Episode 57: Collider bias

Among patients hospitalized for COVID, smokers had better outcomes. Among people with cardiovascular disease, those with obesity live longer. Among NBA basketballers, taller players don’t do any better. These are all facts. But the interpretation you might immediately draw is completely wrong.

It turns out that these findings (and many more) might be due to the weird and under-discussed phenomenon of “collider bias”. Everyone who’s interested in scientific methods knows what a confounder is—but do they know what a collider is? In this episode of The Studies Show, Tom and Stuart attempt to explain.

We’re delighted to announce our sponsor for the next month: GiveWell. They’re the organisation who use rigorous evidence to point you towards the highest-impact charities. Want to make sure your donation goes as far as possible, maximising the lives that you’ll save and improve? GiveWell.org is the place to go.

And here’s a fantastic opportunity: if you’re a new donor, GiveWell will match up to $100 of your charitable donation if you go to GiveWell.org, then choose “PODCAST” and enter “The Studies Show” at checkout.

Show notes

* French study on COVID and smoking rates

* French doctors handing out nicotine patches during the pandemic

* Review of 13 studies in China showing lower smoking rates in those hospitalised for COVID

* Among heart attack sufferers, smokers have better subsequent health

* Obesity linked to improved survival among patients with a wide range of diseases

* Within the NBA, tall basketball players do no better than short ones

* Standardized testing doesn’t predict how well graduate physics students do

* The same but for biology

* The same but for STEM in general

* Do neurotic people actually live longer, once you correct for self-rated health?

* Julia Rohrer’s blog article on collider bias, using the conscientiousness/IQ relation

* The “collider scope” paper - one of the best explanations of the phenomenon

* Article on “the obsesity paradox”

* Follow-up arguing that it might not be a paradox at all

Credits

The Studies Show is produced by Julian Mayers at Yada Yada Productions.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit sciencefictionspod.substack.com/subscribe