Reliability and Validity of a Substance Craving Scale: Testing Invariance across Substances

Craving for alcohol or drugs has been studied for many years by scholars and clinicians in an effort to improve the efficiency of substance abuse treatment across individuals and settings. The reduction of craving has been conceptualized as a target of intervention in addiction treatment programs and post treatment relapse prevention programs. The purpose of this research was to develop a substance craving scale that is reliable and valid for use across substance use disorders. The scale was examined via the Rasch model, an item response theory model that provides the analysis of data through use of a single, difficulty parameter. To date, there has been no Rasch analysis conducted of a substance craving scale that assesses cravings for all substances. The current Substance Craving Scale was developed based on several of the prior validated, drug-specific scales that are used to measure substance craving. An additional craving item, used on the National Longitudinal Alcohol Epidemiologic Survey and a key basis of the DSM-V decision to add craving as a criterion, was also included in the final scale. Analysis of the data employed both Rasch and traditional test theory techniques. The measure was analyzed with a Rasch dichotomous model with Winsteps statistical software, and other techniques used SPSS statistical software. Analyses were performed to provide summary statistics of the data, to determine item fit, dimensionality of the measure, reliability, validity, and differential item functioning (DIF) for substance of choice. The results showed that all but one item on the measure were appropriate based on a preliminary item analysis, and the final psychometric analyses were conducted on the revised 11-item scale. The craving scale was found to be unidimensional, reliable for research purposes, and showed evidence of validity of construct. Most importantly, there was no significant differential item functioning by substance of choice. Several items were discussed regarding their theoretical importance, and the validity and reliability of the scale were critiqued. Limitations of the study included homogeneity of the sample, hence reliability and validity of the Substance Craving Scale should be replicated in different samples, especially for those individuals entering substance treatment.