College and Research Libraries Bibliographic Verification for Interlibrary Loan: Is It Necessary? JoAnn Bell and Susan Speer The model Interlibrary Loan Code states that "All items requested shall be verified ... and sources of verification cited." During a three-month study involving approximately 900 re- quests, the Health Sciences Library at East Carolina University examined the necessity and effectiveness of prerequest bibliographic verification. This paper reports the results of there- search and how interlibrary loan policies might be affected. ibliographic verification is a standard interlibrary loan (ILL) practice required by the ALA Interlibrary Loan Code. A veri- fied request is assumed to reduce the time necessary for the lending library to locate the needed item, thus reducing the cli- ent's wait. However, the verification pro- cess appears to require excessive time, be- cause the borrowing library must attempt to verify every request before sending it to another library. This step can be especially time-consuming for those articles pub- lished so recently that they are not yet in- dexed. In the case of new journal titles, even the journal itself may not be indexed. Further, it appears to the verifier that al- most all of the citations are already correct when submitted. Since no published re- search on this topic could be found, the Health Sciences Library undertook a study to determine the degree to which in- terlibrary loan requests were bibliographi- cally correct and complete when submit- ted. Also, could a trained verifier identify incorrect or incomplete requests and thereby limit verification to those items re- ally needing it? The study focused on seven questions. 1. How many interlibrary loan requests are correct upon receipt? 2. Which bibliographic elements are most likely to be incorrect or incomplete? 3. Are some elements more critical than others to the ability of the lending library to supply an interlibrary loan without ex- pending extra effort? 4. How many requests needed critical elements added/corrected during the veri- fication process? 5. What errors occur? a. Are there patterns associated with occurrence of errors? b. Are some departments or clients more likely to submit requests with critical errors than others? c. Does the age of the publication in which the request appears relate to the error rate? d. Is the error rate significantly lower for citations for which a source is given? e. Are requests with one error likely to have additional errors? f. Do errors occur more frequently in ]o Ann Belt is Director, and Susan Speer is Systems Librarian, formerly Circulation Librarian, at Health Sci- ences Library, East Carolina University, Greenville, North Carolina 27858. 494 Bibliographic Verification for Interlibrary Loan 495 some bibliographic elements than in others? 6. What is the mean time needed to ver- ify requests, and is there a statistically sig- nificant difference in the fill time between verified and unverified requests? 7. Is it possible for the interlibrary loan technician to identify accurately those re- quests requiring verification? METHODOLOGY Library clients submit photocopy and interlibrary loan requests on individual re- quest forms. They are not required to ver- ify requests prior to submitting to the Health Sciences Library (HSL) at East Car- olina University. Many clients submit photocopies or cutouts from online searches on the ILL request forms. These requests are .not verified. An attempt is made to verify all other requests prior to sending them to another library. A form was designed to attach to the back of each ILL request to simplify there- cording of data as the request was pro- cessed. The dates on which the request was received from the client, verified, mailed to the lending library, and received were recorded, as were the sources searched in the verification process and the bibliographic elements on each re- quest that were su_pplied or corrected dur- ing verification. FINDINGS During the period of the study~ 835 in- terlibrary loan requests were received. Of that number, 217 came from computer 835 217 110 508 TABLE 1 ANALYSIS OF REQUESTS RECEIVED Requests received From computer search Unable to verify Verified searches; these requests were not ana- lyzed in the study, because we consider such requests already verified. This study dealt with the 618 remaining requests. . Occu"ence of E"ors and Missing Elements Of the 618 requests, 110 were never veri- fied, although they were searched in both manual and online sources. Therefore, for almost 18 percent of the requests, the time spent attempting verification did not im- prove the citations submitted. Our analy- sis of errors and missing data elements was confined to the remaining 508 verified requests. There were 200 requests "":ith one or more bibliographic elements miss- ing, and forty-two had one or more errors corrected. In addition, seventy requests had only part of the article title g_iven .. Of course, in some cases requests with miss- ing data