From “Open access to electronic theses and dissertations (ETDs),” SPARC Open Access Newsletter, July 2, 2006.
I finished my dissertation in 1977, before the web, before the internet, and even before personal computers. I typed it on an Olivetti typewriter and, when my committee accepted it, I paid the department secretary a dollar a page to retype it according to the formatting specs of the university. I was honored when the university made a copy on acid-free paper, bound it in boards, and put it on the open stacks in the main library. It even had a card in the card catalog. I was also honored when I discovered a week later that someone had stolen the copy from the library.
In addition, I sent a copy to University Microfilms International (UMI), which produced priced paper or microfilm copies on-demand. (UMI is now owned by ProQuest.) As far as I know it's still accessible for a price from UMI. I have no idea whether anyone has ever ordered a copy, let alone how many.
Unlike some other Ph.D.s (the majority? the minority?) I never mined my dissertation for publications. I was too eager to get on to other projects to publish it as a book or turn any of its chapters into articles. So it's only accessible today from UMI, on UMI's terms. If I had the text in digital form, I'd certainly want to make it OA through a suitable repository, but I honestly couldn't tell you whether that would violate the agreement I signed with UMI back in 1977. I'd have to research that question, and I don't expect that the research would be easy. But I don't have a digital copy of the text and am not likely to make one any time soon.
The Quality of Dissertations
I know firsthand that dissertation literature is valuable, and not only because my mother and I think I wrote a good one. I wrote on a fairly obscure topic for which there wasn't much existing literature—a fairly common phenomenon, given the assignment. But I found a handful of dissertations on neighboring topics in the UMI catalog and one was better than every book I found on the same subject. Unfortunately, I had to buy these dissertations in order to read them. I had to buy them even to look at them closely enough to evaluate their relevance.
Dissertations are longer than journal articles and cover their topics more comprehensively. They are more responsive to past literature than journal articles and are usually researched, refined, and revised over a longer period of time. And they're not yet salami-sliced into meaningless or trivial snippets. Indeed, they're a prime brand of the salami itself.
Dissertations are more like preprints than postprints in the sense that they're not formally peer-reviewed. But they undergo a kind of review that's at least as rigorous. If you've ever refereed a journal article, you know that the job can be done in an afternoon, and often is. Moreover, your name is rarely associated with the published (or rejected) work, and rarely known to the authors or readers. This frees referees to criticize powerful authors of flawed articles—but it also frees referees to trash powerless authors of brilliant articles. It frees referees from accountability. By contrast, your dissertation was vetted by your faculty committee for months or even years. You know who they are, and so will most readers of the final product. They feel that their own reputations are on the line, almost as much as yours is. That's why they willingly devote time and care to reviewing a dissertation and why they rigorously, almost jealously, enforce a high standard. When they certify that you have satisfied the university's requirements for originality, contribution to knowledge, and mastery of the relevant literature, their judgment is at least as well-considered, authoritative, and useful as a thumbs up from a journal referee.
Instead of devaluing dissertations because they are not formally peer-reviewed, we should see a beautiful win-win situation here. They undergo a review that is sufficiently rigorous to make them good, or to make them worth disseminating and using. But at the same time, their review is sufficiently unconventional (or sufficiently unlike journal review) to carry no publisher's investment and therefore no publisher's resistance to OA.
The Invisibility of Dissertations
Dissertations are not just good, they're largely invisible. Libraries rarely hold dissertations not written by their own students. Dissertations are not well indexed. They're available for purchase, but difficult to evaluate before purchasing. Moreover, many details from dissertations never make it into journal articles, and many dissertation topics are too narrow to justify book publication.
In short, dissertations are high in quality and low in accessibility, In fact, I'd say they constitute the most invisible form of useful literature and the most useful form of invisible literature. Because of their high quality, the access problem is worth solving.
You know what I'm building up to, but let me get there step by step.
Three Degrees of Difficulty in Achieving Open Access
Because OA to copyrighted literature requires the copyright-holder's consent, we can rank different bodies of literature according to the ease or difficulty of obtaining that consent. The low-hanging fruit—in the words of the BOAI—is the literature that “scholars give to the world without expectation of payment.” Let's say that Phase One of the OA movement is to provide OA to this kind of royalty-free literature. Because its authors don't expect to be paid, and write for impact rather than money, they can consent to OA without losing revenue. That makes it much easier for scholars to consent to OA than musicians or movie-makers.
