fields. Further research should be conducted to determine whether the observed relationships in this study will hold up under different research conditions. The findings of this study help sup- port Hirokawa�s theory. They demon- strate that certain communicative behav- iors and patterns of behaviors do have an impact on group decision-making ef- fectiveness. As professional librarians, we must be aware of these behaviors and use them to improve group decision- making in our own institutions. Notes 1. Robert S. Taylor, �Question-Negotiation and Information Seeking in Libraries,� College & Research Libraries 29 (May 1968): 178�94; Bernard F. Vavrek, �Communications and the Reference Interface� (Ph.D. diss., Univ. of Pittsburgh, 1971); Helen Gothberg, �Immediacy: A Study of Com- munication Effect on the Reference Process,� Journal of Academic Librarianship 2 (July 1976): 126�29; Mary Jo Lynch, �Reference Interviews in Public Libraries,� Library Quarterly 48 (Apr. 1978): 119�42. 2. Marynelle DeVore-Chew, Brian Roberts, and Nathan M. Smith, �The Effects of Reference Librarians� Nonverbal Communications on the Patrons� Perceptions of the Library, Librarians, and Themselves,� Library and Information Science Research 10 (Oct. 1988): 389�400. 3. Jinnie Y. Davis and Stella Bentley, �Factors Affecting Faculty Perceptions and Academic Libraries,� College & Research Libraries 40 (Nov. 1979): 527�32; Marian Shilstone, �Faculty and the Academic Library,� Connecticut Libraries 21 (spring 1979): 39�40; Elizabeth Kleinhans, �Commu- nicating with Classroom Faculty,� Urban Academic Librarian 3 (spring 1985): 11�18. 4. Daniel Callison, Judy Fink, and Greg Hager, �A Survey of Cooperation and Communica- tion between Public and School Librarians in Indiana and Beyond,� Indiana Libraries 8, no. 2 (1989): 78�86. 5. Deborah Campana, �Information Flow: Written Communication among Music Librarians,� Notes 47 (Mar. 1991): 687�707. 6. T. Whitehall, L. G. Durbidge, and A. J. Meadows, �Communication between Library Prac- titioners,� Library Review 38, no. 1 (1989): 7�33. 7. Dennis S. Gouran, �The Paradigm of Unfulfilled Promise: A Critical Examination of the History of Research of Small Groups in Speech Communication,� in Speech Communication in the 20th Century (Carbondale: Southern Illinois Univ. Pr., 1985), 90�108; John F. Cragan and David W. Wright, �Small Group Communication Research of the 1980s: A Synthesis and Critique,� Commu- nication Studies 41 (fall 1990): 212�36. 8. Randy Y. Hirokawa, �A Function-Oriented Analysis of Small Group Interaction within Effective and Ineffective Decision-Making Groups: An Exploratory Investigation� (Ph.D. diss., Univ. of Washington, 1980). 9. Loyd H. Goodall, Small Group Communication in Organizations, 2nd ed. (Dubuque, Ia.: W. C. Brown, 1990). 10. Stephen Littlejohn, Theories of Human Communication, 4th ed. (Belmont, Calif.: Wadsworth, 1992). 11. Ernest G. Bormann, Small Group Communication: Theory and Practice (New York: Harper & Row, 1990). 12. Cragan and Wright, �Small Group Communication Research of the 1980s,� 212�36. 13. Randy Y. Hirokawa and Kathryn M. Rost, �Effective Group Decision Making in Organiza- tions: Field Test of the Vigilant Interaction Theory,� Management Communication Quarterly 5 (Feb. 1992): 267�88. 14. Ibid., 269. 15. Steven A. Beebe and John T. Masterson, Communicating in Small Groups: Principles and Prac- tice, 4th ed. (New York: HarperCollins College Publications, 1994). 16. B. Aubrey Fisher and Donald G. Ellis, Small Group Decision Making: Communication and the Group Process (New York: McGraw-Hill, 1990). 17. Littlejohn, Theories of Human Communication. Small Group Interaction 171 18. Charles Pavitt and Elen Curtis, Small Group Discussion: A Theoretical Approach (Scottsdale, Ariz.: Gorsuch Scarisbrick, 1990). 19. Ibid. 20. Fisher and Ellis, Small Group Decision Making; Pavitt and Curtis, Small Group Discussion; Beebe and Masterson, Communicating in Small Groups. 21. Fisher and Ellis, Small Group Decision Making. 22. Littlejohn, Theories of Human Communication. 23. David P. Fisher, �Is the Librarian a Distinct Personality Type?� Journal of Librarianship 20 (Jan. 1988): 36�47; Sandra M. Black, �Personality�Librarians As Communicators,� Canadian Li- brary Journal 38 (Apr. 1990): 65�71; Indra David and Mary Jane Scherdin, �Librarians in Transi- tion: Profiles of the Strong Interest Inventory,� in Discovering Librarians: Profiles of a Profession (Chi- cago: ACRL/ALA, 1994), 102�21. 24. Hirokawa, �A Function-Oriented Analysis of Small Group Interaction,� 53. 25. John Dewey, How We Think (Lexington, Mass.: Heath, 1910; reprint Buffalo, N.Y.: Prometheus Bks., 1991); James H. McBurney and Kenneth G. Hance, The Principles and Methods of Discussion (New York: Harper, 1939); Robert F. Bales and Fred L. Strodtbeck, �Phases in Group Problem- Solving,� Journal of Abnormal and Social Psychology 46 (Oct. 1951): 485�95. 26. Hirokawa, �A Function-Oriented Analysis of Small Group Interaction,� 124.