As shown in table 3, without adjusting for covariates, a significant two-way interaction is observed.1 When covariates are included—for example, extraversion, neuroticism, tough-mindedness, and social nonconformity; frequency of church attendance and religious affiliation; educational level; and history of major depression and conduct disorder (see Heath et al. 1997b for further details)—the effect of the two-way interaction is reduced but remains significant.2 In other words, the influence of sociodemographic, personality, and psychiatric risk factors on both the risk of alcohol dependence and the risk of regular smoking is not sufficient to explain the observed genetic association between smoking and alcohol dependence.
