Annotation for Transparent Inference (ATI): Selecting a platform for qualitative research based on individual sources

Social scientists working in rule-bound and evidence-based traditions need to show how they know what they know. The less visible the process that produced a conclusion, the less one can see of the conclusion. A sufficiently diminished view of that process undermines the claim. What an author needs to do to fulfill this transparency obligation differs depending on the nature of the work, the data that were used, and the analyses that were undertaken. For a scholar arriving at a conclusion using a statistical software package to analyze a quantitative dataset, making the claim transparent would include providing the dataset and software commands. Research transparency is a much newer proposition for qualitative social science, especially where “granular” data are generated from individual sources, and the data are analyzed individually or in small groups. Because the data are not used holistically as a dataset, however, new ways have to be developed to associate the claims with the granular data and their analysis. The Qualitative Data Repository has been working on annotation for transparent inference (ATI) for some time, and has made considerable progress, particularly in specifying what information needs to be surfaced for readers to be able to understand and evaluate published claims. With these requirements in mind, this paper will develop a list of functional specifications and a set of criteria for choosing an annotation standard to use as the basis for ATI.