Essays on structural equation modelling with applications of the basic univariate twin model

Abstract: Structural equation models have proven useful for studies of genetic and environmental effects forvarious traits end the classical twin study has been considered as a powerful design for such studies. This thesis consist of five papers dealing with the analysis of structural equation models with applications of the basic univariate twin model.Monte Carlo simulation is a useful method for assessing the properties of an estimator for situations which are often found in real-world applications. In the first paper a method to generate data for simulation of structural equation models is presented.The second paper deals with the robustness of two commonly used estimators assuming the basic univariate twin model. The robustness is analysed from a theoretical point of view and through simulations.In the third paper it is shown how closed form expressions for a class of estimators can be derived assuming the basic univariate twin model. Expressions for robust standard errors and a robust test statistic for test of model fit are derived using these closed form expressions. A simulation study is presented where the properties of the robust standard errors and robust fit statistic are analysed for this model.The fourth paper considers Full Information Maximum Likelihood estimation of the basic univariate twin model when the observed trait is measured on an ordinal scale. Identification and estimation of the underlying mean and covariance structures as well as test of model fit are discussed. Numericalillustrations are presented.The fifth paper examines children liability to accidental injuries. This empirical study is based on a twin study and the method of estimation proposed in the fourth paper is applied.