Cochran Q Test

In conducting meta-analyses of estimates from a set of studies, researchers often use a statistic denoted by Q to assess the homogeneity of the estimates. When used as the basis of a formal statistical test of homogeneity, Q is commonly referred to a chi-square distribution, and the test is called the Cochran Q test. The name of the test, however, embodies a misunderstanding: Although William G. Cochran wrote about the statistic Q, he did not propose a test based on it. Also, the test generally uses an incorrect null distribution. This entry describes the Q statistic, examines its statistical behavior, and discusses implications for the heterogeneity measure I2 and for a popular method of random-effects meta-analysis.