(PS8-C51) Expanding the Clinical Utility for the Strengths and Difficulties Questionnaire in an Ethnically Diverse Sample

Within the broader movement toward strengths-based psychology, a method for gathering nuanced information about youths’ risk and protective factors for emotional and behavioral difficulties is through assessments that target multiple dimensions of their social and emotional development. One such assessment is the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1998), a 25-item behavioral screening tool that assesses for both youth strengths (e.g., prosocial behavior) and problem areas (e.g., emotional symptoms, conduct problems, hyperactivity-inattention, and peer problems). To date, the SDQ self-report form has shown good reliability and validity among the 11- to 17-year-old youth for whom it was initially developed. Recent evidence suggests the SDQ self-report may also be a reliable and valid tool for youth as young as age nine (e.g., Hobbs & Laurens, 2020); however, published studies have found mixed findings for a three- versus five-factor structure. Additionally, existing clinical cut-offs for this form do not include youth below age 11 and were derived from somewhat arbitrarily chosen percentile rankings. The current study had two aims. First, we investigated the results of two competing SDQ factor structures. Second, we explored the potential for refining the SDQ’s screening utility by comparing youths’ established score cut-offs with new, mean-based cut-offs generated in this study. Specifically, we compared the degree to which youths’ membership in the proposed “high,” “borderline,” and “low” SDQ score groups was associated with their scores on a strengths-based youth self-report, the Social-Emotional, Evidence-Based Developmental Strengths Assessment (SEEDS; Ebesutani, 2016).

Participants included 504 ethnically diverse fourth through sixth graders from five elementary schools. Youth were 51% male and were an average of 10.99 years old (SD = 1.00; range = 9.23 - 13.58). Confirmatory factor analytic results pointed to the five-factor model fit indices as indicating improved fit over the three-factor model (Χ²(265) = 592.544). Although two model fit indices for the five-factor model fell below benchmark for good model fit (CFI = .804, TLI = .778), two other indices were found to be acceptable (RMSEA = .049, SRMR = .060). Concerning Cronbach alpha coefficients, the Emotional Symptoms scale and the Prosocial Behavior scale fell in the questionable range, and the Conduct Problems, Hyperactivity-Inattention, and Peer Problems fell in the unacceptable range (George & Mallery, 2003). Given the model results, additional analyses examining item relationships for the five-factor model within the context of our mean-based score cut-offs will be explored and discussed. Although full results regarding the comparison of youth SDQ scores are forthcoming, for the mean-based score cut-offs, using a Total Difficulties SDQ score cut-off of 15 (0.5 SD above the sample mean) for the “borderline” group and 21 (1.5 SD) for the “high” group resulted in 69.7% of sample youth falling in the “low,” 21.4% in the “borderline,” and 9.0% in the “high” group, respectively. Full study results, limitations, and future directions, such as potentially creating norms for an American/ethnically diverse sample, will also be discussed.