Generalized Estimating Equations in Longitudinal Studies: A Non-Parametric Alternative for Two-Way Repeated Measures Mixed ANOVA

• Published in: Research journal of pharmacy and technology Vol. 16; no. 5; pp. 2381 - 2384
• Main Authors: M Karun, Kalesh, M S, Deepthy
• Format: Journal Article
• Language: English
• Published: Raipur A&V Publications 01.05.2023
• Subjects: Estimates; Generalized linear models; Infections; Intervention; Knowledge; Longitudinal studies; Medical research; Nonparametric statistics; Variables; Within-subjects design
• ISSN: 0974-3618; 0974-360X; 0974-306X
• DOI: 10.52711/0974-360X.2023.00392

Two group pre-post designs are very commonly used in medical research to study the effect of interventions on numerical outcome variables. Sometimes these measurements don’t follow the fundamental statistical assumption of normality and two-way repeated measures mixed ANOVA cannot be used. Generalized Estimating Equation (GEE) with Gamma log link function is a non-parametric analogue that can be used when data is skewed. When compared to other methods GEE has fewer assumptions and provides precise estimates.In the present study, the application of GEE is demonstrated using a simulated data. Different steps involved in the GEE analysis using SPSS software were also provided as an easy guide to researchers. This study could serve medical researchers understand, perform and interpret GEE in a better way.