In this paper, we advocate the use of “effective sample size” (ESS) as a simple method to capture the effect of sample correlation and variance inflation. The term effective sample size has appeared in the literature before (Kish, 1965; Thiébaux and Zwiers, 1994; Rosner and Milton, 1988; Rao and Scott, 1992; Madden and Hughes, 1999) but has not become a commonly used tool in genetic analysis. We define effective sample size NE as the equivalent number of independent samples that leads to the same variance of an intensive quantity, i.e., a quantity that does not change with the sample size.
