Figure 3 depicts expected trajectory class membership as a function of CHRM2 genotype (for a single, representative SNP, rs978437) and peer group antisocial behavior (which was modeled as a continuous variable, but illustrated using a median split). Among adolescents perceiving relatively little antisocial behavior within the peer group, CHRM2 genotype appeared to play a smaller role in differentiating between those with decreasing low (GG = 82%, AG = 74%, and AA = 72%) and high stable externalizing trajectories (GG = 6%, AG = 4%, and AA = 4%) as compared with those reporting high levels of peer group antisocial behavior, where there were much stronger differences in trajectory membership as a function of genotype (GG = 73%, AG = 57%, AA = 32% versus GG = 0%, AG = 11%, AA = 17% in the decreasing low and stable high trajectory classes, respectively). Likewise, genotype was better able to discriminate between probabilities of membership in the low (GG = 73%, AG = 57%, AA = 32%) versus moderate (GG = 27%, AG = 32%, AA = 51%) decreasing trajectories among