also contained errors. There were 242 (47.6 percent) of the requests w~th a~ error, missing data, or part of the article ti- tle missing; approximately 52 percent were correct as submitted. To train interlibrary loan technicians to spot errors in requests, it is impo~tant to identify those request elements _likely to be incorrect or incomplete. A review of ta- ble 3 reveals that the month and the issue number were most often missing and that the article title was the most frequently in- complete item. The journal title was most often incorrect, but this was true for les:; than 4 percent of the requests. Critical Bibliographic Elements The importance of the various biblio- graphic elements varies. Therefore, an er- ror in the second author's name-or for that matter in the first author's name-is not as important as an error in t~e _journal title. Further, some errors or omissiOns are unimportant if other key elements are ac- curate and complete. For the purposes of this study, a critical bibliographic element was defined as o~e that if missing or incorrect would result m the lending library spending more time lo- cating the article than would have been re- quired had that element been present ~nd/ or correct.lt was decided that, for optrmal efficiency, the lending library must have the author's last name or the title of the ar- ticle the title of the journal, the volume and the issue or month in which the article appeared and the beginning page. How- 496 College & Research Libraries November 1988 TABLE2 DISTRIBUTION OF "ERRORS" IN INTERLffiRARY LOAN REQUESTS Percent of Requests Number of Errors With Missing Missin~ Part Total with Per Request Elements With Errors ofT1tle Corrections 0 60.6 91.7 86.2 52.3 1 17.3 6.5 13.6 18.9 2 13.6 1.4 0.2 14.2 3 7.1 0.2 0.0 11.0 4 1.0 0.0 0.0 2.0 5 0.2 0.2 0.0 1.4 7+ 0.2 0.0 0.0 0.2 TOTAL 100.0 100.0 100.0 100.0 TABLE3 DISTRIBUTION OF CORRECTIONS AMONG BffiLIOGRAPIDC ELEMENTS Bibliographic Element Journal title Author: last name Author: first name Second author Volume Issue Month Year Beginning page Article title Corrected (%) 2.8 1.6 0.6 0.2 1.0 0.6 0.4 0.8 2.6 0.4 0.2 1.8 9.1 3.5 1.0 19.1 22.4 2.6 1.6 10.8 Minor change made (%) 0.2 0.2 0 0 0 0 0 0 0.4 13.4 TABLE4 ANALYSIS OF CORRECTED AND MISSING/KEY BffiLIOGRAPIDC ELEMENTS Incorrect Bibliographic Element (%) No incorrect/missing element 93.4 Journal title 2.8 Beginning page 2.6 Author last name/title Volume/ 1.0 Issue 0.2 Month ever, not all of these elements may be nec- essary for locating an article. Minimally the lending library needs only the au- thor's last name or the title of the article, the title of the journal, the volume, and the beginning page. It was hypothesized that although the lack of an issue number or the month might slow up location in journal volumes Combination of Missing (%) Error/Missing (%) 62.2 98.0 0.2 1.6 0.4 0.4 1.0 18.9 0.6 15.7 1.0 not continuously paged, these elements are not always necessary because most journals are continuously paged. To test this assumption, the fill time for unveri- fied requests for which the issue/number/ date was missing was compared with the fill time for the verified requests. The mean time required for lending libraries to supply requests for which the issue and/ or Bibliographic Verification for Interlibrary Loan 497 0PTIHAL [LEHENTS HINIHAL [LEHENTS AUTHOR'S LAST NAME OR AUTHOR'S LAST NAME OR TITLE OF ARTICLE TITLE OF ARTICLE TITLE OF JOURNAL TITLE OF JOURNAL VOLUME VOLUME ISSUE NUMBER OR MONTH BEGINNING PAGE NUMBER BEGINNING PAGE NUMBER FIGURE 1 Key Bibliographic ElemP.nts month was given was slightly more than sixteen days, while the mean fill time for requests without these elements was ap- proximately eighteen and one-half days. After the key bibliographic elements were selected, the requests were analyzed to determine how many contained errors or omissions in these key elements. One hundred and forty-six requests were missing at least one of the key biblio- graphic elements. Fifteen of these re- quests had an incorrect or incomplete journal title, and twenty-one had an incor- rect beginning page number or were miss- ing the page number. None of these "er- rors" was in the same request; therefore, thirty-six requests contained errors in these primary bibliographic elements. Other errors included incorrect or missing volume and issue numbers. "Among the requests with mini- mally complete citations, that is, cita- tions without the issue or month, only 8. 