3.2            IR Budget Line Items

Questionnaires for IMP and PPT institutions asked respondents what percentage of their IRs’ annual budget is allocated to various line items. The two questionnaire versions listed the same line items, and respondents could write in items that were not listed. SurveyMonkey was programmed so that it required respondents to enter percentages that added to 100. Figure 3.1 gives the results.

Costs for staff and vendor fees represent about 75% of the IR budget, with staff costs exceeding vendor fees during PPT and vice versa during implementation. Hardware acquisition represents about 10% of the IR budget. Software costs represent 7% and 2.5% of PPT and IMP budgets, respectively. Costs for software maintenance, hardware maintenance, and system backup account for 12.5% of the IR budget.

Although many respondents volunteered open-ended comments pertaining to costs, only two comments cite line items that MIRACLE Project investigators failed to include in their original list:

• “Marketing and PR [public relations] activities.”
• “Server farm charges, 1%, hosted by Central Computing; storage farm charges, 11%; indirect costs (15%).”

Despite having a fully functional and operational IR, several IMP respondents write about the informality of their IR’s budget:

• “Our IR does not have a budget.”
• “No specific budget for IR. It is absorbed in the library budget.”
• “We do not really have a budget for this. The fee to the vendor is paid out of our library’s operating costs. Three staff members each spend a few hours a week working on this project. It is impossible to estimate the staff cost.”
• “We only budget for the subscription to our hosted product. We don’t budget the staff time.”
• “IR isn’t budgeted separately anymore and was only partially budgeted separately from the library in year 1 and year 2.”
• “[T]his question is difficult to answer. Staff responsible for the repository [are] doing repository work [and other unrelated tasks]. Vendor fees are shared between the library and central administration . . .  I don’t know the full operating budget of [either the library or central administration] nor am I interested to know.”

Here is a comment from an IMP respondent who is exceptionally precise about her institution’s IR budget:

• “This coming year will be an exception: $100,000 has been allocated for initial purchase and migration of commercial service provider. Our operating budget alone without the above would be staff, 49%; hardware acquisition, 27%; hardware maintenance, 3%; software acquisition, 16%; software maintenance and updates, 2%; vendor fees, 3%.”

Several PPT respondents comment on the shared nature of the IR initiative:

• “We are not funding this project with dollars from our [library] budget; system administration is picking up all hardware and software costs. We [the library] are providing only human resources.”
• “Our IR software and hardware were a special allocation from Instructional Technology Services (ITS) and the central administration. Maintenance and upgrade of server and software will be absorbed by ITS regular budget process. Implementation of the IR will be absorbed into regular library workflow.”
• “The software license of ContentDM was purchased by central computing. The annual maintenance license agreement is paid by the library. All labor is carved from staff time in the Library and Institutional Technology Departments, with faculty involvement supervising work-study students. We are small scale, concentrating on unique content when faculty want something digitized.”
• “Our IR is distributed among departments on campus—it has no separate budget.”

Other PPT respondents could not break down IR costs into listed line items because they did not know or were unsure about the breakdown, or had not yet budgeted for the IR.

3.3            Chapter 3 Summary

PO, PPT, and IMP respondents agree about the top-ranked funding sources for IRs—funding comes or will come from the library (see Table 3.1). They also agree that funding is not coming from academic units.

Costs for staff and vendor fees represent about 75% of the IR budget, with staff costs exceeding vendor fees during PPT and vice versa during implementation (see Figure 3.1). Hardware acquisition represents about 10% of the IR budget. Software costs represent 7% of PPT and  2.5% of of IMP budgets. Costs for software maintenance, hardware maintenance, and system budget account for 12.5% of the IR budget. Underlying the write-in responses of several IMP respondents is a certain informality about the IR budget. We did not speculate on reasons for this informality.

 

4                IMPORTANT INVESTIGATIVE ACTIVITIES

Chapter 4 explores important investigative activities that institutions conduct to determine whether to implement an institutional repository (IR).

4.1            Important Investigative Activities

Planning only (PO), planning and pilot testing (PPT), and implementation (IMP) questionnaires asked respondents to rate the importance of various investigative activities in terms of influencing their decision to initiate planning, pilot testing, and implementation. To simplify results, MIRACLE Project staff assigned weights to response categories as follows: (+2) very important; (+1) somewhat important; (0) no opinion, don’t know, or not applicable; (-1) somewhat unimportant; and (-2) very unimportant. They totaled the weights. These results were then compiled to rank order all the activities. Table 4.1 uses ranked activities in the “Total” column to order top-, middle-, and bottom-ranked activities.

Table 4.1. Important investigative activities

Top-ranked investigative activities (1 to 4)	PO	PPT	IMP
Learning about successful implementations at comparable institutions	1	2	1
Learning from reports of other institutions’ PO, PPT, and IMP activities	2	1	2
Learning about successful implementations at a wide range of academic institutions	(5)*	3	3
An analysis of a thorough literature review of IRs	(9)	(5)	4
Middle-ranked investigative activities (5 to 8)	PO	PPT	IMP
Using other institutions’ operational IRs	6	8	5
Results of your institution’s needs assessment	7	7	6
Demonstrating operational IRs to my institution’s decision makers	(3)	6	7
Learning about available expertise and assistance from a library consortium, network, group of libraries, etc.	(4)	(4)	8
Bottom-ranked investigative activities (9 to 12)	PO	PPT	IMP
Demonstrating IR metadata harvesters such as OAIster and Google Scholar to my institution’s decision makers	10	10	9
Identifying better digital preservation techniques	(8)	9	10
Waiting for a critical mass of IR implementation at comparable institutions to happen	11	12	11
Waiting for a critical mass of IR implementation generally to happen	12	11	12
*   Parentheses indicate PO and PPT investigative activities that deviated from IMP top, middle, or bottom ranks.

At the top of the ranked list are investigative activities concerning learning about IRs from the experiences of others. For PPT and IMP respondents, this includes analyzing literature reviews. PO and PPT respondents rank “Learning from a library consortium …” higher than IMP respondents do, most likely because the latter charged ahead with IR implementation before consortia, networks, and comparable groups had begun their involvement with IRs. PO respondents rank “Demonstrating operational IRs to my institution’s decision makers” much higher than PPTs and IMPs do. Such demonstrations probably make IRs more tangible to decision makers. They increase decision makers’ understanding of system functionality, IR contributors, contents, users, and uses. They help decision makers understand how IRs are in keeping with the institution’s mission and thereby make them more favorably inclined to the IR initiative in terms of both funding and rhetoric.

At the bottom of the list are two activities about “Waiting for a critical mass of IR implementation to happen.” Because only 28% of both PO and PPT respondents rate it “very” or “somewhat” important, it is clear that these respondents want to get involved with IRs now rather than to follow the crowd.

A wait-and-see attitude is evident in this write-in comment:

• “Waiting for clear leaders to emerge in the vendor or open-source IR options. Waiting for options that better meet our needs. Many products have potential but [are] not ready for prime time.”

Write-ins by several PO respondents reveal three investigative activities that, had they been listed on the questionnaire, would have received high ratings: (1) finding funding for IR hardware and software; (2) finding funding for IR staffing; and (3) finding expertise for IR staffing.

