sjstats (version 0.17.3)

re_var: Random effect variances

Description

These functions extracts random effect variances as well as random-intercept-slope-correlation of mixed effects models. Currently, merMod, glmmTMB, stanreg and brmsfit objects are supported.

Usage

re_var(x, adjusted = FALSE)
get_re_var(x, comp = c("tau.00", "tau.01", "tau.11", "rho.01",
  "sigma_2"))

Arguments

x

Fitted mixed effects model (of class merMod, glmmTMB, stanreg or brmsfit). get_re_var() also accepts an object returned by the icc function.

adjusted

Logical, if TRUE, returns the variance of the fixed and random effects as well as of the additive dispersion and distribution-specific variance, which are used to calculate the adjusted and conditional r2 and icc.

comp

Name of the variance component to be returned. See 'Details'.

Value

get_re_var() returns the value of the requested variance component, re_var() returns all random effects variances.

Details

The random effect variances indicate the between- and within-group variances as well as random-slope variance and random-slope-intercept correlation. Use following values for comp to get the particular variance component:

"sigma_2"

Within-group (residual) variance

"tau.00"

Between-group-variance (variation between individual intercepts and average intercept)

"tau.11"

Random-slope-variance (variation between individual slopes and average slope)

"tau.01"

Random-Intercept-Slope-covariance

"rho.01"

Random-Intercept-Slope-correlation

The within-group-variance is affected by factors at level one, i.e. by the lower-level direct effects. Level two factors (i.e. cross-level direct effects) affect the between-group-variance. Cross-level interaction effects are group-level factors that explain the variance in random slopes (Aguinis et al. 2013).

If adjusted = TRUE, the variance of the fixed and random effects as well as of the additive dispersion and distribution-specific variance are returned (see Johnson et al. 2014 and Nakagawa et al. 2017):

"fixed"

variance attributable to the fixed effects

"random"

(mean) variance of random effects

"dispersion"

variance due to additive dispersion

"distribution"

distribution-specific variance

"residual"

sum of dispersion and distribution

References

• Aguinis H, Gottfredson RK, Culpepper SA. 2013. Best-Practice Recommendations for Estimating Cross-Level Interaction Effects Using Multilevel Modeling. Journal of Management 39(6): 1490<U+2013>1528 (10.1177/0149206313478188)

• Johnson PC, O'Hara RB. 2014. Extension of Nakagawa & Schielzeth's R2GLMM to random slopes models. Methods Ecol Evol, 5: 944-946. (10.1111/2041-210X.12225)

• Nakagawa S, Johnson P, Schielzeth H (2017) The coefficient of determination R2 and intra-class correlation coefficient from generalized linear mixed-effects models revisted and expanded. J. R. Soc. Interface 14. 10.1098/rsif.2017.0213

See Also

icc

Examples

# NOT RUN {
library(lme4)
fit1 <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy)

# all random effect variance components
re_var(fit1)
re_var(fit1, adjusted = TRUE)

# just the rand. slope-intercept covariance
get_re_var(fit1, "tau.01")

sleepstudy$mygrp <- sample(1:45, size = 180, replace = TRUE)
fit2 <- lmer(Reaction ~ Days + (1 | mygrp) + (Days | Subject), sleepstudy)
re_var(fit2)

# }