Because P300 amplitude is known to differ for males and females, analyses were conducted separately by gender (Hoffman and Polich, 1999; Euser et al., 2012). Analyses were conducted for age ranges consisting of ages 8–12, 13–18, and 19–29 to illustrate the development and stabilization of P300 amplitude as well as test for possible age dependence for any association of P300 amplitude development with familial risk status. Within each age range, subjects were included if at least two P300 recordings were available (Table 2). Approximately 10% of cases within an age range were excluded from analysis because a single recording was available within that range. Missing P300 recordings were assumed to be missing at random. With the missing at random assumption, the maximum-likelihood parameter estimation approach used in PROC TRAJ provided unbiased parameter and error estimates. Robust estimates of error in the parameter estimates were obtained using a clustered sandwich estimator that corrects for correlations in measurements from offspring within families.
