Open Scholar Stars interview series: Interview with Kathleen Gregory

In the next episode of our Open Scholar stars series, I am talking with Kathleen Gregory, Researcher and PhD Candidate at Data Archiving and Networked Services (DANS). Why is scholarly data discovery a social practice? What are the skills that are important in effective data discovery and evaluation for reuse? How the data landscape could and should change in the near future and what are the special flavors of data search in the arts and humanities disciplines? Click to read what Kathleen’s research reveals us about all these exciting issues. 

Hi Kathleen, and thanks for joining us here! Could you start off by letting us know a little bit about your background? What brought you to study scholarly data discovery practices and what are you working on now?

I have always been interested in scientific knowledge and scholarly communication, although this interest has been present in my work in different ways.

After getting an MA in Education, I spent a few years teaching chemistry before earning an MSc in Library and Information Sciences. In my work as an academic librarian, in both Germany and the United States, my research and teaching focused on scientific information literacy, where I helped students to find and evaluate scientific information and understand the process of scientific knowledge production. It was also during this time that I became involved in developing services around research data.

My current doctoral work, which is part of a multi-institutional project funded by the Netherlands Organization for Scientific Research (NWO), brings together my interests and experiences. In this project, I have been studying how researchers discover, make sense of and use data which they do not create themselves (so-called secondary data), with the overall aim of informing search solutions for research data.

After three years of working on this project, I am now in the final phase of my PhD. This means that I am busy writing my dissertation and am beginning to play with new research ideas, brainstorming projects around data metrics, the role of data in academic evaluation and data visualization literacy, to name a few.

In your recently published study, Understanding data search as a socio-technical practice, you explore data users’ strategies for locating data and evaluating it for reuse. Part of your methodology is to conduct interviews with participants involved in data search from a variety of disciplines. What are your biggest take-aways from these interviews? 

One of the most important findings from these interviews was that social interactions play a key role throughout the process of discovering and understanding data, even within disciplines which have established data infrastructures. For some participants, speaking with colleagues is simply seen as the most expedient route to finding data. 

For others, the situation was a bit more complex. For one participant, finding data to use in her computational models required visiting government agencies and establishing relationships with officials there to help her locate the data she needed. 

These types of personal interactions are not only important in finding and accessing data, but they also extend to sensemaking processes. This is a finding that I followed up on in another study (https://doi.org/10.1162/99608f92.e38165eb), where I conducted a global survey investigating data discovery practices of researchers (n=1677).

Here, I found that attending conferences and having conversations with personal networks were most commonly reported as being used to discover data, while conversations with peers was the most often selected way of making sense of data. There were also some disciplinary differences observed in these patterns. Computer scientists in the study, for example, reported using internet forums and mailing lists more often than did researchers in other disciplines. These differences were most likely the result of different disciplinary infrastructures as well as practices for communication and collaboration within the field.

Figure 1.  Social interactions used in discovering (n=3311), accessing (n=3589) and making sense of data (n=3031). Percents represent the percentage of responses for each option; multiple responses were possible. Figure from https://doi.org/10.1162/99608f92.e38165eb. 

Are scholarly data discovery practices still embedded in the traditional genre of literature search (or review) or is it becoming a scholarly activity on its own? Are researchers well connected to data repositories?

This question addresses the second takeaway I would like to highlight from both the interview study and the survey. At times, data discovery exists as a separate practice, where data seekers are searching directly in data repositories or search engines to find relevant data. However, at other times,  the practice of data discovery is closely interwoven with other data and search practices. Participants in the interview study, for example, reported finding data in the course of sharing or managing their own data.

The results of the survey study foreground the relationship between data discovery and the academic literature. Ninety-four percent of researchers in the survey indicated using the academic literature either often (75%) or occasionally (19%) to find data. The majority of those who did so reported that they follow citations from the literature to the data. This is an especially interesting finding, as data citation is still an emerging and unstandardized practice in many disciplines, and the infrastructures for citing and linking data to literature are still being developed.

Across disciplines in the survey, domain data repositories are an important source for discovering data. General search engines (i.e. Google)  tend to be equally (and sometimes even more) important. Survey respondents use search engines in many ways, including to locate known data (for example by searching for the name of a particular dataset) or to locate a particular data repository. In this instance, the search engine takes on a mediating role between the data repository and the individual searching for data.

The survey results also indicate some disciplinary differences in the sources used to find data. Arts & humanities respondents, for example, reported relying more on research support professionals for assistance in discovering data than did individuals in other disciplines. 

Do you think you can name/suggest skills that are important in effective data discovery and evaluation for reuse?

There are a few general tips that can be helpful, and some great resources already exist for researchers wanting to begin searching for data (i.e. https://www.cessda.eu/Training/Training-Resources/Library/Data-Management-Expert-Guide/7.-Discover ). Before beginning to search online, I think it is important to first get a feeling for the data repositories and sources commonly used in a field. While search engines and resources such as re3data.org can provide useful starting points to locate repositories, speaking to colleagues is probably the most valuable and effective way of locating often-used repositories within a discipline. Also, it is important to realize that because of differences in granularity and indexing, not everything may be findable using search strategies and keywords that a person commonly uses for searching the literature.

