However, the association between a candidate gene and the disease may be spurious—that is, the gene actually is not genetically linked to the disease. For example, if the control population is not properly matched (e.g., with respect to ethnicity) to the population from which the affected people are drawn, spurious linkage may occur due to a phenomenon called population stratification. This means that for some genes the frequency with which different gene variants occur in a population may differ between ethnic groups. Consequently, if the samples of people with and without the disease do not have the same ethnic composition, it may appear as if certain gene variants are associated with the disease when, in fact, the frequency of those variants is determined by ethnicity and is unrelated to the disease under investigation. In the COGA study, however, researchers can study candidate genes for both alcohol dependence and habitual smoking without running the risk of spurious findings resulting from population stratification by using family-based tests, such as the Transmission Disequilibrium Test (TDT).4 This approach eliminates the need for a separate,
