D-Lib Magazine
November/December 2008

Volume 14 Number 11/12

ISSN 1082-9873

A Study of Institutional Repository Holdings by Academic Discipline

Peter A. Zuber
Engineering Librarian
Brigham Young University
2321 HBLL
Provo, Utah 84602
<peterzuber@byu.edu>

1. Introduction

General Background for the Study

Studies have shown the advantage of open access publication in terms of citation rate and research impact (Antelman, 2004). These studies have quantified citation rates through various methods, including comparing identical articles in both closed (meaning paid) access and open access, and historical averages within disciplines. Although the data suggest a higher citation rate with open access, some disciplines have demonstrated a reluctance to embrace open access. Issues of premature disclosure, plagiarism, fear of upsetting the current system, indifference, long term storage and retrieval, and intellectual property rights are among the most noted (Yiotis, 2005). At the same time, traditional print services, straining to keep up with the volume of material being produced, have created dissatisfaction with delays and extended wait times common in print publication. Consequently, networking and digital publication practices have, over time, found willing users. Some disciplines, notably astronomy, have embraced open access completely, having almost all scholarly content from current as well as past publications freely available in online form (Kurtz et al., 2005).

The involvement of early adopters in pre-print publication, e-print initiatives, and author self-archiving created a precedent for open access practices that might have resulted in high participation rates in institutional repositories once they were created. However, through citation analysis surveys, trends have revealed that early successes in specific discipline-based repositories have yet to be repeated among other disciplines that have little prior experience in digital publication (SPARC, 2002).

If an academic library is considering providing support and resources for an Institutional Repository (IR), it faces significant challenges, among them the ability to persuade faculty to contribute important research representing large investments of time (van Westrienen & Lynch, 2005). If a preference for traditional publication workflows and practices vary for each academic discipline, it is reasonable to assume that motivations and concerns vary as well.

Statement of the Problem

From a survey of 45 Institutional Repositories, collections show an average of approximately 1,200 items and demonstrate sporadic contribution or self-archiving practices (Ware, 2004). Although some studies of citation analysis have been categorized by discipline, no current survey of contribution level as a function of discipline was readily found. Knowing the weaknesses in a system is required in order to address and strengthen the system. Librarians, knowing national tendencies as well as philosophical positions with regard to open access, can more effectively solicit content by knowing beforehand which disciplines tend to embrace or reject IR's. Exceptions to national trends that show strong contributions from certain disciplines to IR's, if known, can be a valuable resource in providing specific rationale, justification, and practical advice in preparing incentives and soliciting materials.

The purpose of the study reported in this article was to determine nationally which academic disciplines demonstrate a greater tendency to publish in academic institutional repositories. The study focused on colleges and universities culled from the fifty states that matched specific criteria with the intent to provide equal representation nationally. Three hypotheses were determined.

Hypotheses

H1. Institutional repositories do not yet demonstrate broad, discipline diverse contributions.

H2. Academic disciplines having prior history in pre-print and e-print practices contribute the greatest percentage of content.

H3. The majority of institutional repositories do not provide incentives for publication, such as highlighting recent additions.

Limitations

• The study measured content on a national scale and assumed its results were generally representative of content distributions in other domestic, liberal arts institutions.
• Institutions were selected according to specific criteria; therefore, it was possible that some states would not be included.
• For the purposes of this study, academic disciplines that are considered to have a history in pre-print, author self-archiving, and e-print practices include the physical and social sciences, such as physics and mathematics, economics, the cognitive sciences, astronomy, astrophysics, geophysics, and computer science. (SPARC, 2002).
• Only academic disciplines having content in a repository were categorized, therefore, not all disciplines were represented in the survey. Only one instance of a digital object from any discipline in any repository was required in order for that discipline to be included.
• The study renamed similar academic disciplines in order to insure manageable data and create a sufficiently diverse categorization scheme. The disciplines renamed depended on the naming practices at each institution. For example, electrical engineering and electronic engineering were considered the same discipline for the purposes of this study.
• For simplifying some comparisons, disciplines were grouped together similar to a university's collection of academic disciplines organized under schools or departments. For example, academic disciplines such as electrical engineering, mechanical engineering, chemical engineering, etc., were placed in an "academic group" called "Engineering". Math, Statistics, Chemistry, Physics, etc., were place in a "Physical Sciences" academic group and so on.
• Content not considered germane to an academic discipline, such as inventory lists, license agreements, generic newsletters, test materials for repository functionality, were placed in a "Miscellaneous" category.

