Hello there,
I am using the robumeta package in R to fit a meta - regression model.
One of my categorical moderators has three levels. Everything is dummy coded.
AFAIK regression coefficients robumeta gives out indicates whether there is a difference between a single level of a moderator and a single comparison level. No problem with categorial moderators with only two levels. But not really helpful for a categorical moderator with 3 levels To create an omnibus test and get an F that indicates whether the moderator as a whole is significant, Tanner & Smith (2016) suggested using the club sandwich package.
I have read through the tutorial and the package descriptions, but I assume my knowledge about Wald-Tests is lacking. In the tutorial I am reading they use different constraints like 2:3. / 2:4 / 2:6.
Can someone explain to me what constraints to use / how to choose the constraints?
R fitted model looks as follows (Occupation being the moderator with three levels):
Fit
Saskia Annen for your kind perusal
Pustejovsky, J. (2017). clubSandwich: Cluster-Robust (Sandwich) Variance Estimators with Small-Sample Corrections. R Package Version 0.3.0. Available at: https://CRAN.Rproject.org/package=clubSandwich
Saskia Annen for your kind perusal
Pustejovsky, J. (2017). clubSandwich: Cluster-Robust (Sandwich) Variance Estimators with Small-Sample Corrections. R Package Version 0.3.0. Available at: https://CRAN.Rproject.org/package=clubSandwich
Pradeep Paraman thank you very much for the answer.
This is the package I am using, correct.
I was asking for information on how to choose the correct constraints using that package.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R