Premium
Univariate analysis of dichotomous or ordinal data from twin pairs: A simulation study comparing structural equation modeling and logistic regression
Author(s) -
Ramakrishnan Viswanathan,
Meyer Joanne M.,
Goldberg Jack,
Henderson William G.
Publication year - 1996
Publication title -
genetic epidemiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.301
H-Index - 98
eISSN - 1098-2272
pISSN - 0741-0395
DOI - 10.1002/(sici)1098-2272(1996)13:1<79::aid-gepi7>3.0.co;2-1
Subject(s) - univariate , statistics , heritability , logistic regression , categorical variable , ordered logit , mathematics , ordinal regression , ordinal data , structural equation modeling , regression analysis , additive model , sample size determination , econometrics , twin study , multivariate statistics , biology , genetics
The univariate analysis of categorical twin data can be performed using either structuralequation modeling (SEM) or logistic regression. This paper presents a comparison betweenthese two methods using a simulation study. Dichotomous and ordinal (three category) twindata are simulated under two different sample sizes (1,000 and 2,000 twin pairs) andaccording to different additive genetic and common environmental models of phenotypicvariation. The two methods are found to be generally comparable in their ability to detect a“correct” model under the specifications of the simulation. Both methods lack power to detectthe right model for dichotomous data when the additive genetic effect is low (between 10 and20%) or medium (between 30 and 40%); the ordinal data simulations produce similar resultsexcept for the additive genetic model with medium or high heritability. Neither method couldadequately detect a correct model that included a modest common environmental effect (20%)even when the additive genetic effect was large and the sample size included 2,000 twinpairs. The SEM method was found to have better power than logistic regression when there isa medium (30%) or high (50%) additive genetic effect and a modest common environmentaleffect. Conversely, logistic regression performed better than SEM in correctly detectingadditive genetic effects with simulated ordinal data (for both 1,000 and 2,000 pairs) that didnot contain modest common environmental effects; in this case the SEM method incorrectlydetected a common environmental effect that was not present. © 1996 Wiley‐Liss,Inc.