Using the Rasch model to evaluate the South Oaks Gambling Screen for use with nonpathological gamblers.

Despite its utility for the dichotomous differentiation of pathological and nonpathological gamblers, some have questioned the use of the South Oaks Gambling Screen (SOGS) for discriminating individuals across the entire range of gambling problems. Using methods from item response theory, we derived a six-item version of the SOGS. This shortened version of the SOGS performed uniformly across a sample of pathological gamblers and a sample of students and resulted in five levels of discrimination and the following interpretation: a score of 1 reveals potential for problems; 2 reveals likely problem gambling; 3 or 4 items endorsed represent significant levels of problems noticeable by others; while 5 or 6 represents severe problems with significant financial involvement. We discuss the implications and remaining limitations of using this shortened measure for the continuum-based assessment of gambling problems.