MMPI-2 Profile Code Types and Measurement Error

• Published in: Journal of personality assessment Vol. 82; no. 2; pp. 179 - 188
• Main Authors: Munley, Patrick H., Germain, Jean M., Tovar-Murray, Darrick, Borgman, Amy L.
• Format: Journal Article
• Language: English
• Published: Philadelphia, PA Lawrence Erlbaum Associates, Inc 01.04.2004; Taylor & Francis; Routledge, Taylor & Francis Group
• Subjects: Biological and medical sciences; Female; Humans; Male; Medical sciences; Minnesota Multiphasic Personality Inventory-2; MMPI - standards; Personality; Personality tests; Psychological tests; Psychology. Psychoanalysis. Psychiatry; Psychometrics. Diagnostic aid systems; Psychopathology. Psychiatry; Reliability; Stability; Techniques and methods; United States; United States; Human; Personality test; Stability; Minnesota Multiphasic Personality Inventory; Mental disorder; Psychometrics; Typology; Revised drafting; Inventory
• ISSN: 0022-3891; 1532-7752
• DOI: 10.1207/s15327752jpa8202_6

We studied simulated MMPI-2 (Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 1989) code type stability and change expected with measurement error for 12 MMPI-2 well-defined mean code type profiles. Profile scores for the 2 scales defining the code type were systematically varied to represent target code type profiles at 9 different levels of T-score profile definition. We randomly generated samples of 50 simulated, estimated true score profiles at each level of profile definition for each code type around the estimated true scores for each scale at each level of profile definition. Two sets of simulated profiles were developed. The first simulation was based on the reported means, test-retest reliabilities, and the standard errors of measurement for the MMPI-2 normative group. The second simulation was based on the means, standard deviations, and estimated retest stability for a clinical group of psychiatric patients. We calculated frequencies and percentages of simulated profiles with the highest estimated true scores on the same 2 scales as the original code type profile. Percentages of simulated profiles with the same 2 highest scales as the original code type profiles increased from 27% to 37% for the 3-point level of definition, 37% to 49% for the 5-point definition, 46% to 61% for 7-point definition, 63% to 78% for 10-point definition, 78% to 89% for 13-point definition, 83% to 93% for 15-point definition, and greater than 90% for profile definition greater than 15 points.