Open-Ended Versus Closed-Ended Responses: A Comparison Study Using Topic Modeling and Factor Analysis

For practical reasons, surveys that aim for a large number of respondents tend to restrict themselves closed-ended responses. Despite potentially bringing richer insights, the use open-ended questions poses great challenges in terms extracting useful information while significantly increasing analysis time. Nevertheless, automatic text techniques speed up In this research, we explore potential topic modelling [Latent Dirichlet Allocation (LDA) and Supervised LDA (sLDA)] extract from This is compared obtained responses, accomplished using questionnaire measures intention shared autonomous vehicles (SAVs). Two versions questionnaire- Ver_OE Ver_Lk were used, with Likert scales measuring same attitudes alternative versions. Factors extracted questions. common both questionnaire, answering had higher positive attitude towards vehicles. These attitudinal placed after When evaluating performance models predict SAVs, estimated performed better. increased further inclusion responses both, unsupervised supervised (sLDA) methods. No improvement was observed model Ver_Lk. indicate measure these