Allelic variants that could have a large effect on a phenotype may not be present in a small and unrepresentative strain set. The inbred laboratory strains are reproductively isolated populations, which are derived from several different ancestral founders that diverged ~1 million years ago [21]; and thus contain a substantial amount of genetic variation that could affect many phenotypic traits. The observed phenotypic variance would be increased if a larger percentage of the 450 different inbred strains [23] were evaluated. Thousands of unrelated individuals are randomly sampled in a human association study, which ensures that phenotypic and genetic variation present in the human population is well represented. In contrast, the inbred mouse strains used in linkage or GWAS studies are not a random sample of the mouse population. These studies usually examine only a limited number of strains, and have a strong selection bias toward strains analyzed in previous studies, which may not be relevant for the current phenotype. Analysis of a small number of strains with a restricted phenotypic range will produce a reduced estimate of the genetic effect
