Integrating open- and closed-ended questions on attitudes towards outgroups with different methods of text analysis

Abstract

Researchers in behavioral sciences often use closed-ended questions, forcing participants to express even complex impressions or attitudes through a set of predetermined answers. Even if this has many advantages, people’s opinions can be much richer. We argue for assessing them using different methods, including open-ended questions. Manual coding of open-ended answers requires much effort, but automated tools help to analyze them more easily. In order to investigate how attitudes towards outgroups can be assessed and analyzed with different methods, we carried out two representative surveys in Poland. We asked closed- and open-ended questions about what Poland should do regarding the influx of refugees. While the attitudes measured with closed-ended questions were rather negative, those that emerged from open-ended answers were not only richer, but also more positive. Many themes that emerged in the manual coding were also identified in automated text analyses with Meaning Extraction Helper (MEH). Using Linguistic Inquiry and Word Count (LIWC) and Sentiment Analyzer from the Common Language Resources and Technology Infrastructure (CLARIN), we compared the difference between the studies in the emotional tone of the answers. Our research confirms the high usefulness of open-ended questions in surveys and shows how methods of textual data analysis help in understanding people’s attitudes towards outgroup members. Based on our methods comparison, researchers can choose a method or combine methods in a way that best fits their needs.