A Data-Driven Approach to Appraisal and Selection at a Domain Data Repository

Abstract

Social scientists are producing an ever-expanding volume of data, leading to questions about appraisal and selection of content given finite resources to process data for reuse. We analyze users’ search activity in an established social science data repository to better understand demand for data and more effectively guide collection development. By applying a data-driven approach, we aim to ensure curation resources are applied to make the most valuable data findable, understandable, accessible, and usable. We analyze data from a domain repository for the social sciences that includes over 500,000 annual searches in 2014 and 2015 to better understand trends in user search behavior. Using a newly created search-to-study ratio technique, we identified gaps in the domain data repository’s holdings and leveraged this analysis to inform our collection and curation practices and policies. The evaluative technique we propose in this paper will serve as a baseline for future studies looking at trends in user demand over time at the domain data repository being studied with broader implications for other data repositories.

How do properties of data, their curation, and their funding relate to reuse?

“…All studies with question text also received level 3 curation; the regression results indicate that attaching question text leads to roughly 30% more downloads than level 3 curation alone. Earlier research emphasized the importance of subject terms in data reusers searches (Pienta et al, 2018). Our findings confirm that subject terms are especially important for connecting reusers with data: each new subject term was related with a 3% increase in downloads.…”

“…Earlier research emphasized the importance of subject terms in data reusers' searches (Gregory et al, 2019; Pienta et al, 2017). Our findings confirm that subject terms are especially important for connecting reusers with data: each new subject term was related with a 3% increase in downloads.…”

“…What about the data and its curation impacts how often it is downloaded? When users look for data to use, they search by keyword or phrase (and not study name or data producer) more than two‐thirds of the time (Pienta et al, 2017); this pattern suggests that attaching subject terms to data will make them more discoverable. Data users also often turn to data that are produced by researchers or institutions they know and who have provided information about the context of the data's collection and production (Birnholtz & Bietz, 2003; Faniel et al, 2019).…”

Re-use of research data in the social sciences. Use and users of digital data archive

“…Social science data archives providing a hub for sources for secondary analyses have been established in the 1960ies in the US as well as in Europe [18,27] and for example, the CESSDA, Consortium of European social science data archives, exists since 1976 [27]. Established data archives, provide support and curation for long-term data preservation for the entire data life cycle and tools for data search [28,29].…”

DataChat: Prototyping a Conversational Agent for Dataset Search and Visualization

“…ICPSR makes a number of linked resources – including datasets, variables, and publications – available for search and discovery (Levenstein and Lyle 2018); however, links between these resources are not made directly visible to users as they search (Lafia et al 2022; Fan et al 2022). Users can currently use third‐party aggregators (e.g., Google Dataset Search) or ICPSR's web search system, which is built on a Solr index, to search through study‐level metadata, codebooks, variables, and publications (Pienta et al 2018). Most users initiate their searches through ICPSR's “Find Data” webpage, which provides a search box, a word cloud of popular search term topics, a list of the most downloaded datasets, and other features (“Find Data” 2023).…”