Description of Session
The use of open-ended questions can be a blessing or curse for mobile health platforms and the researchers who want to improve them. Qualitative data are rich, offering unexpected insights and perspectives that standard closed questions are likely to miss. But analysing open-ended questions, comments, or focus groups transcripts can also be time consuming to process, analyse and interpret. Praekelt.org’s Patient Engagement Lab have explored the use of open-ended questions in a series of surveys over the past two years. This session will provide simulated data and template code to enable participants to apply the analytical approach we developed as a way to efficiently capture key insights from these open-ended questions. Our approach uses simple Natural Language Processing techniques and descriptive statistics. Coded in a generalizable way, these techniques allow easy processing of unstructured data. The aim is to offer a quick snapshot of data content, enabling informed and selective decision-making about which data warrant in-depth qualitative analysis. We will kick off the session explaining how we have used this approach, highlighting cases where open-ended questions were found to be valuable and cases where the data were too sparse or otherwise weak.
