meta-workshop

Meta-analysis workshop for the Society of Experimental Biology (SEB) 2022

Part 1

Introduction: Why do meta-analysis?

• Quantitative synthesis that provides effect magnitude that can be used for:
  1. power analyses,
  2. predicting effects across space and time etc.
• Philosophical
  • One study is almost never enough (sampling theory)
  • Studies in ecology and evolution are underpowered so provides a way to improve power.
• Heterogeneity.
  • We can measure how variable an outcome is across studies
  • Attempt to explain this variability
  • Provide important new avenues for research and test critical hypotheses
  • Prediction intervals

Effect sizes, what to use and when to use them, what they mean

• Key characteristics of an effect size
• Comparability issues
• 'Nuisance heterogeneity' and how to deal with it?

Meta-analytic models

• Random effects and multi-level meta-analysis
• Meta-regression models
• Mostly need multilevel meta-analysis -Multi-level is important! Add obs in metafor
• Visualising
  • Funnel plots
  • Orchard plots
• Interpretation of and reporting from models

Part 2

Advanced Meta-analytic Topics

• Publication bias
  • how to deal with it.
• Meta-analysis of magnitudes
  • Correctly analysing unsigned effect sizes
• Controlling for Non-independence
  • Phylogeny
  • Shared control / shared trait
  • VCV matrices; phylogeny, shared control etc
  • Robust variance estimators when unknown.
• Missing standard deviations
• Arm-based meta-analysis