Research
Nonlinear Growth Models
Some forms of change in longitudinal responses can pose analytic challenges to researchers. Although many longitudinal responses can be summarized by a polynomial function such as a linear or quadratic function, there are many instances in which a nonlinear function provides a better approximation to the response. In this line of research, methods to model nonlinear responses are developed and evaluated using real data.
Missing data
Statistical inference from applications of popular multivariate methods including structural equation models and mixed-effects models is valid when data are missing completely at random or missing at random. In some applications, data are missing not at random, a problem that can lead to biases in parameter estimates if the missing data process is not addressed. A variety of approaches are available to study mechanisms that may give rise to missing data. We are currently developing such procedures for mixed-effects models that are applied to longitudinal data.
Some forms of change in longitudinal responses can pose analytic challenges to researchers. Although many longitudinal responses can be summarized by a polynomial function such as a linear or quadratic function, there are many instances in which a nonlinear function provides a better approximation to the response. In this line of research, methods to model nonlinear responses are developed and evaluated using real data.
Missing data
Statistical inference from applications of popular multivariate methods including structural equation models and mixed-effects models is valid when data are missing completely at random or missing at random. In some applications, data are missing not at random, a problem that can lead to biases in parameter estimates if the missing data process is not addressed. A variety of approaches are available to study mechanisms that may give rise to missing data. We are currently developing such procedures for mixed-effects models that are applied to longitudinal data.