 Catherine Peter is the Social Science Data Librarian at the University of Southern California. She has a B.A. in Sociology, a Master's in Library and Information Science, and a graduate certificate in Geographic Information Science and Technology. And she is one of our leading specialists in all things data and research data management. Well, thank you all for coming. For today I put together a general introduction to research data management, and this is not a workshop on how to manage data, which I think will be a relief for most of you. The intention here is to answer three main questions. What is research data management? Why is it important, especially as it relates to scholarly communication? And then what are libraries doing to support it? So where can you start? And I usually begin this presentation when I'm talking to librarians discussing a specific research study. For the sake of your lunch hour, I'm going to present it just to get through it a little bit quicker. This slide has a description of a study that looks at the experiences of people in Uganda who are living with HIV and have access to antiretroviral therapy, so ART. And as librarians, we can familiarize ourselves with the study by asking some questions so we can identify the purpose of the study, the research questions, what they hope to learn. So in this case, the researchers know in the second sentence that they aim to understand how people have responded to a new chance of life, at life, and what factors enable people to adjust to living with HIV as a chronic condition. So that's what they're trying to find out. And we could also look at the data they're going to collect. So in this brief summary, they mentioned that they're using both qualitative and quantitative methods to compare people affected and unaffected by HIV and ART. So if we were to read further and have more time, you could learn that this data they're collecting, it includes life history and illness narratives. They did semi-structured interviews that were taped, transcribed, and then translated into English. And then they had survey questionnaires. So that was the data they're going to collect. And at this point, I'd like to open up and ask you as librarians to think about this data I've just presented and feel free to respond in the chat box to this question. I want you to tell me if you think there are some issues or questions that need to be considered before sharing these data. So are there any concerns or challenges you foresee? I encourage you to type these into the chat box. This is data about people's lives. So right off the bat, we've got quite a few people saying privacy, privacy, privacy, safeguarding privacy. Yeah, this is a huge concern, confidentiality. People are sharing their personal medical information. So how will they be projected if we were to share the data? So another question, or my last question on this at least, is can you think of any research that might benefit from access to these data? So who might want to reuse the data? Are there other research questions it could address? This one's a little harder. So this is a study that's asking people about their lives. There's some interview data, transcript data, the survey questionnaires, quality of life, their health experience. So one suggestion is people might want to look at long-term care. Public health. So somebody studying public health either generally or in Uganda specifically. The effectiveness of treatment. Great. This one is someone suggesting that somebody might want to replicate this study. So maybe they want to use their methods and compare and do it in another country or another aspect of health. And then you can do comparative using the methods. Linguistics and speech. This is kind of any time you have qualitative transcript data, the linguists are going to be really interested in looking at that. And especially this, which is data that's been transcribed and then translated. Great. Lots of good suggestions. This is a really interesting and diverse data set that I think has a lot of potential uses. And this study was self-archived by the researcher using open access options available through the UK data archive. You can see this is UK data share. And this is a screenshot from the item record for the data, which is quite long. In addition to this summary and the citation that we have here, if I were to scroll down the record, you'd see a lot more information. We have information on the coverage of the data and that collection method, so how they collected as we already talked about. We have information on who owns the data, the copyright holders. And in this case, they have details on some processing that was done to anonymize the data. If you notice here, it says that the consent to archive was not explicitly requested of participants. So most likely when they were collecting the data, they weren't even thinking about what they would do with it once they published their research. They were thinking about answering their research question. So they make it very clear in two places that the data was anonymized to protect the participants. If we scroll down a little further, we have the actual data files. We have two files. The first one is a .sav, which is an SPSS file, so it's the survey responses. This is a file that would be somewhat akin to a spreadsheet. And then we have these qualitative data in the form of the interview transcripts as a zip file. Below the actual data, we have documentation. And the documentation files will ideally go along with the data set to provide all the information someone would need to use the data correctly. So for a typical social science data set, this would mean you would need information on how the data was collected. So a description of your methods. You would want copies of any research instrument. So if it was a survey, you would want to share the actual survey instrument. And then a code book that might help you interpret the actual data files. Sometimes the documentation could include programming code. And you can see that these files, the naming of them, show that this is really the name the researchers were using. This isn't something that's been standardized as an archivist might. We have a lot of different file formats. So this is where I'm pretty sure that this was self-curated and not done by an archivist. And one last thing in this record is that we have links to related publications. And this is what's so fun about these sorts of archives is that we can see if we were to reuse this data, we could look at these publications and see how it's already