Two write-ins by PPT respondents describe how they are taking the initiative to study their institutions’ digital output:

• “IRs were starting to be formed on an ad hoc basis across the campus; we wanted to provide a single gathering space and search engine for these documents.”
• “Conducted study of [our] institution’s Web presence, which demonstrated a stewardship need and identified an extensive amount of potential IR content. Worked with pilot departments to add content and gauge interest.”

Two write-ins by IMP respondents capture of experience of early adopters of IR technology:

• “We agreed to become a member of the original DSpace Federation in order to test a repository system and position ourselves to engage in e-publishing activities.”
• “We were very early in our implementation, so there were few fully implemented repositories to examine. It was our provost’s desire to start a ‘faculty e-archive’ that was the primary deciding factor.”

4.2            The Needs Assessment

MIRACLE Project investigators expected that a needs assessment would be an important investigative activity that institutions would undertake before deciding to get involved with IRs. For that reason, the questionnaires featured as many as three additional questions about the needs assessment.

To our surprise, respondents ranked the needs assessment in the middle of the pack (Table 4.1). Although most evidently felt that the needs assessment was relatively unimportant compared with other activities, their answers were revealing.

About one in sixteen PO, one in four PPT, and one in three IMP institutions, respectively, have conducted a needs assessment (Figure 4.1). The percentage of respondents who do not know whether their institution conducted a needs assessment ranges from 5% to 12%. Asked whether they would be likely to conduct a needs assessment prior to making a decision about implementing an IR, about 70% of PO respondents and 44% of PPT respondents say they are “very” or “somewhat” likely to do so.

Questionnaires asked IMP respondents how important the needs assessment was for accomplishing 11 IR-related tasks. Table 4.2 lists these tasks and the percentages of respondents who told us that the needs assessment is “very” or “somewhat” important for accomplishing them. More than 75% of respondents rate all but four tasks very high in importance. At the top of that list is “Formulating IR policies.” Because “Making the decision to implement an IR” is close to the bottom of the list, we can presume that census institutions were not conducting the needs assessment to help them decide whether to implement an IR. Instead, they were conducting the assessment to discover the reception their IR would get from their institution’s learning community. One write-in comment says as much:

• “This was not a traditional needs assessment. We knew were going to implement an IR and some of the needs assessment was carried out while planning the IR.”

Table 4.2. Importance of the needs assessment

Rank	IR-related tasks	% Important
1	Formulating IR policies	90.0
2	Identifying first adopters of an IR	84.2
3	Recruiting digital content for the IR	83.3
4	Choosing an IR software package	82.4
5	Streamlining IR planning and implementation	82.4
6	Increasing faculty awareness of the IR	79.0
7	Identifying especially active contributors to the IR	77.8
8	Identifying new services to build onto the IR	72.2
9	Scheduling the rollout of various IR services	68.8
10	Making the decision to implement an IR	68.4
11	Identifying preservation techniques	62.5

A handful of respondents told us that faculty interest was key to proceeding with an IR  and that they did not necessarily have to conduct a needs assessment to find signs of such interest. Here are their comments in this regard:

• “There was no needs assessment but the IR was very much faculty driven. Leadership was taken by the University Library Council (a senate-provostial advisory group) that pushed the agenda and prepared the report that led to provost funding and support.”
• “Our assessment was more dynamic and ongoing … it involved response to innovative faculty requests and ongoing outreach from librarians regarding changes in scholarly communication practices, e.g., an e-publishing symposium hosted by the library author’s rights issues.”
• “Our former dean of faculty was particularly interested in DSpace and secured funding for the university libraries to implement and support its use here.”

4.3            Pilot Testing IR Software Packages

The PPT and IMP questionnaires asked respondents to rate the importance of various benefits of pilot testing one or more IR-system software packages. To simplify results, MIRACLE Project staff assigned weights to response categories as follows: (+2) very important; (+1) somewhat important; (0) no opinion, don’t know, or not applicable; (-1) somewhat unimportant; and (-2) very unimportant. They totaled the weights. These results were then compiled to rank order all the positions. Table 4.3 lists all 10 benefits in rank order. Except for the bottom-ranked benefit, the percentages of respondents rating benefits “very” or “somewhat” important are very high, ranging from 67% to 93%. Respondents are positive even about the bottom-ranked benefit, giving demonstrations to prospective partners, because almost 50% of them rate it “very” or “somewhat” important.

Table 4.3. Important benefits of pilot testing

Important benefits (1 to 5)	PPT	IMP
Identifying the strengths and shortcomings of available IR software	2	1
Developing the requisite technical expertise for IR implementation	1	2
Estimating costs for the technical implementation of an IR	3T*	3
Giving demonstrations to people involved in the IR implementation decision	5	4
Identifying first adopters of an IR at your institution	(6)	5
Less important benefits (6 to 10)	PPT	IMP
Preservation of your institution’s intellectual output	(3T)	6
Gauging the interest of potential contributors to the IR	7	7
Control over your institution’s intellectual output	9	8
Gauging the interest of potential IR-system users	8	9
Giving demonstrations to an institution(s) interested in partnering with us to encourage them in IR implementation	10	10
†  Parentheses indicate PPT benefits that deviated from IMP top- and bottom-ranked benefits. *   T indicates a ranked benefit that tied another benefit’s weight.

The three top-ranked benefits—developing the requisite technical expertise, learning about IR software, and estimating costs—are very practical in terms of implementing an IR. Middle- to low-middle ranked benefits pertain to potential contributors and users of the IR. MIRACLE Project investigators thought benefits pertaining to contributors especially would be ranked higher in view of the difficulty in recruiting each (see Subchapter 6.5 and Appendix F8.4), but they were not. PPT and IMP respondents’ lists of ranked benefits are almost the same. The only difference is that PPT respondents give greater importance to preserving their institution’s intellectual output. Two write-ins comment on the importance of pilot testing for collection building:

• “Building an IR collection prior to production so that on [the] release [of our operational IR] it has apparent value.”
• “Expanding student access to teaching materials in particular courses such as archaeology and botany.”

Questionnaires asked PO respondents whether they were likely to pilot test one or more IR software packages prior to implementing an IR. Figure 4.2 graphs the results.

Almost two-thirds of PO respondents are likely to pilot test. Whether the one-quarter of PO institutions that are not pilot testing or the one-eighth of PO institutions that do not know whether pilot testing is in their future are skipping directly to implementation or terminating IR-related activities is revealed by their answers to a question about their next steps pertaining to the IR effort (see Subchapter 4.4).

4.4            Next Steps Pertaining to the IR Effort

The questionnaires asked PO respondents what steps they plan to take next as a direct result of their IR planning and asked PPT respondents what their next steps are as a direct result of their IR planning and pilot testing. Table 4.4 gives the results.

Table 4.4. Next steps pertaining to the IR effort

Next steps	PO	PPT
Your institution supports implementation of an IR software package	2	1
Your institution widens the scope of its investigation into IRs	1	2
Your institution seeks funding for the next step of investigation of IRs	3	3
Your institution seeks a partner institution(s) to share in an IR	4	4
Your institution waits for a consortium, network group, or similar to implement an IR	5	5
Your institution terminates its investigation of Irs	6	6

Ranked at or near the top for PO and PPT respondents are widening the scope of their IR investigations and implementing an IR software package, respectively. Both are logical next steps given their current stages in the IR effort.

Examining percentages of respondents’ ratings conveys the strength of their convictions. Figure 4.3 graphs respondents’ ratings for the top-two ranked answer categories—implementing IR software and widening the scope of planning investigations.

