Xi Chen, Ying Zhang, Amanda M Fretts, Tauqeer Ali, Jason G Umans, Richard B Devereux, Elisa T Lee, Shelley A Cole, Yan Daniel Zhao
The generalized estimating equations method (GEE) is commonly applied to analyze data obtained from family studies. GEE is well known for its robustness on misspecification of correlation structure. However, the unbalanced distribution of family sizes and complicated genetic relatedness structure within each family may challenge GEE performance. We focused our research on binary outcomes. To evaluate the performance of GEE, we conducted a series of simulations, on data generated adopting the kinship matrix (correlation structure within each family) from the Strong Heart Family Study (SHFS)...
March 29, 2024: Journal of Biopharmaceutical Statistics