Currently I am doing or at least trying to do a meta-analysis of some analytical studies in the field of smokeless tobacco and cancer, a few papers have given adjusted odds ratios, now when I calculate the standard error, should I take strata wise figures of exposed and unexposed among the two groups, or the whole study sample for calculation of standard error, because if I take the log of an adjusted odds ratio and then calculate the standard error from crude figures (don't know the right word), I am not sure if this would make sense, or would it ?