Two-thirds of PPT respondents at institutions said implementing IR software is “very likely” to be their next step. None says that IR software implementation is “very unlikely,” and only a small percentage (2.6%) say such implementation is “somewhat unlikely.” Clearly, almost all PPT respondents in the MIRACLE census will be going ahead with IR implementation.

About one-sixth of PO respondents say implementing IR software is “very likely” to be their next step. Almost 50% said it is “somewhat likely.” Compared with PPT respondents, PO resondents are lukewarm about implementing IR software as their next step. Instead, widening the scope of their investigation into IRs is “very” (17.7%) or “somewhat” likely (54.4%) to be their next task.

Figure 4.4 graphs respondents’ ratings for the two middle-ranked answer categories—seeking funding and seeking partners. Large percentages of PO (65.8%) and PPT (55.5%) institutions will be seeking funding as their next step. Although large percentages of PO (42.3%) and even larger percentages of PPT (51.8%) respondents say they are unlikely to seek partners for IR implementation, several write-in responses mention possible participation in state-funded IRs. Not knowing their next step is more characteristic of PO respondents, about 10% of whom are not sure whether seeking funding or partners will be their next step.
Results for the bottom-ranked next steps—waiting for consortial developments and terminating IR involvement—are shown in Figure 4.5. Almost equal percentages of PO institutions are likely and are not likely to wait for a consortium or other group to implement an IR. PPT institutions appear to be speeding ahead with IR implementation—hardly 16% are waiting for a consortium or other group to implement an IR while about 70% are not waiting.

Percentages of PO and PPT institutions likely to terminate their investigations of IRs are very low (10.3% and 11.2%, respectively). Some respondents misinterpreted a “very” or “somewhat” likely answer to this question to mean that they would be turning their IR investigation in a different direction, for example, toward IR pilot testing or actual IR implementation, instead of terminating all IR-related activities; consequently, the percentages of census respondents who are truly terminating may be even lower than the percentages represented in Figure 4.5. For the most part, PO and PPT respondents in the MIRACLE Project census will be continuing their institutions’ IR efforts.

Figure 4.6 shows how long it will take PO and PPT respondents to make the decision to implement an IR. Overall, PO respondents will be taking longer than PPT institutions. For example, about three-quarters of PPT respondents will be making the decision within six months. The same proportion of PO respondents will be making this same decision within 12 months.

4.5            Chapter 4 Summary

Asked to rate a list of 12 investigative activities, PO, PPT, and IMP respondents put those associated with learning about IRs from the experiences of others at the top. For PPT and IMP respondents, this includes analyzing literature reviews (see Table 4.1). PO respondents rank “Demonstrating operational IRs to my institution’s decision makers” much higher than PPTs and IMPs do. Such demonstrations probably make IRs more tangible to decision makers, and, possibly, more favorably inclined to support the IR effort in rhetoric and funding. In the middle of the pack is the needs assessment. Follow-up questions reveal one in sixteen PO, one in four PPT, and one in three IMP institutions, respectively, have conducted a needs assessment (see Figure 4.1). Between 5% and 12% of respondents do not know whether their institutions have conducted a needs assessment. Asked whether they are likely to conduct a needs assessment prior to making a decision about implementing an IR, about 70% of PO respondents and 44% of PPT respondents say they are “very” or “somewhat” likely to do so.

Questionnaires asked IMP respondents how important the needs assessment was for accomplishing 11 IR-related tasks (see Table 4.2). More than 75% of respondents rate all but four tasks very high in importance. At the top is “Formulating IR policies.” Because “Making the decision to implement an IR” is close to the bottom of the list, it is likely that census institutions are not conducting the needs assessment to help them make the decision to implement an IR. Instead, they are conducting it to discover how their institution’s learning community will react to the IR.

Rating the importance of various benefits of pilot testing one or more IR-system software packages, most PPT and IMP respondents choose benefits that are very practical in terms of implementing an IR—developing the requisite technical expertise, learning about IR software, and estimating costs. Middle- to low-middle ranked benefits pertain to potential contributors and users of the IR (see Table 4.3).

Census respondents in the PPT stages of the IR effort are downright positive about implementing an IR at their institutions (see Subchapter 4.4). Their next steps are widening the scope of planning activities or implementating an IR. Most will not be waiting for a consortium, partner, or group of libraries; instead, they prefer to do IR implementation on their own. Very few will be terminating all IR-related activities.

5 IR SYSTEMS AND FEATURES

Chapter 5 tells how many institutions with institutional repositories (IRs) are implementing the IR-system software packages they have chosen, describes system features that respondents believe are satisfactory and less than satisfactory, and explains why and when IMP respondents would migrate to a new IR.

5.1            Number of IRs at Institutions

On planning and pilot testing (PPT) and implementation (IMP) questionnaires, the first question asked respondents how many IRs were available or would be available to their institution’s learning community in the near future. Table 5.1 lists the results.

Table 5.1. Number of IRs

Number of IRs	PPT		IMP
	No.	%	No.	%
1	50	72.5	37	77.1
2	12	17.4	8	16.6
3	4	5.8	3	6.3
4	1	1.4	0	0.0
5 or more	2	2.9	0	0.0
Total	69	100.0	48	100.0

Most PPT and IMP institutions have one IR, but almost a quarter have two or more IRs. PPTs probably have more than one IR because they are pilot testing IR-system software packages. Also, PPTs and IMPs may be counting the academic departments and research units that have launched IR-like software to preserve, exchange, and distribute research and teaching objects among themselves, to colleagues at other schools, and to Web searchers generally. This project’s phone interviews and case studies should ask follow-up questions to determine whether institutions with multiple IRs will eventually centralize IR services and, if so, which IR they will choose for centralization.

After respondents answered the first question, questionnaires instructed them to answer the remaining questions with the one IR in mind that offered the widest array of services to the most people and greatest number of constituencies.

5.2            IR Software

Table 5.2 enumerates the IR-system software packages that PPT and IMP respondents have pilot tested and implemented.

Table 5.2. Pilot-tested and implemented IRs

System	PPT Pilot-tested IRs		IMP Pilot-tested IRs		IMP Implemented IRs
	No.	%	No.	%	No.	%
Dspace	31	27.9	13	40.7	19	46.4
ContentDM	22	19.8	2	6.2	2	4.9
Fedora	15	13.5	3	9.4	0	0.0
Greenstone	6	5.4	3	9.4	0	0.0
Luna	6	5.4	0	0.0	0	0.0
Bepress	5	4.5	4	12.5	11	26.8
ProQuest	4	3.6	0	0.0	5	12.2
Innovative Interfaces	4	3.6	0	0.0	0	0.0
ExLibris	4	3.6	1	3.1	1	2.4
Virginia Tech ETD	3	2.7	2	6.2	1	2.4
GNU Eprints	2	1.8	1	3.1	0	0.0
Custom-made IR	2	1.8	0	0.0	2	4.9
Other	7*	6.3	3†	9.4	0	0.0
Total	111	100.0	32	100.0	41	100.0
*   California Digital Library’s Preservation Repository System, Confluence, Documentum, Dpubs, Endeavor, Opus Storage resources broker. †  Dpubs, Microtek Scanmaker 5, MS Access database and Cold Fusion script.

