In the ordinal regression analyses designed to mimic Caspi et al. [2003], there was no evidence of either a main effect of 5HTTLPR or an interaction with SLE, but the reduced sample size necessarily means that these analyses have substantially less power than our full analyses.