Making sense of data is a complex process on its own; it is further complicated by the fact that information about the data is often distributed in different sources – in the metadata, in related journal articles, perhaps in a data paper etc. In order to create a full contextual picture of the data, it will often require some digging, as these pieces of the puzzle are not always fully linked and integrated with each other. 

There is a recognizable political drive in the European Union to facilitate free and public access to research data hosted at publicly funded institutions. What do you see, where are the biggest gaps between openness as it is portrayed in policy documents and the current data curation, data sharing and resume practices? 

Recent work has suggested that the idea of ‘openness’ tends to be portrayed along three lines in policy discourse. These dimensions emphasize either the technological benefits of openness, the organizational benefits, where new collaborations will supposedly form around open pools of data, or the moral imperatives behind making research open [1]. 

Data curation and data sharing are complicated processes, and they do not always line up with these motivations.

As work by Levin & Leonelli shows, openness is not a one-size fits all concept; it requires that researchers make tough choices about what data to open when and for whom [2]. This recognition of difference, and the different motivations and reasons for not making something open, seems to me to be missing from the larger policy discourse. I believe that continued study of the data practices of particular communities is needed in order to continue to nuance the discussion around ‘openness.’

How we could better incentivize data sharing? 

That is a key question. I am not the first to say that increasing data sharing requires both top-down and bottom-up strategies. I would point out, however, that this language assumes that this is a process which can be incentivized, that if the right incentives are provided, more data will be shared. One person’s or communities’ incentive to share may not be the same incentive for another person or community. Understanding data communities more deeply, not just disciplinary communities, but communities which develop around, for example, shared uses of data or shared data methodologies, is an important step in trying to understand what incentives may or may not work. 

In your aforementioned study, you remark that: “Within disciplines, ideas of data ownership, data sharing regulations, data types and cultural differences also affect participants’ practices.’ – Do you also work with humanities scholars? What are your experiences about their needs and the challenges they face? 

One of the interesting findings from the survey was that many researchers identifying as arts and humanities scholars also identified as being computer scientists. I believe this reflects a larger trend, which we also see in the projects we work on at DANS (https://dans.knaw.nl), towards experimentation with new technologies in the context of digital humanities research. As arts and humanities scholars work with new types of data, which are organized and structured in different ways and require different skill sets, they are also forming collaborations and working with individuals from other disciplines. It is interesting to consider how ideas of openness and data sharing are developing and being modified in these cross-disciplinary situations – that would be an interesting area for future research!

Speaking of openness, on the other hand, we are aware that especially in the context of cultural heritage, a solid portion of the collections cannot be digitized or has not been digitized yet. How do you think you can avoid research being skewed to what is already and easily available online? 

This is indeed a challenge. Detailed descriptions, with accompanying metadata used to enhance discoverability (e.g. schema.org), could help to ensure that potential data reusers can at least learn of the existence of these data when searching online. However, we know that ease of access plays an important role in deciding whether or not researchers use data.

As more data become increasingly accessible in a digital environment, it is possible that less accessible data will not be used, and that the research lines which make use of those data will not be pursued. It is important for research communities and funders to continue to support research that makes use of these non-digital collections to counter this tendency. 

Finally: if you could give one piece of advice to a scholar or team of scholars wishing to start their data discovery, what would it be?

This may not be the most practical advice, but I would encourage them to realize that the data discovery landscape will likely be very different in a short time. New search technologies will be developed, data will be described in different ways, and data infrastructures and metadata schemas will continue to evolve. All of this will impact how researchers discover data. Some of these changes will be invisible to the data seeker, but others will result in more noticeable changes in researchers’ own practices. Secondly, I would like to encourage them to realize the value of their own personal networks in discovering and helping to make sense of data, and to make use of these resources in addition to digital technologies. 

Thank you for your time, Kathleen! It has been great to learn from your insight and experience!

[1] Noorman, M., Wessel, B., Sveinsdottir, T., & Wyatt, S. (2018). Understanding the “open” in making research data open: Policy rhetoric and research practice. In A. Rudinow Sætnan, I. Schneider, & N. Green (Eds.), The Politics and policies of big data: Big data Big Brother (pp. 292–317). Routledge.

[2] Levin, N., & Leonelli, S. (2017). How does one “open” science? Questions of value in biological research. Science, Technology, & Human Values, 42(2), 280–305. https://doi.org/10.1177/0162243916672071

Cite this article as: Erzsébet Tóth-Czifra, Kathleen Gregory: "Open Scholar Stars interview series: Interview with Kathleen Gregory," in DARIAH Open, 30/06/2020, https://dariahopen.hypotheses.org/886.

---

---