More institutions have pilot tested and implemented DSpace than any other IR system. This is not unexpected. DSpace was one of the first software packages specifically developed for IR services (see Appendix F2). Most Coalition for Networked Information (CNI) members, Canadian Association of Research Libraries (CARL) members, and Association of Research Libraries (ARL)-member libraries have implemented DSpace (Lynch and Lippincott 2005; Shearer 2004; Bailey et al. 2006) (see Appendix F7).

Among census respondents, Fedora and ContentDM are popular for pilot testing but not yet for implementation. Perhaps respondents have not yet had enough time to reach implementation with these packages. Developed by Berkeley Electronic Press, bepress is popular for IR implementation possibly because it hosts clients’ IRs. ProQuest recently partnered with bepress to market Digital Commons, a combination of bepress and ProQuest’s electronic theses and dissertations (ETDs). Academic institutions are starting to work with commercial firms such as Innovative Interfaces, ExLibris, and ProQuest in connection with their IR.

Two IMP respondents wrote open-ended comments about replacing their in-house IRs with commercial IR products.

• “In 1998 … only the VT-ETD system existed [and] we did not have the campus expertise to utilize [it] at that time, hence we went [with an] in-house option … In the coming year, we will be considering commercial based IR systems.”
• “[We have been] using in-house system[s] … Presently, [we are] considering outsourcing to [a] commercial platform. Ideally, [we] would migrate to system such as Ex Libris Digital. We’ll be soliciting RFPs in the coming year.”

Asked how long their IR has been operational, 52.1% of IMP respondents say 12 months or less, 27.1% from 13 to 24 months, 4.2% from 25 to 36 months, and 16.6% for more than 36 months.

IMP respondents characterize their IRs’ hosts as follows: (1) their institution only, 51.2%; (2) a for-profit vendor, 31.7%; (3) a partnership that joins their institution with one or more comparable institutions, 9.8%; and (4) a regional or state-based consortium, 7.3%.

5.3            IR-system Features

Questionnaires asked PPT and IMP respondents to rate IR systems generally or their chosen IR system, respectively, with regard to various features. To simplify results, MIRACLE Project staff assigned weights to response categories as follows: (+2) very adequate; (+1) somewhat adequate; (0) no opinion, don’t know, or not applicable; (-1) somewhat inadequate; and (-2) very inadequate. They totaled the weights. These results were then compiled to rank order all the positions. Table 5.3 uses IMP ranks to order top- (1 to 5), middle- (6 to 10), and bottom-ranked (11 to 14) features.

Table 5.3. Ranking IR-system features

Top-ranked IR-system features (1 to 5)	PPT	IMP
Supported file formats	1	1
Adherence to open-access standards	2	2
Browsing, searching, and retrieving digital content	4	3
Technical support	(10)†	4
Scalability = system growth and enhancement	(7)	5
Middle-ranked IR-system features (6 to 10)	PPT	IMP
Formulating metadata for digital documents	(5)	6
Customization	8	7
User authentication	(3)	8T*
End-user interface	9	8T
Digital preservation	6	10
Bottom-ranked IR-system features (11 to 14)	PPT	IMP
Technical documentation	11	11
Extensibility = access to other campus systems and data	12	12
Controlled vocabulary searching	13	13
Authority control	14	14
†  Parentheses indicate PPT features that deviated from IMP top, middle, or bottom ranks. *   T indicates a ranked feature that tied another feature’s weight.

PPT and IMP respondents agree about the two top-ranked IR-system features—supported file formats and adherence to open-access standards. PPT respondents rank technical support and the scalability of their pilot-test systems lower and rank user authentication higher than IMP respondents do. IR-system functionality for browsing, searching, and retrieving digital content is generally satisfactory; however, the user interface receives middle-ground grades. Because the user interface is usually connected to two bottom-ranked features, controlled vocabulary searching and authority control, IR systems could benefit from improvements to system features that users rely on to retrieve digital content.

Asked how likely they are to modify their IR’s software, about 75% of IMP respondents said that they are “very” or “somewhat” likely to do so (Figure 5.1). About 6% are “very unlikely” to make such modifications, and 6% do not know whether they will modify it.

5.4            Migrating to a New IR

When asked how long they thought their institutions would stick with their present IR software before migrating to a new system, about half the respondents skipped the question. Answers from those who responded average 3.4 years. About 56% of IMP respondents think they will migrate to new IR software within the next three years. Forty percent think they will migrate in the next four to six years. The remaining 4% may stick with their present system for seven or more years.

A follow-up question asked IMP respondents to identify reasons they would migrate to a new IR. In the  “% Important” column in Table 5.4 are respondents who gave a listed reason a “very” or “somewhat” important rating.

Table 5.4. Reasons for migrating to a new system

Rank	Reasons	% Important
1	Greater capacity for handling preservation	90.3
2	Greater opportunities for customization	86.7
3	Greater versatility with the wide range of digital formats	80.7
4	Advanced searching features	80.7
5	Friendlier user interface	77.4
6	Better tools for assisting contributors with metadata creation	74.2
7	Friendlier digital content submission procedure	74.2
8	Greater versatility for linking to other campus systems and data	71.0
9	Around-the-clock technical support	44.0

Except for around-the-clock support, IMP respondents think every reason is important. Their top-ranked reason targets improved preservation capabilities, a feature that PPT respondents rank in the top middle of the pack and IMP respondents rank in the bottom middle of the pack with respect to their current systems (see Table 5.3). The few write-in reasons cite future availability of more commercial IR systems, opportunities to participate in an IR on the consortium level, and the cost of available software.

5.5            Chapter 5 Summary

Most institutions involved with IR PPT or implementation have one IR but almost a quarter have two or more IRs (see Table 5.1). Some PPTs may have multiple IRs because they are engaged in pilot-testing activities. In addition, respondents may be counting IR-like systems at their institutions that academic and research units have launched to share research and teaching production.

More institutions have pilot tested and implemented DSpace than any other IR system; bepress is popular for implementation (see Table 5.2). Fedora and ContentDM are popular for pilot testing but not yet for implementation. Overall, respondents have implemented more than two dozen different IR systems.

Asked how long their IR has been operational, 52.1% of IMP respondents say 12 months or less, 27.1% 13 to 24 months, 4.2% from 25 to 36 months, and 16.6% for more than 36 months.

IMP respondents characterize their IR’s host as follows: (1) their institution only, 51.2%; (2) a for-profit vendor, 31.7%; (3) a partnership that joins their institution with one or more comparable institutions, 9.8%; and (4) a regional or state-based consortium, 7.3%.

PPT and IMP respondents agree on the two top-ranked IR-system features—supported file formats and adherence to open-access standards (see Table 5.3). At the bottom are controlled vocabulary searching and authority control, two features that pertain to end-user searching of IR content. Asked how likely they are to modify their IRs software, about three-quarters of IMP respondents say that they are “very” or “somewhat” likely to do so (see Figure 5.1).

IMP respondents think they will stick with their present IR system for about three-and-a-half years. Presented with a list of reasons for migrating to a new system, IMP respondents tell us that all but one (around-the-clock technical support) are important (see Table 5.4). Their top-ranked system-migration reason is greater capacity for handling preservation, a feature in their current systems that they rated in the middle (see Table 5.3).

6 IR PRACTICES AND POLICIES

Chapter 6 gives details on institutional repository (IR) practice and policies, such as the number and types of digital documents in IRs, experiences with contributors and recruiting content for the IR, managing the IR’s intellectual property (IP) rights, and the extent to which IR policies are implemented in institutions with pilot-test and operational IRs.

