MSc Project Presentation: Analyzing Check-All-That-Apply (CATA) Data Using Continuous and Discrete Multivariate Methods

Kartheeka Hebbale

ABSTRACT: This project explored the analysis of Check-All-That-Apply (CATA) data - a survey data where respondents select multiple attributes that apply to them - using continuous and discrete-based statistical methods. Two datasets, Classroom data and BC Health data, were analyzed to understand how different groups’ responses varied and the associations within those responses. Traditional methods such as correspondence analysis and multidimensional scaling are utilized alongside multinomial hypothesis testing and multivariate Bernoulli logistic regression to evaluate and compare these methods. The findings showed that the continuous methods provided broad insights into group patterns and choice associations, while the discrete methods offered a more nuanced understanding of binary data. The choice of method depends on the research goal: continuous methods are ideal for general group-choice patterns whereas discrete methods are ideal for capturing the intricacies of the multivariate binary responses. These methods can also complement each other, offering a more comprehensive understanding of CATA data.