Research Article   E-Preferred Approval Books at the University of Manitoba: A Comparison of Print and Ebook Usage   Jan C Horner Senior Scholar, Collections Management University of Manitoba Libraries Winnipeg, Manitoba, Canada Email: j.horner@umanitoba.ca   Received: 23 Aug. 2016  Accepted: 13 Feb. 2017      2017 Horner. This is an Open Access article distributed under the terms of the Creative Commons‐Attribution‐Noncommercial‐Share Alike License 4.0 International (http://creativecommons.org/licenses/by-nc-sa/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly attributed, not used for commercial purposes, and, if transformed, the resulting work is redistributed under the same or similar license to this one.     Abstract   Objective – To compare the usage of print and ebooks received on University of Manitoba’s e-preferred YBP approval plan as well as to examine cost per use for the approval print books and ebooks. Methods – Usage data was compiled for books received on approval in 2012/2013 to December 31, 2014. Counter reports were used to determine use and non-use of ebooks, while vendor reports from EBL and ebrary were used for the cost per use analysis. Print usage information was drawn from SIRSI and then ALMA when UML switched systems at the beginning of 2014. Results – Ebooks received more use than p-books overall, but when examined by subject discipline, significant differences could not be found for the “STM” and “Other” categories. With ebooks, university press books tended to be used more than those from other publishers, but the same result was not found for print books. Ebrary ebooks tended to be used more often than EBL, EBSCO, and Wiley ebooks, and single-licence books tended to be somewhat more used than multi-user ones. Cost-per-use data was much lower for print books, though the comparison did not look at staffing costs for each medium. Conclusions – This study finds that of approval books matching the same profile, ebooks are used more, but print books receive more substantial use. Both formats are needed in a library’s collection. Future comparisons of cost per use should take into account hidden labour costs associated with each medium. Usage studies provide evidence for librarians refining approval plan profiles and for budget managers considering changes to monographic acquisition methods and allocations.   Introduction   Libraries are changing the way they acquire monographs. Demand-driven and publisher front-list acquisition options are competing with traditional methods of firm ordering and approval plan delivery.1 Approval plans are intended to save staff time and can operate despite staff absences or changes. But libraries are looking more critically at approval plans since purchasing a publisher’s front list or using demand-driven acquisition can also save staff time.    This study looks at the performance of the YBP e-preferred approval plan at the University of Manitoba Libraries (UML). UML is an ARL and CARL member with an enrollment of approximately 30,000 and faculties of Agricultural & Food Sciences, Architecture, Arts, Engineering, Health Sciences, Law, Management, Music, and Sciences. It is the major research library for the province.   UML is facing budget challenges similar to those at other ARL libraries where monograph budget funds have decreased on average 4% over the last three years reported. UML’s monograph budget has decreased 8.5%.2 During the same period the financial commitment to demand-driven ebook acquisitions has remained steady and UML has purchased the Springer ebook collection and subscribed to other ebook collections (e.g. EBSCO, Knovel). Budget challenges were thus a strong motive to review the performance of UML’s approval plan.   Literature Review   This study is quantitative; therefore, qualitative studies of ebook versus print use, although of interest and value, were not included in the review.   Ebook v Print Comparisons   Most quantitative studies of ebook and print monograph usage have examined the same titles in both formats.  In addition, most of the early studies (e.g., Littman & Connaway, 2004) looked at Netlibrary titles as a basis of comparison with print. After 16 months of use the Littman & Connaway study at Duke University found 40% of the ebook versions had been used and 36% of the print (p. 259).   Kimball, Ives, and Jackson, in their study (2009), identified 4,288 Netlibrary books in the sciences that were also available in print. Although the ebooks were used more times per book, 24% of the ebooks were accessed while 23% of the print versions circulated (p. 23-24).   