Assumptions

• It was assumed that data acquired from IR's sponsored by the universities included in this study were accurate.
• It was assumed that the universities included in this study were representative of universities in general having broad curriculum offerings and the potential for institutional repositories.
• It was assumed that an institutionally sponsored repository referred to a repository that was unbiased in its collection intent. In other words, the IR was sponsored by the institution or the institution's library and available for submission by all disciplines on campus. In this context, a subject repository sponsored by a department or a university program to hold subject specific material was not considered.
• It was assumed that the lack of an institutionally sponsored repository did not infer that subject repositories were also absent from the institution's set of resources.

II. Methodology

Research Design

The study used a quantitative research design in order to support or not support the hypotheses. The study collected data from academic colleges and universities that were selected according to the following criteria:

1. Four-year, co-ed institutions with no preference for urban/suburban/rural location, or minority, international, or out-of-state population,
2. Either public or private, without regard to religious affiliation,
3. Resident within the 50 states of the United States of America,
4. Had over 15,000 students enrolled, and
5. Offered a bachelor's level Liberal Arts & Sciences degree and/or a bachelor's level General Studies degree.

The first four criteria were intended to cull a population of institutions whose student and faculty population were sufficient in size and diversity to warrant support, interest, and need for an institutional repository. The fifth criterion, schools that offered liberal arts and general studies degrees, inferred equal importance and priority was given to all academic disciplines on campus in contrast to agricultural or engineering schools, for example, whose charter was clearly directed to specific disciplines. Using Collegeboard.com, a list of eighty-three colleges and universities having met these criteria were generated and tabled alphabetically in the Appendix (Table 4). For convenience in keeping the survey of manageable size and impartially selected, a random sampling method to generalize the population was employed whereby each even numbered institution on the list was selected. This created a survey set of forty-one institutions from which data were collected. The table of selected institutions along with all data collected from each is given in the Appendix (Tables 5a-5f).

Data Description

The data collected were primarily numerical. In most instances, descriptive data were collected solely to identify numerical content. The data consisted of seven elements:

1. The name of the college or university,
2. The college or university's current student enrollment (total of both undergraduate and graduate),
3. The type of technology used for the repository (DSpace, BePress, etc.),
4. The name and location of the repository,
5. The total number of repository holdings,
6. The presence, nature and location of incentive content,
7. The total number of repository holdings reported for each academic discipline.

Data Organization and Analysis

Data were organized in an Excel spreadsheet with the college or university name comprising a single row with successive columns on that row containing the collected data.

Using this approach, summations were made at the end of each column to produce national totals for each discipline, while college or university totals for holdings were generated row-wise. Academic disciplines were added column-wise as they were encountered in a repository. Only one instance of a digital object in any repository was required in order to include that academic discipline. Other content not considered germane to an academic discipline was placed in a "Miscellaneous" category including "Event proceedings," "Library Misc/License/Newsletters/Tests," "e-Books," and "University Archives."

To simplify some comparisons, "academic groups" were created that were comprised of a collection of related academic disciplines. These academic groups were organized in a way similar to a university's collection of academic disciplines organized under schools or departments. For example, academic disciplines such as electrical engineering, mechanical engineering, chemical engineering, etc., were placed in an academic group called "Engineering". Similarly, Math, Statistics, Chemistry, Physics, etc., were place in a "Physical Sciences" academic group. In total, twelve academic groups were created from forty-seven academic disciplines and are listed in Table 2.

To create a measure that would ascertain the first hypothesis, that repositories do not yet provide discipline diverse collections, a value of one was given for each academic discipline when an institution's repository contained content of any size in that area. Accordingly, since forty-seven disciplines were identified during the study, a maximum count of forty-seven was possible if all disciplines were represented within a single repository. A measure of "percent discipline coverage" was created by generating a percentage by dividing the individual scores awarded by the total possible, forty-seven, and multiplying by one-hundred. For example, if a school had holdings in nine disciplines, the percent discipline coverage would be calculated by dividing nine by forty-seven, multiplying by one-hundred for a result of nineteen percent. An average of all schools' percent discipline coverage was also calculated.

