College and Research Libraries More Benefits of Automation Malcolm Getz More benefits of an automated library system are described, measured, and valued in dollar terms. The benefits include the automated circulation system as well as an automated catalog and are measured from the library user's point of view. The dollar value of the annual benefit of the automated circulation system is substantial. Additional benefits of the automated catalog are found over those described in May 1987. n May, 1987, this journal re- ported the author's effort to chronicle ''Some Benefits of the Online Catalog. 1 In this essay, a measure of the benefit of the online circu- lation system is added and the first assess- ment is reviewed. As before, the goal is to develop and test methods for measuring how users benefit from changes in library services. If the method proves persuasive, then the benefits can be repeatedly mea- sured in a number of libraries and increase our understanding of these particular actions. In 1984, the Heard Library at Vanderbilt University used a card catalog and a key- sort card circulation system. In 1985, they introduced the online catalog called Acorn using NOTIS software, and in 1986, they introduced the online circulation system component of the integrated system. To measure the effect of these developments, the time required to find and charge a known item was observed before and after the change, as well as the success of li- brary users in finding specific items. The improvements are valued in dollar terms. LIBRARY USE Library users were asked to complete forms indicating how they used the li- brary. The survey form was revised from the previous years to allow users to record the nature of their use of the electronic tools now in the library. Surveys were conducted during randomly chosen two- hour time blocks of the 1987 Spring semes- ter, during day, evening, and weekend hours. In all, 1,125 responses were usable as shown in table 1. Table 1 indicates the proportion of re- spondents who engaged in finding and reading tasks. The proportion who en- gaged in finding tasks is higher in 1987 (51.6 percent) than in 1986 (41.9 percent) and the increase might be attributed to the growing popularity of the electronic sys- tem. The proportion of respondents who engage in reading tasks is also higher, but there is no ready explanation for this change. Table 2 compares the pattern of finding activities in 1986 with those observed in 1987. Catalog use increased sharply in the second full year of Acorn operation, as the system grew and users gained confidence in it. Tables 3 and 4 report changes in pat- terns of reading and other tasks. Table 5 reports the low card catalog use in 1987. The extent of catalog searching in 1986, Malcolm Getz is Associate Provost for Information Services and Technology at Vanderbilt University, Nash- c."ville, Tennessee 37203. Angela Maddux led the survey and statistical work underlaying this study, and her efforts are much appreciated. 534 early in the life of Acorn, is compared with its second year, 1987. In 1986, 17.6 percent of the respondents engaged in author-title searches of the catalog, while in 1987, 31.6 percent used Acorn for author-title catalog searching. The increase in known-item searching suggests that the electronic cat- alog has had a significant effect on how us- ers, especially students, use the library. A value is assigned to this increase below. Similarly, the subject searching has in- creased from 12.5 percent of visits to 23.1 percent of visits. ACCESS TIME Paul Kantor's ''Measure of Access Time by Simulation"2 method is again used to record the average time to locate a known item and charge it, given the citation, and a group of relatively inexperienced stu- dent users is again compared with a group of experienced student library employees. Table 6 compares the simulation times for 1986 with those for 1987. According to the results for inexperi- enced users on lines A and B of table 6, the catalog time looks lower in 1987 than in 1986. However, the 1986 observations had very high variances, SQ the test for the dif- ference in means is not valid. Perhaps in the second year of the electronic catalog even