Citation

Kühlhorn E, Leifman H. J. Stud. Alcohol 1993; 54(5): 542-554.

Affiliation

Department of Sociology, University of Stockholm, Sweden.

Copyright

(Copyright © 1993, Rutgers Center of Alcohol Studies)

DOI

unavailable

PMID

8412144

Abstract

Two Swedish alcohol surveys were compared in a search for a reasonable explanation of the large difference in their coverage rates, namely 75% and 28%. In many respects both surveys conducted in the late 1980s by large, well-known institutes, are of a similar type with rather large samples of Swedes. The technique used in the survey with a very high coverage rate (Survey A) takes into consideration the actual drinking pattern of the population studied (i.e., the concentration of drinking on weekends). By dividing a "normal week's consumption" into four units (Monday-Thursday, Friday, Saturday and Sunday), the technique allows one to average periods with varying drinking habits. In the survey with a low coverage rate (Survey B) a "normal week's consumption" was not so divided. A test of internal validity within Survey A underlined the general finding that its higher coverage rate was due to this division. A test of the external validity at aggregate level did not support assumptions about "telescoping" effects in A. Both A and B had a normal week as a basis of measurement for investigating typical drinking habits. The literature concerning differences in coverage rates focuses on the measurement of modal habits versus mean habits. The main explanation of differences is that methods that focus on modal habits (i.e., the Quantity-Frequency Scale) generate a lower coverage rate than do methods that elicit the arithmetic mean (i.e., the last-week recall). Since A and B both belong to the former type of scale, this does not explain our results.

Language: en