I spent some time reading through this (managed to access via Sci-Hub) but I lack the statistical background to fully understand their Regression Discontinuity

I spent some time reading through this (managed to access via Sci-Hub) but I lack the statistical background to fully understand their Regression Discontinuity Design. Some things I find interesting: 1. The study seems fairly large, with 444 children included in the IQ comparison. 2. Taking the claim of 0.5% in IQ per month in daycare at face value, and the included age range of 4 months to 36 months, you have an expected drop of 16%. They include daycare from ages 4 months to 36 months, yielding an expected drop of 16% for a child who spends the full time in daycare. That's more than a full standard deviation in IQ and equivalent to the drop expected for >250 micrograms/dL of lead in the bloodstream (for reference, lead poisoning is defined as 50 micrograms/dL). 3. For an effect this strong, I'd expect you could simply scatter plot "months in daycare" as one axis and "IQ" as the other and see the effect. I don't see such a chart. 4. The ".5%" per month claim seems to be derived from a single drop associated with acceptance into the child's preferred daycare program. There is a 3% drop and also a ~6mo increase in typical attendance as a result. They seem to have turned this into a claim of .5% per month without evidence of a continuous month-by-month effect. I am perhaps misunderstanding. 5. There also seems to be little to support the notion that less affluent parents will provide lower-quality home care, which is the entire hypothesis behind the measured effect. Frankly from 0-2 you aren't teaching kids physics. I'm not sure less affluent parents are really that much worse at being stay-at-home caregivers. You take care of the kid, feed them, stack some blocks and maybe read some books. This whole thing seems really iffy to me. The supposed change is massive. [1] https://tbiomed.biomedcentral.com/articles/10.1186/1742-4682-10-13 https://tbiomed.biomedcentral.com/articles/10.1186/1742-4682...