Abstract
We developed the R package SimCorMultRes to facilitate simulation of correlated categorical (binary and multinomial) responses under a desired marginal model specification. The simulated correlated categorical responses are obtained by applying threshold approaches to correlated continuous responses of underlying regression models and the dependence structure is parametrized in terms of the correlation matrix of the latent continuous responses. This article provides an elaborate introduction to the SimCorMultRes package demonstrating its design and usage via three examples. The package can be obtained via CRAN.
Highlights
Fitting marginal models with correlated binary or multinomial responses is required in many applications in which the responses are assumed to be correlated
We discuss a simple and straightforward simulation method that is based on the NORmal To Anything (NORTA) method, and we present a general algorithm that combines the threshold approaches with the modified NORTA method, enabling us to generate correlated categorical responses subject to a marginal model specification in a unified manner
The output of any core function is displayed as a list with three items: (i) a matrix with the simulated responses such that the (i, t)-th element corresponds to the realization of Yit (Ysim), (ii) a data frame that contains the simulated responses (y), the covariates specified by xformula, subjects’ identities and the measurement occasions, and (iii) the NORTA generated or user-defined latent random vectors
Summary
Fitting marginal models with correlated binary or multinomial responses is required in many applications in which the responses are assumed to be correlated. We demonstrate the use of SimCorMultRes by considering the problems of evaluating two estimation methods for marginal models with correlated nominal multinomial responses, of assessing the quality of an approximation that links the uniform local odds ratios structure with the correlation parameter of an underlying bivariate normal distribution, and of simulating correlated categorical random variables under no marginal model specification.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.