다중집단 인자분석에 관한 연구 = A Study on the Multi Group Factor Analysis

다국어 초록 (Multilingual Abstract)

Multi group factor analysis model enables concurrent estimation of a common factor in both the whole group and each group when the multivariate statistical analysis is identical or the data with similar variable structure is repeatedly established.
When the data with an identical variable structure is repeatedly established in a same object group or when the data is derived from a similar group, we term the multi group factor analysis model as a common multi group factor analysis model. Therefore the common multi group factor analysis model is a model under the assumption that the factor loading of the common factor is identical.
The rejection of the null hypothesis was proved to be inappropriate in the significant level 0.05 because the test statistic(7.0289) is smaller than the critical value(12.59). This conclusion was made after the test for goodness of fit on the estimator of the factor loading matrix and the specific factor variance matrix, in order to compare the appropriateness of models between the multi group factor analysis model and the common multi group factor analysis model. Therefore the common multi group factor analysis model is more adequate than the multi group factor analysis model.
Since the test statistic rejects the null hypothesis in some circumstances depending on the size of a sample, the comparison was made according to the goodness of fit index number and the modified goodness of fit index number, which is independent of size of a sample. The goodness of fit index number was 0.9559 in the multi group factor analysis model and 0.9673 in the common multi group factor analysis model, showing a higher index number in the common multi group factor analysis model. The modified goodness of fit index number was 0.6916 in the multi group factor analysis model and 0.8537 in the common multi group factor analysis model, showing higher index number in the common multi group factor analysis model. Therefore we can conclude that the common multi group factor analysis model is more appropriate than the multi group factor analysis model.