The second hypothesis stated that academic disciplines having prior history in pre-print and e-print practices contribute the greatest percentage of content. For the sake of this study, those academic disciplines having prior history in pre-print and e-print practices include those in the physical and social sciences such as physics and mathematics, economics, the cognitive sciences, astronomy, astrophysics, geophysics, and computer science (SPARC, 2002). Those disciplines were placed into one of two academic groups, named "Physical Sciences" and "Social Sciences." Adding the contributions from both of these groups allowed their total to be compared against the contributions from all other academic groups created in the study.

With regard to the third hypothesis, the presence of incentive content, three types were recorded:

• Most recent – a listing of most recently uploaded documents
• Most popular – a listing of documents having the highest viewing statistics
• Paper of the Day – a paper that has been selected for exposure by the repository administrator or some other process.

In addition, the location of the incentive was recorded as being either on the main page of the repository, or on a sub-page, typically where a collection or a community's content was ordered below the site's main page.

Once data were collected, the various disciplines were sorted, organized, and placed under generalized academic "schools" or "groups," creating categories similar to the organizational structure found at most Universities. For example, an "Engineering" group was created to include Chemical Engineering, Civil Engineering, Electrical Engineering, Engineering Management, and Mechanical Engineering. Fifteen groups were created, including Engineering, Physical Sciences, Biology and Agriculture, Business, Education, Technology, Social Sciences, Fine Arts and Communications, Humanities, Medicine, Law, Athletics, Thesis, and Miscellaneous. The creation of academic groups was done simply to allow generalized comparisons. Table 2 and Tables 5a-5f are provided if a more specialized examination of specific discipline contributions is desired.

III. Results

IR National Trends

Survey techniques can help to identify differences, growth, and emergence in areas of interest. In the area of academic publication, potential differences can exist in both publication practice and acceptance of open access alternatives by academic disciplines. The identification of academic involvement by discipline in institutional repositories led to this research as no survey of this type was readily found. Knowing strong and weak contributors by discipline can assist institutional repository sponsors to better formulate recruitment strategies and generate incentives as a means to increase contribution levels. In addition, identifying institutional repositories that demonstrate strong contribution levels in various academic disciplines can provide an important resource to an IR sponsor who needs advice in improving their collection. This study may indicate relative growth or reduction of interest in institutional repositories holdings by discipline nationally. It may also tend to lead developers toward specific repositories for advice and direction in recruitment, subject content and/or organization. The complete set of data collected including colleges surveyed, student enrollments, IR technology used, IR name and location, total holdings, incentive type, and disciplines represented are found in Tables 5a-5f in the Appendix.

An analysis of forty-one randomly selected universities, whose curriculum offerings were considered broadly based in order to reduce collection bias, revealed that eighteen universities currently sponsored institutional repositories for a forty-four percent acceptance rate. One of the repositories was currently off-line and unavailable at the time of the survey. Of the remaining twenty-three, nine are evaluating or in the process of launching a repository. Assuming all launched within the next few years, the percentage would grow to sixty-six percent. Regarding content, numerous academic disciplines were represented, although the total contribution level showed a distinct advantage in the engineering disciplines.

Three groups were created and excluded from the comparison, "Thesis," containing undetermined interdisciplinary content, "Miscellaneous," containing no discipline specific content, and "Interdisciplinary/Peer-Reviewed." In several repositories, holdings placed under a "Thesis" or "Dissertations" category were found and considered small enough in number to warrant the additional effort to retrieve, read, and categorize each title. However, in the case of Texas Tech University and the University of Connecticut, over sixteen hundred and three thousand titles respectively, representing numerous academic disciplines, were found under a single "Thesis/Dissertation" category. This made the effort required to categorize each title beyond the scope of this article. Therefore, the "Thesis" category in this study includes the titles found in the "Thesis/Dissertation" category from these two universities.

The later category is a special instance by the University of Michigan's "DeepBlue" repository. This singular group contained over twenty-two thousand interdisciplinary, peer-reviewed articles generated by the University's faculty over several decades. This one collection represents almost half of the total holdings sampled in this survey. It consists of articles produced at the university that were previously processed through the traditional scholarly publication cycle and are now made available through open access. Although both the "Thesis" and "Interdisciplinary/Peer-Reviewed" groups represent important collections with regard to open access, none were included or organized according to discipline as again, the effort required to categorize each article was beyond the scope of this paper.

