Properties of structured association approaches to detecting population stratification

Objective: To examine the properties of the structured association approach for the detection and correction of population stratification. Method: A method is developed, within a latent class analysis framework, similar to the methods proposed by Satten et al. (2001) and Pritchard et al. (2000). A series of simulations illustrate the relative impact of number and type of loci, sample size and population structure. Results: The ability to detect stratification and assign individuals to population strata is determined for a number of different scenarios. Conclusion: The results underline the importance of careful marker selection.