Phase Two is to provide OA to royalty-producing literature like books. This is harder because the copyright holder must be persuaded that OA will either increase sales or bring benefits that outweigh the loss of sales. If you've been following the book-digitizing wars, you know that some authors are persuaded and some are not.
Phase Three is to reform copyright law in order to reduce permission barriers. It would help to shorten term of copyright, extend the first-sale doctrine to digital content, restore fair-use rights, nullify clickwrap licenses as contracts of adhesion, and safeguard the public domain from further prospective or retroactive enclosure. But because these steps require legislation, and are opposed by well-funded industries, they are the most difficult of all. Fortunately, they're merely desirable and not necessary for OA. We can get all we need from Phases One and Two. For cutting-edge research published in journals, we can get all we need from Phase One.
First point: Dissertations are Phase One literature, just like journal articles. Graduate students are not paid for their dissertations and can consent to OA without losing revenue. Their consent is even easier to obtain than the consent of faculty members, since dissertations are already subject to the terms and conditions of the university.
If there's a difference, it's that authors of journal articles know they'll never be paid for those texts, but some grad students plan to turn their dissertations into books that generate (or could generate) revenue. I'll return to this possibility. But note that it's the future book that's Phase Two; the dissertation is still Phase One.
Mandating OA for ETDs
I've read about 30 university web pages on ETD policies. What's remarkable is the way they list the benefits of OA (wider visibility and greater impact) among the benefits of ETDs as if OA were a natural consequence of creating the work in digital form.
In principle, universities could require electronic submission of the dissertation without requiring deposit in the institutional repository. They could also require deposit in the repository without requiring OA. But in practice, most universities don't draw these distinctions. Most universities that encourage or require electronic submission also encourage or require OA. What's remarkable is that for theses and dissertations, OA is not the hard step. The hard step is encouraging or requiring electronic submission.
For dissertations that are born digital and submitted in digital form, OA is pretty much the default. I needn't tell you that this is not at all the case with journal literature.
There are two lessons to draw from this. First, anything that fosters ETDs (as opposed to paper TDs) fosters OA to ETDs. Second, the call for OA to ETDs is not new. It's been part of the ETD movement since the beginning. If there's anything new here, it's that I'm arguing for an OA mandate, not just for OA.
Notable, explicit calls for OA to ETDs have already been made by Edward Fox and Gail McMillan (1997), Edinburgh's Theses Alive project (2004), JISC's Electronic Thesis project (2005), Richard Jones and Theo Andrew (2005), and Arthur Sale (2006). UNESCO's ETD project called for “equal access” to ETDs in 1999, but this is just another way of calling for OA, since priced access cannot be equal access. The international Digital Access to Research Theses (DART) project is committed to OA for ETDs but is just starting up its advocacy efforts.
Nine Reasons to Mandate OA for ETDs
(1) Nowadays most theses and dissertations are born digital. They're already ETDs even if the university only wants to deal with printouts.
(2) ETDs are Phase One, royalty-free works of research literature. Their authors lose no revenue by consenting to OA.
(3) ETDs are not formally published. Hence there are no publishers in the picture to resist or oppose OA. There are no publisher fears of lost revenue to answer. There are no publisher permissions to seek. There are no publisher negotiations to delay or deter OA archiving.
(4) Mandates work and exhortations don't. This is the universal lesson from OA mandates to date, whether at funding agencies or universities.
The US National Institutes of Health (NIH) has encouraged but not required OA to NIH-funded research since May 2005. It hoped that the increased flexibility would increase participation, but it had the opposite effect. In February 2006, the NIH reported to Congress that the compliance rate by its grantees was only 3.8%. The low rate led the agency's own Public Access Working Group to recommend a mandate (November 2005). The Board of Regents of the National Library of Medicine reaffirmed the call for a mandate in February 2006. And last month, the House Appropriations Committee instructed the NIH to adopt a mandate. […]
By contrast, the Wellcome Trust has mandated OA to Wellcome-funded research since October 2005 and has enjoyed a nearly 100% compliance rate.