6.1            The Number of Digital Documents in IRs

Questionnaires asked respondents from institutions involved in IR planning and pilot testing (PPT) and IR implementation (IMP) to estimate the total number of digital documents that are published or in process in their operational or pilot-test IR. Figure 6.1 gives results.
Both pilot-test and operational IRs are very small. About 80% of PPT respondents and 50% of IMP respondents report that their IRs contain fewer than 1,000 digital documents. Only four (8.3%) IRs in the PPT stage and seven (19.4%) in the IMP stage contain more than 5,000 documents.

Thinking that older IRs would be more likely to contain more digital documents, MIRACLE Project staff compared IMP respondents’ answers on this question to their responses to questions about IR age and size. Surprisingly, we did not find a relationship between IR size and age. We added PPT data to the mix and were still unable to find a relationship. On one hand, young and old IRs may have several thousand digital documents; on the other, both young and old IRs may have only a few hundred digital documents.

6.2            Digital Document Types in IRs

PPT and IMP questionnaires listed three dozen digital document types and asked respondents to estimate how many documents per type were in their respective IRs. Table 6.1 cites averages per document type and lists them in four categories from high to low. In the center “ID” column, document types in implemented IRs are numbered from 1 to 36. In the far left-hand “ID” column, these same document-type ID numbers are repeated for document types in planning and pilot-test IRs. Except for the first-listed document type, “Doctoral dissertations,” document-type ID numbers hardly ever match. Averages for some document types are surprisingly high, so we included the number of respondents who volunteered estimates. For example, “Other learning objects … prepared by faculty, lecturers, teaching assistants, etc.” is listed third in the “PPT high” list and averages a whopping 550.0 documents per pilot-test IR; however, only four PPT respondents gave estimates and these were 0, 0, 200, and 2,000, which average to 550.0. In this case, the estimate (i.e., 2,000) given by one of a handful of respondents inflates the document type’s average.

Table 6.1. Document types in pilot-test and operational IRs

ID	PPT high: More than 200 documents	#	ID	IMP high: More than 200 documents	#
1	Doctoral dissertations (n=9)	1,288.2	1	Doctoral dissertations (n=18)	1,518.3
7	Preprints (n=7)	900.4	2	Working papers (n=18)	716.3
12	Other learning objects … prepared by faculty, lecturers, teaching assistants, etc. (n=4)	550.0	3	Journal articles (n=19)	461.5
5	Master’s theses (n=10)	229.8	4	Raw data files that result from doctoral dissertation research (n=11)	456.6
ID	PPT medium high: 51 to 200 documents	#	5	Master’s theses (n=16)	418.8
3	Journal articles (n=14)	172.2	ID	IMP medium high: 51 to 200 documents	#
2	Working papers (n=18)	124.0	6	Committee meeting agenda and minutes (n=8)	90.0
30	Your institution’s course catalogs (n=7)	109.4	7	Preprints (n=10)	84.2
21	Books (n=4)	96.3	8	Your institution’s newspapers (n=7)	80.9
20	Video recordings of performances (n=6)	76.2	9	Senior theses (n=12)	68.1
9	Senior theses (n=7)	68.3	10	Committee meeting documents, e.g., budgets, reports, memos (n=8)	67.5
ID	PPT medium low: 6 to 50 documents	#	11	Maps (n=9)	61.1
18	Faculty senate agendas and minutes (n=5)	50.6	ID	IMP medium low: 6 to 50 documents	#
24	Interview transcripts (n=6)	48.5	12	Other learning objects … prepared by faculty, lecturers, teaching assistants, etc. (n=9)	31.0
33	Your institution’s alumni publications (n=4)	42.5	13	Written papers or transcripts of conference presentations (n=12)	27.4
22	Sound recordings of interview transcripts (n=7)	35.9	14	Undergraduates’ class notes, outlines, assignments, papers, and projects (n=10)	17.6
8	Your institution’s newspapers (n=2)	33.8	15	Conference presentations (e.g., summaries, abstracts, notes, outlines) (n=10)	16.1
27	Course syllabi, class notes, handouts, etc., prepared by faculty, lecturers, etc. (n=3)	33.3	16	Interim and final reports to funders (n=9)	12.8
19	Regent, trustee, board meeting agenda and minutes (n=4)	30.0	17	College, departmental, and school alumni publications (n=9)	12.6
10	Committee meeting documents, e.g., budgets, reports, memos (n=5)	28.2	18	Faculty senate agendas and minutes (n=8)	12.5
6	Committee meeting agenda and minutes (n=5)	27.8	19	Regent, trustee, board meeting agenda and minutes (n=9)	11.1
17	College, departmental, and school alumni publications (n=4)	22.3	20	Video recordings of performances (n=12)	9.4
23	Journals (n=16)	22.0	21	Books (n=15)	6.3
13	Written papers or transcripts of conference presentations (n=3)	17.0	22	Sound recordings of interview transcripts (n=11)	6.0
15	Conference presentations (e.g., summaries, abstracts, notes, outlines) (n=4)	7.8	ID	IMP low: 5 or fewer documents	#
34	Undergraduate student e-portfolios (n=2)	7.7	23	Journals (n=6)	4.0
ID	PPT low: 5 or fewer documents	#	24	Interview transcripts (n=10)	3.3
11	Maps (n=3)	5.0	25	Raw data files that result from master’s thesis research (n=9)	2.3
31	Raw data files from senior thesis research (n=3)	5.0	26	Software (n=9)	2.2
29	Raw data files from faculty research projects (n=5)	3.3	27	Course syllabi, class notes, handouts, etc., prepared by faculty, lecturers, etc. (n=9)	1.8
26	Software (n=2)	2.5	28	Software documentation (n=9)	1.7
14	Undergraduates’ class notes, outlines, assignments, papers, and projects (n=2)	0.0	29	Raw data files from faculty research projects (n=7)	1.4
4	Raw data files that result from doctoral dissertation research (n=1)	0.0	30	Your institution’s course catalogs (n=7)	1.4
16	Interim and final reports to funders (n=1)	0.0	31	Raw data files from senior thesis research (n=8)	1.3
25	Raw data files that result from master’s thesis research (n=1)	0.0	32	Graduate students’ class notes, outlines, assignments, papers, and projects (n=8)	0.8
28	Software documentation (n=1)	0.0	33	Your institution’s alumni publications (n=9)	0.0
32	Graduate students’ class notes, outlines, assignments, papers, and projects (n=1)	0.0	34	Undergraduate student e-portfolios (n=2)	0.0
35	Graduate student e-portfolios (n=2)	0.0	35	Graduate student e-portfolios (n=7)	0.0
36	Blogs (n=1)	0.0	36	Blogs (n=8)	0.0
PPT average and total (n=4.9)		4,038.0	IMP average and total (n=9.9)		4,206.4

Although MIRACLE Project investigators were skeptical about including such a long list of digital types in the questionnaires, we are glad we did because the results show a wide range of document types in both pilot-test and operational IRs. Estimates of the various document types in both pilot-test and operational IRs are generally low, seldom exceeding 50 documents per type. Estimates for pilot-test and operational IRs are not that much different; in fact, totaling the two estimates results in hardly a 200-document difference in favor of operational IRs. With a few exceptions (e.g., the “Other learning objects…” type described above), greater numbers of respondents volunteering estimates result in document types listed in Table 6.1’s “high” and “medium-high” categories and lower numbers of respondents volunteering estimates result in document types listed in Table 6.1’s “medium-low” and “low” categories.