More recently Levine-Clark & Brown (2013) compared the use of Duke University Press titles online and in print published between 2009 and 2012. A greater number of print titles had been acquired before the ebook version, but there were 1,150 titles held in both formats. Of those titles, 54% of the print circulated and 39% of the ebooks had been used, leading the authors to conclude that when both formats were available, users preferred the print (slide 25). Goodwin (2014) looked at the 2011 collection from Duke University Press, 285 ebooks and 275 print (10 matching print titles had not been received by the study start). Usage was counted to October 2013 (two years, five months) and resulted in 73% of the ebooks being used and only 29% of the print, although only 12% of the ebooks received “substantive use.”   A Kent State study (Downey et al., 2014) did not compare the same titles in print and ebook. Rather it looked at the performance of 20,000 of the most recently acquired print books (up to January 2012) added between July 2009 and January 1, 2012 in relation to 20,018 discovery records which were loaded in their catalogue for a patron-driven acquisition (PDA) project that ran January to December 2012 using ebrary ebooks. By the end of 2012, 8% of the books in the PDA discovery pool had been used (p. 148) with only 2% triggering a purchase (p. 154), but a relatively high number (62.5 %) (p. 149) of the purchased print books had circulated.   Ebook Usage   Levine-Clark (2014), working with major ebook vendors EBL and ebrary, examined data for about 625,000 ebooks across roughly 800 academic libraries. This study could not be used to predict rates of use in a single library, since the titles had a massive pool of potential readers. It found that usage of Social Science ebooks was higher than that for Humanities and Science, Technology & Medicine (STM) titles in terms of percentage of titles used and average amount of use (slides 31-36). Patrons spent more time in Humanities ebooks per online session than for ebooks in the other disciplines. STM ebooks involved more downloading. EBL and Ebrary titles were examined separately. In general Humanities ebooks were used more than STM titles on ebrary, and STM titles were used more than Humanities ones on EBL (slides 31-32).   Comparisons of Ebook Types of Acquisition   Many authors have published reports on patron-driven or demand-driven acquisition (PDA, DDA) programs at academic libraries. However, because PDAs/DDAs count usage of never purchased titles as well, they are not relevant to the present study of titles acquired by approval in print and electronic format.   Carrico et al. (2015, pp. 106-107) compared the usage of ebooks acquired in large publisher packages with those that were firm-ordered by librarians. In terms of the packages, 50% were used compared to 52% of the firm-ordered titles. This figure was somewhat skewed by the use of the medical titles in which 63% of the package titles and 84% of the firm-ordered titles were used. For the Humanities/Social Sciences, package titles and firm orders were used 47% and 45% respectively. For Science & Technology package titles and firm orders were used 49% and 57% respectively.   Print Usage and Substantive Use of Ebooks   A number of studies have shown that the percentage of print collections that circulate is low. Rose-Wiles (2013) examined the circulation for 2005 to 2009 of the entire collection of 443,577 print books at Seton Hall University and found 21.5% had circulated (p. 137). However, only 17.7% had been published in the 2000s. She examined the subset of science books published since 2000 and found 34.5% circulated 2005 to 2009 (p. 141).  A Cornell University study (2010, p. 2) looked at the circulation of its collection published between 1990 and 2010 (1.6 million titles) and found that 45% had circulated.   A 2009 study (Alan et al.) looked at the usage of print titles acquired through approval plans, examining books received at University of Illinois at Urbana-Champaign (UIUC) and Pennsylvania State University (Penn State) July 2004 to June 2005 where usage was gathered from July 2004 to March 2007. It found that 69% of Penn State’s and 60% of UIUC’s approval books had circulated at least once (p. 70). It was suggested that the higher percentage of use at Penn State was based on the larger user population, 98,000 versus 45,000 at UIUC. No breakdown of use/non-use was provided by subject.   