Population-based genetic association studies are increasingly used to explore the association between genetic polymorphisms and outcomes such as disease-status and disease-related quantitative traits. Because multiple polymorphisms are typically available, there are several statistical analysis strategies that might be appropriate depending on the goal of the study. In this paper, we compare several linear model parameterizations that might be used to perform a test of association between a genomic region defined by multiple SNPs and a quantitative trait. We compare via simulation the type I error and power of the omnibus F-test to detect association. As expected, there is no one most powerful test across the genetic models we considered, although tests based on simple parameterizations that do not rely on phase information can be as powerful as more complicated haplotype-based tests even when it is a haplotype that is truly associated with the trait.