It’s almost as if “significant” coefficients from your multivariable regression model *aren’t* inherently causal or biologically meaningful 
https://twitter.com/RottenInDenmark/status/1456197431192924161 …
-
This Tweet is unavailable.
-
Replying to @EpiEllie
Why are we comparing coefficients from completely different models from different populations with completely different statistical controls? The first one, from Sweden, doesn't control weight, the second one, from the US, doesn't control for marriage/household status.
2 replies 1 retweet 29 likes -
Replying to @zeynep
Neither study actually assesses the causal effect of either of the variables highlighted, is my point. Regardless of any differences between the studies, neither can be interpreted as a biological or sociological effect.
1 reply 1 retweet 30 likes -
Replying to @EpiEllie
Right, and hence the importance of taking into account the existing and fairly extensive literature on causal pathways for both, rather than dismissing either through a statistically invalid comparison between two very different studies, no? Both are important questions/issues.
1 reply 0 retweets 5 likes -
Replying to @zeynep
If you dont like the comparison between them, take it up with the OP. All I care about here is the damage that the
#Table2Fallacy has done, during the pandemic & otherwise. For example, see:https://journals.lww.com/epidem/Fulltext/2021/01000/Comment_on_Williamson_et_al___OpenSAFELY___The.38.aspx …1 reply 1 retweet 8 likes -
Replying to @EpiEllie
People can get confused by stats, so seemed useful to remind people comparing coefficients A and B from different models from different populations with different controls to declare A is a "better predictor" than B—when neither controlled for both A and B—is obviously invalid.
1 reply 1 retweet 4 likes -
Dr Tufekci, you would not be so hung up on this point, if you read the whole original Hobbes thread, where he says the comparison is invalid but he's trying to illustrate the point that neither association is proven causal while being broadly misinterpreted as causal
1 reply 0 retweets 1 like -
You can underline the importance of the extensive examination of the potential causal pathways for both in existing and pretty broad literature, and object to some or question studies, but both factors are way beyond glib/simple misinterpretations at this point.
1 reply 0 retweets 1 like
The relationship between social support and morbidity, mortality and quality of life is among the most robust findings in literature, for example. There are open questions and complexities, but it is way beyond the researchers saw a correlation and lost their ability to reason.
-
-
Hobbes' point, if you read the whole thread or some of his well respected journalism, is that active discrimination and shaming of fat ppl is a bigger driver of these outcomes than the inherent biology of a certain amount of fat
1 reply 0 retweets 6 likes -
e.g. this already-classic piece:https://highline.huffingtonpost.com/articles/en/everything-you-know-about-obesity-is-wrong …
0 replies 0 retweets 2 likes
End of conversation
New conversation -
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.