Most studies and guidelines (NISO, 2014, p.34) have exercised caution in drawing conclusions about the higher number of uses/transactions/sessions of ebooks in comparison with print usage. Many libraries do not capture in-house use of their print collections, and some ebook accesses may be equivalent to the act of taking a book off the shelf, glancing through it, and then replacing it.   Some authors have looked at measures of substantive ebook use as a means of comparison with print circulation. Rose-Wiles used viewing of 10+ pages in an ebook as equivalent to a print check-out (p. 146). Goodwin used viewing 11+ pages as a measure of substantive use (p. 103). Ahmad & Brogan (2012, p. 198) in their case study of EBL ebooks usage, set the standard of more than 9 minutes spent in a book as an indication of reading, since the EBL platform uses that time to differentiate reading from browsing. In their study 29.12% of transactions were spent in reading under this definition and 70.88% in browsing.   None of the studies mentioned downloading as an equivalent to print check-outs. However, download could provide a more equivalent measure of comparison to the print check-out. Some effort is required to download the book. Though shorter than most print loan periods, a patron can use the ebook offline for one to three days. With both ebook downloads and print check-outs, there is no way of knowing how much a book is used while ‘out’ or in ‘offline’ mode.   Aims   This study assumes that a comparison of print book (p-book) and ebook usage is possible, especially when substantive use of ebooks is compared with print checkouts.   Given that the YBP approval plan will deliver a greater proportion of ebooks over time, this analysis attempts to answer these questions:   Are approval ebooks being used at an equal or higher rate than print approval books? Is the usage or non-usage of approval ebooks and print books affected by their broad subject discipline? Is the usage/non-usage the same or different for the two media? Are university press books used more than non-university press books, and is usage the same for print and ebook? Did the vendor or type of license affect the usage of ebooks? What is the cost-per-use of an approval ebook and the cost-per checkout for a print approval book?   Method   Most comparisons of the performance of p-books and ebooks have been done comparing the same titles in both formats. Unless an analysis looks at the same titles purchased at the same time in the two formats, it is challenging to compare performance over a similar time period. Differing time periods for access may have affected the results of earlier studies (Levine-Clark & Brown, 2013, slide 3; Downey et al., 2014, p.145).   Although the titles are different in the UML comparison of approval books, approval books, no matter what format, are purchased over the same time period, match the same subject profile, and come from a similar set of publishers with similar publication dates. There is a wide variety of ebook platforms on the market, but the UML ebooks come on four platforms prescribed by the approval profiles: EBL, ebrary, EBSCO, and Wiley.   In 2012/2013, all subject profiles within the YBP approval plan were converted to e-preferred with the exception of pharmacy. The e-preferred mode meant that if an ebook were published within eight weeks of the print version, an ebook would be received by UML.  In 2012/13, 35% of the approval receipts were ebooks. Usage of the approval books received in 2012/13 was counted to December 31, 2014, so both p- and ebooks were available for use between 20 and 32 months.     Table 1 Overall Comparisons of Print- and Ebook Usage   No. of titles No. used % used Print Books 5,237 2,003 38% Ebooks 1,855 872 47% P-Value two-sided proportion of usage different between p- & ebooks 0.0011 P-Value one-sided proportion of usage greater for ebooks 0.00056 Note: P = < 0.05         Source of data   Lists of ebooks and p-books received on approval were generated from SIRSI’s Director’s Station. Some call number and publisher data was retrieved from YBP’s GOBI database.   