Conclusions and Recommendations (Chapter 9)
Our explorations into genome-wide association studies (GWAS) led us to believe that a number of governing principles were absent from the manner in which GWAS were conducted. The descriptions of methods that researchers have provided in the literature commonly assumed an additive inheritance model or were agnostic with respect to an inheritance model assumption. Furthermore, there were many examples in which GWAS provided inconsistent results and researchers could not replicate study findings. Our own experiences investigating GWAS indicated that different statistical methods assuming specific inheritance properties also produced different association results. Because the inheritance properties are generally not known a priori, we favored a statistical model that was agnostic with respect to inheritance, and we turned to simulation studies to confirm our beliefs.