Latent Class Modeling of Children's Preference Profiles on Tangible and Graphical Robot Programming

Contribution: Prior studies on tangible versus graphical user interfaces have reported controversial findings concerning children's preferences. This paper shows that their preference profiles in the domain of introductory programming are associated with gender and age for both interfaces. Background: The relevant literature mainly consists of empirical research into the possible advantages and weaknesses of tangible user interfaces versus graphical user interfaces in the domain of programming for children. Research designs have considered aspects of usability, attractiveness and users' collaboration. Existing findings are not clear. Methodological issues, e.g., limitations of simple statistics, may have limited the clarify and usefulness of findings to date. The present study explores children's preferences profiles. Research Questions: Two research questions were posited: the first focused on the latent profiles of children's preferences for the two programming interfaces (tangible or graphical), and the second focused on the association of these profiles with gender and age. Methodology: Children ( N=148 ) aged from six to 13 years old participated in this paper. Eight variables related to the attractiveness, collaboration, and usability factors for the tangible and graphical programming interfaces were measured and analyzed by Latent Class Analysis, a model-based clustering method that can reveal distinct preference-profiles and associate them with individual differences. Findings: The patterns of responses are consistent across items for both interfaces. The graphical programming interface strongly trisects children's preferences profiles; the three latent profiles encountered were associated with gender and age.