While the Counter Book Report 2 was only available from ebrary, only the Counter Book Report 1 was produced for EBSCO, Wiley, and EBL ebooks. The Counter Book Report 2 records the number of uses of a section or sections of a book by month and title, whereas the Counter Book Report 1 counts only the uses of a title by month, without considering the use of chapters within it. The Counter 2 Book Report will therefore yield higher uses per title since it counts every section or chapter of a single title as a separate use, whether they were viewed in separate sessions or not. Because of this discrepancy, this study does not attempt to measure rates of use based on Counter reports, only use and non-use. More detailed information about use was obtained from vendor reports for EBL and ebrary ebooks for cost per use analysis; EBSCO and Wiley did not provide detailed information for their ebooks so were not included. Cost per use was calculated per book and then averaged, rather than the collection cost being divided by overall use as in other studies (Bucknell, 2010, p. 133; Bucknell, 2012, p. 53).   For p-books, circulation data was derived from SIRSI to the end of 2013, and then when UML changed systems, from ALMA for 2014. UML does not have consistent data on in-house use of its print collections. Therefore usage of print books was based simply on number of check-outs.   2012/2013 approval results   In this fiscal year UML received 5,237 p-books and 1,855 ebooks on approval. Of those, 38% of the p-books (2,003) circulated and 47% of the ebooks (872) were used. A Pearson’s chi-square test for comparing two proportions showed that there was sufficient statistical evidence to conclude the usage of print and ebooks was different and that ebook use was greater than print.   A further analysis was done by broad subject discipline based on the following LC call number ranges, using the same subject breakdown as Levine-Clark (2014, slide 30).   ·         Arts & Humanities: B-BD, BH-BX, C-F, M-P ·         Social Sciences: BF, H-L, U-V ·         Science, Technology & Medicine (STM): Q-T   Use of ebooks by broad discipline   This analysis indicates that usage of ebooks in the broad disciplines is remarkably similar.   Table 2 Ebook Usage by Broad Subject Discipline   No. ebooks No. used % used Proportion of total ebooks Proportion of total uses Arts & Humanities 635 300 47% 34.2% 34.4% Social Sciences 669 306 46% 36% 35% Science, Technology & Medicine (STM) 494 236 48% 26.6% 27.2% Other 57 30 53% 3.1% 3.4% Total 1,855 872 47%         Use of p-books by broad discipline   UML received many more “Arts & Humanities” print books through the approval plan, but their usage was lower, compared to the size of the collection, while use of “STM” print books appeared to be higher relative to the size of its collection.   The results of the chi-square tests (Table 4 below) provided sufficient statistical evidence to find that usage of print and ebooks was different and higher for ebooks in the “Arts & Humanities” and “Social Sciences” categories. However, there was not statistical evidence to claim the same for the “STM” and “Other” categories.     Table 3 Print Book Usage by Broad Subject Discipline   No. p-books No. used % used Proportion of total p-books Proportion of total uses Arts & Humanities 2,263 807 36% 43% 40% Social Sciences 1,740 657 38% 33% 33% Science, Technology & Medicine (STM) 1,063 469 44% 21% 24% Other 171 70 41% 3% 3% Total 5,237 2003 38%         Figure 1 Comparison of print and ebook usage by broad subject.     Table 4 P- and Ebook Usage Comparison Chi-square Test Subject Proportion of Usage Ebook Proportion of Usage P-book P-Value two-sided P-Value one-sided Decision Arts & Humanities 0.472 0.357 <0.0001 <0.0001 Significant difference Social Sciences 0.457 0.379 0.0004 0.0002 Significant difference STM 0.478 0.439 0.1474 0.07372 Insufficient stat. evidence Other 0.526 0.41 0.1262 0.06309 Insufficient stat. evidence Note: P = < 0.05     Ebooks use by type of publisher   For ebooks, usage of books published by university presses was higher, especially considering use relative to the collection size.   Table 5 Ebook Usage by Type of Publisher   No. of ebooks No. used % used Proportion of total ebooks Proportion of total uses University presses 878 449 51% 47% 51% Other publishers 977 423 43% 53% 49% Total 1,855 872 47%     P-Value two-sided     0.0007 Proportion used is different between university presses & other publishers P-Value one-sided     0.0004 Proportion used is greater for university presses than other publishers Note: P = < 0.05             P-books use by type of publisher   In comparison with ebooks, usage of print books from university presses did not appear different from that of print books from trade or other publishers. Proportionally UML received more university press books (47%) as ebooks while university press books accounted for only 39% of the print titles received.   