Pairwise likelihood estimation in longitudinal data analysis: a comparison with full likelihood approach

Abstract

The importance of longitudinal studies has grown tremendously over the past twenty years due to their unquestionable advantages. While it would appear that they are now considered foundational for drawing causal inference, they are not without limitations. Many times high dimension of the joint covariance matrix results in computational complexity and requires special handling in dealing with this situation. Subject of this article is the presentation of the most prominent ways of analyzing longitudinal continuous data and the proposal of pairwise approach in order to confront computational problems that might appear.