JASP meta-analysis

Comments

• EJ

  February 2021

  Hi Ravi,

  1. A covariate is something continuous, like GDP; a factor refers to a discrete category, like country. Sometimes there are no covariates or factors of interest, so it is purely optional.
  2. Most people would recommend a random effects model; however, with few studies and very low heterogeneity, there is something to be said for the fixed effects model.
  3. I believe the factors and covariates allow you to do fixed-effects meta-regression (I'll check). Not sure about random-effects meta-regression (I'll check that too). https://en.wikipedia.org/wiki/Meta-regression

  Cheers,

  E.J.
• raviteja85

  February 2021

  Thanks a lot E.J. This helps.
• Raoul

  March 2021

  Hi Ravi,

  As a follow up to EJ:

  1. I have nothing to add
  2. It's true that most people mindlessly recommend random effects over fixed effects. A good discussion of this choice can be found in (Hedges & Vevea,1998). In summary it comes down to whether you want to answer the question of whether the mean effect of existing studies is different from zero, or whether you want to test the hypothesis that the mean of all possible replications of those studies are different from zero. It's a somewhat subtle discussion, perhaps https://www.meta-analysis.com/downloads/Intro_Models.pdf helps.
  3. meta-regression is what you are doing in JASP as soon as you enter explanatory variables in the covariates and/or factors boxes.

  Good luck,

  
  

  Raoul
• raviteja85

  March 2021

  Thanks a lot Raoul.
• Nik

  March 2021

  Hello together

  maybe you can help me with my problem. I have performed a subgroup analysis in the course of my meta-analysis. So I wanted to ask you how I can calculate the heterogeneity (I^2 = %) between the subgroups?

  
  

  Thanks,

  Nik