Generalized Linear Mixed Models for the Analysis of Categorical Data

Abstract

The forced-choice response in the experiment is very common in cognitive psychology. This study provided in informal introduction to a method (generalized linear mixed models) for analysis a case about categorical data in binocular vison. The results showed that the two important factors (vertical disparity gradient and cue conflict) in binocular vision has significant influence in depth perception. Moreover, the results also revealed that the two factors have a positive effect on double fusion, but a negative effect on no fusion or single fusion. Additionally, the subjects' random effects should be involved in the model to estimate probability. These findings systematically suggested the relationship between the two factors and different depth perceptions, which corroborated and extended the previous researches.