Our study sample for this analysis is one of the largest to date and uses both the original definition of 5HTTLPR and a re-definition of the L allele subdivided by the SNP rs25531, which based on evidence from functional studies [based on Hu et al., 2005; Wendland et al., 2006] is expected to be more associated with depression than 5HTTLPR alone. However, we find no evidence for a main effect of genotype nor an interaction between stressful events and genotype, either on the observed scale of disease using ordinal regression or on the underlying liability scale or risk to depression using IRT. This implies that individuals exposed to more stressful environments are no more susceptible to depression if they have either one or two S alleles than if they have the L allele. We observe this irrespective of whether we fit an additive or S-dominant model. Further, we attempted to replicate Caspi et al. [2003] by matching their sample and analyses as best we could but still found no evidence for an interaction. Finally, our data suggest that the reported
