because the effects of treatment or stabilization in a temporary crisis are likely to be reflected in a patient's test performance (J. Graharn, Smith , & Schwartz, 1986).
Bergin (1971) has demonstrated that Scale 2 (Depression) is particularly likely to be lowered after successful treatment.
Similarly, Scale 7 (Psychasthenia) would be likely to alter according to a person's external situation. Thus, test-retest reliability may actually be an inappropriate method of evaluating these scales for certain populations.
This defense of the test's reliability is somewhat undermined by the observation that test-retest reliability is actually slightly more stable for psychiatric populations than for normals. Whereas the median range for psychiatric patients is about .80, median reliabilities for normals are about .70.
Split-half reliabilities are likewise moderate, having an extremely wide range from .05 to .96, with median correlations in the. 70s (Hunsley et al., 1988).