Table 6 Print Book Usage by Type of Publisher   No. of p-books No. used % used Proportion of total p-books Proportion of total uses University presses 2,055 788 38% 39% 39% Other publishers 3,182 1,215 38% 61% 61% Total 5,237 2,003 38%     P-Value two-sided     0.4205 Proportion used is not significantly different between publisher type P-Value one-sided     0.2103 Proportion used is not significantly greater Note: P = < 0.05             Ebooks use by vendor platform   The analysis by vendor platform suggested that ebrary was the preferred platform.     Table 7 Ebook Usage by Vendor Platform   No. of ebooks No. used % used Proportion of total ebooks Proportion of total uses EBL 867 371 43% 47% 43% ebrary 820 434 53% 44% 50% EBSCO 77 28 36% 4% 3% Wiley 91 39 43% 5% 4% Total 1,855 872 47%     P-Value two-sided     <0.0001 Proportion of usage differs significantly among vendors Note: P = < 0.05             Table 8 Ebook Usage by Type of License   No. of ebooks No. used % used Proportion of total ebooks Proportion of total uses Single user  (ebrary) 483 250 52% 26% 29% 1 user  (EBSCO) 77 28 36% 5% 4% Non-linear lending  (EBL) 863 368 43% 46% 52% Multi-user (EBL) 4 3 75% .02% .03% Multi-user (Wiley) 91 39 43% 5% 4% Multi-user (ebrary) 337 184 55% 18% 21% Total 1,855 872 47%     P-Value two-sided     <0.0001 Proportion of usage differs significantly among the different vendor licenses Note: P = < 0.05             Figure 2 Comparison of usage of single user and multi-user licences.     Ebooks use by license   Ebrary ebooks seemed to receive higher usage independent of license type.   When results were separated into single-user and multi-user licenses, single-user licensed books were used somewhat more.   Substantive use of EBL and ebrary ebooks   Both EBL and ebrary provided detailed reports on usage, allowing a better comparison between p-books and ebooks. The two providers accounted for 91% of the ebooks received on approval in 2012/2013. Wiley and EBSCO ebooks were excluded from cost and cost-per-use calculations. For the purpose of this study, substantive use was determined to have occurred either when a title had been downloaded at least once, or when 11+ pages had been viewed or read. The level of substantive use of ebooks (32%) is lower than the overall mean usage (38%) of approval print books based on check-outs.   Table 9 Substantive Use of EBL and Ebrary Ebooks   No. ebooks No. used No. of sessions/ transactions Mean session per book Mode session per book No.  of ebooks down- loaded No. of ebooks with substantive use % ebooks with substantive use EBL 868 371 1,519 4 2 149 309 35.6% ebrary 819 434 1,610 3.7 2 81 238 29% Total 1,687 805         547 32%                       Mean costs for EBL and ebrary ebooks   The overall cost of the EBL collections is slightly higher than that of ebrary’s because it has about 40% more “STM” books, which tend to be more expensive. In addition, pricing of an EBL non-linear lending ebook tends to be higher than a single-user title from ebrary.   A further analysis was done by broad discipline on the two platforms. The results showed that in general the EBL books were more expensive than ebrary’s in all disciplines except “Social Sciences”.     Table 10 Mean cost of EBL and Ebrary Ebooks   No. of ebooks Cost of all ebooks Mean cost of all ebooks No. of ebooks used Cost of ebooks used Mean cost of ebooks used EBL 868 $87,737* $101 371 $36,826 $99 ebrary 819 $70,556 $86 434 $36,063 $83 Total 1,687 $158,293 $94 805 $72,889 $91 * all costs given are in Canadian dollars   Table 11A EBL Ebooks Mean Cost by Discipline   No. of ebooks Cost of all e- books Mean cost of all ebooks No. of ebooks used Cost of ebooks used Mean cost of ebooks used Arts & Humanities 284 $25,168 $88 116 $10,792 $93 Social Sciences 313 $24,801 $79 123 $9,726 $79 STM 245 $31,800 $130 119 $14,863 $125 Other 26 $3,043 $117 13 $1,446 $111 Total 868 $84,812 $98 371 $36,827 $99   Table 11B Ebrary Ebooks Mean Cost by Discipline   No. of ebooks Cost of all ebooks Mean cost of all ebooks No. of ebooks used Cost of ebooks used Mean cost of ebooks used Arts & Humanities 322 $24,426 $76 173 $12,944 $75 Social Sciences 322 $27,574 $86 174 $15,062 $87 STM 148 $17,059 $115 70 $7,133 $102 Other 27 $2,310 $86 17 $1499 $88 Total 819 $71,369 $87 434 $36,638 $84     Mean Print Book Costs   The mean purchase price of all EBL and ebrary ebooks ($94) is much higher than the mean cost of all print books ($55).   A further breakdown of this information was done by broad discipline.     