A three-level mixed model to account for the correlation at both the between-day and the within-day level for ecological momentary assessments

Longitudinal within-person variability around personality trajectories.

“…Individual differences in autocorrelations have indeed already been shown to exist for personality constructs in intensive longitudinal data (Jahng & Wood, 2017; Nestler, 2022; Vansteelandt & Verbeke, 2016). Furthermore, models that constrain autocorrelations to be equal across people will have biased parameter estimates if they are truly heterogeneous (Hamel et al, 2012; Jahng & Wood, 2017; Ma et al, 2020). Thus, quantifying the degree of individual differences in autocorrelations for longitudinal personality development, in conjunction with individual differences in sigma, could similarly prove to be a fruitful endeavor.…”

Guest Editorial: Articles selected from the 2020 International Conference on Health Policy Statistics

“…Such data can have complex within-subject correlation structures that need to be modeled correctly to avoid bias. Ma et al ( 2020 ) propose a linear mixed effects model for such data that accounts for autocorrelation at both the within-day and between-day levels, providing a better fit to the data. This approach represents a practical solution to the issue of modeling EMA data, which we expect to become increasingly common.…”