Construct interpretation of Minnesota Multiphasic Personality Inventory (MMPI) clinical scales: An orthogonal transformation

The implications for personality test construction of the revolution in testing caused by construct validity considerations are outlined, with particular relevance to the assessment of psychopathology. These include (a) substantive definition of constructs; (b) concern for internal consistency reliability as well as generalizability; (c) evaluation of structural relationships among items and scales; (d) suppression of response biases; (e) emphasis on minimum redundancy among scales; (f) evaluation of convergent and discriminant validity of scales and profiles; and (g) evaluation of criterion validity for configurations of scales and profiles, as well as single scales. Benefits are seen as accruing to an increased understanding of psychopathology and higher levels of validity. Prior, and subsequent, to the forthcoming revision of the Minnesota Multiphasic Personality Inventory (MMPI), one approach to realizing some of the aims of construct measurement with an empirically based test is through an orthogonal transformation of the scales. Preliminary results for the extant MMPI clinical scales are reported, yielding evidence of (a) scale independence while retaining high correlations with uncorrected scales, (b) an appropriate pattern of correlations with a separate set of new scales of psychopathology, (c) a possible basis for new item analyses, and (d) freedom from correlations with a putative measure of response bias. Implications of the orthogonal transformation for profile interpretation are discussed.Portions of this paper were presnted at an invited address, 18th Annual Symposium on Recent Developments in the Use of the MMPI, Minneapolis, April 9, 1983. This paper was written while Douglas N. Jackson was distinguished visiting professor at the College of Education, The University of Iowa. This research has been supported by Research Grant 895-84/86 from the Ontario Mental Health Foundation, Research Grant 411-83-0014 from the Social Sciences and Humanities Research Council of Canada, and the Alberta Hospital Edmonton.