Nice article from @stephmurrayyyy on demographic forecasting.
My only quibble would be the quote that short-run forecasting is easy. That's wrong. It's often harder! theatlantic.com/family/archive…
*Nobody* disputes that fertility in, say, Nigeria, will be lower 20 years from now than it is today. And while there's disagreement by how much, the band of error is not THAT wide.
But will it fall *next year*? And by how much?
Demographers do not even model that question!
To find guidance on short-run forecasting like that the demographic literature is almost worthless; it just doesn't concern itself with that kind of thing. For that you need economics papers and such, which estimate elasticities directly.
Also, the bit about it being easy to say how many people of childbearing age there are right now is.... hopelessly naive.
Here's a thread I did on the catastrophically bad forecasts for India by UNPD AND IHME:
Short version is, none of the sources even remotely agree on historic births, none of the sources agree on current population, and both major forecasters have outright-impossible "first year forecasts." They were falsified in the *first year* of the forecast!
So, overall, nice piece by @stephmurrayyyy on demographic forecasting and the challenges in it.... but if anything it understates the difficulty. Doing demography is even harder than it looks.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
US excess deaths appear to be remaining stubbornly high.
Could still just be a time lag before falling cases generate falling deaths. Could also be excessive upward revisions in my model. Hard to say exactly.
Here's deaths in the rolling 12 month period vs. expected deaths in that period from previous trends. As you can see, our current pandemic is really edging towards 1918.
look, it's just too insane a theory to not believe it. theories this nuts must be believed merely on the premise that the possibility of them being true makes the world vastly more entertaining.
google polls no longer allow you to ask any questions about race, ethnicity, politics, sex, health, pregnancy, or religion
this is irritating because google polls are the super-cheap option. for a 1-question thing you could recruit respondents for $0.15 per response, which made it an incredibly good option for doing "tracking polls" of various kinds
but google is making it clear now that their cheap polls are basically *only* for marketing research. that's it.
This survey report has some clear typos in it, and the methodology is a convenience sample, so, caveats.... but the cross-cultural gaps it shows in birthing norms are interesting nonetheless. drive.google.com/file/d/1Kn1CQg…
Especially comparing Europeans and North Americans, two groups which should both have pretty similar access to the sampling frame, a pregnancy tracking app.
North American women are more likely to say the father of their child is highly involved in the pregnancy:
in general, when protestors seize public roadways and deny general access to them, especially without prior provision of permits, it is reasonable for the security forces of the state to clear them from those roads with appropriate force.
this applies in the winter of 2022 and the summer of 2020 and in fact it just sort of applies generally.
when protestors seize public spaces and then deny their usage to the public, and do so continuously and without legal permission, they should be cleared.
parks, roads, whatever.
now, if legal authorities are denying permits for reasonable protests, or creating unduly onerous permitting processes, or otherwise denying basic assembly rights, this general rule can of course be suspended.
With the first month of abortion data from Texas' abortion restrictions, we can see that the number of abortions fell.
What happened to births?
Basically nothing.
Now, most abortions are early on, so in principle you'd expect several months of delay before any meaningful effects.
Right now, given the usual timing of abortions, we can really only reliably estimate the share of monthly conceptions which ended in abortions through December 2020. Here's what it looks like.