inexperienced users have somewhat more experience. Time to find the item in the stacks remains unchanged and noth- ing has changed to affect stack time. 3 The time to return from the stacks with the item and present it for charge falls by one minute, fourteen seconds from 1986 to 1987. This decrease is statistically signif- icant and reflects the introduction of the bar code-based automated circulation component of Acorn (i.e., readers need hot fill out cards). Time to check out the book, once it is presented to the library staff person, is unchanged by the auto- mated circulation system. The times for the experienced users, lines C and D of table 6, show no signifi- cant change except for the "return from stacks." Here experienced users save one minute, twenty-three seconds, on average from not having to complete a card to charge a book. This is 78.5 seconds on av- erage for experienced and inexperienced users. More Benefits of Automation 535 The times of the experienced with the inexperienced users can be compared. In 1987, with an integrated online catalog and circulation system, the time differ- ence at the catalog for a known-item search no longer differs between experi- enced and inexperienced users. The forty- eight-second average for the experienced users is not a statistically significant differ- ence from the fifty-four-second average for the inexperienced users. The conver- gence of search times for inexperienced and experienced users reflects the power of a well-designed electronic search tool. It allows inexperienced users to search with the speed of experienced users. Experienced users do record a thirteen- second shorter average time to return from the stacks than the inexperienced us- ers (one minute, thirteen seconds versus one minute, twenty-six seconds). Perhaps experienced users know their way around the library better. THE VALUE OF TIME SAVED Assuming that, on average, library us- ers save 78.5 seconds for each book charged from the library because of the au- tomated circulation system (the mean be- tween experienced and inexperienced us- ers), multiplying this number by the number of items charged and converting to hours gives the number of hours saved per year by not completing cards. By valu- ing the time at appropriate hourly rates as discussed in the 1987 essay, an annual dol- lar benefit of the automated circulation component of the integrated system can be computed. 4 Figure 1 shows the calcula- tion. Column A reports the number of per- sons entering the library per year, figures from the most recent year for which the entrance turnstiles give complete data. Column B repeats from table 4 the propor- tion of respondents who say they checked out a book during their visit. Column C multiplies column A by column B times 2.45 on the assumption that each user who charges out books averages 2.45 books per visit. Our estimate of the number of visits is divided by the books charged into the annual circulation figure to arrive at the 2.45 books charged per transaction. Column C, then, indicates the number of TABLE 1 LIBRARY USE COUNT OF RESPONDENTS U1 ~ 0'1 No. of patrons 1986 Finding 1987 Reading Other Oass related Research 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 Faculty 114 77 73 45 68 58 43 30 18 20 85 59 (") Percent 0 of visits 64.0 58.4 59.6 75.3 37.7 39.0 15.8 26.0 74.6 76.6 = Graduate 613 418 314 273 439 370 336 228 321 242 322 222 ~ QQ Percent ~ of visits 51.2 65.3 71.6 88.5 54.8 54.5 52.4 57.9 52.5 53.1 ~ U-Grad 769 571 227 224 594 474 257 227 569 490 93 91 ~ Percent fll of visits 29.5 39.2 77.2 83.0 33.4 39.8 76.6 85.8 12.1 15.9 ~ Other 194 59 94 39 115 46 92 32 45 24 91 36 e: n Percent ::r" of visits 48.5 66.1 59.3 78.0 47.4 54.2 23.2 40.7 46.9 61.0 ~ .... Total 1,690 1,125 708 581 1,216 948 728 517 973 776 591 408 0" ~ Percent e: of visits 41 .9 51.6 72.0 84.3 43.1 46.0 57.6 69.0 35.0 36.3 .... ~ fll TABLE2 FINDING TASKS z 0 Catalog Acorn < ~ Author/ Catalog Author/ Acorn Search Ask Total a No. of Title Subject Title Subject Periodical Librarian Total Fr=g Patrons Search Search Search Search Index or Staff Browse f:rnd~~~ 0" ~ 1986 1987 1986 1987 1986 1987 1987 1987 1986 1987 1986 1987 1986 1987 1986 1987 ~ Faculty 