It doesn't matter how big your sample is if you haven't at least tried to address systemic biases!
-
-
Show this thread
-
A good example - this is from a study looking at working hours and stroke. Huge potential for unmeasured confounding. But the analysis only controlled for these factors (logistic model, y/n stroke)pic.twitter.com/zhv9zREOQw
Show this thread -
Totally fine that they didn't control for everything - you can never have ALL the information - but there's no control for ethnicity, only a very meagre control for income, and none for education either That may not be a problem, but it should be addressed!
Show this thread -
As it is, I'd have to say that the observed association between stroke and working hours is just as likely to be a result of any number of unmeasured confounders as a causal relationship
Show this thread -
In other words, I'm not convinced that working too many hours will give you a stroke, outside of the societal pressures that force people to work 50+ hour weeks in the first place
Show this thread -
Show this thread
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.