Multilevel modeling for longitudinal data: concepts and applications

Abstract

Purpose This paper aims to discuss multilevel modeling for longitudinal data, clarifying the circumstances in which they can be used. Design/methodology/approach The authors estimate three-level models with repeated measures, offering conditions for their correct interpretation. Findings From the concepts and techniques presented, the authors can propose models, in which it is possible to identify the fixed and random effects on the dependent variable, understand the variance decomposition of multilevel random effects, test alternative covariance structures to account for heteroskedasticity and calculate and interpret the intraclass correlations of each analysis level. Originality/value Understanding how nested data structures and data with repeated measures work enables researchers and managers to define several types of constructs from which multilevel models can be used.

Motivation states to move, be physically active and sedentary vary like circadian rhythms and are associated with affect and arousal

“…Still the current approach is valid given that modeling level two variance (accounting for unmeasured individual differences) is particularly important, especially when the numbers of clusters is small (83). Our models also used restricted maximum likelihood (REML) for estimation-a method shown to perform well even with 10 or fewer clusters (83)(84)(85). Thus, our statistical approach is consistent with recent suggestions in the literature for analyzing multilevel data with small level two sample sizes.…”

Hazardous alcohol use among Danish adolescents during the second wave of COVID-19: Link between alcohol use and social life

“…Due to the hierarchical structure of the study design – where students were nested within schools, time was measured within students and the non-normality of several dependent variables – we applied multilevel models (MLM) with random intercepts that are developed to allow for the nesting of multiple individuals within a group and can be fitted for both continuous, count and dichotomous dependent variables ( Curran et al, 2010 ; Hair Jr & Fávero, 2019 ). In addition, we were unable to measure how the different regions administrated the COVID-19-related resources and how each school may have different ways of adjusting to the national recommendations on preventive measures and restrictions; hence, modelling the nesting in schools was important.…”

“…The final multilevel regression models included time as a temporal variant (level 1) within-subject independent variable with fixed effects nested within students, students (level 2) nested within schools with random effects, and schools (level 3). We included random intercepts for students given the potential variability in alcohol consumption between students ( Hair Jr & Fávero, 2019 ). Further, the models controlled for age, gender and COVID-19 national risk levels.…”

Seeking an adversary: The radical right and the saliency of pro-immigration positions in green party speeches

“…This may bias our results as each observation may not actually be independent, resulting in an atomistic fallacy whereby group-level inferences are drawn from individuallevel data (Luke, 2004;Steenbergen and Jones, 2002). In order to account for the data's structure and ensure that all variables are assessed at the monthly level, we use multilevel regressions, whereby each monthly observation is nested within speakers and years (Hair and Fávero, 2019). Consequently, the multilevel regressions will ensure independent observations and remove time-related sources of bias through the year fixed effects component.…”