been analyzed. And that would be really helpful. And this complete record, this archive record, is an end product of research data management. This is one of the things we are working towards. So it's a well-documented data set that is available for reuse in a data archive or a data repository. And I don't see any questions. So I'm going to move on to what I think are the takeaways from this case study. And I've identified two things I think are, I'm hoping are communicated, which is initially, for librarians, research data management is about asking questions. We really need to be listening to researchers to produce data. And in looking at this case study, we went through four questions. We looked at the purpose, we talked about the data, any sort of issues with sharing the data, and who can access it and why they might want to. In talking to researchers, in sort of the vernacular, we might not ask the questions like this. We might say, tell me about your data. Or we could flush these questions out and ask them to tell us about their research. How do you collect your data? What size is it? How is it stored? What are your data management challenges? We can ask researchers, what will you do with your data after it's complete? Who owns it? Can it be shared? So these are the kinds of questions librarians should be asking researchers, in my opinion. The second takeaway from the case study is the RDM is not something you do at the end of a project. It must be planned for before the data is collected and then integrated into the research workflow. So I'm going to talk a little bit about this some more throughout the presentation today. So what is research data management anyways? That's today's fundamental question. And in order to answer it, I think we really need to first consider what exactly research data is. Because this word data on its own is used to describe so many different things that it's almost meaningless. It's like the word thing, I would say. And research data is not statistics. Statistics are data that have already been compiled, aggregated, or calculated. This screenshot shows statistics or findings that have been grouped into a table to illustrate the findings from the study we discussed. So essentially this people shows that people in the study we read about on ART had a higher quality of life and lower depression as compared to the community controlled groups. Now research data, on the other hand, would be the raw material used to calculate these types of statistics. For example, in this screenshot, it's an excerpt of a survey data file. And it's probably similar to the survey data component in the case study and what that looks like. With each row here it's going to represent a person. So the person who has completed the study and the columns represent our variables or the things we know about them. You can see that this is probably a survey of people in the United States because the third column is responded state. But this is what the survey would look like for the case study as well. And much of the data I work with looks like this. As a social science data librarian I work with a lot of numeric data arranged in row and columns. So this sort of tabular data. But there are many other types of data as well and not to overly pander to my audience. We can look to government documents for some guidance here. So the White House defines research data as recorded factual material commonly accepted in the scientific community as necessary to validate research findings. Of course research data isn't just limited to the sciences. In the case study there were two types of data. Survey responses, so a numeric data set. And then the interview transcripts as well as the narratives. So we had this qualitative textual data. And these are fairly common data types used in the social sciences and also in our own field library science. But there are many other data formats. Too many to mention here, but some of them include experimental observations, instrument measurements, audio images and video as well as specimens, samples, artifacts, sequences. And data can be created digitally so it could be born digital or it could be digitized, something that's converted to digital format after the fact. And we've heard a lot in the news lately about big data, but it can also just be large or small. The key concept is that research data is data collected in the course of research and used as evidence of findings. So building on this definition of research data we can now look towards defining research data management. So looking at this first half of this quote from White and Ted's, research data management concerns the organization of data from its entry to the research cycle through the dissemination and archiving of valuable results. So as librarians, you're probably intimately familiar with the research lifecycle. When looking or talking about data, we often talk about this from the perspective of the data lifecycle. And the early stages of the data lifecycle look very similar to the research cycle. You start your research by planning or you design your study, design your research. You collect your data, you experiment, you measure, you observe. You process your data so you enter it, you digitize it and transcribe it, and then you analyze or interpret your data. For the data lifecycle, there are several research data management components for each of these stages. For example, at the planning stages, in addition to planning your research, you have to plan for your data management. So this might mean obtaining consent for data sharing, you might consider how you're going to maintain data security and integrity, how you're going to set up your file structure, what the naming conventions are going to be, how you're going to deal with version control and backing up your data. When you collect your data, you also, for data management, need to capture and create any metadata. So you need to document how the data is collected and your research methods. Processing includes validating and cleaning your data, anonymizing your data when necessary, and tracking any changes to the data. And of course, you also need to document your analysis. So for these initial stages, research data management concerns best practices for research workflow. So ICPSR, to demonstrate this, they're one of the Social Science Data Archives, and they have compiled their own best practice recommendations for social science data in this publicly available document. So on the right hand is a shot of a page on best practices specifically for creating quantitative