Something that I keep seeing pop up is the idea that meta-analysis somehow eliminates issues with the underlying research This is just confusingly incorrect
-
Show this thread
-
Replying to @GidMK
That’s a straw man. No one says it eliminates all issues with underlying research. The argument is that it can help account for variations in things like dosage and methods. Large scale crts are also worthless if you give the thousands of participants the wrong dosage.
1 reply 0 retweets 2 likes -
Replying to @chad_senger
That's simply not something that meta-analysis can do. It does not account for variation in dosage and methods, all it does is weight studies by their variance and provide a weighted mean/median
1 reply 0 retweets 1 like -
Replying to @GidMK
Are you saying that a single study with a single treatment method can account for variability in treatment as well as a meta-analysis of many studies with different treatment methods?
1 reply 0 retweets 0 likes -
Replying to @chad_senger
While there are some exceptions, in almost every situation putting many studies with divergent treatment methods into a single meta-analytic model gives you an entirely meaningless result
1 reply 0 retweets 1 like -
Replying to @GidMK @chad_senger
Think of it like an average (because it essentially is one) - if one study measured weight, another height, and yet another blood pressure, averaging the results from them all would give you a number with narrow confidence intervals that had literally no meaning
1 reply 0 retweets 1 like -
Replying to @GidMK
Your straw-manning again. Yes, if you measure completely different things it’s worthless. But if the studies are similar, you can eliminate noise and outliers.
1 reply 0 retweets 2 likes -
Replying to @chad_senger
If you have multiple very similar small trials that measure the same thing in a very robust way, meta-analysis is great at giving you a more certain estimate. But if the studies are bad, or very divergent, it gives you no more certainty than you have to begin with
2 replies 0 retweets 1 like -
Replying to @GidMK
This is of course true. But not unique to meta analysis, so it’s not clear why your focusing on them. A poorly run large scale double blind randomized control trial is just as worthless if it’s run poorly. A meta analysis could at least overcome a bad study with the aggregate
1 reply 0 retweets 0 likes -
Replying to @chad_senger
That is not true actually. Meta-analyses can be biased by bad studies in a number of ways, often to do with the weighting. Sometimes a single bad study can push the entire model in a single direction
1 reply 0 retweets 0 likes
The rest of your statement confuses me. I talk all the time about bad study design, but *this* conversation is about meta-analysis
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.