At the top of both the PPT and IMP lists are the traditional text-based document types that are the result of the research enterprise of faculty and students at postsecondary institutions, e.g., doctoral dissertations, master’s theses, working papers, preprints, and journal articles. Large numbers of doctoral dissertations, raw data files that result from doctoral dissertation research, and master’s theses in the IR may be the result of institutional monitoring of student compliance with mandatory submission of these document types. Respondents do not always give high estimates for document types that would be packaged in numeric and multimedia files (e.g., video recordings of performances, e-portfolios, raw data files, software, sound recordings of interview transcripts, maps), but there is evidence that numbers for nontext files will grow in the years to come.

Census respondents volunteered document types we missed. These included government documents, archives, institutional historical documents (including photographs and art history slide collections), faculty spatial data sets, staff project reports, research reports from centers and institutes, self-study reports, and other documentation from academic accreditation events, posters, newsletters, musical scores, and scrapbooks.

6.3            Status of IR Policies

PO, PPT, and IMP questionnaires asked respondents to characterize the status of 16 policies as follows: (1) no policy, (2) drafted policy, (3) implemented policy, (4) do not know, and (5) not applicable.

Sixteen IMP respondents skipped the policy question and three chose the “do not know” or “not applicable” categories. As a result, about 60% of the total 48 IMP respondents answered this question. Figure 6.2 shows the status of policies at IMP institutions.

High percentages of IMP respondents report implemented policies for (1) acceptable file formats (73.3%), (2) determining who is authorized to make contributions to the IR (68.8%), (3) defining collections (63.6%), (4) restricting access to IR content (61.3%), (5) identifying metadata formats and authorized metadata creators (61.3%), and (6) determining what is acceptable content (60.6%). Generally, IMP institutions have implemented or drafted policies for all but these four policies: (1) charging for IR services (16.7%), (2) licensing IR content (29.2%), (3) formulating a privacy policy for registered IR-system users (41.4%), and (4) providing access management services (37.1%).
It may be not necessary for all IR policies to be in place for IR implementation. IMP institutions may take a wait-and-see attitude, evaluating what transpires after a period of time, and then firming up existing policies and implementing new ones as needed. Whether this is the case will be verified in the phone interviews and case studies to be conducted in later phases of the MIRACLE Project.
Sixteen PPT respondents skipped the policy question altogether and another nine chose the “do not know” or “not applicable” categories. As a result, about 64% of the total 70 PPT respondents answered this question. Figure 6.3 shows the status of policies for PPT institutions

Except for metadata formats, percentages of implemented policies are in the single digits at PPT institutions. Between one-third and one-half of such institutions have, however, drafted policies in the following six areas:

1. Defining collections (1.9% implemented, 50.9% drafted)
2. Intellectual property (5.9% implemented, 47.1% drafted)
3. Determining what is acceptable content (2.0% implemented, 44.0% drafted)
4. Determining who is authorized to make contributions to the IR (5.8% implemented, 40.4% drafted)
5. Acceptable file formats (4.4% implemented, 37.0% drafted)
6. Metadata formats and authorized metadata creators (13.7% implemented, 35.3% drafted)

Between 60% and 95% of PPT institutions have no policies whatsoever for the bottom 10 policies listed in Figure 6.3. A handful of respondents who replied to the open-ended component of this question say as much:

• “Nothing implemented yet—still investigating.”
• “Too soon to tell.”
• “We are just in the planning stage now. We are just starting.”
• “No policy yet written.”

A few PO respondents who have gotten started on policy formulation report that they have drafted policies. One or two have implemented policies. They started with the same policies as PPT respondents did, namely:

1. Intellectual property (2.8% implemented, 12.5% drafted, 84.7% no policy)
2. Determining what is acceptable content (1.4% implemented, 11.3% drafted, 87.3% no policy)
3. Acceptable file formats (11.4% drafted, 88.6% no policy)
4. Defining collections (11.4% drafted, 88.6% no policy)
5. Determining who is authorized to make contributions to the IR (11.1% drafted, 88.9% no policy)
6. Identifying metadata formats (9.9% drafted, 90.1% no policy)

The percentage of respondents who report no policy for the remaining 10 policies ranges from 93% to 100%. Respondents at PO institutions say that it is too early to get started on policy formulation.

• “Given the stage at which the institution is at [with regard to IRs], no policies are in place.”
• “We are truly at the very beginning stages of examining this issue. We are sending a group of librarians/faculty to a scholarly communication workshop next month and will use that core group to begin investigating what we will do here.”
• “We are not yet far enough along in the planning process to be able to effectively and prudently draft policies.”

PO, PPT, and IMP questionnaires asked respondents who manages the IR’s IP rights. Because respondents could choose more than one response category, responses exceed 100%. Figure 6.4 gives the results. It reveals that library staff manage the IR’s property rights. IP rights are also in the hands of the contributors themselves. Here is what respondents had to say in this regard:

• “We basically ask our contributors to sign a license agreement that says they’ve cleared any copyright issues. So, managing the IR’s intellectual property rights is not a huge task.”
• “Authors posting their materials [in the IR] should clarify copyright status.”
• “Contributors decide.”
• “Contributor, especially regarding copyright.”
• “Submitters.”

PPT and IMP respondents volunteer “Others,” such as the university counsel and the institution’s IR vendor. Most PO and about half of PPT respondents who checked “Other” confess that they do not know, are unsure, or are not far enough along in the planning process to know the answer.

Managing IP rights can be a partnership involving IR staff who profile their institution’s IR, the system’s IP process, and the contributor who encounters this process when depositing content into the IR. Respondents’ uncertainly about IP rights may be a result of their lack of familiarity with how their IR systems usher contributors through the IP process.

The domain of IP rights in IRs deserves more coverage than can be done through Web-administered questionnaires. MIRACLE Project staff will learn more about IP in their subsequent activities.

6.4            File Formats Guaranteed in Perpetuity

PPT and IMP questionnaires listed 34 file formats and asked respondents to check which ones their IR guarantees in perpetuity. Few PPT respondents were prepared to be definitive about their responses, checking instead answer categories such as “do not know,” “no opinion,” or “not applicable.” In write-in responses, they said they could make guarantees at this time:

• “We don’t guarantee anything, at least not until we have a preservation plan in production.”
• “No guarantees yet. Too soon to say.”
• “I don’t know. ProQuest will handle the technical part.”
• “We will preserve many of these formats, but not in perpetuity. Storage options are changing too rapidly.”

Because PPT respondents are not yet far enough along in the planning process to discuss guaranteed file formats with a high degree of certainty, we excluded them from Table 6.2. This table enumerates digital file formats that at least 12.5% of IMP respondents guarantee in perpetuity. Because IMP respondents also appear to be uncertain about guaranteeing file formats, Table 6.2 includes the number of IMP respondents who failed to answer the question.