data. And just to give you an example of what these are, one of the best practices is to carefully check the first 5% to 10% of the data records created and then choose random records for quality control checks throughout the process. Another best practice, in this case, anonymizing qualitative data. It provides guidelines for replacing names with generalized text. So these are just some of those best practices. And going back to our lifecycle, at this point in the research lifecycle, the researcher traditionally disseminates their findings and moves on. This isn't the end for our dataset. It doesn't belong on a thumb drive in someone's desk or it's an attachment to an email. Much research data really should be preserved. And this might entail migrating data to the best format for long-term preservation. It could be backing up and storing the data, creating any metadata and documentation like the stuff we saw in the record for the case study. Distributing data, this includes sharing the data. It is data that can't be shared with everyone such as health data or confidential data. You might provide controlled access where people apply to use the data. You might share a public version of the data that is completely anonymized. Here is where you would establish copyright and you would promote your data. And then the hope is that the data is reused and it could be used to scrutinize the findings, to conduct follow-up research, or to answer new research questions. Sometimes this is called secondary data analysis. And of course, completed studies can also be used for teaching and learning. They are very valuable for that. So research data management is a set of best practices for research workflow. And it also includes data sharing. Data is increasingly being considered part of the research output. It's a key component of scholarly communication. And as such, it should be preserved. So in concluding this section, I just want to restate that research data management practices and the best practices that they really vary according to discipline, research methods, data structures, data science, and they often are going to reflect the culture of a lab group, a research group, a university. So it's not sort of, it's not homogenous. Okay, I don't see any questions. So what's the big deal? Going back to this research data management definition, and this time looking at the second half, the point of research data management, the point of doing this is to ensure reliable verification of results and to permit new research built on existing information. So having reliable findings and creating innovative research are not new goals for research. So why are we talking about this now? And I think there's a few drivers for data management and why it's kind of something we're talking about now. And to start off, librarians are talking a lot about funder mandates for data sharing. You may have heard of the OSPP memo from 2013 that outlines the government's policy that the results of federally funded research be made available to the public. And this includes peer review publications and digital data. So to this end, several funding agencies, such as the NSF and the NIH, have begun to require things called data management plans as part of the research proposal process. So essentially a data management plan is a brief formal document that you submit with the grant application and it outlines what you're going to do with your data during and after a research project. So this here is a sample plan shared on the NIH website. Please don't try reading it. I mainly wanted to show you how short it is and that it includes a few things. It has a description of the data and how it is collected. So the data set will include self-reported demographic and behavioral data and laboratory data. And then it also includes a discussion of under what conditions they'll provide access to the data to other researchers as well as how they're going to protect the research participants. So they note the possibility of deductive disclosure of subjects. And then they identify the components of the data sharing agreement they're going to develop. So this is what other researchers who want to use the data would have to agree to. And specific data management plan requirements vary by agency, but they generally don't seem to be asking for much. Some of the plans might ask for more specific details on how or where the data will be archived or preserved. And some plans require researchers to identify the metadata standards they'll be used to document their data. But as you see here, the NIH doesn't require that much. And there are some questions in the discussions about this, about how rigorously the plans are being evaluated by various funding agencies and whether or not researchers are actually following through with the plans once they've completed their research. So there's a lot being written on this in the library literature and also quite a bit of studies being undertaken now. So if you're interested in knowing more, I'd encourage you to really look into that as well. But for our purposes, I think knowing what a data management plan suffices. Okay, so funder requirements are this stick that many librarians are talking about with research data management. But researchers seem to be talking about this a little differently. And you've probably seen the headlines about research results that couldn't be replicated or falsified and non-existent research data. And these are a huge blow to the credibility of scholarly research. And reacting to this, some researchers and universities are starting to talk about rigorous research, research transparency, and the importance of research that can be reproduced or replicated. And an example from my own university, the Academic Senate at USC recently created guidelines for researchers toward the goal of reproducibility. And this is a sort of memo that includes detailed recommendations with the intent to raise the quality and impact of research produced at USC. And they identify several guiding principles, including one that I wanted to highlight, which says that these types of activities in support of reproducible research that they should be, basically they should be, people should get credit for them. They should be considered in the tenure process. And in the actual memo, which you can read if you're interested, they discuss the importance of maintaining public trust in research investments and producing accurate results. So this is something that is on researchers' minds. And as such, it's also showing up in scholarly journal. So