Table 12 Mean Cost of Print Books No. of p-books Cost of all p-books Mean cost all No. of p-books used Cost of p-books used Mean cost p-books used 5,237 $289,672 $55 2,003 $112,318 $56     Table 13 Mean Cost of Print Books by Discipline   No. of p- books Cost. of all p-books Mean cost all p- books No. of p- books used Cost of p- books used Mean cost (p- books used) Arts & Humanities 2,263 $109,462 $48 807 $39,735 $49 Social Sciences 1,740 $90,654 $52 657 $33,383 $51 STM 1,063 $79,681 $75 469 $34,774 $74 Other 171 $9,875 $58 70 $4,426 $63 Total 5,237 $289,672 $55 2003 $112,318 $56     Cost per Use Comparison   Cost per use for EBL ebooks, ebrary ebooks, and print books ($57, $51, and $34) was about 40% less than the mean purchase cost per book ($101, $86, and $56 respectively).   The cost-per-use data has a non-normal or right-skewed distribution for p- and ebooks and for all disciplines. Therefore, the nonparametric Kruskal-Wallis test was used to determine where cost-per-use differs among the three groups: p-books, EBL ebooks, and ebrary ebooks.   As an alternative strategy, the square-root transformation was applied to the cost-per-use figures in all groups in order to reduce the right-skewedness in the data. A t-test (to compare two groups at a time) and the ANOVA (to compare all three groups) were conducted on the transformed data and the results were compared with those from the Kruskal-Wallis test.  The reported conclusion is based on the Kruskal-Wallis test.   EBL and ebrary ebooks are not significantly different in terms of their cost-per-use.  The mean cost per use is definitely lower for print books with the exception of books in the “Other” discipline, where the small sample size fails to lead to a conclusion.     Table 14 Ebrary, EBL, and Print Comparison – Cost Per Use/Checkout   Arts & Humanities Social Sciences STM Other All EBL no. used 116 123 119 13 371        cost $10,792 $9,726 $14,863 $1,446 $36,827        uses 396 522 521 82 1,521        CPU $57 $45 $72 $38 $57 Ebrary no. used 173 174 70 17 434        cost $12,944 $15,062 $7,133 $1,499 $36,638        uses 676 641 230 63 1,610        CPU $47 $50 $64 $39 $51 Print no. used 807 657 469 70 2,003       cost $39,735 $33,383 $34,774 $4,426 $112,318       uses 1,857 1,725 1,374 173 5,129       CPU $32 $31 $42 $38 $34     Discussion   In terms of the comparison of print and ebook usage, although the overall results favoured ebooks, in the “STM” and “Other” categories the results were inconclusive. The sample size was too small for the Other category, and usage of STM books in the two formats was not significantly different. The Sciences and Technology libraries at UML have always had a comparatively small print book budget, and their smaller collections may account for greater print usage than in other categories.   Usage of print approval books was low, but a number of libraries have indicated equally low results (e.g., Rose-Wiles, Cornell University). These libraries studied a larger number of books over a longer time period, but nonetheless their findings point to a similar trend. A number of factors may be affecting print use: changes in teaching requiring fewer research papers, the availability of information sources outside of libraries, changing reading habits, an increased ability of students to buy books, and student reluctance to use the physical library.   As others have pointed out, it is difficult to compare p- and ebook usage, especially if the print usage does not include in-house use.  Although the comparison of substantive use of ebooks to check-outs of print books is an attempt to mitigate the lack of in-house information, ‘substantive use’ is not an objective measure. The results do, however, seem to agree with the often cited belief that print and ebooks are used for different purposes; ebooks for quick searches and reading shorter texts, and print books for more extensive reading.   In addition, the cost per use data for both formats in the study does not consider the labour cost of the processing and maintenance of print books versus those for ebooks. During the time studied, UML paid $14.87 per p-book for outsourced cataloguing and processing and nothing for ebook catalogue records. Full consideration of any such ‘hidden’ costs is beyond the scope of this study, but nonetheless deserves further inquiry.   The ebook results by broad subject are similar to those for package titles in the “Humanities/Social Sciences” (47%) and “Science/Technology” (49%) areas in the Carrico study (2015, p. 106). Carrico found much higher use of medical books; however, the only categories of medicine covered by the UML’s approval plan were nursing and public health. Levine-Clark’s (2014, slides 31-36) meta-analysis of EBL and ebrary ebooks found that social sciences titles were used more than “STM” ones, but in the UML study of approval ebooks, the usage was quite similar for the three main categories.    