(1989) examined the ADI's reliability and validity across three independent samples of adolescents between the ages of 12-17 years. The first sample included 261 adolescents referred for evaluation of psychological, emotional, or behavioral problems. This sample yielded reliability coefficients of .87 for the MED subscale, .88 for the REB subscale, and .95 for the total ADI scale. The second sample included 583 adolescents recruited from schools, and yielded reliability coefficients of .80, .80, and. 93, respectively. The third sample included 233 adolescents in treatment for substance abuse. This sample yielded reliability coefficients of .85 for the MED subscale, .88 for the REB subscale, and .94 for the total ADI scale. Further, the intercorrelations across these three samples ranged between .63 and .74 for the total ADI and MED subscale, .74 and .84 for the total ADI and REB subscale, and .53 and .63 for the MED and REB subscales. Further, Harrell and Wirtz (1989) assessed the utility of the ADI by comparing ADI results from a sample of 264 adolescents to ratings from clinicians with training in adolescent assessment across the

four domains (e.g., loss of control, social and interpersonal, psycholog- ical, and physical indicators) of the ADI. Correlations on each of the four

domains were between .75 and .79 (p b .0001), and all the domains showed significant positive correlations with the clinician severity rating, which ranged between .64 and .71 (p b .0001).

What is the reliability in these statistics?