Australia registers all accepted dissertations, giving it a good sense of the denominator, or the number of dissertations eligible for OA. The OA repositories themselves give a good sense of the numerator, or the number that are actually OA at a given time. In April 2006, Arthur Sale summarized the results of different university policies on OA for ETDs: “[V]oluntary ETD deposition results in repositories collecting less than 12% of the available theses, whereas mandatory policies are well accepted and cause deposit rates to rise towards 100%.”
(5) OA solves the invisibility problem for ETDs. Without OA, there is almost no access, visibility, or indexing for dissertations. They are hard to retrieve even if discovered, and they are hard to discover. When they are OA, ETDs are not only searchable by cross-archive search tools that index the ETD repositories, they are also indexed (in growing numbers but jerky stages) by Google, Yahoo, and Microsoft. Scirus already indexes the ETDs held by the Networked Digital Library of Theses and Dissertations (NDLTD).
By making ETDs visible, OA helps the readers who wouldn't otherwise have ready access. But it also helps the ETD authors, boosting their visibility and impact just as it does for the authors of journal articles. I don't believe that anyone has studied the OA citation advantage for ETDs, but for journal articles it ranges from 50% to 250% and it's likely to be comparable (not necessarily identical) for ETDs.
(6) Universities are in a good position to mandate OA. They can make it a simple condition of submission and acceptance.
In fact, if universities mandate OA for ETDs, their compliance rates should be higher, and grumbling lower, than mandating OA for faculty research articles. Graduate students are not as anarchical as faculty, or at least not tenured; graduate students won't be subject to countervailing pressures from publishers, at least not as often; and graduate students more likely to see the benefits of OA and the obviousness of taking advantage of the internet to disseminate research.
Universities that don't have institutional repositories can still mandate OA. The best way is to launch their own IR. But they could use a consortial or regional ETD repository, or they could have their students submit ETDs directly to NDLTD, which functions as a universal or fall-back OA repository for universities without their own. They could use the universal repository I'm setting up with the Internet Archive (delayed but still coming). Or they could use ProQuest's UMI, which will offer OA to ETDs when the authors or institutions request OA.
(7) Mandating OA for ETDs will educate the next generation of scholars about OA, when they don't already know about it. Young scholars are already more familiar with OA than older ones, at least in the sciences. But even knowledgeable young scholars may not have much experience providing OA to their own work, let alone support and reinforcement from an important research institution. An OA mandate will teach new scholars how easy it is, how beneficial it is, and how routine and expected it ought to be. It will teach them that OA is not incendiary and countercultural, but mainstream and simply useful. It will help create lifelong habits of self-archiving.
The greatest obstacle to routine self-archiving is unfamiliarity with the process, including groundless fears of the time it takes. Familiarity removes this obstacle.
(8) An OA mandate will elicit better work.
All teachers know that students work harder and do better work when they know they are writing for a real audience—large or small—beyond the teacher. The effect is amplified if they are writing for the public. Some teachers try to harness this power by telling students to write as if their work were to appear on the front page of the New York Times. Some arrange to give students a real audience beyond the teacher. In a law course in which I conducted moot court, the quality of student preparation and argument improved dramatically after I started videotaping them. I didn't even have to put the videos online; I just put them on reserve in the library for the rest of the semester.
OA gives authors a real audience beyond the dissertation committee and real incentives to do original, impressive work.
I wrote my dissertation on Kierkegaard's dissertation. The members of my committee were strong on Kierkegaard in general, but comparatively weak on his dissertation. There were many spots in my dissertation where I could have bluffed if wanted to. But even when grad students think it's safe and easy to fool their committee, it's risky and difficult to fool the world.
(9) Finally, an OA mandate shows that the university takes the dissertation seriously.
The university asks for a new and significant work of scholarship and most students deliver one. But because the university doesn't disseminate the dissertation publicly, it sends a subtle signal that it doesn't take it seriously as a work of scholarship. Of course the dissertation committee takes it very seriously as a work of scholarship, but the university itself doesn't do what it normally does when its scholars produce new and important work: it doesn't apply its publish-or-perish policy. This policy not only proclaims that research good enough for internal recognition is good enough for external distribution. It also proclaims the stronger converse that only research good enough for external distribution is good enough for internal recognition.
