Relative Advantages of Person- and Variable-Based Approaches for Predicting Problem Behaviors from Kindergarten Assessments

Kindergarten teacher ratings of physical aggression, hyperactivity,inattention, anxiety, and prosocial behavior were used to predictself-reported delinquency, peer-rated social withdrawal, and schoolplacement in preadolescence (ages 10 to 12 years) in a large longitudinalsample of boys from low socio-economic neighborhoods. Two analyticstrategies were used: person-oriented and variable-oriented approaches. Inthe person approach, eight clusters, based on the kindergarten behaviors,were used to predict delinquency, social withdrawal, and schoolplacement. In the variable approach, the kindergarten behaviors were used asdimensions in logistic regressions. Family adversity was used as the firstpredictor in both approaches; it significantly predicted all the outcomes inpreadolescence. The results obtained using the two approaches were partly inaccordance. In both sets of results, kindergarten teacher-ratedexternalizing behavior problems were most related to later self-reporteddelinquency, and internalizing problems to peer-rated socialwithdrawal. However, the person approach showed that all patterns ofkindergarten behavior problems increased the risk for placement out of anage-appropriate regular classroom, while only inattention and lack ofprosocial behavior were the significant dimensional predictors of thisnegative outcome according to the variable approach. Also, the personapproach showed that the Multiproblem kindergarten boys had the highestpercentage of comorbidity of preadolescent problems. The ROC curvesindicated that prediction of delinquency, social withdrawal, and schoolplacement were equally accurate using the cluster and variableapproaches. Advantages and limits of both approaches are discussed withreference to their usefulness for clinicians.