Reading Between the Lines of Qualitative Data – How to Detect Hidden Structure Based on Codes

Reading Between the Lines of Qualitative Data – How to Detect Hidden Structure Based on Codes

While qualitative research is experiencing broad acceptance in the information systems discipline, growing volumes of heterogeneous data pose challenges to manual qualitative analysis. We introduce an unsupervised machine learning approach based on graph partitioning to detect hidden information and structure in qualitative data samples. With the clustering technique, we map coded data to a graph and formulate a partitioning problem which is solved by integer linear programming. As a result, clusters of information sources are identified based on similarities given in the coded data. We demonstrate the approaches’ ability to detect hidden information in coded qualitative data by application on coded interview transcripts. With the approach, we draw on a technique from the operations research discipline and expand the repertoire of approaches being used to analyze qualitative data in the context of information systems.