We fit a general linear model in which continuous scores on the EXT factor was the between-subjects variable, the number of standards preceding the target (1, 2, 3, or 4) the within-subjects variable, and P3 amplitude was the dependent variable. We also performed this analysis with P3 latency as the dependent variable. Because twins violate the assumption of independence of observations, we used Generalized Estimating Equations in SAS PROC GENMOD. These take into account the nested nature of the sample and produce appropriate standard errors. We conducted a Type 3 analysis, which is similar to the standard ANOVA Type III sums of squares used in PROC GLM, except that likelihood ratios are used instead of sums of squares. This procedure produces a chi-square statistic for each effect.