The higher usage of university press ebooks at UML is in agreement with Levine-Clark’s (2014, slides 23-27) study. However, type of publisher did not seem to affect print book usage at UML, perhaps reflecting the higher number of print approval books coming from important, non-university press publishers (e.g., Wiley, Routledge, Palgrave Macmillan).    The greater usage for ebrary ebooks was somewhat surprising given that EBL is the preferred format for many users and librarians at UML (Warren, 2015, p. 15). However, during the study UML’s proxy server was dropping users of EBL ebooks. These access problems have since been remedied but could have skewed these ‘historical’ results. In addition, this result may reflect a different database structure and method of counting use.     Table 15 P-Value: EBL, Ebrary, and Print Comparison – Cost Per Use/Checkout Discipline Comparison Average difference t-test ANOVA Kruskal-Wallis Conclusion Arts & Humanities EBL to ebrary 9.30 0.163   0.322 Not sig. different EBL to print 23.27 < 0.0001   < 0.0001 Significantly different ebrary to print 14.97 < 0.0001   < 0.0001 Significantly different EBL-ebrary-print     < 0.0001 < 0.0001 Significantly different Social Sciences EBL to ebrary -4.28 0.170   0.068 Not sig. different EBL to print 14.76 < 0.0001   0.024 Significantly different ebrary to print 19.04 < 0.0001   < 0.0001 Significantly different EBL-ebrary-print     < 0.0001 < 0.0001 Significantly different STM EBL to ebrary 7.55 0.326   0.281 Not sig. different EBL to print 29.92 < 0.0001   < 0.0001 Significantly different ebrary to print 22.37 < 0.0001   0.005 Significantly different EBL-ebrary-print     < 0.0001 < 0.0001 Significantly different Other EBL to ebrary -0.89 0.811   0.802 Not sig. different EBL to print 0.35 0.822   1.00 Not sig. different ebrary to print 1.24 0.899   0.931 Not sig. different EBL-ebrary-print     0.965 0.931 Not sig. different Overall EBL to ebrary 6.46 0.201   0.474 Not sig. different EBL to print 22.93 < 0.0001   < 0.0001 Significantly different ebrary to print 16.47 < 0.0001   < 0.0001 Significantly different EBL-ebrary-print     < 0.0001 < 0.0001 Significantly different Note: P = <0.05     Conclusions   Of approval books that matched the same subject profile, this study finds that ebooks are used more, but print books receive more substantial use. When the data is broken down by broad discipline, the greater usage for ebooks is significant for the “Arts & Humanities” and “Social Sciences” category. However, results for the “STM” and “Other” categories were inconclusive. Clearly both formats of books are needed in a library’s collection.   In terms of type of publisher, university press ebooks were more highly used than ebooks from other publishers, while for print books there appeared to be little difference in usage of the two publisher types. This difference may have occurred because a greater number of important non-university press books arrived in print. Cost per use for print books ($34) was lower than for EBL and ebrary ebooks3 ($53), in part because ebooks are generally more expensive; the mean purchase cost of ebrary and EBL titles was $91, but $55 for print books. These differences may not tell the full story since print usage did not include in-house browsing or reading. On the other hand, turnstile counts at UML, as at other academic libraries, have decreased over the last decade suggesting that in-house use may not be as significant as it once was. Finally, cost per use in this study did not reflect the cost of processing and maintenance of the different media, but the cost is likely higher for print.   It is recommended that future usage studies work with a larger sample of ebooks relative to print and take into account the hidden costs of access and maintenance of the two formats in order to produce more compelling results. Many books are published in the arts & humanities and social sciences. For a library with limited resources and similar low use of print in these areas, it may be more cost-effective to acquire monographs in these disciplines, through publisher front-lists or PDAs/DDAs than through an approval plan.   