Rasch Analysis of a Measure of Alcohol-Related Problems in Undergraduate Students

Abstract

This study aimed to evaluate the psychometric properties of an instrument measuring alcohol-related problems typical of undergraduate students. The participants were 401 undergraduate students that reported using alcohol at least occasionally (mean age = 22.44; SD = 3.43), which also answered a preliminary 24-item measure. Rasch modeling was used to estimate items regarding in terms of difficulties (δi), local independence and and Differential Item Functioning (DIF) among items for each gender. After excluding two items with DIF, results showed adequate psychometric properties for a relatively non-biased, unidimensional 20-item measure. Item reliability was .99 and person reliability was .73. Evidence suggests that the new instrument adequately addresses a large continuum of drinking problems that present themselves in the university context. Recommendations for use, along with a logit and standardized logit scales are presented for the interpretation of the scores