Drug addiction and quality of life: methodological issues /

Abstract

Objective: Substance abuse research data containing categorical and continuous variables often violate assumptions of parametric statistical methods. Clustering of individuals, lack of repeated measurements, missing data, non-representative samples, etc. aggravates the problem. Quality of life (QoL) measures suffer from meaningful application of statistical methods. The paper describes statistical approaches which fit well with structure of drug addiction data and measures of QoL, facilitating better analysis and interpretations of results. Method: Converting ordinal item scores to normally distributed continuous scores in the range [1-100], irrespective of number of response-category in items. Such transformations fit well with structure of drug addiction data and measures of QoL and facilitate better analysis and interpretations of results. Results: Proposed method enables parametric statistical analysis leading to meaningful comparisons and inferences, finding equivalent scores, computation of responsiveness of the scale i.e., ability to assess changes across time and psychometric qualities like reliability, as per definition, Factorial validity reflecting the main factor for which the test was developed. Conclusions: Considering theoretical advantages, the proposed method generating normally distributed scores is recommended. Future studies with longitudinal data suggested finding sensitivity with emphasis on progression of disease and to different therapeutic interventions.