Open AccessACQUIESCENCE AND DESIRABILITY AS RESPONSE DETERMINANTS ON THE MMPI 1
Author(s) -
Jackson Douglas N.,
Messick Samuel
Publication year - 1960
Publication title -
ets research bulletin series
Language(s) - English
Resource type - Journals
eISSN - 2333-8504
pISSN - 0424-6144
DOI - 10.1002/j.2333-8504.1960.tb00101.x
Subject(s) - acquiescence , minnesota multiphasic personality inventory , psychology , social desirability , variance (accounting) , social psychology , scale (ratio) , sample (material) , response bias , statistics , econometrics , personality , mathematics , economics , chemistry , physics , accounting , chromatography , quantum mechanics , politics , political science , law
In order to evaluate on the MMPI the respective contributions of consistent responses to item content on the one hand and of the stylistic determinants of acquiescence and desirability on the other, five new MMPI scales were developed which varied randomly with respect to content but systematically in level of judged item desirability. These scales thus represented the tendencies to endorse very desirable, somewhat desirable, neutral, somewhat undesirable, and very undesirable items, respectively. Scores for these five desirability measures and for separate true and false keys of MMPI clinical and validity scales, obtained from a sample of 201 prison inmates, were intercorrelated, factor analyzed, and rotated analytically to orthogonal simple structure. The five desirability scales correlated with each other and with the true and false MMPI keys in a manner predictable from knowledge of the average item desirability values of the scales. Two large orthogonal factors, accounting for more than 75 per cent of the common variance, were clearly identifiable as acquiescence and as desirability, since all the true MMPI keys showed higher loadings on the acquiescence factor than any false key and factor loadings for scales on the desirability factor correlated .95 with independently judged mean desirability values. Three very small factors were apparently the result of content consistencies, and three others were considered due largely to item overlap. In only one case did a true and a false key for a particular MMPI scale receive a substantial loading in the same direction on any of the small factors. Implications drawn from the results included suggestions for further evaluation and revision of MMPI scoring methods, based upon homogeneous keying and a consideration of response set effects. The importance of utilizing the multidimensionality of desirability judgments in accounting for item similarities and for response variance was also suggested.