Citation

Harburg E, Gunn R, Gleiberman L, Roeper P, DiFranceisco W, Caplan R. J. Stud. Alcohol 1988; 49(6): 522-531.

Affiliation

Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor 48109.

Copyright

(Copyright © 1988, Rutgers Center of Alcohol Studies)

DOI

unavailable

PMID

3236884

Abstract

The Short Michigan Alcoholism Screening Test (SMAST), often given to clinical samples to detect alcoholics, was used in a survey of a randomly selected sample (N = 1,266) in a small midwestern town, with a population of about 10,000. The specific aims were: (1) to test if scores on the SMAST would be related to these normal subject's levels of drinking; (2) to test if the SMAST could identify a drinking group who had never sought treatment for alcohol-related problems, thus defining a potential subset for help; and (3) to appraise the SMAST as a research tool for use in a general population. Results show that the number of SMAST items endorsed is significantly but modestly correlated to drinking levels, with higher volume drinkers having higher scores. This prediction was improved by removing two items reporting "not normal" drinking that had caused SMAST scores to be spuriously high. Factor analysis of test items allowed construction of five rank-ordered groups: those (1) reporting no alcohol-related SMAST problems; (2) described by two items, self-reported "not normal" drinking; (3) with one problem: "can't stop" drinking; (4) reporting problems but had not received treatment for them (the predicted group); and (5) reporting both problems and attempts at help with them (e.g., Alcoholics Anonymous). These latter two groups had significantly higher scores on several other measures of alcohol-related behavior, as well as tests measuring emotional adjustment, notably Eysenck's Neuroticism Scale and Buss and Plomin's measure of Impulsivity. The SMAST demonstrates potential usefulness in selecting problem, no-problem and patterns of use in drinkers from a normal drinking population when clusters of items are constructed to be exclusive in a Guttman-like rank order, rather than by sheer number of items endorsed or use of simple factor scores.

Language: en