A1035
Title: Integration of textual and quantitative responses in multidimensional item response theory
Authors: Thomas Chan - Hong Kong University of Science and Technology (Hong Kong) [presenting]
Amanda Chu - The Education University of Hong Kong (China)
Mike So - The Hong Kong University of Science and Technology (Hong Kong)
Abstract: In the study of psychosocial health, it is common to collect responses with questions in closed and open-ended formats. Compared to the closed-ended quantitative responses, the open-ended textual responses provide extra opportunities for the respondent to provide additional information on the research of interest. However, due to the distinct nature of these two types of responses, we often separate them in analysis. The aim is to propose a multidimensional item response model that analyses textual and quantitative data together. Following the item response theory framework, the underlying latent traits of respondents are explored using the variational Bayes method. It is also studies which items contribute most to a respondent's latent traits so that we can pay extra attention to these items. The model is applied to study psychosocial health among respondents and determine the most influential items of their psychosocial health.