Universities have the same interests in promulgating excellent research by grad students as they have in promulgating excellent research by faculty, the same reasons for taking pride in it, and the same reasons for applying a publish-or-perish policy or public dissemination mandate. It wants the world to know about the quality of the work done there and it wants other researchers to benefit from it. By adopting a serious public dissemination mandate for faculty and not for doctoral students, universities invite students to draw the cynical inference that the dissertation is not so much real scholarship as a hoop to jump through, a final piece of disposable “student work,” an admission ticket to the profession, or a rite of passage.
Of course the dissertation is also an admission ticket and a rite of passage. Writing a dissertation is a lot like entering the wilderness alone, fasting to delirium, killing a wild animal, and then returning to civilization where one is welcomed as an adult. But universities should do more to send the signal that it's an admission ticket and rite of passage because it's a significant work of scholarship, not the other way around.
Students may regard the dissertation as fodder for some truly significant, adult scholarship they might publish in the future. But if so, the incentive to make it significant, adult, and public comes from a future employer, not from the institution that assigned, supervised, and approved the research.
Without an OA mandate, the university is saying that it doesn't care whether the dissertation is publicly disseminated. But if the dissertation is really a new and significant work of scholarship, then the university should care.
The message should be: If we approve a dissertation, then we think it's good. If we think it's good, then we want others to be able to find it, use it, and build on it.
Note that this message is about the purpose of universities and the value of scholarship, not about coercion. The school doesn't have to say “we're requiring OA for your sake” or even “we're requiring OA for our sake.” It's saying, “We'll do all we can to help you do good work, and then we'll do all we can to make your good work available to others.” It's about the mission of a research university.
Mandates, Coercion, and Consent
Our experience in advocating and implementing OA comes largely from the world of faculty, not the world of grad students. In the world of faculty, the best rationale for an OA mandate is to get the attention of authors. Authors control the rate of OA growth, but they're not paying attention to OA. We can't appeal to them as a bloc because they don't act as a bloc. It's not hard to persuade them, or even excite them, once we catch their attention, but it's very hard to catch their attention because they are so anarchical, overworked, and preoccupied. So we have to work through the institutions that have the greatest influence on authors.
These arguments apply even more easily to grad students than to faculty: the benefits are just as valuable and the barriers much lower.
One objection is that a mandate paternalistically coerces students for their own good. If true, this would be a serious problem for me, though perhaps not for everyone who defends mandates. I cannot support paternalism over competent adults, and I certainly put graduate students in that category. Fortunately, the paternalism objection misses the target and is easily answered. (The answer also applies to faculty mandates but here I'll elaborate it only for grad students.)
First, I only support mandates that are conditions on voluntary contracts. They might be funding contracts: if you take our money, you'll have to provide OA to your research; if this bothers you, then don't take our money. They might be employment contracts: if you work here, you'll have to provide OA to your research; if this bothers you, then don't work here. An OA mandate for ETDs would belong to the same family. If you attend this university, you'll have to provide OA to your dissertation; if this bothers you, then don't attend this university. Students who see this as a threat will go somewhere else; students who see it a promise are getting the idea.
Second, I only support mandates with reasonable exceptions. Grad students who have good reasons to be exempt from the mandate should be exempted, not coerced. (More on the exceptions themselves in the next section.)
Third, an OA mandate for ETDs advances the university's interest, not just the student's. The student interest is greater visibility and impact. The university interest is that an OA mandate will elicit better work, better show students that the university is taking the dissertation seriously as scholarship, better fulfill the university mission to share the knowledge it produces, and better assist researchers elsewhere who could benefit from this knowledge.
In short, the paternalism objection doesn't apply because the kind of OA mandate I'm talking about is fundamentally consensual, not coercive, and aims at benefits far beyond the student-authors themselves.
An OA mandate for ETDs is no more problematic than other academic requirements and considerably more mission-critical. Today universities seem more interested in mission-trivial details like the margins and font sizes of a dissertation than in its availability to others who could use it, apply it, or build on it.
