Comparison of the Psychometric Characteristics of the Adolescents Depression Scale Based on the Item Response Theory with those based on the Classic Test Theory

The objective of this study was to assess the psychometric properties of the Adolescent Depression Scale (ADS) based on the item-response theory and compare the results with those based on the classic test theory.
A total of 750 students (364 males and 386 females) were selected through multistage random clustering (levels proportional to size) and completed a questionnaire comprising 46 questions developed by the investigators.
Results of factor analysis and questions structure showed that the important item-response theory presumptions, namely, unidimensionality and local independence, are contained in ADS. As regards the psychometric characteristics of the questions, results from the two theories were similar, except that the item-response theory provided more valid indices for assessing the questions. Further analysis of the data indicated that the two theories showed close values for internal consistency with regard to psychometric characteristics, such that values of 0.933 and 0.947 were obtained for Alpha Cronbach and marginal validity, respectively. Finally, the normality scores were calculated based on the Item Response Theory with those based on the Classic Test Theory. Ranking the individuals by the two theories showed considerable differences as regards the depression construct continuum, such as a difference of 86, 92, or 93 individuals.
Based on the findings, it can be concluded that the item-response theory models provides a higher number of more exact indices for judgment and making decisions about questions, tests and individual ranking. Therefore, application of the item-response theory for assessment of other educational and psychological constructs is recommended.