this is an example of a policy from the PLOS, the Public Library of Science, Open Access Journals. And it states that authors are required to make data underlining the findings available. And this is becoming increasingly common in sciences and social sciences journals to a varying degree. So when you submit your article, you have to submit the data you use to come up with your findings. And this is an example from the PLOS One journal, and I picked it because the article mentioned Dingo's. And as you notice, this is the article navigation on the side. You have that traditional science, you know, intro methods, results, discussion, conclusions. And then after that, they have a section of supporting information where they have an Excel file embedded into the online journal. And this is the data for their underlying findings. And just to clarify here, this data file here, this embedded Excel file, is different from the case study we looked at earlier. This is not a complete data set with the supporting documentation housed in an archive or data repository. This is an Excel file and it's a replication data. So it has all the variables that are needed to replicate the statistics or the findings in the article, but it's probably not the complete data set. They have other variables that maybe they didn't discuss in this article and they do not need to share them. So as a librarian, we wouldn't consider an embedded Excel file to be archived or preserved data, but it is an important component of scholarly communication and research data management. Looking at these two sticks for data management or drivers for data management, you might assume that researchers are going to be totally on board with this. And spoiler alert, just because there are variations in types of data, there are also variations in researchers' willingness to share data. And there's several reasons why a researcher might be reticent to do this. And just some of them, I think, first of all, it takes a lot of time to make sure your data is clean, well documented, anonymized, and in a format that can be shared. And this isn't something that's generally being taught in graduate school. So some researchers might not know how to do this or they might not have the time to do it. Some researchers might be insecure about sharing their data. They could be worried that someone will judge their data or worse that they might find an error. And then there could also be concerns that someone will scoop your data before you've had a chance to publish your findings. So they analyze your data and put out their own article and get all the glory. And then I think that there are some researchers who just don't see that their data has any reuse value. So given all these anti-drivers, if that's a word, what's the benefit for researchers? Where's the carrot? And if we were all altruistic, I think our hearts would be warmed by this shoulders of giants' argument of permitting new and innovative research built on existing information. But closer to earth, there is an argument to be made to researchers for sharing data in order to get credit for your work, making sure what you created is attributed to you, and that it's discoverable, that people are finding it, and seeing that you're the one who created it. And this ties into an argument that by sharing your data, you can increase your citations and the impact of your research. This is done in Web of Science. It's the data citation index, and they harvest citation details for some completed studies. I think all the ones here are ICPSR studies. But this requires that researchers be conscientious in citing data sets in their own work, and that journal require it when you use someone else's data that you fully cite it. And I guess, you know, I didn't mention it, but the other thing that comes up a lot with thinking about data management is these horror stories of people losing their data, you know, of people not backing it up correctly or not having it in a format that can be accessed, you know, once computers evolve. So that is another sort of aspect of research data management. So what are librarians doing about it? I think an unstated question here is probably should libraries be involved? And of course my answer is yes. If libraries want to support scholarly communication, they must be involved in research data management in some way. If we, you know, want to support discovery, creation, and preservation of knowledge, we can't ignore research data. And why would we want to? It's so fun. I think we need to be part of the conversation for our own relevance so that we are still, you know, relevant to modern scholarly communication. But also because we are in a position to help. You don't need to be a data scientist or even have ever opened a data file to offer support of research data management. And our librarian skillset and our university connections with researchers make us a great advocate and partner for research data management. So what is possible? Data discovery and access. This is the equivalent to what reference librarians already do, helping researchers find and evaluate information. And this is a big component of my work as a social sciences data librarian and a lot of GovDocs librarians do this as well. In my case, I've done a lot of assignment support with data, so actually working with faculty who want to incorporate data into their assignments. Libraries already deal with acquisition, licensing, and access issues for scholarly databases, and these are tasks that apply to research data sets as well. In addition to just finding data, libraries might also help users with the analysis of data. So using software, programming, creating data visualizations, maybe setting up their files when they're collecting their own data. And this level requires a higher skill set than just discovery and access. And some libraries do provide this type of support, but it's also really common for this to be provided in another unit on campus at a university. And then other terms you might have heard or you might hear related to data wrangling include software carpentry or data carpentry. So moving on up, we have data management planning and best practices. And librarians who want to support this can offer a variety of levels of support. This isn't an all or nothing thing. At the most basic, this could be something that you promote and you advocate for. So this might be more of an informational service, so you direct researchers to fund the requirements. You make them aware of data management plans. You direct them to data repositories and other