I am now conducting a meta-analysis for depression in a certain disease. The prevalence of depression is available for the studies (n=21), and I would like to conduct meta regression with the prevalence of depression as the dependent variable, and baseline sociodemographic and clinical data as the independent variables.
However, some of the baseline data, such as female or male, is in terms of proportions (as opposed to continuous data like age). Is it valid to conduct a meta regression with the logit transformed proportions for both my independent and dependent variables? I have also conducted Q-Q plot and Shapiro-Wilk test for normality to ensure the logit transformed proportions are approximately normal.
