Meta-analysis for a single mean estimate

Steve Simon


Someone noted that the usual meta analysis is carried out for the study on two treatment groups, usually for a difference in means. What if you had several studies estimating not a difference in means, but just estimates of a single mean. Could you conduct a meta-analysis in this situation?

Yes, you can do this. The statistics change slightly, but in a predictable and logical way. The general concept and the general procedures stay the same.

It gets a bit trickier when the statistic of interest is a single proportion (e.g., sensitivity/specificity), especially when there are some publications where the proportion equals 1.0. If you try to weight by the inverse variance, these studies get infinite weight.

You can find an earlier version of this page on my original website.