Hello, my name is Joachim Schapfer, I'm a scientist at the University of Lille in France, and my research field is information and communication science,

and I'm an expert specifically in the field of scientific information, scientific communication, academic publishing, open science, and grey literature.

I will present you results of a study we conducted in 2014, together with Ellen Kost from France.

Our observation at the time was that in spite of the growing success of the Open Access Initiative, a significant part of scientific and technical information remains unavailable

or circulates with restrictions. Even in institutional repositories created to disseminate the scientific production of an academic institution,

broad and open access to more or less important sectors of the scientific production is restricted. In order to provide new empirical evidence, 25 large institutional repositories from different continents

were selected in the international directory OpenDUR, and for each repository, we evaluated the access to the full text for different document types, and we analyzed also the statistics for each site and cumulated.

Building on our past work and new empirical data from large institutional repositories on different continents, we distinguished between different degrees of openness, which are the main reasons, which are the stabilizing functions of these situations.

Our communication tried to provide some elements of understanding, together with good practices and recommendations. The slides I will comment now are available on the gray guide.

Data are available and the full paper, too. So let me show what we did nearly 10 years ago.

It was part of research on different types of documents, especially of research on electronic theses and dissertations. We called our study,

degree of openness, gray literature in institutional repositories. And it was presented at the 2014 Gray Literature Conference at Washington.

The first slide shows a global overview of more than 2000 institutional repositories. And we just checked the presence of different types of documents, of resources in these repositories.

What you can see here is that 70% of institutional repositories contain articles,

60% electronic theses and dissertations, and then about one third books, chapters, communications, reports, working papers, very few, some persons, patents and data sets.

So what we did, we made a sample of some large institutional repositories from research organizations and universities from different countries.

And we had to look on these document types, items, and we analyzed if they were available, if the document was really deposit and available as a file, full text or not.

Reposit type, institutional content type, PhD thesis and articles at least.

So these were the both criteria. And then, of course, we had to look on the other, but we wanted only repositories with articles and thesis. And we preferred large institutional repositories, 10,000 or more items.

There are some which are a little bit smaller, but most of them are large institutional. Here are the different repositories you can see from Uppsala, Sweden, Geneva, Switzerland,

Ghent, Belgium, Stellenbosch, South Africa, Prod-Inra and Hal in France, and then other Macari and so on,

Sweden, Australia. I think this is one of the main results of our study. Each point here is one institutional repository.

And you can see on the left side the degree of openness, the little points on the upper side of the figure are institutional repositories

with a high degree of openness. All items are available, or most of them. Downside are the repositories with a low degree of openness.

Only 20 or less percents of the items are available with full text. And below is the size of the institutional repository.

On the left, smaller repositories. On the right side, larger repositories. What you can see immediately are three different groups or clusters. On the left side, on the upper side are smaller institutional repositories

with a high degree of openness. That means that most or all of the items are freely available. Downside, on the right side, you have large institutional repositories

with a very low degree, 30%, 20%, 10% or less, of openness. This means that these large repositories are only a small part of items

available with full text. And then downside, on the left side, there are some small institutional repositories with a low degree of openness.

And then there are some other which were in between, of course. For the whole sample, we compared the degree of openness of each document type.

And here are the results for what can be considered generally as gray literature, communication, conference papers and so on, thesis and dissertation, reports, working papers.

So what you can see immediately, there's of course a different communication area, thesis and dissertation, less and even less reports and working papers.

What is important here is the degree of openness. You can see for the communication, conference papers and so on, low degree of openness, similar to articles and book chapters and so on, and high degree of openness for thesis and dissertations, reports and working papers.

Especially high for working papers. Nearly all working papers in these 25 institutional repositories are available with their full text. Not only the metadata, not only the records, but the full text, the file, the document file.

You see there's a difference, a large difference between communications, dissemination of communications in these repositories and dissemination of working papers.

But there's another difference, which you can see here, between especially working papers and communications.

This is the variance, the distribution, statistical distribution. And you can see that the difference between the institutional repositories for communications are much higher. There's much more variation between repositories than for working papers.

