4 ms·

They claim the pandemic has dropped kids IQ scores by a full standard deviation? N=40? Hmmm. Nothing about that paper inspires confidence.

gwern 5y ago
> Leveraging a large on-going longitudinal study of child neurodevelopment, we examined general childhood cognitive scores in 2020 and 2021 vs. the preceding decade, 2011-2019. We find that children born during the pandemic have significantly reduced verbal, motor, and overall cognitive performance compared to children born pre-pandemic. Moreover, we find that males and children in lower socioeconomic families have been most affected. Blatant, blatant, family-level confounding is driving this. Having kids dropped like a stone in covid, because who would be so irresponsible as to have a child during a pandemic? During lockdown? And with global warming the way it is? (The opening of _Idiocracy_ is justly famous.) So, who then is having kids...? It's like maternal smoking. If you look at kids with mothers who smoke vs the kids with mothers who don't smoke, the latter are healthier and more functional; however, if you look at siblings whose mother smoked during one sibling's pregnancy but not the other's, they are often identical. That's because what you are measuring is the effect of 'having a mother who will smoke', not the effect of 'smoking'.
- heartbeats 5y ago
  That's true in the United States, but not in other countries: https://www.pnas.org/content/118/36/e2105709118 https://www.pnas.org/content/118/36/e2105709118 There were statistically insignificant increases in Slovenia, Norway, Denmark, South Korea, Finland, Sweden, Germany, Netherlands, and Switzerland. Going by this, I also think that you may have the correlation wrong. If the study is correct (more generous welfare systems => less decrease in birth rates), that seems like it would mean that people who can afford it will keep having kids. Granted, so will people who aren't in the planning business. But it's hardly cut-and-dry like that. (extra confounder: Northern Europe had less harsh lockdowns, maybe that plays a role?)
  - dragonwriter 5y ago
    > There were statistically insignificant increases “Statistically insignificant” means that you have data from a sample from which, at the desired level of confidence, you cannot be sure of any effect, or that there was not the reverse effect, in the universe from which the sample is drawn.
  - gwern 5y ago
    > Going by this, I also think that you may have the correlation wrong. If the study is correct (more generous welfare systems => less decrease in birth rates), that seems like it would mean that people who can afford it will keep having kids. It does not matter direction what the relationship is, only that it exists at the family level and so family-level correlations are non-causal for the effect of coronavirus as claimed by OP. We should be equally skeptical for the same reason if the result had instead came out as "kids during coronavirus have 10 higher IQ points because their parents played with them more thanks to remote working and subsidies"; there is no reason to believe that changes in childbearing during coronavirus has been, in any place, exogenous and randomized, and we know that pretty much all of these relationships are partially to completely confounded, thanks to studies which do include family controls (like all the Scandinavian population registry studies which exploit siblings). The burden of proof is on anyone who wants to argue that their correlation is the one special snowflake immune to this general trend, and should be taken at 100% face value.

gwern 5y ago

> Leveraging a large on-going longitudinal study of child neurodevelopment, we examined general childhood cognitive scores in 2020 and 2021 vs. the preceding decade, 2011-2019. We find that children born during the pandemic have significantly reduced verbal, motor, and overall cognitive performance compared to children born pre-pandemic. Moreover, we find that males and children in lower socioeconomic families have been most affected. Blatant, blatant, family-level confounding is driving this. Having kids dropped like a stone in covid, because who would be so irresponsible as to have a child during a pandemic? During lockdown? And with global warming the way it is? (The opening of _Idiocracy_ is justly famous.) So, who then is having kids...? It's like maternal smoking. If you look at kids with mothers who smoke vs the kids with mothers who don't smoke, the latter are healthier and more functional; however, if you look at siblings whose mother smoked during one sibling's pregnancy but not the other's, they are often identical. That's because what you are measuring is the effect of 'having a mother who will smoke', not the effect of 'smoking'.

heartbeats 5y ago
That's true in the United States, but not in other countries: https://www.pnas.org/content/118/36/e2105709118 https://www.pnas.org/content/118/36/e2105709118 There were statistically insignificant increases in Slovenia, Norway, Denmark, South Korea, Finland, Sweden, Germany, Netherlands, and Switzerland. Going by this, I also think that you may have the correlation wrong. If the study is correct (more generous welfare systems => less decrease in birth rates), that seems like it would mean that people who can afford it will keep having kids. Granted, so will people who aren't in the planning business. But it's hardly cut-and-dry like that. (extra confounder: Northern Europe had less harsh lockdowns, maybe that plays a role?)
- dragonwriter 5y ago
  > There were statistically insignificant increases “Statistically insignificant” means that you have data from a sample from which, at the desired level of confidence, you cannot be sure of any effect, or that there was not the reverse effect, in the universe from which the sample is drawn.
- gwern 5y ago
  > Going by this, I also think that you may have the correlation wrong. If the study is correct (more generous welfare systems => less decrease in birth rates), that seems like it would mean that people who can afford it will keep having kids. It does not matter direction what the relationship is, only that it exists at the family level and so family-level correlations are non-causal for the effect of coronavirus as claimed by OP. We should be equally skeptical for the same reason if the result had instead came out as "kids during coronavirus have 10 higher IQ points because their parents played with them more thanks to remote working and subsidies"; there is no reason to believe that changes in childbearing during coronavirus has been, in any place, exogenous and randomized, and we know that pretty much all of these relationships are partially to completely confounded, thanks to studies which do include family controls (like all the Scandinavian population registry studies which exploit siblings). The burden of proof is on anyone who wants to argue that their correlation is the one special snowflake immune to this general trend, and should be taken at 100% face value.