What's driving patterns in social distancing as seen by lower open table reservations? Could be:
- conservative aversion to social distancing
- COVID-19 shows up first in coast
- city size
- conservative aversion to social distancing
- COVID-19 shows up first in coast
- city size
Here's an argument that social distancing has a media component:
And a rejoinder that it's about coastal cities being affected first
Let's look into this with city-level Open Table data
Let's look into this with city-level Open Table data
First specification: share of 2016 vote (0-1) that went to Trump instead of Clinton. The trend is the most important part; but does seem that Red states don't show as much of a decline. 
Next, we add in (log of) active COVID-19 cases. Probably some data errors here for sure. Seems like COVID cases and vote share matter.
Last, we also add a control for city size. The trend is super important; COVID case load super important too. City size sort of meh. Vote share is marginally significant.
To conclude: the trend in declining open table visits is pretty strong. In addition to that, exposure to COVID cases is driving things further. There may be an additional partisan dimension, though a little harder to parse out.
cc @paulgp @marthagimbel
cc @paulgp @marthagimbel
If my regression output is ugly, it's because I didn't have time to make them prettier
