I am working on a meta-analysis of diagnostic value and one included study used several sets, comprises training set, validation set, replication set, and asymptomatic set. Each of them yielded different diagnostic values. How should I extract the data for quantitative analysis? Should I extract all of them and made them into individual data?
What specifically is your research question? If it is defined well enough, it should answer your "data set" question for you. It is your research question. You should be able to defend your selection.
The question is not very clear.
However, if I understood you correctly, from my limited experience, it is better to go back to raw data and re-analyze, in case the studies you choose have followed different analytical methods. But it is practically difficult. Data extraction is not an easy thing to do.
The research question is not clear.
If I understood you, one study reported diagnostic value for each set, and you are confused about whether you should choose. If so, I would recommend you better define your research question and use the most appropriate dataset from the available data and studies.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts