E-ISSN : 2733-4538
An aim of the study was to examine the Korean version of MMPI in a psychometrical perspectives. The study attempted to more accurately reestimate the theoretical reliabilities of the MMPI subscales utilyzing the cases in the restandardization. It was found that although the computed α reliabilities for most of the scales were similar to those in the restandardizaion, α for the Mf sclae was much lower than before. It was pointed out that because the α values less than .7 estimated for many of the MMPI scales, which were not high enough for the individual disgnoses, may cause the scale scores to be unstable, diagnoes based on the scores should be done very cautiously. The item analysis revealed that the 52 items were negatively correlated with the scale scores. To explore the possiblities that the negative item-total correlations were due to the negative correlations among the dimensions of each scale, an exploratory principal component analyses were performed. The present results suggested that most of the negative item-scale score correlations did not result from the negatively correlated scale dimensions. A cross validation using the response proportions of the criterion group and the normal group confirmed the results of the item-total correlations. The results indicated that an item analysis be indispensible prior to norming. Some considerions for future research study and the limitations of the present study were discussed.