Usage patterns take time to unfold, and libraries like the UML often have to make abrupt changes to monographic acquisition methods and allocations for budgetary reasons. Thus UML   altered and reduced its YBP approval plan before the results of this study were determined. Nonetheless usage studies can provide evidence for librarians refining approval plan profiles and for budget managers considering changes to monographic acquisition methods and allocations.   Endnotes   1 For a discussion of demand-driven acquisition options, see NISO’s “Demand Driven Acquisition: A Recommended Practice” (9-13) and for a discussion of ebook acquisition methods see Kumbhar (pp. 786-787, 2012) and Bucknell (2012).   2 ARL Statistics 2011/2012 to 2013/2014 provide statistics on amounts spent on one-time purchases, of which monographs would be a large portion.   3Although EBSCO (77) and Wiley (91) ebooks came on approval, neither vendor supplied detailed usage reports that could be used for cost-per-use comparisons.   References   Ahmad, P., & Brogan, M. (2012). Scholarly use of e-books in a virtual academic environment: a case study.  Australian Academic & Research Libraries, 43(3), 189-213. http://dx.doi.org/10.1080/00048623.2012.10722277   Alan, R., Chrzastowski, T. E., German, L., & Wiley, L. (2010). Approval plan profile assessment in two large ARL libraries, University of Illinois at Urbana-Champaign and Pennsylvania State University.  Library Resources & Technical Services, 54(2), 64-76.   Association of Research Libraries. (2013). ARL statistics 2011-2012. Retrieved from: http://publications.arl.org/arl_statistics   Association of Research Libraries. (2014). ARL Statistics 2012-2013. Retrieved from: http://publications.arl.org/arl_statistics   Association of Research Libraries. (2015). ARL Statistics 2013-2014. Retrieved from: http://publications.arl.org/arl_statistics   Bucknell, T. (2012). Buying by the bucketful: a comparative study of e-book acquisition strategies. Insights, 25(1), 51-60. http://dx.doi.org/10.1629/2048-7754.25.1.51   Bucknell, T. (2010). The ‘big deal’ approach to acquiring e-books: a usage-based study. Serials, 23(2), 126-134. http://dx.doi.org/10.1629/23126   Carrico, S. B., Cataldo, T. T., Botero, C., & Shelton, T. (2015). What cost and usage data reveals about e-book acquisitions: ramifications for collection development. Library Resources & Technical Services, 59(3), 102-111. http://dx.doi.org/10.5860/lrts.59n3.102   Cornell University Library.  (2010). Report of the Collection Development Executive Committee Task Force on Print Collection Usage. Retrieved from http://staffweb.library.cornell.edu/system/files/CollectionsUsageTF_ReportFinal11-22-10.pdf and http://staffweb.library.cornell.edu/node/1781   Downey, K., Zhang, Y., Urbano, C., & Klinger, T. (2014). A comparative study of print book and DDA ebook acquisition and use. Technical Services Quarterly, 31(2), 139-160. http://dx.doi.org/10.1080/07317131.2014.875379   Goodwin, C. (2014). The e-Duke Scholarly Collection: e-book v. print use. Collection Building, 33(4), 101-105. http://dx.doi.org/10.1108/CB-05-2014-0024   Kimball, R., Ives, G., and Jackson, K. (2009). Comparative usage of science e-book and print collections at Texas A&M University Libraries. Collection Management, 35(1), 15-28. http://dx.doi.org/10.1080/01462670903386182   Kumbhar, R. (2012). E-books: Review of research and writing during 2010. The Electronic Library, 30(1), 777-795. http://dx.doi.org/10.1108/02640471211282109   Levine-Clark, M. (2014, June 27). Diving into ebook usage: An ALA update. American Library Association Annual Meeting, Las Vegas, NV, USA. Retrieved from: http://www.slideshare.net/MichaelLevineClark   Levine-Clark, M., & Brown, C. (2013, March 19). E or p? A comparative analysis of electronic and print book usage. Electronic Resources & Libraries Conference, Austin, TX, USA. Retrieved from: http://www.slideshare.net/MichaelLevineClark/erl-e-books-2013   Littman, J., & Connaway, L. S. (2004). A circulation analysis of print books and e-books in an academic research library. Library Resources & Technical Services, 48(4), 256-262. Retrieved from: http://www.oclc.org/content/dam/research/publications/library/2004/littman-connaway-duke.pdf   National Information Standards Organization. (2014). Demand-driven acquisition of monographs: A recommended practice (NISO RP-20-2014). Retrieved from: http://www.niso.org/apps/group_public/download.php/13373/rp-20-2014_DDA.pdf   Rose-Wiles, L. M. (2013). Are print books dead? An investigation of book circulation at a mid-sized academic library. Technical Services Quarterly, 30(2), 29-152.   Warren, R. (2015). E-book assessment report [for usability]. University of Manitoba Libraries.