“…1,50,000 words). Therefore, motivated by advocacy in recent work [26,12,13,6,7] for combining topic modeling [9] and qualitative reading to analyze free-text survey questions, this study used the Latent Dirichlet allocation (LDA) model [43], the most commonly used topic model in HCI and related communities, along with close inductive qualitative reading, to identify the underlying themes and label "topics" identified by the model. Specifically, LDA is an unsupervised computational approach in which documents (in context of our study, the responses to open-ended questions) are represented as random mixtures of latent topics, where each "topic" is characterized by a distribution of words [9] that summarizes the main themes in the documents.…”