114 77 43 34 5 12 30 11 8 9 16 9 17 10 73 45 89 74 ~ I.C Percent of visits 37.7 44.2 4.4 15.6 39.0 14.3 7.0 11.7 14.0 11.7 14.9 13.0 64.0 58.4 78.1 96.1 (X) (X) Percent of finding tasks 48.3 45.9 5.6 16.2 30/6 11.2 9.0 12.2 18.0 12.2 19.1 13.5 Grad/Professional 613 418 137 189 98 116 182 103 118 73 74 53 90 64 314 273 517 495 Percent of visits 22.3 45.2 16.0 27.8 43.5 24.6 19.2 17.5 12.1 12.7 14.7 15.3 51.2 65.3 84.3 118.4 Percent of finding tasks 26.5 38.2 19.0 23.4 28.6 16.2 22.8 14.7 14.3 10.7 17.4 12.9 Undergraduate 769 571 80 120 83 118 114 110 64 36 91 64 65 79 227 224 383 417 Percent of visits 10.4 21.0 10.8 20.7 20.0 19.3 8.3 6.3 11.8 11.2 8.4 13.8 29.5 39.2 49.8 73.0 Percent of finding tasks 20.9 28.8 21.7 28.3 22.5 21.7 16.7 8.6 23.8 15.3 17.0 18.9 Other 194 59 37 14 26 18 11 10 18 6 27 8 26 15 94 39 134 61 Percent of visits 19.1 23.7 13.4 30.5 18.6 17.0 9.3 10.2 13.9 13.6 13.4 25.4 48.5 66.1 69.1 103.4 Percent of finding tasks 27.6 23.0 19.4 30.0 14.3 13.0 13.4 9.8 20.1 13.1 19.4 24.6 Total 1,690 1,125 297 355 212 260 347 235 208 125 208 134 198 168 708 581 1,123 1,047 Percent of visits 17.6 31.6 12.5 23.1 30.8 20.9 12.3 11.0 12.3 11.9 11.7 14.9 41.9 51.6 66.4 93 .1 Percent of finding tasks 26.4 34.1 18.9 25.0 26.3 17.8 18.5 12.0 18.5 12.9 17.6 16.1 TABLE3 READING TASKS Total No. of Own Reference Reserve Other Microform Government Total Reading Patrons Materials Materials Materials If~ary~~ Journals Materials Documents ~~nd;~~ Tasks 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 Faculty 114 77 19 13 20 20 4 1 14 10 31 30 4 1 5 1 68 58 97 76 Percent of visits 16.7 16.9 17.5 26.0 3.5 1.3 12.3 13.0 27.2 39.0 3.5 1.3 4.4 1.3 59.6 75.3 82.5 98.7 Percent of library materials 25.6 31.7 5.1 1.6 17.9 15.9 39.7 47.6 5.1 1.6 6.4 1.6 Grad/Professional 613 418 175 147 169 119 70 41 105 65 187 131 38 18 28 10 439 370 772 531 Percent of visits 28.5 35.2 27.6 28.5 11.4 9.8 17.1 15.6 30.5 31.3 6.2 4.3 4.6 2.4 71.6 88.5 125.9 127.0 Percent of library materials 28.3 31.0 11.7 10.7 17.6 16.9 31.3 34.1 6.4 4.7 4.6 2.6 Undergraduate 769 571 447 328 115 92 88 73 79 68 65 46 21 10 16 9 594 474 831 626 Percent of visits 58.1 57.4 15.0 16.1 11.4 12.8 10.3 11.9 8.5 8.1 2.7 1.8 2.1 1.6 77.2 83.0 108.1 109.6 Percent of library materials 29.9 30.9 22.9 24.5 20.6 22.8 16.9 15.4 5.5 3.4 4.2 3.0 Other 194 59 39 12 31 12 7 2 25 10 39 10 6 4 3 1 115 46 150 51 Percent of visits 20.1 20.3 16.0 20.3 3.6 3.4 12.9 16.9 20.1 16.9 3.1 6.8 1.5 1.7 59.3 78.0 77.3 86.4 Percent of library materials 27.9 30.8 6.3 5.1 22.5 25.6 35.1 25.6 5.4 10.3 2.7 2.6 Total 1,690 1,125 680 500 335 243 169 117 223 153 322 217 69 33 52 21 1,216 948 1,850 1,284 Percent of visits 40.2 44.4 19.8 21.6 10.0 10.4 13.2 13.6 19.1 19.3 4.1 2.9 3.1 1.9 72.0 84.3 109.5 114.1 Percent of library materials 28.6 31.0 14.4 14.9 19.1 19.5 27.5 27.7 5.9 4.2 4.4 2.7 TABLE4 ~ OTHER TASKS 0 """ Check ~ Number of Microform Int=ary Books Return = Patrons Photocopy Printer Out Books Other Total ~ 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 1986 1987 ::s ~ Faculty 114 77 18 16 0 0 10 5 13 13 5 6 6 2 52 42 ,..... .... .. Percent til of visits 15.8 20.8 0 0 8.8 6.5 11.4 16.9 4.4 7.8 5.3 2.6 45.6 54.5 0 ,..... Graduate 613 418 185 145 14 17 26 21 120 92 82 51 49 15 476 341 > Percent r:: .. of visits · 30.2 34.7 2.3 4.1 4.2 5.0 19.6 22.0 13.4 12.2 8.0 3.6 77.7 81.6 0 Undergraduate 769 571 174 154 11 5 6 3 83 76 50 53 16 "20 340 311 a Percent et. of visits 22.6 27.0 1.4 .9 .8 .5 10.8 13.3 6.5 9.3 2.1 3.5 44.2 54.5 s· Other 194 59 40 11 1 4 2 3 25 14 24 12 18 6 110 50 ::s Percent of visits 20.6 18.6 .5 6.8 1.0 5.1 12.9 23.7 12.4 20.3 9.3 10.2 51.5 84.7 Total 1,690 1,125 417 326 26 26 44 32 241 195 161 122 89 43 978 744 U1 Percent ~ ....:I of visits 24.7 29.0 1.5 2.3 2.6 2.8 14.3 17.3 9.5 10.8 5.3 3.8 57.9 66.1 538 College & Research Libraries November 1988 TABLES 1987 CARD CATALOG SEARCHES Author/ Number of Title Patrons Searches Faculty 77 8 Percent of visits 10.4 Percent of finding tasks 8.2 Graduate/Professional 418 35 Percent of visits 8.4 Percent of finding tasks 5.5 Undergraduate 571 21 Percent of visits 3.7 Percent of finding tasks 4.1 Other 59 4 Percent of visits 6.8 Percent of finding tasks 5.2 Total 1,125 68 Percent of visits 6.0 Percent of findings tasks 5.2 TABLE6 MEAN TIME TO FIND AND CHECK OUT ITEMS, ONE AT A TIME MINUTES: SECONDS (STANDARD DEVIATIONS ARE IN PARENTHESES) AtCatalo8 To Stacks Return Checkout I. Inexperienced Users A. Acorn, 1986 1:16 2:34 2:40 :45 (133) (146) ( 29) ( 42) B. Acorn, 1987 :54 3:12 1:26 :38 ( 44) (123) ( 25) ( 56) T - Statistic for difference 1.55 -1.65 . 