If I want to put categorical variables in with continuous variables in my meta-regression run, (using the in-built -meta- function) do I need to transform them or tell Stata they are categorical somehow, because they are currently numerically coded. E.g. variable = country, 1 = UK, 2 = USA, 3 = China, etc...
I have already performed sub-group analyses on them, but am now trying to find the model of best fit that explains the most of my between-study heterogeneity by combining continuous and categorical covariates as potential moderators. I have entered them without transforming them in any way and gotten outputs that seem to make sense, but would like to check I have followed the correct process. (As other guidance talks about dummy-coding, which I attempted, but then encountered errors when trying to include all the dummy-coded levels in a meta-regression.)
Many thanks for any help.
I have already performed sub-group analyses on them, but am now trying to find the model of best fit that explains the most of my between-study heterogeneity by combining continuous and categorical covariates as potential moderators. I have entered them without transforming them in any way and gotten outputs that seem to make sense, but would like to check I have followed the correct process. (As other guidance talks about dummy-coding, which I attempted, but then encountered errors when trying to include all the dummy-coded levels in a meta-regression.)
Many thanks for any help.
