More direct evidence, however, is now available from several studies. Leon, Gillum, Gillum, and Gouze (1979) examined the 30 year stability of MMPI scores for a sample 001 men initially tested at ag~ 49. Stability coefficients ranged from .28 for hypochondriasis and schizophrenia scales to .74 for social introversion, with a median correlation of .40. All correlations were statistically significant. The wide range of retest correlations in this data set might be interpreted to mean that some traits, like introversion, are stable, whereas others, like schiz.ophrenia, are not. But an alternative explanation is to be foui,ld in the nature of the instrument and the sample. The MMPI is a measure of psychopathology, not normal personality. In a normal sample (like Leon's), the range on many scales is too restricted to show substantial correlations. &cales with more "normal" content, including masculinity!femininity (r = .58) and social introversion (r = .74) show much higher correlations.