This function calculates values of \(I^2\) and the variance distribution for multilevel meta-analysis models fitted with rma.mv.

mlm.variance.distribution(x)

Arguments

x

An object of class rma.mv. Must be a multilevel model with two random effects (three-level meta-analysis model).

Value

Returns a plot summarizing the variance distribution and \(I^2\) values, as well as a data frame for the results.

Details

This function estimates the distribution of variance in a three-level meta-analysis model (fitted with the rma.mv function). The share of variance attributable to sampling error, within and between-cluster heterogeneity is calculated, and an estimate of \(I^2\) (total and for Level 2 and Level 3) is provided. The function uses the formula by Cheung (2014) to estimate the variance proportions attributable to each model component and to derive the \(I^2\) estimates.

References

Harrer, M., Cuijpers, P., Furukawa, T.A, & Ebert, D. D. (2019). Doing Meta-Analysis in R: A Hands-on Guide. DOI: 10.5281/zenodo.2551803. Chapter 12.

Cheung, M. W. L. (2014). Modeling dependent effect sizes with three-level meta-analyses: a structural equation modeling approach. Psychological Methods, 19(2), 211.

Examples

# Use dat.konstantopoulos2011 from the "metafor" package library(metafor) # Build Multilevel Model (Three Levels) m = rma.mv(yi, vi, random = ~ 1 | district/school, data=dat.konstantopoulos2011) # Calculate Variance Distribution mlm.variance.distribution(m)
#> % of total variance I2 #> Level 1 4.812686 --- #> Level 2 31.862476 31.86 #> Level 3 63.324838 63.32 #> Total I2: 95.19%