Subject selection biases in alcoholic samples: effects on cognitive performance.J Clin Psychol. 1988 Sep; 44(5):831-6.JC
The effects of subject selection bias in research on cognitive deficits in sober alcoholics were studied in a sample of 523 subjects (98 controls, 276 ineligible alcoholics, 40 eligible alcoholics who declined to participate, and 144 eligible and participatory alcoholics). All subjects received the Shipley Institute of Living Scale (Vocabulary and Abstracting subscales) and measures of anxiety, depressive symptoms, childhood hyperkinesis and attentional deficit disorders were obtained. Results indicate that current guidelines for alcoholic subject selection are biasing analyses toward support of the null hypothesis. Declined alcoholics performed more poorly on the Shipley Abstracting than did the "Used" group, yet did not differ significantly from the Used alcoholic groups on depression or anxiety. Declined alcoholics did, however, report significantly fewer Hk/MBD symptoms than did other alcoholic groups. ANCOVAs that used the affective and childhood disorders as covariates did not alter the differences in cognitive performance described above.