The purpose of this study is to explore the possibility and the usefulness of Cognitive Diagnostic
Models (CDMs) when it is applied to psychological tests. This study uses the Multi-group
G-DINA with gender and school achievement as group classifying variables to analyze the
emotion and behavior problem test results. This is because, the Goodness of Model fit indicated
that Multi-group G-DINA is more applicable than G-DINA. Using data from the 2012 Korean
Children and Youth Panel Study (KCYPS), this study shows that girls have more internalized
behavioral problems than externalized behavioral problems, and boys have opposite characteristics.
The emotion and behavior problems become severe as the school achievement get lower, but the
highest school achievement group have strong academic anxiety. Multi-group CDMs can estimate
parameters by controlling the group effect and applicable when one item measures two or more
factors. Furthermore, using the Multi-group CDMs, researchers can compare the group
characteristics meaningfully because it offers individual and group level results simultaneously.
These features of Multi-group CDMs assure that it is practical for applying to psychological
tests.