Exploring associations between complex traits and genetic variants plays an important role in biostatistics and many approaches have been developed in this area. Moreover, a large amount of data were collected from familial and longitudinal studies in recent years. The major challenges to analyze this type of data are population stratification and the correlation within the family. However, most of the current methods are based on data of independent individuals and cannot handle the population stratification effect well. In this paper, we present a semi-parametric strategy (called PC-GEE) based on family data to detect associations. We use extensive simulation experiments to evaluate the performance of the PC-GEE method. Besides, we apply the proposed method to analyze GAW 17 mini-exome data.