Is It Time to Revise the SDQ? The Psychometric Evaluation of the Strengths and Difficulties Questionnaire

Despite the wide use of the Strengths and Difficulties Questionnaire (SDQ) to assess adolescent mental health, its psychometric functionality is still under debate. This study investigated the structural validity and reliability of the SDQ scores, and the resemblance of the SDQ sum scores and factor scores. Factor one-dimensionality and competing multifactor structures were tested against data. With the best acceptable models, measurement invariance was tested between genders and over time. Subscale reliability and correspondence between subscale sum scores and factor scores were estimated. The nationally representative self-report data from 23,980 Finnish early (12-13 years) and mid- (15-16 years) adolescents (50.4% girls) were collected from two cohorts in 2008 and 2013. The results showed that among early adolescents, the revised SDQ with a controlled method effect had an excellent fit. In contrast, none of the tested models had an acceptable fit among the mid-adolescents. Among early adolescents, strong measurement invariance was achieved between genders and over time. Three of the five subscales were one-dimensional, and all subscales had low reliability. The resemblance between the subscale sum scores and factor scores was alarmingly low. Researchers should be cautious when using the SDQ Total Difficulties sum score or the subscale scores as they may be substantially biased, and practitioners should desist from using the SDQ as a screening tool in its current form. This study strongly supports the revision of the SDQ. In line with the previous findings, we suggest rewording the worst functioning items and revising the reverse-worded difficulties items.

Automated summary

This study examined the structural validity and reliability of the SDQ scores, as well as the resemblance between the sum scores and factor scores. The results showed that the revised SDQ had a good fit among early adolescents, but not among mid-adolescents. Measurement invariance was achieved between genders and over time among early adolescents. However, the subscales had low reliability and the resemblance between the subscale sum scores and factor scores was alarmingly low. This study highlights the need for caution when using the SDQ as a screening tool and supports its revision.