Atividade física e saúde na infância e adolescência

Abstract

At Duke, the ICPSR Learning to Curate project provided an opportunity to learn best practices in social science data curation from ICPSR while evaluating the feasibility of providing data curation as a service in Duke libraries. By participating in this project, we hoped to get a better sense of the amount of effort required to identify, process and publish data collections created by Duke researchers using ICPSR's curation standards. As a result of the curation project, we have created a section of our repository explicitly for datasets and now have a much better sense of the opportunities and challenges inherent in incorporating faculty created data collections into the library's institutional repository. Data for this project came from a team of political scientists wishing to share their survey data on the characteristics of presidential donors during the 2000 and 2004 presidential elections (http://hdl.handle.net/10161/7882). The research team that conducted the study had always intended to share the data and had already documented the content of the associated data files. Additionally, we had direct access to one of the principal investigators, Alexandra Cooper, who frequently provided invaluable context for processing the study. Despite the initial work by the research team to document their work, the curation team encountered a range of challenges in processing the files. First, the documentation occasionally omitted information necessary for secondary use and occasionally did not match the values that we found in the dataset. We expect that these alignment errors are probably present in most data projects produced by large research teams since it is difficult to document for secondary usage when your goal is to produce the primary research. Another challenge we encountered on the project was dealing with issues of confidentiality. The IRB requirements prohibited directly revealing the respondents of the survey, yet the dataset initially contained a large amount of demographic and geographic information on respondents. After some negotiation, the principal investigators agreed to provide some of the demographic information as long as it could be aggregated at a level that would make disclosure much less of a concern. Overall, the Learning to Curate project benefited our data curation workflow in three ways. First, the experience allowed us to see the curation process at ICPSR firsthand, providing a much better appreciation of data curation at ICPSR and the efforts of their staff to produce quality datasets. Second, it provided us with a much better sense of the resource implications of providing quality data curation services to researchers on campus. At the end of the project, we realized that providing human mediated data curation proved extremely resource intensive. However, the project also raised many questions about the quality of data collections that do not receive a high level of processing/screening as they are archived. Finally, the project reaffirmed that the libraries could play a valuable role consulting with faculty on the best ways to manage and preserve data as a scholarly object.