Development of the Reactive Aggression Assessment (RAGA-16) Part I: Item Selection and Initial Validation

ObjectiveSevere behavioral outbursts characterized by reactive aggression (RA) are common and highly impairing for young children and their families. A growing body of research suggests RA may characterize a distinct group of youth with behavior problems; however, RA is not specific to a single diagnosis in the current clinical nosology. The challenges of assessing the severity of RA and elucidating its diagnostic role as a transdiagnostic symptom are compounded by a lack of validated, RA-specific assessment tools. The current work presents psychometric evidence for a novel screening tool and outcome measure: the 16-item Reactive Aggression Assessment (RAGA-16). It is the first of two manuscripts presenting evidence for the RAGA-16.MethodData comprised a nationally representative US sample (N=1,162) of parents with children ages 6 to 19 (M=11.4; SD=3.98) completed an as-yet unpublished assessment of mood and behavior problems. Latent class analysis (LCA) of expert ratings identified a pool of RA-specific items. We compared two approaches to developing the RAGA-16 from this item pool: a Data Informed approach combining item response theory (IRT) techniques and expert judgment of item characteristics; and an automated test assembly (ATA) approach that selected items based on optimizing test information. Scores from the resulting forms were evaluated based on multiple facets of reliability, differential item functioning (DIF), and associations with parent-reported health information.ResultsLCA identified 43 RA items representing two distinct but highly correlated (r=.63) content domains: Temper Loss and Aggressive Behaviors. Both the Data Informed and ATA-Derived forms showed high reliability (rxx>.80) across a wide range of symptom severity and excellent internal consistency despite containing largely different items. The Data Informed version showed less evidence of DIF across child race and sex as well as slightly higher detection of parent-report externalizing disorders (AUC=.76 vs. .73). Percentile and T-Score norms for scores on the Data Informed form are reported.ConclusionsThe Data-Informed version was selected as the final version of the RAGA-16. Having a brief, highly reliable, targeted assessment of RA with epidemiological norms allows researchers and clinicians to better characterize RA behaviors in youth. Forthcoming work will validate the RAGA-16 as a screening tool in inpatient and outpatient behavioral health settings.