Descriptive Metadata for Scientific Data Repositories: A Comparison of Information Scientist and Scientist Organizing Behaviors

Abstract

AbstractThe integration of scientific data sets into repository collections is a continuing challenge for research institutions. This study used a concurrent triangulation mixed-methods approach to gather basic quantitative and qualitative data about how 11 information professionals and 16 scientists organized data sets for personal and repository use. The results incorporate comments from study participants that elaborate on descriptive metadata use, data curation practices, and organization styles/preferences.KEYWORDS: data repositoriesDryadinformation organizationmetadata applicationpersonal information management APPENDIX ASeven Steps Performed by ParticipantsThis process is outlined in the seven steps that follow.1. Scientists and information professionals were divided into two groups.2. Participants were sent study instructions, two data sets, and two questionnaires.3. Each participant answered the basic demographic questionnaire.4. Each participant simulated integrating the data set into the participant's own collection.5. Each participant answered a questionnaire describing the organizing output process performed in Step 4.6. Each participant sent questionnaires, data sets, and any other output created in steps 4 and 5 to the researcher.7. Each participant submitted data sets into Dryad and created descriptive metadata and subject terms.Once participants were recruited and consent forms were completed, participants were assigned to either Group L or Group S. This grouping follows a quasi-experimental methodology where the two populations are divided into naturally occurring groups. Group L members were information professionals and Group S members were scientists. Group names did not affect the study implementation but were used to generalize research results during analysis.APPENDIX BAtlas.ti CodingUsing the Coding Analysis Toolkit (CAT), coding from both coders was compared for basic frequency, exact match, overlap, and Kappa scores. Table B1 highlights the results of the CAT output. The first column lists the codes used to mark up the narratives. The second column lists the number of times the corresponding code was applied by the first coder. The third column lists the number of times the corresponding code was applied by the second coder. The fourth column lists the number of exact matches for the corresponding code between Coder 1 and Coder 2. Exact match includes not only the use of the code, but the exact highlight length used on a sentence or word that was coded. The fifth column lists the number of times coding overlapped between the two coders for the corresponding code. Overlap is not an exact match, but it means that part of the two coders' highlights overlapped when assigned that specific code. The sixth column lists the Kappa score for each code. Kappa is measured in percentages. The last column lists the Kappa score with overlap as a consideration.The codes highlighted above were used to gain more insight into participant descriptive metadata creation and subject term application. Coding in relationship to narrative responses is discussed in more depth in the Discussion section.CAT Analysis of Atlas.ti Codes Applied by Two CodersDownload CSVDisplay TableAPPENDIX CNarrative Coding InstructionsThere are 27 documents that need to be coded. Documents are a combination of narrative, short answers, and lists. The question prompt is included for each response. Questions are italicized and should not be coded. Only the plain font (no italics) should have coding applied to it.How to apply codingEach Word document represents one participant's response. Each response needs to be coded using the codebook provided.Please use the Word Commenting feature to add a code. Highlight the section of text that the code will apply to, then under the Review tab, select New Comment. Write the appropriate code in the comment box. When finished coding a document, resave it. Please send all newly coded documents back to Hollie once you are done.CodebookA total of 22 codes have been used for this study. Codes can be used multiple times in one narrative. Codes applied should come from the codebook provided below. TableDownload CSVDisplay TableNotes1Coding Analysis Toolkit: http://cat.ucsur.pitt.edu/