search

christinakent / lotteryBabies

0 stars 0 forks source link

Analysis: Babies over time for lottery winners, with zeros and interpolation #2

Closed christinakent closed 4 years ago

christinakent commented 4 years ago

I edited the dofile so that now it is a full time series, so it also includes months with zero births, set to zero in those months. I also interpolate average age of parents and average number of siblings in observations with no data.

christinakent commented 4 years ago

Results after filling in data

Results look mostly the same

image image image image
christinakent commented 4 years ago

A thought: might be good to do the binscatters with the RD at 9 months, since that is when we would expect the effects to start kicking in (assuming the changes are not due to migration)

christinakent commented 4 years ago

Restricting to towns with no months of zero births, and using logBirths

175 winning towns have nonzero births in every month 60 months before and 60 months after a lottery win.

Here is the binscatter when we plot log births for only these towns. Effect is less drastic/more noisy but still there.

image