Another perspective on autocorrelation in intensive longitudinal data: Polynomial (straight line, quadratic…) ‘vs’ autoregressive models

Abstract

While the theory of longitudinal data analysis (LDA) has a solid foundation, there are instances where the assumptions of the analytical model remain unverified. Failure to examine autocorrelation in residuals (ACR) can elevate the risk of committing a Type I error, leading to the rejection of a true null hypothesis. This study compares two distinct analytical models within LDA: the polynomial (straight line, quadratic…) model and the autoregressive (AR) model. Three separate studies were conducted to investigate this comparison.