13.7* 0.7 in means (degrees of freedom) ( 88) F - statistic in variances 4.47* 0.74 1.34 0.56 (degrees of freedom) (41,47) II. ExAerienced Users C. com, 1986 :39 2:25 2:36 :35 ( 23) (107) ( 45) ( 25) D. Acorn, 1987 :48 2:50 1:13 :34 ( 26) (190) ( 18) ( 29) T - statistic for difference -1.64 -.76 11.8* .17* in means (degrees of freedom) ( 82) F - statistic in variances 0.78 0.12 6.25* 0.74 (degrees of freedom) (35,47) m. T - statistic for comparison of inexperienced and experienced users E. Acorn, 1986 2.71 * 0.75 0.40 1.29 (compariny A and q (degrees o freedom) ( 76) ( 70) ( 70) ( 70) F. Acorn, 1987 0.90 0.70 2.93* 0.51 ~compariny B and D) ~ deS!ees o freedom} ( 72} ( 81} ( 84} ( 71} *Statistically significant at 0.01level. Subject SearChes 3 3.9 3.1 27 6.5 4.2 17 3.0 3.4 11 18.6 14.3 58 5.2 4.4 Total n 7:15 42 6:10 48 6:15 36 5:25 48 More Benefits of Automation 539 1986·1987 Value of tiine saved in "return from stacks" Faculty Grad/Prof Undergrad Other Total A. Visits/year 31,509 265,250 326,568 33,977 657,304 8. Book charge per visit 0.169 0.22 0.133 0.237 0.173 Seconds saved: 78.5 per book charged Books charged: 2.45 per visit c. . .. D. Books charged Hours saved per year 13,046 626 142,970 6,859 106,412 5,105 19,729 946 278,598 13,365 A. 1984·1985 attendance count, latest available annual figures. B. Respondents charging books/visit C. A. times B. times books charged per visit D. C. times 78.5 second$ per book charged e. F. Value of time Value of time for per hour checkout saved S25.00 $15,647 $7.50 S51 ,439 S5.00 S25,524 $7.50 $7,098 $8.63 $115,340 78.5 seconds is average savings for experienced and inexperienced users e. Roughly, wage rate for each group F. D. times e. G. e. times (78.5 seconds/3600) G. Value of time saved/charge $1.20 $0.36 $0.24 $0.36 $0.41 FIGURE 1 books charged to users per year. To find the number of hours saved, mul- tiply column C by 78.5 seconds per book and convert to hours. Column E repeats the hourly value of time used for each class in the 1987 essay on benefits. Column F multiplies the number of hours in D by the value per hour in E to arrive at the total dollar value of the time saved by library users through the automated circu- lation system. Figure 1 then reports a total benefit to users of the automated circulation system of $115,340 per year. This excludes poten- tial benefits in the law, medical, and music libraries because implementation varies in those libraries. Also excluded are the ben- efits of prompt overdue notification and the reporting of circulation status. This benefit is in addition to the benefits of the online catalog estimated in the 1987 essay. The benefit can be disaggregated by cate- gory of user as in column F of table 7. The benefit can also be represented on a per transaction basis as in column G of fig- ure 1, by using the system at the hourly rate of each class of user. For faculty, the time saved is worth about $1.20; for un- dergraduates, $0.24 per book charged. AVAILABILITY With the automated circulation compo- nent in use along with the automated ac- quisition tracking in the integrated sys- tem, the measure of availability can be traced more easily. The method involves asking persons using the catalog whether they are looking for a specific item. Each person who is, is asked to complete a form giving the author and title, call number if it is found, and to indicate whether the book is found. With the integrated system, are- port of the status of the book can be shown on the system. Figure 2 shows our modifi- cation of Paul Kantor's form. Figure 3 reports the measure of avail- ability for spring 1985, fall 1985, and for spring 1987. The conditional success rate is the proportion of users who success- fully complete a particular step in search- ing, given that the preceding steps are successful. For example, 92.2 