Arthur Sale argues that the OA mandate should apply to all dissertations submitted as of a certain date rather than all dissertations by students who enroll as of a certain date. The two methods differ because students finish their dissertations at different rates. The first method jumps instantly to 100% compliance while the second phases in compliance over a few years. If the primary goal is rapid growth in the body of OA ETDs, then Sale is right to recommend the first method. The drawback of course is that it would change the rules for students who are already enrolled. Hence, if it's important to preserve a consent or contract basis for the OA mandate, then it's better to use the second method, announce the new policy to all new applicants, and apply it only to those who choose to enroll. On the other hand, the possibility of exemptions (see next section) may introduce a sufficient consent element to let us take Sale's recommendation as well.
Snags and Solutions
(1) Some students fear that providing OA for their ETD will disqualify it for future publication. In the world of journals, the policy to disqualify works that have already circulated as preprints is called the Ingelfinger rule. I haven't heard a special name for the analogous rule applied to ETDs but for convenience I'll use the same name here. Some students fear the Ingelfinger rule. Some fear it even though for decades most universities have submitted dissertations to UMI, which distributed copies on demand by xerography or microfilm to paying customers, a process that certainly counts as “publication” for journals that still follow the Ingelfinger rule.
The fear is justified in a small number of cases and unjustified in most. But we shouldn't harm the students whose fears are justified or simply override the fears of the rest. The solution is straightforward. Universities should require students of approved dissertations to deposit the full-text and metadata in the institution's OA repository. This should take place immediately upon final approval (say, within a couple of days or a week). The university should require immediate OA to the metadata. For the text of the dissertation, immediate OA isn't necessary, although it should be the default. Students may apply to the relevant dean for permission to delay OA to the text. They can seek a delay for the whole dissertation, when they plan to publish it as a book, or for specific chapters, if they only plan to publish journal articles. Deans should approve delays only for the affected chapters and require immediate OA for the rest of the dissertation. Deans should only approve temporary delays and make them as brief as possible. During the period of the delay, deans may temporarily block access to outside users, but they should not block access to everyone. For example, access should still be open to the student, the dissertation committee, the administration, and perhaps all authenticated users affiliated with the university.
The OA metadata helps the dissertation become known to others working in the field and could even help the author gather citations, impact, and reputation while submitting chapters to journals. More critically, most Ingelfinger fears are groundless. In 2001, Gail McMillan reviewed the literature and concluded that “if one looks at the results of the Dalton and Seaman surveys in combination with Virginia Tech's surveys of graduate student alumni, the ready availability of ETDs on the Internet does not deter the vast majority of publishers from publishing articles derived from graduate research already available on the Internet.”
(2) Some students make patentable discoveries during their doctoral research and want time to apply for a patent.
We don't have to force students to disclose their research before they've had a chance to patent it. We can use the same solution that we used for students who fear the Ingelfinger rule. The only difference may be the length of the approved delay.
(3) Some sections of the dissertation may be under copyright by others.
In one kind of case, students quote extensively from a copyrighted work, or reproduce a copyrighted illustration, and don't have permission to redistribute it. In another kind of case, a student has already published a chapter as a journal article, has transferred copyright to the journal, and doesn't have the journal's permission to redistribute it.
Here we can use the solution to the Ingelfinger problem with a few tweaks. Some OA delays may have to be permanent rather than temporary—i.e., for the life of the copyright rather than for some fairly short period like six months. Universities could require students to seek permission to reproduce the copyrighted material rather than to give up without trying. They could also require students who publish articles before finishing their dissertations either to retain key rights or to give up hope of using the articles in their dissertations. Students who would like to use the articles in their dissertations should retain the right of OA archiving. Students who try and fail to retain these rights could be required to delay journal publication until after their dissertation is approved. This would not be as onerous as it may look. Students could publish and get the rights they need either by publishing in an OA journal (gold journal) or in a non-OA journal that permitted postprint archiving (green journal), and about 70% of subscription journals already fall into the latter category. Or, with the dean's permission, students could include published articles in the version of their text used for internal review and approval, but replace the articles with citations and links in the version used for distribution and storage.
(4) Finally, a snag of a different kind. The largest obstacle to mandatory electronic submission and OA for ETDs seems to be faculty opposition. When universities give students the option to submit their dissertation electronically, well-meaning faculty advisors often caution students against it. They are thinking of the Ingelfinger rule and preservation. They want to protect their student's shot at future dissertation-based publications and they want to be sure the student's dissertation is well-preserved.
