The latent trait structure of the New Japanese Version of the Minnesota Multiphasic Personality Inventory (MMPI) was examined with item response theory (IRT). Item difficulty and item discrimination in the two-parameter logistic model were estimated from the data of a national norm group of 522 women and 500 men, who individually filled a booklet form. Of all clinical scales, except Mf and Si, test informations curves (TICs) obtained from all the items and from only items of good fit both had maximum information around 2 on trait estimate (θtheta), which corresponds to a T score of 70. The T score of 70 is generally considered a high scale score, or an index of abnormality. Thus, with the results from an IRT analysis, it became clear that all clinical scales, except Mf and Si, can discriminate a clinical group from a normal one at the high T score, and that an adaptive test of the MMPI could be developed.
Key words: item response theory, Minnesota Multiphasic Personality Inventory (MMPI) the New Japanese Version, test information curve, adaptive testing