A Facet Analysis Approach to Content and Construct Validity

The major purpose of the study was to demonstrate the utility of Guttman's facet analysis in serving as a methodological approach for instrument development and validation. A Likert-type survey for measuring college students' attitudes toward alcohol was the vehicle for the demonstration. Guttman's facet analysis was used to construct items according to a carefully designed mapping sentence which specified the basic attitude components of interest to the study. Namely, four basic facets were hypothesized to underlie students' attitudes toward alcohol use. Likert-type survey items were written to match all combinations of the elements of the hypothesized facets in an attempt to sample the content domain representatively. These efforts, as well as an interjudge agreement process, were undertaken to address issues of content and construct validity during instrument construction. The survey was administered to a sample of college students (n = 1049). The coefficient alpha reliability estimate was found to be .91. Subsequent multidimensional scaling analyses provided evidence in support of the four hypothesized facets measured by the instrument and thus its construct validity. The substantive findings related to students' attitudes toward alcohol are also discussed.