Twin concordance for a binary trait. II. Nested analysis of ever-smoking and ex-smoking traits and unnested analysis of a 'committed-smoking' trait

Abstract

Twin concordance rates for a binary trait can provide information about causes of trait variation. However, if trait prevalence varies with age (or birth cohort) or between the sexes, trait concordance rates will be artificially inflated because of the matching within pairs of twins. Our previous paper showed how to minimize the effects of such confounding by using logistic regression to model trait prevalence as a function of age and sex and that the binary correlation coefficient was useful as a measure of concordance that can be adjusted for trait prevalance. This method is extended here to allow for nested analyses and is applied to the smoking habits of a sample of 3,807 pairs of adult ..