relevant resources. And this is something I think most public service librarians could be doing right now. And we're going to return to this in a minute. But stepping up the level of service, libraries could also offer or sponsor trainings for researchers on how to manage their data. And then at the highest level, they could offer individual researcher support and consulting on the production of data. And of course, the documentation. So the highest level of service a library might offer is actual data curation and preservation. And data curation is often defined as adding value to data so that it can be used by others. And it can be highly technical labor intensive process of preparing research for sharing and preservation. And it will generally require some sort of infrastructure. And it follows that sort of standard archival steps of appraising a data set you decide if it warrants preservation, you ingest it, arrange it, process it, you create metadata and then you archive and provide access. And there was this pretty interesting discussion on the ISS LISCER, which is a lot of data librarians are members of ISS. And they were talking about how we define data curation. And my favorite suggestion was from Tumas Alatera at the Finnish Social Science Data Archive, who shared a possible Finnish word for data curation, which he indicated would translate into English as nursing, taking care of or tending to data, which I thought was just lovely. And going back to these level of services, when we look at research data management, especially curation, a lot of librarians seem to be sweating about it, thinking that we need to be able to curate these giant data sets, big data. But these large research projects probably have the resources and expertise to do this themselves. I think it's the smaller studies and the individual researchers that don't have the resources or the time and skill sets to address this on their own. So this is where I think there are a lot of opportunities for libraries to make an impact. And it's also where we can get some really unique data. I mean, even just thinking about the data that undergraduates collect for their senior thesis, I mean, that stuff, where does it go after they graduate? And is there some sort of, is that stuff worth sharing and preserving? I would say some of it might be. So if you want to sort of come up with a framework for which services your library provide, you could easily tie these levels of service to the research cycle or the data life cycle. You could also tie these to the data information literacy competencies. I'm not going to go into these here, but I did want to make sure you were aware of them. Let's see. Anything else on this? Oh, I presented these as levels of service, but they aren't actually hierarchical. You could do any one of these or any combination of these. So it's not required that you do all four or any of them, actually. So how do you decide which services your library should offer? And to go back to my very first point in this presentation, you really need to talk to researchers. You need to ask what their needs are, what their obstacles are. So we don't want to develop services ahead of an evidence of a need for them. And this argument was well articulated by Jake Carlson, who's written quite a bit on data management. And he talked about that libraries would be well advised to learn from their experience with institutional repositories. And the literature on institutional repositories demonstrate that the services that do not align with real-world needs of researchers will not be used. So before you decide what to do for data, you need to talk to researchers. So how should librarians talk to researchers? What can you do right now? And to start, before approaching a researcher, you can become better informed, bolster your discipline-specific data knowledge. And this isn't as hard as you probably think. I like to start before I talk to a faculty or a researcher about their data with a healthy dose of online stocking. I familiarize myself with their researcher's work, and I especially take note of the data they use in their publications or their recent publications. So you could look at the method sections of any published articles, which will do tell how they collect their data. The other things you can do include familiarizing yourself with the specific data management plan requirements for the funders in your discipline, as well as the primary data repositories and then any archives. You might want to take a look and explore any of the archives that researchers are using for publishing their data. And really quickly, for websites you might not be aware of, Spark has compiled the data sharing requirements by federal agency. You might want to take a look at that if you aren't sure which federal agencies are in your discipline. There is a pretty well known data management planning tool, which is sort of a template for creating data management plans with examples from various funders and researchers or universities can create an account. It's a free source. There are also several directories for finding out about data archives and data repositories. This is the open access directories page on data repositories. I just want to point out that they have them organized by discipline. So a data repository might focus on a specific type of data, a specific discipline, but it could also be qualitative data repository or interdisciplinary. We saw some examples of national data repositories, which the U.S. doesn't have, but many countries in Europe do. And then this is just another registry of research data repositories that you might want to look into. So once you're informed, you can then listen and ask questions. And I mentioned that for librarians, research data management is asking questions at the beginning of the presentation and provided quite a few suggested questions in those earlier slides. I favor a really informal approach to asking researchers about their data and any challenges they have. But if you want something a little more structured, I'd encourage you to take a look at the data curation profile toolkit. It includes semi-structured interview templates that are designed to assist librarians in identifying data curation needs of researchers. So they also on their website share examples of data profiles that are based on the toolkit. So it could be a great learning resource before you talk to researchers or before you ask questions. Lastly, just because we're asking questions, it doesn't mean we don't have something to contribute to the conversation. I