Table 6.2. Guaranteed digital file formats

File format	Skipped	*Other	Guaranteed by IMP respondents
	No.	No.	No.	%
PDF	15	6	27	56.3
JPEG	24	6	18	37.5
TIFF	22	8	18	37.5
GIF	24	10	14	29.2
XML	24	11	13	27.1
Microsoft Word	25	10	13	27.1
Microsoft Excel	26	11	11	22.9
PDF/A	22	15	11	22.9
Rich text	26	12	10	20.8
Microsoft PowerPoint	26	12	10	20.8
Postscript	24	15	9	18.8
MPEG audio	26	14	8	16.7
Plain text ANSI X3.4/ECMA–6/US-ASCII (7-bit)	26	15	7	14.8
Plain text UTF–8 (Unicode)	25	16	7	14.8
Plain text ISO 8859–x (8-bit)	26	16	6	12.5
Plain text (all other encodings, including, but not limited to, ISO 646, national variants)	26	16	6	12.5
PNG	25	17	6	12.5
TeX	25	17	6	12.5
*   Other includes responses for “do not know,” “no opinion,” and “not applicable.”

Except for PDF files, about half of IMP respondents skipped the question. Most Table 6.2 file formats handle text, numerical, or image data. Only one handles audio data. Generally fewer than 12.5% of IMP respondents guarantee multimedia formats such as QuickTime (10.4%); MPEG-4 (10.4%); Windows Media Video (6.3%) and AVI (6.3%); sound formats such as AIFF (8.3%), Real Audio (6.3%), and Wave (6.3%); and image formats such BMP (10.4%) and PhotoCD (6.3%).

One respondent comments specifically about MIRACLE census questions pertaining to preservation and about digital preservation generally:

• “It … depends on what you mean by ‘preserved.’ We will be providing bit-level preservation for all [listed] formats, but aren’t promising that the files will be usable in terms of software available. There needs to be a better definition of digital preservation in this question—it’s not just a matter of supported and unsupported formats—we have a rather complicated system for determining levels of support. I’m a little put off by the questions about digital preservation in this survey since no IR software that I know of (nor any digital content management software for that matter) provides off-the-shelf digital preservation capabilities. It takes a lot of work to build the additional infrastructure and identify the additional resources and policies needed to actually preserve items.”

MIRACLE Project investigators wanted to question census respondents about digital preservation issues. Although questionnaire drafts contained a number of long and complicated questions about preservation, only one question on this subject survived the editing and review process. Although simple, it revealed the large measure of uncertainty about preservation. Here is what we learned from it:

• Few PPT respondents are prepared to address long-term preservation issues.
• Except for PDF files, percentages of IMP respondents guaranteeing file formats in perpetuity are low.
• Percentages of IMP respondents guaranteeing in perpetuity some image formats and almost all audio and video file formats are very low (i.e., less than 12.5%).

Having learned from census results, MIRACLE Project investigators will make digital preservation a major theme in subsequent activities.

6.5            Contributors to IRs

6.5.1         Authorized Contributors

All questionnaires asked respondents about authorized contributors to IRs, but they asked them in different ways:

• NP: If your institution eventually does make the decision to implement an IR, who do you think would be authorized contributors?
• PO and PPT: If you could foretell the future, who will be authorized contributors to your institution’s IR?
• Who are authorized contributors to your institution’s IR?

Questionnaires listed a dozen choices, and respondents checked ones that were or were likely to be authorized IR contributors. Table 6.3 gives the results. It uses IMP percentages to list contributors in order from high to low, and its four “Rank” columns to the left of the “%” columns rank percentages for each of the four listed respondent types.

Table 6.3. Authorized contributors to IRs

Contributor	NP		PO		PPT		IMP
	Rank	%	Rank	%	Rank	%	Rank	%
Librarians	2	82.2	2	85.9	2	85.7	1	79.2
Faculty	1	89.0	1	87.0	1	91.4	2	77.1
Graduate students	5	45.8	5	63.0	4T	64.3	3T	56.3
Research scientists	11	26.7	7T	46.7	7	57.1	3T	56.3
Archivists	4	61.4	3	80.4	3	78.6	5	54.2
Undergraduate students	6	40.7	6	51.1	6	58.6	6	43.8
Your institution’s administrators	3	65.3	4	65.2	4T	64.3	7	39.6
Academic support staff	9	34.7	7T	46.7	9	50.0	8	37.5
Your institution’s central computer services staff	8	37.3	10	40.2	11	31.4	9T	16.7
Your institution’s press	10	27.5	11	31.5	10	44.3	9T	16.7
External contributors	12	11.9	12	14.1	12	21.4	9T	16.7
Your institution’s news service	7	38.1	9	45.7	8	52.9	12	12.5

Faculty and librarians top the list for all four respondent types. Librarians and archivists are especially likely to be active contributors on their own because they have work assignments connected with digitizing and depositing special collections in the IR. In addition, they may be proxies for faculty and research scientists who want to deposit content in the IR but have no time to do it. Surprisingly, at IMP institutions, archivists fall in the middle of the pack—below research scientists and graduate students. Why the archivist is a middle-of-the-pack authorized contributor is something that MIRACLE Project staff will pursue in subsequent activities.

Large percentages of IMP respondents acknowledge only “Research scientists” as contributors. Middle-of-the-pack percentages come from PO and PPT respondents, and NP respondents put research scientists well toward the bottom of the list. Research scientists typically staff research institutes, centers, and clinical units at research-intensive universities. Such scientists would be especially prolific at generating data sets and writing reports, white papers, conference presentations, and journal articles that would be appropriate for deposit in IRs. A large percentage of participating IMP institutions are classed as research universities (see Table 2.3); for this reason, many IMP respondents authorize research scientists as IR contributors. Research scientists are less likely to be members of the learning communities at baccalaureate and master’s institutions, and such institutions are more typical of NP, PO, and PPT institutions in the MIRACLE Project census.

The percentages of NP, PO, and PPT respondents authorizing college and university administrators as IR contributors are somewhat higher than the percentages of IMP respondents. Maybe the former are “playing up” to administrators because they need their support to implement an IR. It could also be that the administrators who serve on IR planning committees express greater optimism about the potential of IR contributions from fellow administrators than is actually present.

Less likely to be authorized as IR contributors are the institution’s news service, press, central computer services staff, academic support staff, and external contributors. External contributors rank at or almost at the bottom. A few write-in responses from IMP respondents mention that anyone—even external contributors—could submit material to their IRs as long as a faculty member or academic department is willing to sponsor them. Several write-in responses from NP, PO, and PPT respondents mention that alumni may be authorized contributors to their IRs.

6.5.2         The Major Contributor to the IR

Questionnaires asked respondents who they thought would be (PO and PPT) or who is (PPT) the major contributor to their IR. Respondents could choose only one answer category. Table 6.4 gives the results.

Table 6.4. The major contributor to the IR

Major contributor	PO		PPT		IMP
	No.	%	No.	%	No.	%
Faculty	39	48.1	37	59.7	13	33.3
Graduate students	4	4.9	2	3.2	8	20.5
Librarians	9	11.1	12	19.4	4	10.3
Undergraduate students	7	8.6	3	4.8	3	7.7
Research scientists	2	2.5	2	3.2	3	7.7
Archivists	16	19.8	5	8.1	3	7.7
Academic support staff	0	0.0	0	0.0	2	5.1
Central administrators	2	2.5	1	1.6	1	2.6
Other	2	2.5	0	0.0	2	5.1
Total	81	100.0	62	100.0	39	100.0

Although IMP respondents credit faculty with being the major contributors to their IRs, they are not overly optimistic about faculty contributions. In fact, only 33.3% of IMP respondents choose faculty as the major contributors to their IRs. PO and PPT respondents are much more positive about faculty contributions, with percentages coming close to 50% and 60%, respectively.