percent of the users found items in the catalog when they in fact were described in the catalog; 7.8 percent did not find the items in the . catalog. Similarly, 77.3 percent found that items owned by the library had not been checked out; 22.7 percent looked for items that were in circulation. The product of the conditional success rates is the overall success rate. 5 The probability of a user finding a spe- cific item sought on the shelf at central li- brary was 57.5 percent in period one, 64.1 percent in period two, and 45.0 percent in period three. The sources of variation in . this aggregate measure are uncertain and will be explored in future investigation. Heretofore, the availability survey has been conducted at times convenient to the 540 College & Research Libraries $/Search Vo --- V' November 1988 --:=::::: ·-- . ·--·-· MC' MB 0 Qo O' Items Sought MB : Marginal benefit in dollar terms from one more unit of search, given that a particular amount of search has already been com- pleted . MC : marginal cost, primarily in user time, of additional units of search . MC' : the lower schedule of searching cost resulting from improvements due to automation. V0 : The value of the last search before automation . Q0 : the quantity of search performed before automation . V' : the value of the last unit of search given automation . Q ': the quantity of search performed given automation . The area of the trapezoid under MB from Q 0 to Q' : the value of the increased search performed as a consequence of the fall in cost (user time) to complete a search given automation . FIGURE2 Marginal Benefits and Costs of Search library staff members who must find out why users can not find the items they seek. In the future, randomly chosen two- hour time blocks over a semester will be surveyed so that all the users are equally likely to be included. Also in the future, we will ask participants to indicate their status: faculty, graduate and professional school student, undergraduate. In this way, conditional success rates for known- item searches by status of user will be available. The success rate for known-item searches by library users reflects both the . behavior of the library in selecting and managing its collection and the behavior of library users in seeking materials. The large increase in searching from 1985 to 1987 is associated with a decline in the rate of successes per search. The absolute number of successful searches increased markedly. The number of unsuccessful searches grew even more rapidly. To un- derstand the sources of change in the suc- cess rate, a more sophisticated description of both the library operations and of the users' searching behavior must be devel- oped. What determines the amount of searching by library users? In the 1987 essay, the only benefit from improved known-item searching confi- dently ascribed to Acorn was due to the fact that it was a union catalog. In the 1987 result, again a significant number of searches would have been failures if lim- ited to a card catalog of the central library's More Benefits of Automation 541 COUNT CONDITIONAL SUCCESS Spring 1985 Fall 1985 1987 Spring 1985 Fall 1985 1987 ============================================================================== 1. n (total-discards) 637 325 420 2. Found on shelf at Central 367 208 189 57.6X 64.0X 45.0X 3. Not found on shelf at Central 253 114 209 42.4X 35.1X 55.0X Illegible titles 17 22 Never ordered 72\ 22\ 45\ Ordered, not available 6 > 0 > 3 > 87.8X>82.4X 93. 5X>85. sx 88.6X>72.8X At other divisions 34/ 25/ 66/ I In catalog but missed 17 14 24 96.7X 95.2X 92.2X Checked out 51 30 64 89.9X 88.6X 77.3X On reserve or missing 65 22 20 85.7X 90.6X 90.8" On shelf but missed 25 9 93.6X 98.1" 95.4X Overall success at Central 57.5X 64.1X 45.0X Overall success at system 61.2X 70.1X 54.5X *Of 66 at other divisions, only 47 have potential to be hits, and if the percentage rate of hits at Central on available i terns is the same for the other divisions, then only 40 will be available. FIGURE 3 Measures of Availability collection, but that, in fact, are successes from the point of view of the university li- brary system. The gain in success rate due to the union catalog effect