For working papers, most of the institutional repositories have a high degree of openness, which is not the case for communication. Some of them have a high degree, others have a low degree of openness,

and then most are somewhere in between. You see for thesis and reports, situation is again somewhere in between lower average degree of openness and more differences than for working papers.

Well, you can see, working papers, most of them are published openly, freely, but are available in repositories.

When we analyze the differences between the different repositories, we distinguish between three clusters.

A high level of openness, with seven repositories, with more than 80% items with full text in each category of Kree literature,

type one in the figure. On the other side, a low level of openness, with six repositories.

These repositories obviously prefer exhaustiveness, a comprehensive collection of records to openness.

Large repositories, low degree of openness. And then, in between 12 repositories,

with average level of openness, especially for thesis and working papers, there are more open in communications.

We compared then, for each institutional repository,

the degree of openness between Kree and White literature. Kree, communication thesis, reports, working papers, White, articles, books, chapters. What you can see here is that the degree of openness

is not so different between Kree and White. There is a relatively high correlation between both variables.

What is interesting is when you take out the communications from the Kree literature,

and you consider all, only the rest, working papers, thesis, and you see here, reports, the situation is quite different.

There is no correlation between White and Kree degree of openness,

which I think can be interpreted, and you will see it immediately, that the communications are not so Kree. At least there is one part which is not Kree, more White than Kree.

Based on these results, we tried to understand the reasons why access to so many items

is restricted. There is only the record, not the document file, not the full text.

For the communications, there are three reasons. So far as we can see, the rights are transferred to publishers.

Communications are often published in conference proceedings, special journal issues or monographs, and then published not as Kree literature, but as White literature, commercially published.

And a significant number of communications are under embargo. For the PhD thesis, so far as we know from literature and from other studies,

one main reason is that the young scientist wants to publish the dissertation as a book,

and so prefers not to disseminate it on an institutional repository. There are probably other reasons, sensitive data, personal data, privacy, confidential information,

but the main reason, so far as we can see, we know from other studies,

it's a kind of competition between open access and book publishing. Reasons on the institutional side, on the side of the hosting organizations,

the main reason is that they consider, partly at least, their repository as a tool for assessment of research performance, where metadata are much more important than the full text.

So they prefer to have many metadata, even if one part of them is not linked to full text. These are reasons for restricted access. Best practices, we observed with these 25 repositories,

best practices for a high degree of openness are mandatory policies,

institutional policy in favor of open access, commitment to open access principles, partly a selection of deposit, not accepting all deposits, but to select,

similar to a digital library if you want, a specific approach for different document types, so they do not consider all document types as the same, they have different policies, for instance, for working paper and thesis, then for articles or books, and institutional workflows with curation and management

by librarians and other information professionals. These are best practices where we observed a high degree of openness.

Recommendations to improve access to gray literature are based on this study. Yes, we have five recommendations regarding typology. Metadata should clearly index the document type.

This is not always the case in the repository, so it's not always easy to distinguish between working papers, reports, and so on. Regarding discovery, the repository should allow browsing and searching

with the different document types. Regarding access rights, the availability or access restriction should be indicated, mentioned for each document.

Regarding policy, institutions should foster the deposit of metadata with the full text, and finally, they should distinguish between the different document types.

I have a different approach following this special characteristic of working papers, of reports, of dissertations, of communications, and there are other document types.

This was our study from 2014. Now we are in 2023, and when I think about the time since our study,

I think the situation has not really changed. I think there are today maybe more identifiers in institutional repositories.

I don't think that the percentage of available documents is much higher, especially for commercial items,

but the situation is changing. I think also that what we mentioned is one reason of restricted access,

low degree of openness. This is the use of repositories for research performance assessment. This is an aspect which is even more important today than nine years ago.

Repositories and research information management systems are converging, and this has an impact on the importance of metadata and the availability of documents.

So, thank you again. The paper, the presentation, and the data of the studies are all available. The link, you will find them in the grey guide.

If you have any questions, we are there. Thank you.