9 percent of all the citations re- · quired verification.'' Examining the requests in light of the definition of optimally complete citations, it was found that 28.8 percent were miss- ing at least one key bibliographic element. However, among the requests with mini- mally complete citations, that is, citations without the issue or month, only 8.9 per- cent of all the citations required verifica- tion. Patterns Associated with Occu"ence of E"ors Although there was interest in measur- ing the occurrence of errors in interlibrary loan requests, it was also important to de- termine if patterns are associated with the occurrence of errors. If such patterns exist, it should be possible to train an ILL assis- tant to spot the likely occurrence of errors, even unobvious ones. The authors exam- ined data on the clients and departments submitting requests, the age of the journal issue in which the requested article ap- peared, as well as the occurrence of errors in citations for which the source is given and the probability of multiple errors in a request when one error is detected. All requests in which one or more of the critical bibliographic elements was miss- ing or incorrect were examined. Almost a quarter of the errors in the critical ele- ments were made by one client: no other client had more than two requests with er- rors/omissions in critical elements. Fur- ther, it should be noted that the client with the highest number of critical errors is known as "usually submitting accurate, but incomplete requests." Slightly more than 10 percent of his requests contained 498 College & Research Libraries errors in critical data elements. Therefore, it may not be possible to make assump- tions about a given client's accuracy with- out closely monitoring the requests over a period of time. A similar conclusion can be drawn about departments. If one con- siders only the critical data elements, one department (that of the client already de- scribed above and supplemented by one additional client) accounted for 25 percent of the requests with errors in critical data elements. The occurrence of errors, omissions, and minor changes was analyzed by the age of the article. A one-way analysis of variance indicated that differences among the groups were significant. For the error rate the difference is significant at the .04 level and for missing data at the .007level. An exact linear relationship does not exist. As the age of the articles increases the number of citation errors or missing ele- ments increase, but for articles published more than ten years ago, the number of er- rors or missing elements decreases. Requests for which the citation sources were not given were significantly more likely to have errors and/or omissions in them. T -tests of the differences in the mean number of errors/ omissions per re- quest were significant at the .0009level. Referring to table 1 it can be seen that the greatest number of requests had only one missing or incorrect element. Therefore, it does not appear that requests with one er- ror are particularly likely to have another error. However, the occurrence of multi- ple missing items is likely. In many cases these multiple missing items consisted of the issue, month, and year of publication. That is, many requests lacked at least two if not all three of these elements. To assess the effectiveness of verifica- tion, the fill time needs to be examined for verified and unverified requests. Table 6 shows the mean processing times for those requests verified and those not veri- fied. The internal and external processing times were greater for the unverified re- quests. A t-test was performed to test for significant difference between the proc- essing times for verified and unverified re- quests. The differences in these times were significant at the .0001level for the November 1988 rJ) c ~ 1.5 jjj Ol c: ·c;; (/) ~ 5 (/) e w 0 .5 Qj .D E :J z • Element Missing 0 Error in Element @ 0 -.:.... __ ......._ __ ......._ __ _._ __ __. 2-3 4-6 7-10 10 ~ Age FIGURE2 Relationship of Age to Missing and/or Incorrect Elements time to verify the requests, the total inter- nal processing time, and the total process- ing time. However, the difference be- tween the time required for the lending library to supply unverified requests and verified requests was not significant. Verification Decision Accura.cy As each request was received, the inter- library loan technician indicated whether she felt that request should be verified. Ta- ble 7 shows the interlibrary loan techni- cian's decision and the number of errors/ omissions in each of these requests. The means are higher for those interlibrary loans the technician indicated should be verified. Further, the mean time searching to verify was almost double for those the interlibrary loan technician indicated should be verified. Her verification deci- sions in fact paralleled very closely our definition of a minimally accurate citation-that is, she did not consider the issue/month necessary. SUMMARY OF FINDINGS While almost 48 percent of the citations had an error or were missing data, approx- imately 71 percent met the standard of an optimally complete and correct citation, and only 8. 