Multiple imputation and posterior simulation for multivariate missing data in longitudinal studies

Area of Science:

• Statistics
• Biostatistics
• Longitudinal Data Analysis

Background:

• Missing data is a common challenge in longitudinal studies, potentially biasing results.
• Incomplete multivariate continuous longitudinal data requires specialized statistical methods for accurate analysis.

Purpose of the Study:

• To present a novel multiple imputation method for handling missing data in designed longitudinal studies.
• To develop a random coefficients model capable of accommodating incomplete multivariate continuous longitudinal data.

Main Methods:

• A hierarchical random coefficients model is employed for time-dependent variables, with an i.i.d. normal model for time-independent variables.
• Multivariate repeated measures are jointly modeled, accounting for heterogeneous error variances across variables and time points.
• Gibbs sampling is utilized for drawing model parameters and imputing missing observations.

Main Results:

• The proposed multiple imputation method is illustrated with an application to startle reaction study data.
• A simulation study demonstrates the performance of the imputation procedure, comparing it to existing weighting methods.

Conclusions:

• The developed multiple imputation method provides a robust approach for analyzing incomplete longitudinal data.
• This method offers a valuable alternative for researchers dealing with missing data in complex longitudinal study designs.