Restatement of Hypotheses

H1. Institutional repositories do not yet demonstrate broad, discipline diverse contributions.

The total number of distinct disciplines discovered to have representative content in IR's surveyed nationwide was forty-seven. A "discipline coverage" percentage was calculated based on awarding a value of one for each discipline having one or more holdings, summing the values awarded, dividing by forty-seven and multiplying the result by one-hundred. If an institution had at least one holding for each of the forty-seven disciplines identified nationally, it would have had a discipline coverage score of one hundred percent. Figure 1 shows the percent coverage of disciplines found in schools sponsoring institutional repositories. Only one school, the University of Kansas, exceeded a fifty percent coverage rate with fifty-seven percent. The next closest was the University of Massachusetts with forty-nine percent followed by the University of Connecticut with a thirty-four percent rate. The average coverage rate for all schools was nineteen percent. Excluding the top three scoring schools, the average percentage dropped to twelve percent.

Figure 1. Percent Disciplinary Coverage by School

H2. Academic disciplines having prior history in pre-print and e-print practices contribute the greatest percentage of content.

Table 2 shows the number of objects found in all institutional repositories sorted by academic groups with their related disciplines. Each discipline lists its total of all collections, and the total for the academic group is given at the top of each group. Figure 2 is a histogram of the total number of digital objects, or the collection count for each academic group. The majority of content, according to discipline, is found in the Engineering group, having thirty-six percent of all holdings surveyed. The Engineering collection size ranged from four hundred and ninety-three times that of Athletics, to almost two and a half times that of Business. Combining the Physical and Social Sciences group produced the number of holdings for disciplines having prior history in pre-print and e-print practices. This number was two thousand, five hundred and fifty holdings, or thirteen percent nationally, compared to sixteen thousand, eight hundred and forty-three holdings for all remaining disciplines combined. Figure 3 shows the percent contributions of all academic groups compared to the combination of the Physical and Social Sciences groups. This combined group ranked third nationally behind Engineering and Business.

Figure 2. Total Holdings by Academic Group

Figure 3. Percent Holdings by Academic Group Compared to Physical/Social Sciences

H3. The majority of institutional repositories do not provide incentives for publication, such as highlighting recent additions.

Table 3 details the incentives present for each institutional repository. A legend defines the IR incentive and location designations.

Legend for the Table

Of the eighteen universities surveyed that had institutional repositories, fourteen have incentive content, or seventy-eight percent, either on the main or sub-pages, and the incentive consists of one or more of the types listed. Much of this is attributable to the technology used. DSpace, which was used for fifty-six percent of installations, and BePress, used for twenty-eight percent, provide incentive features as part of their software package and account for all the incentive-based installations. The remaining installations, using "Greenstone" and "Innovative Interfaces" software technologies, had no incentives.

IV. Conclusions

The results from this research help to illustrate both an overall acceptance rate of institutional repositories as well as collection patterns in institutional repository content based on contributions from various academic disciplines.

H1. Institutional repositories do not yet demonstrate broad, discipline diverse contributions.

The assessment of collection diversity and coverage revealed only one institution whose content exceeded a fifty percent coverage rate. On average, of the institutions surveyed, only nineteen percent of the forty-seven disciplines were represented with at least one holding. Given that the disciplines included were only those encountered during the survey, the percentage would tend to drop even more if all known disciplines were included. The hypothesis is supported.

H2. Academic disciplines having prior history in pre-print and e-print practices contribute the greatest percentage of content.

Engineering contributed the majority of content of any discipline with thirty-six percent. Combining Physical and Social Sciences created a thirteen percent contribution; nearly twenty-three percentage points lower than Engineering. The combined disciplines ranked third nationally, two percentage points below Business, with fifteen percent. The hypothesis is not supported.

H3. The majority of institutional repositories do not provide incentives for publication, such as highlighting recent additions.

In total, eighteen institutions sponsored repositories, and fourteen, or seventy-eight percent of those provided incentives either as "Paper of the Day," "Most recent," or "Most popular." The location of incentive(s) varied depending on the site, with all BePress installations having the incentive(s) on the main page, and most DSpace installations having a single incentive on sub-pages. The hypothesis is not supported.