think we have a really important role that includes advocating for data as a research output that must be managed and preserved. We can also advocate for data sharing and open data just as we advocate for open access scholarly communication. In doing these two things, we can speak the language of research ethics, so emphasizing rigor, transparency, and reproducibility. And when all else fails, we can emphasize the caret of increasing the research impact, getting credit for your work, and maybe hint at some of those data scandals, that lost data, the data that wasn't backed up correctly, or in a format that is no longer machine readable. So just because you aren't a data scientist doesn't mean we don't have something to contribute here. I think libraries have a lot to offer. And if you are watching this with your colleagues or if you wanted to sort of talk about this some more with your colleagues, I did put together some discussion questions that might get you started where you can maybe talk about where you see the library's role and any sort of training needs, things you can do now with what you have resources available. I have a few selected citations if you are looking to get into this a little further. And then for my last slide really briefly, I think we're wrapping up here. I just wanted to talk about the images I used in this presentation. I downloaded them from the National Archives Flickr page because I found them enchanting. But when I looked into the metadata, I discovered that these images, they're part of an initiative in the 70s in which the EPA hired freelance photographers to capture images relating to environmental problems, EPA activities, and everyday life. So for the EPA, these photographs served as a visual baseline for comparing how our environment looked then with how it looks now and in the future. But from a data librarian perspective, I look at these pictures and I see all of the other possibilities for reuse beyond environmental documentation. So these will be great for research on American society and history, urban change, rural societies, industrial development, poverty, social stratification, aging, and more. And it's just a great collection and I think it really illustrates the importance and the possibilities when you preserve research data for further use. So this was my marathon presentation. I would be happy to answer any questions. I did see some data librarians in the participation list. So if you have additional suggestions, things I missed, things I disagree with, I'd love to hear from you. Thank you very much, Katherine. That was great. So yeah, we have some time for questions or if you have ideas to suggest things that you've tried. I think this came privately, so I'm not sure if you saw this, but there was a comment and something that might be interesting to discuss about how we sometimes present the process of cleaning data, something that takes a long time, and that in some ways that we're reinforcing the narrative that might be problematic. Yeah, that's a great comment. I think balancing, like in my own experience as a researcher and just I can be kind of a perfectionist and I would be really nervous to share my own data and what if someone found the error. So I think balancing that openness with that perfectionism and that I could see, I think that's a great comment that I would probably be guilty of going overboard. I'm very, you know, conservative with data. Any tips for getting researchers to talk to you besides the usual? Besides the usual, I don't know. I mean, I think that I guess I have two thoughts, which is the first, my instinct as a librarian is always to offer something and, you know, to put together a workshop and, you know, people love to forward free training opportunities here. And that's been really successful for me. I was like a sort of a long game. Like I'm still nine years in, I'm still getting people who said, oh, I went to your workshop nine years ago. But I really do think that, you know, we got to be listening to researchers. So rather than trying to get them to come to me and do this thing that I think I could do for them, I really think it's about asking them what they need and going to them and saying, what are your problems? So not trying to solve their problems initially, not saying this is what I do, but saying, tell me what you need. Maybe this is something we can do. Yeah. And one of the things we've been doing that was targeted and graduate students would actually have gotten a few researchers coming to me is doing, we do joint presentations with the IRB. Oh, nice. Yeah, it's really been really successful. And so there's one person who talks about research ethics, one person who talks about mentoring, and it's geared toward graduate students. And then I go up and talk about data management and, you know, going at 20-minute presentation. Right. But I've had a lot of researchers, more researchers than graduates. Uh-huh. Come to me and say, oh, can you help me with this issue or that issue? And sometimes it's just data discovery, but other times it's just data management questions. That's great. I tried the first thing last semester. I partnered up with one of our university special collections librarians. She was talking about primary sources and archives and using archival materials. And she was talking to social scientists. So after they finished with their archival materials, I horned in and talked about data archives. And I really framed it in that perspective of comparing it to the great materials they were seeing in print. And it was really a great experience. And I got a lot of interest from the researchers really in thinking, oh, I can go find data that's out there and reuse it. And this is how I do it. Oh, and then I should maybe submit my own data. So that was an unexpected success, but really fun. Yeah. These partnerships are really where you're going to find a lot of things for your thoughts. Because our archives, we started working with a couple of people there. One person here is the born digital artist. Oh, nice. You can talk about that. The other person is the instruction archivist. So we've talked about how do we reach out to humanities, people who don't think that they have things that are data. Yeah. And that's interesting. The first time I gave this presentation, one of the art librarians, her example of what research data was for her was the sketches and the journals used before you create an artwork. And I just thought, wow, that is research data by our definition and certainly something that's worth preserving.