#statstab #354 Modeling Non-Linear Associations in Meta-Regression {metafor}
Thoughts: Modelling non-linear effects is easier than you think with splines!
#splines #metafor #r #metaanalysis #nonlinear #moderator #metaregression #cubic
https://www.metafor-project.org/doku.php/tips:non_linear_meta_regression
Very important work by James Pustejovsky (@jepusto), Jingru Zhang (not on Mastodon as far as I can tell), and Elizabeth Tipton (@stats_tipton) on how to conduct meta-regression analyses when at least some studies provide multiple (possibly dependent) effect size estimates corresponding to different values of the moderator:
https://www.jepusto.com/talk/sree-2023-equity-related-moderator-analysis/
In general, there are a lot of lessons from the multilevel literature we can learn from.
Background/Context In meta-analyses examining educational interventions, researchers seek to understand the distribution of intervention impacts, in order to draw generalizations about what works, for whom, and under what conditions. One common way to examine equity implications in such reviews is through moderator analysis, which involves modeling how intervention effect sizes vary depending on the characteristics of primary study participants.
Dr Mircea Zloteanu ๐บ๐๐
Wolfgang Viechtbauer