Thread by @DashiellShapiro

Did Democrats in blue counties steal the 2020 election?  

As a #YangGang member I wanted to see if #math might help to answer this question. So I looked at red and blue county data in MI and WI from 2020 vs. 2016. The numbers do not seem to support the fraud claim.

A Thread.

First, every credible allegation of voter fraud must be taken seriously.  Fraudulent votes disenfranchise all voters who have cast their ballots legitimately, and as a country we must vigorously guard against threats to the integrity of our system.

Nevertheless, the 2020 election data does not appear to support claims that election officials in big cities in the midwest have stolen this election for Biden.  Let's look at a couple states that have completed their initial canvass.  First, Michigan.

With more than 98% reporting, Biden leads Trump 50.5%/47.9% (+2.6) in Michigan.

In 2016, Trump beat Clinton 47.3%/47.0% (+0.3).

If Democrats in Detroit stole the 2020 election, we might see an unusual spike in Biden's numbers in MI's blue counties (but not in red counties).

But that is not the case.  

Instead, Biden has outperformed Clinton's numbers in MI's red counties by an even greater margin than in the blue counties!  It seems Biden is even more popular than Clinton on a relative basis in red counties (and less popular in the blue counties).

For example, in MI counties Biden won, his share of the heads-up vote was 61.52% (vs. 60.01% for Clinton).  In MI counties Biden lost, his share of the heads-up vote was 40.25% (vs. 38.47% for Clinton).  Here is the breakdown by county: docs.google.com/spreadsheets/d…

So Biden only improved on Clinton's share of the vote by 1.51pts in MI counties he won. Yet he improved on Clinton's share of the vote by 1.78pts in counties he lost. I don't see a spike in Biden's margins in the blue counties.

Let's take a look at another state, Wisconsin.

With 100% reporting, Biden leads Trump 49.4%/48.8% (+0.6) in Wisconsin.

In 2016, Trump beat Clinton 47.2%/46.5% (+0.7).

Let's look at the county-by-county data.

In Wisconsin counties Biden won, his share of the heads-up vote was 67.04% (vs. 66.15% for Clinton). In counties Biden lost, his share of the heads-up vote was 40.84% (vs. 39.83% for Clinton). Here is the Wisconsin breakdown by county: docs.google.com/spreadsheets/d…

Like in MI, Biden is more popular than Clinton in WI's red and blue counties alike. And Biden improved even more on Clinton's margins in WI's red counties than in blue counties! Not what we might expect if election officials in blue counties were indeed stealing the election.

It will be interesting to look at Pennsylvania, Arizona, Nevada, and Georgia once all those numbers are in. There may be larger increases in Biden margins in blue counties vs. red counties in some of these states. But such increases would not necessarily be due to fraud.

Biden performing better in blue vs. red counties may just be expected for a Democrat in a high turnout year. Yet Biden's failure to perform better in blue vs. red counties in MI and WI might undermine the case that fraud is responsible for his victories in these states.

For example, in GA, where Abrams led a big registration and turnout effort, it might be expected for Biden to outperform in blue counties. But it is worth asking whether any fraud alleged (1) can be quantified, and (2) if it could have affected the ultimate margin.

Let's assume for example that Biden wins PA by more than 1.5pts, but outperforms his red county margins in the blue counties by only 0.4pts. It would seem difficult to argue that fraud by election officials in blue counties is responsible given the margins.

Since any dispute may end up in court, the precise margin may matter a great deal. If there is not a sufficient spike in Biden support in blue counties vs. red counties, it may be quite difficult to persuade a court that election officials in blue counties stole the election.

And even if there is a significant spike in Biden support in blue counties vs. red counties in some states, that spike might need to be significantly greater than the margin of victory in order to persuade a court to intervene after-the-fact.

Ultimately, any credible fraud allegations must be taken seriously, and this mathematical analysis probably has a number of limitations. But so far the fraud thesis seems implausible. Curious if anyone has done similar analyses, or has issues with my county numbers? Thanks!