The design effect is a sampling variance under a complex design divided by the corresponding simple-random-sampling variance. For independent equal-size clusters of size with exchangeable intraclass correlation , a common mean-estimation approximation is . It expresses the loss of effective sample size from positive correlation; finite-population and unequal-cluster-size designs require their own calculation.
Articles by others on the same topic
The term "design effect" typically refers to the impact of a study's design on its statistical properties, particularly in the context of complex surveys. It is often used in the field of statistics and research methodology to describe how certain sampling designs can affect the variance of estimates compared to simple random sampling.