Table of Contents
Variable
The K scale was designed as a subtle measure of SDR on the MMPI. It is used to identify persons with psychopathology whose MMPI protocols appear normal.
Description
The 30 items were selected empirically. First, MMPI responses from normals were compared to those of persons with known psychopathology who scored as normals on the clinical scales. This procedure yielded 22 discriminating items. Depressed and schizo phrenic patients, however, scored low on these items. Therefore, eight items were added to differentiate these two groups from normals. In current usage, scores of 16 or above are said to suggest invalid MMPI protocols (Greene, 1980).
Samples
In the original MMPI sample of 610 normals (a cross section of Minnesota residents), McKinley et al. (I 948) reported means of I 2.8 (s.d. = 5.6) and 12.1 (s.d. =5.1) for males and females, respectively. In their mixed sample of968 psychiatric cases, the means were 14.6 (s.d. = 5.9) and 14.3 (s.d. = 5.2) for males and females, respectively. In their sample of 100 university students, the means were 16.1 (s.d. = 5.1) and 15.7 (s.d. = 5.0) for males and females, respectively.
In a more recent sample of college students, Goldberg (I 972) reported means of 15.4 (s.d. = 4.7) and 15.5 (s.d. = 4.3) for males and females, respectively. In a massive sample of 50,000 medical outpatients, Swenson et al. (1973) reported means of 15.4 (s.d. = 4.9) and 15.5 (s.d. = 4.8) for males and females, respectively.
Reliability
Internal Consistency
Gocka (1965) reported an alpha coefficient of .82 on a patient sample.
Test-Retest
Correlations range from .78 to .92 for an interval up to 2 weeks and range from .52 to .67 for intervals from 8 months to 3 years (Greene, 1980; Rorer & Goldberg, 1965).
Validity
Convergent
According to the original test constructors, the scale “was not assumed to be measuring anything which in itself is of psychiatric interest” (Meehl & Hathaway, 1946, p. 544). The validation of the scale was considered to rest on its value as a correction factor. That is, controlling other measures for K should improve their predictive validity.
In this respect, the validational evidence is weak. A few studies have shown improved validity of the MMPI scales after correcting scores as recommended in the MMPI manual (e.g., Wooten, 1984). Most studies have shown, if anything, decreases in validity (e.g., Heilbrun, 1963; McCrae et al., 1989; Yonge, 1966).
The validity of K as a measure of defensiveness is better supported but appears to vary according to the type of respondent. In maladjusted college students, there is evi dence that K indexes defensiveness (e.g., Heilbrun, 1961; Reis, 1966). In normal college students, K appears to tap a healthy positive self-image (e.g., McCrae et al., 1989; Yonge, 1966).
Location
Meehl, P. E., & Hathaway, S. R. (1946). The K factor as a suppressor variable in the Minnesota Multiphasic Personality Inventory. Journal of Applied Psychology, 30, 525- 564.
Dahlstrom, W. G., Welsh, G. S., & Dahlstrom, L. E. (1972). An MMPI handbook (Vol. 1.) Minneapolis: University of Minnesota Press.
Comments
As noted in the introduction, the K scale falls on the first factor of SDR. This association is consistent with the original conception of the K scale as a subtle measure of desirable responding. At the same time, this association suggests that at least some high scores result from a positive bias in self-image.
MMPI K Scale
Sample Items
- I like to let people know where I stand on things. (Tl TRUE FALSE)
- I have very few quarrels with members of my (Tl)
- People often disappoint (Fl)
Complete Scale
The MMPI booklet numbers for the items keyed “False” are 30, 39, 71, 89, 124, 129, 134, 138, 142, 148, 160, 170, 171, 180,183,217,234,267,
272, 296, 316, 322, 374, 383, 397, 398, 406, 461, 502. The sole item keyed
“True” is MMPI booklet number 96