Recommendations for Future Research

Several issues arose from performing the survey that might provide ground for continuing and/or new research. The lack of organized structure for some thesis deposits and the University of Michigan's "DeepBlue" holdings of peer-reviewed materials made surveying that content by discipline difficult. As mentioned, almost half of all the content discovered was represented in the "DeepBlue" collection, but not included in the analysis. Similarly, with the thesis collection, those totals almost matched the level of the "Engineering" group contributions. If the pattern of contributions of the known categorized materials is representative, then organizing the thesis and the peer-review collections may only verify the overall trend discovered in this paper. It remains for additional research to perform this investigation.

Another issue of some value may found in determining the age of the repository versus the collection size. An analysis of launch dates of the several repositories studied here against collection size may or may not tend to show acceptance rates as a function of time online.

Another issue came from the difficulty in finding the institutional repository itself. In some instances, several searches within the institution's website were required to discover if they indeed sponsored a site or if they were contemplating creating one. Typically, if a search contained any reference to an institutional repository, then in-depth reviews of the successful search returns were required to see if a link or URL was provided to the IR's main page. Although content is individually searchable from various web engines, the lack of visibility within an institution's own web site speaks to poor recruitment and incentive strategies. If contributions are desired from faculty, then a more visible presence on the institute's main page, either in the form of a notice or a hyperlink, or on the institution's library page may provide greater awareness and a willingness to contribute. Measuring the impact of incentives of this type can provide important direction to the current and future sponsors of IR's.

This study did not attempt to delineate the type of content found for each discipline. Content can either address the format or the nature of the material. For example, an institutional repository has the capability and potential of archiving multiple formats and is ideal for housing documents, video, sound, and images. Additionally, although born of scholarly communication and open access initiatives, an IR is not exclusive to that goal. Several varieties of content were evident, including not just scholarly articles but inventory lists, newsletters, license agreements, and multi-media. A survey of media types or type of content apart from discipline can provide important insight into how and why IRs are currently used.

V. References

Antelman, K. (2004). Do open-access articles have a greater research impact? College & Research Libraries, 65(5), 372-382.

Chan, L., Cuplinskas, D., Eisen, M., Friend, F., Genova, Y., Guédon, J.-C., et al. (2002, February 14). Budapest open access initiative. Retrieved January 20, 2007, from <http://soros.org/openaccess/read.shtml>.

Kurtz, M. J., Eichhorn, G., Accomazzi, A., Grant, C., Demleitner, M., Henneken, E., et al. (2005). The effect of use and access on citations. Information Processing & Management, 41(6), 1395-1402.

Lynch, C. (2003). Institutional repositories: Essential infrastructure for scholarship in the digital age. portal: Libraries & the Academy, 3(2), 327-336.

SPARC. (2002, August 27). The case for institutional repositories: A SPARC position paper. Retrieved January 20, 2007, from <http://www.arl.org/sparc/IR/ir.html>.

van Westrienen, G., & Lynch, C. A. (2005). Academic institutional repositories: Deployment status in 13 nations as of mid 2005. D-Lib Magazine, 11(9), 1-12. <doi:10.1045/september2005-westrienen>.

Ware, M. (2004). Institutional repositories and scholarly publishing. Learned Publishing, 17(2), 115-124.

Yiotis, K. (2005). The open access initiative: A new paradigm for scholarly communications. Information Technology & Libraries, 24(4), 157-162.

VI. Appendix

Table 4. List of Colleges and Universities Meeting the Study's Criteria
(Schools in bold selected for study).

