Quantifying the distance between cultural groups has received substantial recent interest. A key innovation, borrowed from population genetics, is the calculation of cultural FST (CFST) statistics on datasets of human culture. Measuring the variance between groups as a fraction of total variance, FST is theoretically important in additive models of cooperation. Consistent with this, recent empirical work has confirmed that high values of pairwise CFST (measuring cultural distance) strongly predict unwillingness to cooperate with strangers in coordination vignettes. As applications for CFST increase, however, there is greater need to understand its meaning in naturalistic situations beyond additive cooperation. Focusing on games with both positive and negative frequency dependence and high-diversity, mixed equilibria, we derive a simple relationship between FST and the evolution of group-beneficial traits across a broad spectrum of social interactions. Contrary to standard assumptions, this model shows why FST can have both positive and negative marginal effects on the spread of group-beneficial traits under certain realistic conditions. These results provide broader theoretical direction for empirical applications of CFST in the evolutionary study of culture.