Some more ideas based on infamous examples from other countries

[blajer]

https://projecteuclid.org/euclid.aoas/1458909907 Paper describing various statistical fingerprints for fraudulent elections. Focus and proof of applicability based on several Russian elections.

[1e100]

I've plotted turnout vs vote share for Milwaukee and Allegheny - the lower the turnout, the harder the precincts break for Biden. Don't know what that means, but doesn't seem very logical. Positive correlation would be easy to explain - voters in a heavily democratic county showed up en masse and voted for their candidate. But we're seeing the opposite here - there's negative correlation. That is, Biden vote is overwhelming where voters didn't show up quite as much, and in precincts with high turnout it looks like toss-up.

[pkit]

@1e100 what you plotted is irrelevant to what is discussed here.

[1e100]

Thank you for this valuable contribution, @pkit. 😂 This is one of the graphs used in the paper mentioned above.

[1e100]

BTW, what I plotted can be found here: https://github.com/cjph8914/2020_benfords/issues/17

[pkit]

@1e100 love you too! I'm just seeing too much noise in these issues.

[ghost]

Figure 5 in that paper would be easy to replicate

[1e100]

Fig 5 only helps if tallies are entirely falsified, as they often are in Russia. People who "count" votes there are too dumb to not use a multiple of 5%. :-)

[ghost]

Would be interesting to see figure 5, the diagram to the left that is in black. Just to see whether there could be an anomalous amount of extremely high turnouts.