4 ms·
well if the original studies are inadequate simply because N is too low, then grouping many of them can resolve that problem.
by metalliqaz 5y ago
well if the original studies are inadequate simply because N is too low, then grouping many of them can resolve that problem.
- mahogany 5y agoIs that really true from a statistical standpoint? It seems to me that running ten 20-person studies is different and less valuable than running a single 200 person study, because each of the 20-person studies has a much larger error range that you have to account for. But I'm also not good with stats.
- mcguire 5y agoThat's ideally what the meta-analysis is supposed to correct for. But if two of the studies are just completely made up and another couple have errors in their design that aren't mentioned in the write-up....
- bjornsing 5y agoHow do you correct for publication bias in these meta analyses? I really hope they do correct for it…