Lm. Lewin et al., Childhood social predictors of adolescent antisocial behavior: Gender differences in predictive accuracy and efficacy, J ABN C PSY, 27(4), 1999, pp. 277-292
This study examined the ability of several childhood, school-based, social
variables to correctly classify antisocial adolescents. Children (N = 314;
163 boys, 151 girls) in the 3rd-5th grade were assessed on academic and soc
ial variables (i.e., peer rejection, aggression, withdrawal, and low prosoc
ial behavior) and followed forward for 6-7 years until the 9th and 10th gra
de. Adolescent antisocial outcomes included a consensus measure formed from
diagnostic interviews, contact with juvenile authorities, adolescent self-
report, and mother's report. The gender-differential predictive accuracy an
d efficacy of the early predictor domains to a consensus measure of antisoc
ial behavior were compared with the same estimates found for adolescent sel
f-report of antisocial behavior. Both gender and criterion-method differenc
es were found. For girls, regardless of the measure of antisocial behavior,
early academic problems were the strongest predictors of future problems.
For boys' self-reported antisocial outcomes, peer rejection was the stronge
st independent predictor. For consensus-reported antisocial outcomes, both
early fighting-anger and withdrawn behavior displayed equally strong predic
tive relations. For boys, the combination of early fighting-anger and disru
ptive and withdrawn behavior was the strongest set of predictors for the co
nsensus measure of antisocial functioning. Predictive accuracy and efficacy
estimates are discussed in terms of predictive strength as well as the cos
t-benefit of misidentification.