The best solution here is education for the faculty advisors. They need to know that their own Ingelfinger fears are usually groundless. They need to know that whatever anecdotal evidence they may have is negated by Gail McMillan's systematic survey evidence. (I quoted her above: “… the ready availability of ETDs on the Internet does not deter the vast majority of publishers from publishing articles derived from graduate research already available on the Internet.”)
The preservation objection is equally groundless. Paper dissertations are not like published books that exist in hundreds or thousands of copies (benefiting from the LOCKSS principle). They're usually unique and therefore vulnerable—like mine, which was stolen from the Northwestern University library. Universities could lock them up in special collections, but this is exactly the wrong model of stewardship, as if preservation and access were incompatible when the purpose of preservation is precisely to increase, facilitate, and perpetuate access. Moreover, OA to the ETD is perfectly compatible with the existence of paper copies in the university library and elsewhere and perfectly compatible with microfilm copies at UMI.
Beyond education, the university can use its policies to counteract this bad advice. First, the availability of temporary exemptions should fully answer the Ingelfinger fear. And if necessary, universities could require both electronic and paper submission in order to satisfy everyone that dissertations will be no more vulnerable in the digital future than in the paper past.
BTW, it's because faculty advisors show themselves so backward on these issues that I recommend that exemptions from the OA requirement be sought from a dean rather than from the dissertation committee.
Apart from the web pages of university ETD programs and ETD repositories, here are the sources I found most useful.
Paul Ayris, Chris Pressler, and Austin McLean, The DART-Europe Project: Toward Developing a European Theses Portal, presented at the 2005 ETD meeting, Sydney, Australia, September 28, 2005.
Charles W. Bailey, Jr., Electronic Theses and Dissertations: A Bibliography, DigitalKoans, July 8, 2005.
Charles W. Bailey, Jr., ETD Policies and Procedures at ARL Institutions, DigitalKoans, July 21, 2005.
Tim Brody, Registry of Open Access Repositories (ROAR).
Also see the page of ETD repositories (listing 70 as of June 25, 2006).
Edward A. Fox and Gail McMillan, Request for Widespread Access to ETDs, October 1997.
Edward Fox, It is easy for universities to support free culture with digital libraries: The NDLTD Example, presented at the MetaScholar Initiative (Emory University), May 5, 2005. Only the abstract is free online.
Richard Jones and Theo Andrew, Open access, open source and e-theses: the development of the Edinburgh Research Archive, Program: electronic library & information systems, 39, 3, March 2005, pp. 198–212.
Joan K. Lippincott, Institutional Strategies and Policies for Electronic Theses and Dissertations, Educause Center for Applied Research Bulletin, June 20, 2006. This came out after my June 8 talk in Quebec, but I wish it had come out earlier.
Gail McMillan, Do ETDs Deter Publishers? Does Web availability count as prior publication? A report on the 4th International Conference on Electronic Theses and Dissertations (Caltech, 2001), College and Research Libraries News, v. 62, no. 6 (June 2001).
Arthur Sale, The impact of mandatory policies on ETD acquisition, D-Lib Magazine, April 2006.
Arthur Sale, Unifying ETD with open access repositories, in Tony Carneglutti (ed.), Proceedings 8th International Electronic Theses and Dissertations Symposium, Sydney, Australia, 2005.
Nancy H. Seamans, Electronic theses and dissertations as prior publications: what the editors say, Library Hi Tech, March 2003.
Brian Surratt, ETD release/access policies in ARL Libraries: a preliminary study, presented at the 2005 ETD conference in Sydney, Australia, September 30, 2005.
—Data for this article
I want to thank Richard Fyffe, Arthur Sale, Brian Surratt, Scott Walter, and above all, Sharon Reeves, for answering my many questions about ETDs over the past many months.
This article is based on my keynote address, “Open Access for ETDs,” at the 9th International Symposium on Electronic Theses and Dissertations (Quebec City, June 7–10, 2006). My slides will be online at the site soon.
With one exception, everything from my slide presentation and talk is present and more fully developed here. The exception is that I had two slides in the talk listing universities with different policies on electronic submission and OA for ETDs. To see which universities have which policies, see the slides.