is the difference between the 45.0 percent success at cen- tral alone, versus the 54.5 percent rate if success is defined in terms of the whole system. Despite the underlying unex- plained variability in the conditional suc- cess rate, the benefit of the union catalog effest persists. The value of the increased success rate for known-item searches is estimated in Figure 4, assuming a value of one dollar for each extra success due to the union cat- alog effect as developed in the 1987 essay. The level of author and title searching is substantially higher in 1987, and so the level of benefit from the union catalog ef- fect is commensurately higher, namely, $28,505 if finding a book is worth $1. Given the variability of the known-item success rate, the average union catalog ef- fect over the three years might be our best measure of this benefit, that is 6.4 percent points increase in success. VALUE OF MORE SEARCHING The increased number of searches per- formed as searching becomes more con- venient with the electronic system might be valued in dollar terms. If the added searches were not performed under the older system, the users must have valued the searches at less than the time required to undertake them. The extra searches · now undertaken are worth more than the value of the time required to complete them. Because people allocate their time among competing activities in ways that reveal an implicit dollar valuation on time, a reasonable valuation of time in estimat- ing the value of the increased number of searches can be assumed. As noted above, the level of author-title searching in Acorn increased in the sec- ond year of service. For the base period re- ported in the 1987 essay, a total of 115,686 542 College & Research Libraries Value of Added Availabl ity Author/T iHe Searches Panel 1 Faculty Grad/Prof Undergrad Other Total Panel 11 November 1988 Acorn Author/T i tle Total Acorn Searches /Visit Author/T i tle Annual Visits f r om Table 2A Searches 31 , 509 0.390 12,289 265 , 250 0.435 115,384 26,568 0.200 5,314 33 , 977 0.356 12 , 096 657,304 0.308 202,450 Card Catalog Searches 1987 A/T Card Catalog Nl.mber of Card Search Rate 87 Catalog Searches Faculty 0.104 3,277 Grad/Prof 0.01!4 22,281 Undergrad 0 . 037 983 Other 0.068 2,310 Total 0 . 060 39,438 Panel 111 A. Acorn Author/ Ti tle Searc hes B. Times 2.2 items/search C. Three period average success rate Central Library 202,450 445,389 0 . 555 0 . 619 0.064 D. Three period average success rate Sys tem E. Increased success rate due to Union Catalog F. Increased l"lU\iler of successes-line 8 t i mes li ne E 28,505 S28, 505 G. Value of increased successes at $1 per succe s s Panel 1V Aco r n Car d Tota l 115,686 241,888 126,202 277 , 644 S45,811 A. Search sessions , 1986 B. Search sessions, 1987 202, 450 39,438 C. Change in search sess i ons D. Increased i tem searches , 2 . 2 pe r sessi on E. Value of added sea r che s at SO. 165 pe r i tem FIGURE4 author-title search sessions were per- formed per year. In 1987, the total number of author-title search sessions in Acorn is estimated at 202,450. In addition, 39,438 search sessions used the card catalog. As- suming 2.2 searches per session, the total number of search sessions per year in- creased by 126,202. With 2.2 items per ses- sion, that comes to 277,644 more items searched. The number of searches is com- puted in figure 4. 6 11In 1985, the time 'at catalog' was 84 seconds for inexperienced searchers. In 1987, the time 'at catalog' dropped to 54 seconds." One more author-title search is worth at least the time it takes to complete it. In 1985, the time II at catalog'' was 84 seconds for inexperienced searchers. If users val- ued their time at $8.63 per hour on aver- age, then 84 seconds is worth $0.20. In 1987, the time ''at catalog'' dropped to 54 seconds, worth $0.13, on average. In fig- ure 2, Vo is $0.20 and V is $0.13. The area of the shaded trapezoid reEresents the value of the added searches. 