But those look at effects between mothers. Estimating the effect within mothers by looking at the same mothers over time, short interpregnancy intervals don't seem bad🧵
On the other hand, long interpregnancy intervals do seem bad.
I think this one might be down to aging. The estimates are corrected for maternal age, but that doesn't seem like a correction that should work well within mothers.
Now we can also look at the relationships between maternal characteristics and short interpregnancy intervals.
As it turns out, it's fairly consistently more obese women who get pregnant quickly. That suggests selection matters!
The differences (between/within) in the estimated effects of a short interval were significant excepting neonatal ICU usage. For long intervals, there were no significant differences between/within.
For maternal characteristics, selection was indicated between/within. The design doesn't let us speak to causal effects on gestational diabetes and preeclampsia, but we know the baby doesn't just make people obese at the beginning of pregnancy, while obesity is a risk factor for the other two conditions.
In the public imagination, these things go hand-in-hand.
But the link between poverty and crime is much weaker than people might imagine. It might not even be causal.
A new lottery study shows us just that:
🧵
To understand the causes of crime, there are other things you need to understand first.
For example, you need to understand the roles of sex and age.
In the whole country the lottery study results came from, you get this result when you plot both variables.
The collapse in criminal offending from adolescence is the crux of the "age-crime curve". The gap between men and women that declines with age is another important part.
Unlike age and crime, income and crime are nonlinearly related: after a certain level, income barely matters.