Possibly the greatest NBER abstract ever written.

The Great Man theory of history.... turns out to have some pretty strong statistical support. https://t.co/HIdtRvMSVk
School closures had NO EFFECT (precise null!) on socioeconomic outcomes and future schooling behavior during the 1918 pandemic. https://t.co/VLlGt79wi1
The authors note that school closures were not as long back then.

But also school calendars were far shorter. So idk how it shakes out.
Informing men about the odds their wife will die in childbirth reduces fertility, increases marital satisfaction, and reduces mismatch in fertility preferences within couples (in Zambia). https://t.co/8qF2hn2rwD
This is a neat possible channel for the education-fertility linkage. The effect size they observe is yuuuuuuge.
People with Facebook friends in areas with high COVID cases, and who saw more FB posts about folks getting sick and hospitalized, did more social distancing and were less likely to post about reopening. https://t.co/vze86IGjc4
This is a pretty strong argument that the media should have been broadcasting more images and videos of human suffering in hospitals and elsewhere and fewer videos of wait lines for grocery stores.
This paper also shows an educational reform altering attitudes towards cousin marriage and also reducing its frequency: https://t.co/x2CB7TE7dA

More from Lyman Stone 石來民

So a few days back I was tweeting about SSRIs. The big question with these drugs is: why do controlled trials routinely show such small effects when practitioners and patients report life-changingly-large effects?

So first off, at this point the evidence is pretty clear that SSRIs and other anti-anxiety/anti-depression drugs truly don't do very much. Their average effects are beneath clinical significance, as I tweeted about here:


Basically, the problem these drugs face is that while they actually see relatively LARGE effects.... but that placebos in those trials ALSO see large effects (and most untreated depression improves within a year anyways).

So basically you have this problem where:
1. The condition tends to improve on its own in a majority of cases
2. Placebo effects for the condition are unusually large

Which means the large crude effects of SSRIs get swamped.

So that raises two new questions.
1. (Not my focus here) Are we treating these conditions appropriately given their untreated prognosis is usually (though certainly not always!!) "goes away in a few months"?

2. Why are placebo effects so unusually large?

More from Trading

You May Also Like