Hierarchical Linear Models foR Psychologists

_{Last updated 2024-09-26.}

This Github repo contains all lesson files for Hierarchical Linear Models in R. The goal is to impart students with the basic tools to construct, evaluate and compare various (generalized) linear mixed models, using lme4, based on Lesa Hoffman’s Longitudinal Analysis: Modeling Within-Person Fluctuation and Change. (Materials developed with Yael Bar-Shachar.)

These topics were taught in the graduate-level course Hierarchical Linear Models for Psychologists (Psych Dep., Ben-Gurion University of the Negev; Psych Dep., Tel-Aviv University). This course assumes basic competence in R (importing, regression modeling, plotting, etc.), along the lines of Practical Applications in R for Psychologists.

Notes:

• This repo contains only materials relating to Practical Applications in R, and does not contain any theoretical or introductory materials.
• Please note that some code does not work on purpose, to force students to learn to debug.

Setup

You will need:

1. A fresh installation of R (preferably version 4.3.2 or above).
2. RStudio IDE (optional, but recommended).
3. The following packages, listed by lesson:

Lesson	Packages
01 HLM Basics	tidyverse, lmerTest, emmeans, performance, parameters, merDeriv, haven, sjPlot, afex
02 Model Comparisons	dplyr, ggplot2, lmerTest, performance, parameters, haven, scales, glue, sjPlot, bayestestR
03 Cross level interactions and effect sizes	dplyr, datawizard, ggplot2, lmerTest, performance, parameters, marginaleffects, scales
04 Growth Models	dplyr, ggplot2, lmerTest, parameters, performance, marginaleffects, haven, scales, glue, nlme, glmmTMB, brms, remotes, mixedup, loo, posterior, bayestestR
05 Within-Person Fluctuation Models	tidyverse, datawizard, lmerTest, performance, parameters, haven
06 GLMMs	tidyverse, lme4, performance, parameters, marginaleffects, haven, insight, scales
07 Multilpe random factors	dplyr, ggplot2, patchwork, lmerTest, performance, parameters, marginaleffects, mlmRev, forcats
08 ANOVA	dplyr, ggplot2, lmerTest, datawizard, emmeans, afex, car, effectsize, patchwork, performance, statmod, see
09 Assumptions	tidyverse, lmerTest, performance, DHARMa, scales

You can install all the packages used by running:

# in alphabetical order:

pkgs <- c(
  "afex", "bayestestR", "brms", "car", "datawizard", "DHARMa",
  "dplyr", "effectsize", "emmeans", "forcats", "ggplot2", "glmmTMB",
  "glue", "haven", "insight", "lme4", "lmerTest", "loo", "marginaleffects",
  "merDeriv", "mixedup", "mlmRev", "nlme", "parameters", "patchwork",
  "performance", "posterior", "remotes", "scales", "see", "sjPlot",
  "statmod", "tidyverse"
)

install.packages(pkgs, dependencies = TRUE)

Package Versions

The package versions used here:

• afex 1.4-1 (CRAN)
• bayestestR 0.14.0 (CRAN)
• brms 2.21.0 (CRAN)
• car 3.1-2 (CRAN)
• datawizard 0.12.3 (CRAN)
• DHARMa 0.4.6 (CRAN)
• dplyr 1.1.4 (CRAN)
• effectsize 0.8.9 (CRAN)
• emmeans 1.10.4 (CRAN)
• forcats 1.0.0 (CRAN)
• ggplot2 3.5.1 (CRAN)
• glmmTMB 1.1.9 (CRAN)
• glue 1.7.0 (CRAN)
• haven 2.5.4 (CRAN)
• insight 0.20.4 (CRAN)
• lme4 1.1-35.5 (CRAN)
• lmerTest 3.1-3 (CRAN)
• loo 2.8.0 (CRAN)
• marginaleffects 0.22.0 (CRAN)
• merDeriv 0.2-4 (CRAN)
• mixedup 0.4.0 (Github: m-clark/mixedup)
• mlmRev 1.0-8 (CRAN)
• nlme 3.1-164 (CRAN)
• parameters 0.22.2 (CRAN)
• patchwork 1.3.0 (CRAN)
• performance 0.12.3 (CRAN)
• posterior 1.6.0 (CRAN)
• remotes 2.5.0 (CRAN)
• scales 1.3.0 (CRAN)
• see 0.9.0 (CRAN)
• sjPlot 2.8.16 (CRAN)
• statmod 1.5.0 (CRAN)
• tidyverse 2.0.0 (CRAN)