It'd be great to have a vignette on using g_mlm() to estimate standardized mean differences from multi-level models, such as you might estimate in multi-site trials or cluster-randomized trials. Two key issues to address:
1) The need to use separate models for the numerator and the denominator of the effect size estimate. This is beyond the scope of what's discussed in Pustejovsky, Hedges, & Shadish (2014) for multiple baseline designs.
2) The choice of variance components to use in the denominator of the effect size. Previous research has proposed various alternatives (total variance, within-cluster variance, between-cluster variance) but there's a need for more guidance about which alternative to use in a given application.
It'd be great to have a vignette on using
g_mlm()
to estimate standardized mean differences from multi-level models, such as you might estimate in multi-site trials or cluster-randomized trials. Two key issues to address: 1) The need to use separate models for the numerator and the denominator of the effect size estimate. This is beyond the scope of what's discussed in Pustejovsky, Hedges, & Shadish (2014) for multiple baseline designs. 2) The choice of variance components to use in the denominator of the effect size. Previous research has proposed various alternatives (total variance, within-cluster variance, between-cluster variance) but there's a need for more guidance about which alternative to use in a given application.