ISSN : 1229-0696
Since Campbell and Fiske(1959) proposed the multitrait-multimethod approach as an effective strategy for construct validation, the latent variable model using confirmatory factor analysis has become the choice in analyzing multitrait-multimethod data in recent years. In the present study, we introduce the fixed effect model and the random effect model as an advanced version of the latent variable model and show that interpretations of analysis results can be differentiated depending on whether the measurement method is designed to have a fixed effect or random effect. Furthermore, it is shown that an application of the CU`-UM process in analyzing the fixed effect model allows researchers to obtain the information of individual measures more specifically.