cwbeijer@gmail.com / About / Archive / Other media
Did Russian intelligence inspire Trump to depress Clinton turnout in Michigan and Wisconsin? - 8/2/18
A reader has asked:
I agree with your latest blog post on the points it covers, but what of the allegation that Russia could have provided the Trump campaign with DNC analytics?
This theory was recently popularized by Rachel Maddow, who argues that it was this data which inspired the Trump campaign to hone in on Michigan and Wisconsin with "three major voter suppression operations" targeting Clinton voters. If this campaign were successful, one would expect to see a significant drop in her numbers. Instead, here's what they looked like in both states during the final two months:


In both states, Clinton's numbers were actually higher by the end of October than they were at the beginning. Insofar as there was any real movement, it came from Trump's last minute three-point surge in Michigan, which overwhelmed Clinton's trivial gains.

One can try to salvage Maddow's theory with a few tweaks, but none of them are very convincing. For example, one can imagine that, instead of telling Trump to suppress votes, the Kremlin advised him to win them - but this fares no better at explaining Wisconsin's flat lines. One can also imagine that the Russian strategy (whatever it was) had an effect on the polls, but that this effect was cancelled out or buried inside countervailing trends; perhaps Clinton was going to enjoy a last minute surge, for example, until Trump's voter suppression campaign - which not only nullified her surge, but also nullified any detectable change in the polling. One can also insist, correctly, that the polling we have for these states was clearly incorrect - and then imagine that in a world of accurate polling, we would have seen shifts in Clinton's numbers that directly coincide with Maddow's theory.

Regardless, it should be clear that the point in my previous post holds: the case that Russian intervention was decisive ultimately depends not on anything we can see in the data, but on completely unsubstantiated theories about what's going on inside of the data, buried beneath an massive avalanche of statistical noise, bad polling, underdetermination, and pure fantasy.