3 ms·
Judging by this Twitter feed and Stata output, it looks that we are looking at a classic case of "correlation does not imply causality". Clinton shift might be
by gerty 10y ago
Judging by this Twitter feed and Stata output, it looks that we are looking at a classic case of "correlation does not imply causality".
Clinton shift might be correlated with paper vs. electronic ballot but it looks like districts that opted for electronic ballot have different population groups compared to the ones with paper ballot. Once you control for that, e.g. race and college degree, paper/electronic ballot correlation disappears.
To be fair, data sample is rather small so small effects might be difficult to pick up.
- dillondoyle 10y agoAgreed. Many reports claim that e voting machines are in mostly poor voting locations, which could basically explain the 'discrepancy': http://www.reuters.com/article/us-usa-election-machines-idUSKCN11Q0EU http://www.reuters.com/article/us-usa-election-machines-idUS... But controlling for everything is almost impossible to be perfect: designated location vs county wide, income, race, education, Trump/Clinton modeled support. You'd have to pull actual voting returns by individual for those voting locations and states, and then have a good vote file. Then you can start to control for those variables and I would bet almost everything I have that doing so would reduce any irregularity these scholars are using as an argument that the election was somehow hacked.