Comparison of methods of estimation for a goodness of fit test – an analytical and simulation study

Abstract

Multilevel modelling is a novel approach to analyse data which consist of a hierarchical or a nested structure. With advancements in multilevel modelling, there has been an advancement in the estimation techniques and also in goodness-of-fit tests which are vital to assess the fit of a model. However, these goodness-of-fit tests are not as yet tested to be suitable for models estimated using different estimation techniques. This study aims to conduct a comparison of methods of estimations for use in a goodness-of-fit test which is developed for binary response multilevel models. The comparison is based upon the mathematical background, extensive simulations and an application to a real-life dataset.

KEYWORDS:

• Estimation techniques
• goodness-of-fit
• marginal quasi likelihood (MQL)
• multilevel modelling
• penalized quasi likelihood (PQL)

Disclosure statement

No potential conflict of interest was reported by the author(s).