7 If the mar- ginal benefit curve is a straight line, the av- erage value of the added searches will be approximately $0.165 per search. In the base period, the 115,686 author- title search sessions searched for 254,509 items. In 1987, the 241,888 search sessions searched for 532,153 items. The increase of 277,644 items searched has a value of $45,811 when valued at $0.165. Notice that the searches are valued at $0.165 per j.tem whether successful or not. This benefit is in addition to the benefit of the improved success rate due to the union catalog ef- fect. The increased quantity of successful known-item searches undoubtedly leads to increased use of library materials and that would be a further benefit of the on- line system. At this point, there is no esti- mate of the increased use of materials. SUMMARY The benefits to users of the integrated Sll!lllary of Benefits Value of Time Saved per Year More Benefits of Automation 543 Total Annual Value per Unit Benefit A. Author/Title Searches 1987 essay, Table 8 $11,137 S0 . 10 B. Circulation This essay, Table 4 $115,340 S0 . 41 Value of Increased Availability c. Increased Nllllber of Searches D. Union Catalog Effect This Essay, Table 6 $45,811 $0.17 This essay, Table 6 $28,505 $1 .00 Subtotal Subject Searching No estimate Dial-up Searching no estimate S200, 793 FIGURES Summary of Benefits online system measured so far at Vander- bilt come to $200,793 per year as summa- rized in figure 5. The largest benefit de- rives from the time saved with the circulation system, $115,340 per year. Benefits of about $85,000 per year derive from measured gains in known-item searching. There may be further benefits in known-item searching that remain hid- den because of the inability to explain to a satisfactory degree the variatiott in the known-item success rate. Of course, there are other categories of benefit that are as yet unmeasured. Sub- ject searching constitutes about one third of all Acorn searching. If its value per search were similar to that for known item searches, another $40,000 or so per year of benefit might be found. There will be sig- nificant benefits from all searching done outside the library by use of data com- munication links to Acorn because such network searching may reduce the num- ber of visits to the library and so save users time. There may be benefits from improved management of the library and its collec- tions as we gain more complete informa- tion about patterns of use. The provision of online services delivered to individual desks across campus and beyond may in- duce an even more fundamental shift in the character of information services in a university setting, a link to nationally dis- tributed databases and document delivery that may reduce emphasis on local collec- tions. At this point, the benefits of the online system are significant. As users gain expe- rience with the online system, the level of benefit increases. Added functions add benefits as welL REFERENCES AND NOTES 1. Malcolm Getz, "Some Benefits of the Online Catalog," College & Research Libraries, 48, no.3:224-40. 2. Paul Kantor, Objective Performance Measures for Academic and Research Libraries .. (Washington, D. C.: Assn. of Research Libraries, 1984). 3. We can compare, for inexperienced users, the mean time of 84 seconds at the card catalog in 1985 with the mean time of 54 seconds in 1987, the second year of the electronic world. The variances are not different and the t-statistic to test whether these means are different is 3.25. The 30-second time savings from 1985 to 1987 is a statistically significant difference from zero. 4. Most readers will be willing to spend money to avoid extra time in locating materials. The wage rate is an approximation of how much they would be willing to spend. We, then, value time at the hourly wage rate for undergraduates, graduate students, and faculty. 5. Paul Kantor, Objective Performance. 544 College & Research Libraries November 1988 6. The 202,450 reported sessions with 2.2 items sought per session implies 445,389 items searched per year as implied by our survey. The Acorn system reported 647,074 author and titles search in the year, but 212,693 of these found no match. Acorn, then, reports 647,074 minus 212,693, namely 434,381 author title searches that found some match. The survey information and the sys- tem information, then, seem to show similar magnitudes of searching. 7. Figure 1 is the same as figure 3 in the 1987 essay.