1. Ball State University ; Muncie, Indiana
2. Boise State University ; Boise, Idaho
3. Bowling Green State University ; Bowling Green, Ohio
4. Brigham Young University ; Provo, Utah
5. California Polytechnic State University: San Luis Obispo ; San Luis Obispo, California
6. California State Polytechnic University: Pomona ; Pomona, California
7. California State University: Fullerton ; Fullerton, California
8. California State University: Long Beach ; Long Beach, California
9. California State University: Northridge ; Northridge, California
10. California State University: Sacramento ; Sacramento, California
11. Central Michigan University ; Mount Pleasant, Michigan
12. Colorado State University ; Fort Collins, Colorado
13. East Carolina University ; Greenville, North Carolina
14. Excelsior College ; Albany, New York
15. Florida Atlantic University ; Boca Raton, Florida
16. Florida International University ; Miami, Florida
17. Florida State University ; Tallahassee, Florida
18. George Mason University ; Fairfax, Virginia
19. Grand Valley State University ; Allendale, Michigan
20. Illinois State University ; Normal, Illinois
21. Indiana University Bloomington ; Bloomington, Indiana
22. Indiana University-Purdue University Indianapolis ; Indianapolis, Indiana
23. Iowa State University ; Ames, Iowa
24. James Madison University ; Harrisonburg, Virginia
25. Kent State University ; Kent, Ohio
26. Louisiana State University and Agricultural and Mechanical College ; Baton Rouge, Louisiana
27. Middle Tennessee State University ; Murfreesboro, Tennessee
28. New York University ; New York, New York
29. North Carolina State University ; Raleigh, North Carolina
30. Northern Illinois University ; DeKalb, Illinois
31. Ohio State University: Columbus Campus ; Columbus, Ohio
32. Ohio University ; Athens, Ohio
33. Oklahoma State University ; Stillwater, Oklahoma
34. Penn State University Park ; University Park, Pennsylvania
35. Portland State University ; Portland, Oregon
36. Purdue University ; West Lafayette, Indiana
37. San Diego State University ; San Diego, California
38. San Francisco State University ; San Francisco, California
39. San Jose State University ; San Jose, California
40. Southern Illinois University Carbondale ; Carbondale, Illinois
41. State University of New York at Buffalo ; Buffalo, New York
42. Temple University ; Philadelphia, Pennsylvania
43. Texas A&M University ; College Station, Texas
44. Texas Tech University ; Lubbock, Texas
45. Troy University ; Troy, Alabama
46. University of Akron ; Akron, Ohio
47. University of California: Los Angeles ; Los Angeles, California
48. University of Central Florida ; Orlando, Florida
49. University of Cincinnati ; Cincinnati, Ohio
50. University of Connecticut ; Storrs, Connecticut
51. University of Delaware ; Newark, Delaware
52. University of Florida ; Gainesville, Florida
53. University of Georgia ; Athens, Georgia
54. University of Illinois at Urbana-Champaign ; Champaign, Illinois
55. University of Iowa ; Iowa City, Iowa
56. University of Kansas ; Lawrence, Kansas
57. University of Maryland: University College ; Adelphi, Maryland
58. University of Massachusetts Amherst ; Amherst, Massachusetts
59. University of Memphis ; Memphis, Tennessee
60. University of Michigan ; Ann Arbor, Michigan
61. University of Minnesota: Twin Cities ; Minneapolis, Minnesota
62. University of Missouri: Columbia ; Columbia, Missouri
63. University of Nebraska – Lincoln ; Lincoln, Nebraska
64. University of Nevada: Las Vegas ; Las Vegas, Nevada
65. University of New Mexico ; Albuquerque, New Mexico
66. University of North Carolina at Chapel Hill ; Chapel Hill, North Carolina
67. University of North Texas ; Denton, Texas
68. University of Oklahoma ; Norman, Oklahoma
69. University of Pittsburgh ; Pittsburgh, Pennsylvania
70. University of South Carolina ; Columbia, South Carolina
71. University of South Florida ; Tampa, Florida
72. University of Texas at Austin ; Austin, Texas
73. University of Texas at El Paso ; El Paso, Texas
74. University of Toledo ; Toledo, Ohio
75. University of Utah ; Salt Lake City, Utah
76. University of Washington ; Seattle, Washington
77. University of Wisconsin-Milwaukee ; Milwaukee, Wisconsin
78. Utah Valley State College ; Orem, Utah
79. Virginia Polytechnic Institute and State University ; Blacksburg, Virginia
80. Washington State University ; Pullman, Washington
81. Weber State University ; Ogden, Utah
82. West Virginia University ; Morgantown, West Virginia
83. Western Kentucky University ; Bowling Green, Kentucky

Tables 5a-5e. Spreadsheet of Survey Results

Top | Contents
Search | Author Index | Title Index | Back Issues
Previous Article | ECDL Conference Report
Home | E-mail the Editor

D-Lib Magazine Access Terms and Conditions

doi:10.1045/november2008-zuber