To understand the genetic architecture of complex traits we need theoretical models that make useful predictions that are consistent with empirical observation. Simulations of complex traits are widely used for inference of population parameters and in-silico testing of new experimental or analytical methods. Despite this, the approaches in the field of complex trait simulation are very heterogeneous. One common thread is that classic models consider particular mutations (“SNPS”) as separate loci. However, we wish to model a genomic region as a functional unit or gene. As such, there are important implications of the structure of the relationship between mutations in the region and it’s functional output in a diploid individual. Also, it is well known that demography plays an important role in shaping patterns of DNA sequence variation, but the specifics of how demography interacts with underlying genetic model are unknown. We use forward-time simulation to explore the properties of a co-dominant model and two different recessive models, in constant-sized and recently expanded populations, in the context of heritability estimation and genetic association studies. In particular, we find that the population frequency by effect-size distribution and statistical properties of association studies are both impacted by genetic model. Consequently, when explicitly modeling DNA sequence variation underlying a complex trait it is critical to differentiate between sites within a functional unit (gene) and those in distinct functional units. Comparing the effect of population growth across multiple genetic models suggests that, perhaps, the genetic model is more important than the demographic model.