Dear reader,
I created a dataset for a meta-analysis in which I have effect sizes calculated together with their standard errors. With a random effects model weights are chosen using the (inverse) error variance with the consequence that greater weight is given to more precise studies. However, the amount of estimates vary between studies which results in inappropriate weight being placed on those studies reporting more estimates. Therefore i want to use another weighting scheme in which the weight given to each particular observation corresponds to the square root of the mean sample size of each study. This is calculated for each estimate and has the variable name ''weight''.
When I declare the data as meta data:
meta set EffectsizeT std_error
I can now get the summary statistics for both random model but also with the calculated weights using these commands:
meta summarize, random
meta summarize, wgt(weight)
When I do the meta regression; meta regress x y , I also want to use the calculated weights, however, I do not know how to apply this weighting scheme, as wgt(weight) does not work in the meta regression command. Does someone know how to do this? Your help is much appreciated, thanks in advance
Best,
Daniël
I created a dataset for a meta-analysis in which I have effect sizes calculated together with their standard errors. With a random effects model weights are chosen using the (inverse) error variance with the consequence that greater weight is given to more precise studies. However, the amount of estimates vary between studies which results in inappropriate weight being placed on those studies reporting more estimates. Therefore i want to use another weighting scheme in which the weight given to each particular observation corresponds to the square root of the mean sample size of each study. This is calculated for each estimate and has the variable name ''weight''.
When I declare the data as meta data:
meta set EffectsizeT std_error
I can now get the summary statistics for both random model but also with the calculated weights using these commands:
meta summarize, random
meta summarize, wgt(weight)
When I do the meta regression; meta regress x y , I also want to use the calculated weights, however, I do not know how to apply this weighting scheme, as wgt(weight) does not work in the meta regression command. Does someone know how to do this? Your help is much appreciated, thanks in advance
Best,
Daniël
