Interactively Producing Purposive Samples for Qualitative Research using Exploratory Search

An important step in conducting qualitative research on large collections of text is reducing the size of the collection to one that is manageable. While it is common to use a variety of simple sampling methods, the limitation of these approaches is that their mechanisms do not consider the relevance of the data. We have developed exploratory search methods that leverage visual analytics to produce purposive samples of large qualitative datasets. In this paper, we outline how exploratory search strategies lead to purposive sampling, and put this in the context of the interactive information retrieval literature.