PO and PPT respondents do not foresee graduate students being major contributors to their IRs, but graduate students are major contributors at IRs at some IMP institutions. A large percentage of PO respondents envision archivists being major IR contributors; however, IMP respondents do not perceive archivists to be as active as other contributors to their operational IR. The large percentage of PPT respondents who choose librarians may be a result of the added workload librarians assume during the planning and PPT phase—publicizing the IR, identifying first adopters by submitting content to the IR on behalf of faculty and students, and engaging in similar activities. Only 10.3% of IMP respondents choose librarians as their IR’s major contributor. Into the “Other” category a couple of IMP respondents write about the contributions of “Publishers” and a unique collection contributed by a local association and a couple of PO respondents write about the “Media relations department” and “Academic support staff.”

6.5.3         Early Adopters of IR Technology

Asked about early adopters of IR technology, about two-thirds of PPT and one-half of IMP respondents have worked with their institution’s library or a particular academic college, department, or school. Examples are academic units in the humanities, laboratories, centers, an undergraduate symposium, and the graduate school with regard to dissertations and master’s theses. About two-fifths of PPT and one-quarter of IMP respondents have worked with their institutions’ archives.
Questionnaires asked PO and PPT respondents what digital content–recruitment methods they thought would be most successful at their institution, and asked IMP respondents to assess their methods of recruiting digital content for the IR. The majority of PO, PPT, and IMP respondents gave “very successful” ratings to only one of the nine listed methods—”Staff responsible for the IR working one-on-one with early adopters.” Figure 6.5 shows respondents’ ratings in this regard. IMP respondents were less positive about this method, with 61.1% giving it a “very successful” rating compared with PO and PPT respondents (72.5% and 79%, respectively).

Another successful content-recruiting method is word of mouth from early adopters to their colleagues in the faculty and staff ranks (Figure 6.6). IMP respondents who check the “do not know” or “no opinion” categories may find it difficult to assess this method’s success because, unlike the other recruitment methods above and below, this method does not involve them personally. PO and PPT respondents are more positive about the word of mouth method, but they probably have less recruiting experience than IMP respondents do.

6.5.4         Other Digital Content Recruitment Methods

Ratings respondents give to other digital content recruitment methods are also of interest.  Making personal visits to faculty and administrators and giving presentations about the IR at departmental and faculty meetings are less successful than working one on one with early adopters; however, respondents are still quite positive about their success using these two methods, especially the former (Figure 6.7). Perhaps the small but measurable percentage of IMP respondents checking “do not know” or “no opinion” did not engage in personal visits or presentations.
Questionnaires asked respondents about five other content-recruitment methods listed below. Most IMP respondents rated them “successful”; however, percentages of “unsuccessful” ratings are sometimes sizable, ranging from 20% to 34%. The only method of content recruitment for which most PO and PPT respondents were negative was publicizing the IR in campus newspapers.

• volunteer contributions
• institution-wide mandates regarding mandatory contribution of certain material types, e.g., doctoral dissertations, master’s theses, faculty preprints
• systematic review of faculty, staff, center, and departmental Web sites for potential contributors by staff responsible for the IR publicizing the IR during reference interactions in libraries and archives
• publicity about the IR in campus newspapers

In write-in comments, respondents volunteer a few methods that we failed to include on the questionnaires:

• “Referral from departments doing somewhat related work.”
• “Open house targeted at faculty with content.”

Write-in comments from three respondents tell how they are harvesting publicly available materials for their IRs:

• “Recently, the dean of the library approved harvesting of publicly available materials by the IR staff.”
• “Review of other open-access sites for potential IR content to be added by library staff.”
• “Prepopulation of the repository with materials from PubMed and other open databases that allow IRs to download content.”

One IMP respondent is adamant about the failure of the voluntary method:

• “So far we have made this a voluntary effort for faculty and for undergraduates. It has not really caught on for faculty and it has not been made mandatory for undergraduate theses. Voluntary does not work.”

6.6            Chapter 6 Summary

Both pilot-test and operational IRs are very small, but they contain a wide range of digital document types—text, image, audio, video, and data files (see Figure 6.1 and Table 6.1). About 80% of the pilot-test and 50% of operational IRs contain fewer than 1,000 digital documents. In the MIRACLE Project census are four (8.3%) pilot-test IRs and seven (19.4%) operational IRs containing more than 5,000 documents. There is no relationship between IR size and age. Young and old IRs may have several thousand digital documents or only a few hundred.

IRs in both the pilot-test and operational stages bear the traditional text-based document types that result from the research enterprise of faculty and students at postsecondary institutions, e.g., doctoral dissertations, master’s theses, working papers, preprints, and journal articles (see Table 6.1). Estimates of the various document types in both pilot-test and operational IRs are generally low, seldom exceeding 50 documents per type. Adding up average estimates for PPT and IMP IRs reveals that an average IR bears about 4,100 digital documents representing about 30 document types. The difference between pilot-test and operational IRs may be as small as 200 documents.

Respondents from institutions where IRs have been implemented have made the most progress on IR policy (see Figure 6.2). More than 60% of IMP respondents report implemented policies for (1) acceptable file formats (73.3%), (2) determining who is authorized to make contributions to the IR (68.6%), (3) defining collections (63.6%), (4) restricting access to IR content (61.3%), (5) identifying metadata formats and authorized metadata creators (61.3%), and (6) determining what is acceptable content (60.6%). Large percentages of IMP institutions have implemented or drafted policies for all but these four policies: (1) charging for IR services, (2) licensing IR content, (3) providing access- management services, and (4) formulating a privacy policy for registered IR system users.

At PPT institutions, the emphasis is on drafting, not implementing, policy (see Figure 6.3). Between a third and a half of PPT respondents report they have drafted policies pertaining to collections, IP, acceptable content, metadata, acceptable file formats, and authorized contributors. A few PO respondents report having drafted policies. PO respondents have started with the same policies as those on which PPT respondents report the most progress.

Asked who manages the IR’s IP rights, about 50% of PO, PPT, and IMP respondents cite library staff and about 30% cite the contributors themselves (see Figure 6.4).

Few PPT respondents are prepared to guarantee specific file formats in perpetuity (see Table 6.2). The only file formats for which there is much certainty about long-term guarantees are PDF, JPEG, TIFF, and GIF files in operational IRs.

At PO, PPT, and IMP institutions, authorized contributors to the IR are faculty, librarians, graduate students, research scientists, and archivists (see Table 6.3). Less likely to be authorized as IR contributors are the institution’s news service, press, central computer services staff, academic support staff, and external contributors.

Although IMP respondents credit faculty with being the major contributor to the IR, they are not overly optimistic about faculty contributions, as reflected in the fact that only 33.3% of IMP respondents choose faculty as the major contributor to their IRs (see Table 6.4). PO and PPT respondents are much more positive about faculty contributions, with percentages coming close to 50% and 60%, respectively.

Asked about early adopters of IR technology, about two-thirds of PPT and half of IMP respondents have worked with their institution’s library or a particular academic college, department, or school. About two-fifths of PPT and one-quarter of IMP respondents have worked with their institution’s archives.

The most successful digital content-recruitment method is staff working one-on-one with early adopters (see Figure 6.5). Other successful methods are word of mouth from early adopters to their colleagues in the faculty and staff ranks (see Figure 6.6), personal visits by IR staff to faculty and administrators, and presentations by IR staff at departmental and faculty meetings (see Figure 6.7).