Methodological Issues in Estimating Genetic Correlation among Friendship Pairs

Tianji Cai, University of North Texas

In recently years, studies in social sciences start to collect genetic information, for example, single nucleotide polymorphisms in their regular surveys, and many of those data are now available to social scientists. Given such data, social scientists are able to conduct Genome-wide Association Studies (GWAS) on non-diseases traits. However, many methodological issues associated with GWAS are not well discussed in social sciences. Using the study on genetic correlation among friendship network as an example, this study shows some methodological issues of incorporating genetic markers in regular social science study. Research interest has increased dramatically in the area of genetic correlation among friendship network in recently years. It has been documented that the nominated friendship pairs are genetically correlated. For example, Fowler etc (2010) reported the genotypes in friendship networks are significant correlated. Our study illustrated that such correlation can be sensitive to coding, multiple testing, population stratification, and distribution assumptions.