Cochrane says ordinal data can be meta-analysed as dichotomous or continuous data. I get the dichotomous analysis if you re-group your ordinal data to 2 groups (success/failure). But i dont get how in the world you can analyse ordinal data as continuous. Continuous data is meta-analysed using mean difference between groups. How would that work for ordinal data? There’s no mean difference to compare since ordinal data has no mean. Many thanks if u can explain it in layman terms.
In: Mathematics
The ordinal data could have a mean calculated for the sample/population. It wouldn’t have the same real world meaning as a mean from continuous data, but for certain ordinal variables it would make sense to look at it that way. For example, pain on a scale of one to ten. It’s ordinal, because someone who says their pain is 7 doesn’t objectively have 7 times the amount of pain of someone who says their pain is one. But you could calculate a mean pain score. You could then compare that mean to the mean pain score of people treated with some drug. A meta-analysis could then be done to combine multiple studies to see if the drug reduces pain
Latest Answers