Validity of the Severity of Dependence Scale (SDS) construct applying the item response theory to a non-clinical sample of heroin users

AIMS: The purpose of this work is to study the validity of the Severity of Dependence Scale (SDS) construct by applying Rasch models to a non-clinical sample of heroin abusers. SUBJECTS: 982 (73% men) young people 30 years old or under (mean age 25.9 years) participated. All of them were captured from the community in the metropolitan areas of Madrid, Barcelona and Seville, between April 2002 and December 2003. ANALYSIS: Dimensionality of the scale and calibration of items were studied using the Rating Scale model, which is a Rasch-type model. A factorial analysis was also performed to check the dimensionality of the scale. RESULTS: The analysis of fit shows that all the items have infit and outfit values between +/- 2 logits, indicating that the data fit the model and that it may be assumed to be unidimensional. The principal components analysis also showed the existence of a principal factor that explains 52.5% of the variance observed. Item calibration found that they are between +0.89 and -1.04 logits on the scale. CONCLUSION: The results show unidimensional structure of the SDS scale. Item calibration shows they are distributed along the continuum, which must be taken into account when calculating total scores. The study's limitations are noted.