9 percent of the requests failed to meet the standard of a minimally com- plete and accurate citation. The is- Bibliographic Verification for Interlibrary Loan 499 TABLE 5 ANALYSIS OF ERRORS/MISSING ELEMENTS Total requests analyzed Requests with at least one bibliographic element* missing or incorrect Requests with a key bibliographic element missing or incorrect Requests failing to meet minimal citation standard Number % 508 242 146 45 47.6 28.8 8.9 *Bibliographic Element= Author's last name,t author's first name, second author, title of article,+ journal title,t volume,+ is- sue/month, t beginning page number, t year. tKey bibliographic element TABLE6 MEAN PROCESSING TIMES Time to verify request Total internal processing time External processing time Total processing time N Verified 1.43 Days 3.88 16.15 20.63 508 Not Verified 4.45 Days* 8.63 * 23.16 35.39 * 110 *The differences in these values are significant at the .OOOllevel TABLE7 ANALYSIS OF VERIFICATION DECISION Decision: Verify Decision: Do Not Verify sue/month was the element most often missing, and processing time for the lend- ing library was significantly greater for these requests. It also took longer to ac- quire unverified requests. Data were ex- amined for the requests that were not veri- fied to see if reasons could be identified for their fill time being so much longer. It was discovered that one case required 264 days for filling. By removing this case andre- computing the external processing time it was found that lending libraries required approximately sixteen days to process ver- ified requests and twenty-one days to pro- cess unverified requests. Therefore, even with the extreme case removed, the proc- essing time of unverified requests still re- Mean Number of Errors .20 .10 Mean Number of Missing Elements .97 .71 MeanTime Searching 3.85 1.77 quired a significantly longer time period than did the verified requests. The cost of verifying interlibrary loans for the library and the client is consider- able. The mean time to verify requests even when they are complete and accurate was 2. 9 minutes per request. Additional direct costs occur when it is necessary to use online sources to verify requests. CONCLUSION Whether the lending library really needs a complete bibliographic citation to supply the requested item is questionable. Would libraries realize a net savings of time if re- quests not obviously incomplete or inac- curate are sent without verification? 500 College & Research Libraries Would it be reasonable to submit requests that appear to be accurate and complete . without attempting to verify them and to submit requests that cannot be verified al- · though an attempt has been made to do so? The lending library would take no re- sponsibility for verifying requests and would not hesitate to return any requests that could not be located as cited by the borrowing library. The borrowing library would assume the consequences if the item could not be supplied or if the item received was not the one desired. Because electronic mail is widely used to send re- quests, the lending library could return : immediately any unlocated requests. The accuracy of error recognition is im- proved when the technician begins to search for a source for the requests. Incor- rect journal titles become obvious when one cannot find a lending library for a given title. Errors in volumes and years also become apparent at this point. Thus clients might not experience any greater delay with this approach than with the current approach that requires an attempt to verify all requests. The HSL has an understanding with the three other academic health science li- braries in North Carolina to process re- quests in this manner. This procedure has been followed for two years and no prob- lems have resulted from this modification. The borrowing library sends requests as soon as it can identify a holding library. It would be interesting to replicate this study with the modified procedure to de- November 1988 termine the. impact on fill time. Although modification of current proce- dures may appear beneficial, the ethical implications of these changes should be considered. Is it ethical for the borrowing library to pass on its work to the lending library? In some instances, even if the bor- rowing library has stated that it does not expect the lending library to search for re- quested items, personnel in that library may hesitate to "give up" on a request. Also, the lending library incurs direct ex- pense for returning incorrect requests to the borrowing library. There is also a positive side to procedure modification. All libraries could save the time now spent on searching requests that · appear to be accurate and complete and could limit the time spent attempting to verify any request. In many cases, clients would receive needed information sooner. Also, clients will find through ex- perience that submission of accurate and complete requests results in receipt of needed information more quickly. And, they are more likely to perceive the addi- tional time required when inaccurate and incomplete requests are submitted. At this time, individual libraries should reach agreement with each other to mod- ify the verification procedure on mutually acceptable terms. In this way the value of modifying these procedures can be evalu- ated widely but with open knowledge and acceptance of that modification by all par- ticipating parties.