*This document was prepared some time ago, but could not find a home.*

Ever since I first saw the analysis by Evans, Smith and Shiva of vote leakage in the big Michigan counties, I have thought of it as the single most compelling evidence of electoral fraud, and one with much wider application that just Michigan. Michigan was selected because of the availability of straight part-ticket voting in that State, a bit like the above-the-line voting in the Senate here. That gives a demographic snapshot of the party vote on a precinct-by-precinct basis. (Polling booths are probably the nearest analogy to the precincts.) By comparing the specific Trump/Biden vote to the straight-ticket voting, a picture of the leakage to or from the party Presidential candidate can be constructed.

Here’s the result for Kent County, one of 83 in Michigan. It has 252 precincts. The analysts point out the completely unexpected (and inexplicable) slope of this data, with leakage from Trump steadily increasing –**as a percentage** – as the size of the precinct GOP population increases. Yes, the number can be expected to increase with GOP population size, but not the percentage.

The compelling nature of this demonstration was also recognised on the other side of the aisle, so the debunkers were called in.

Enter Matt Parker. Matt has a YouTube channel called Stand-up Maths. This channel has 792k subscribers, so Matt has quite an audience. And Matt is a mathematician, so he knows what he’s talking about. The Shiva *et al *video was streamed on the 11^{th} of November. On the 15^{th}, Matt the Mathematician posted a video called *Do these scatter plots reveal fraudulent vote-switching in Michigan?* Spoiler alert: No.

Parker introduces his video like so.

…the idea is, this plot should be straight; instead you’ve got this down slope, and that’s meant to represent a nett shift of votes that should have been for Donal Trump over to Biden. However, if you plot exactly the same data from the same precincts in the same County for Biden, which I did, you would expect to see those votes showing up. There should be an anomaly going the other way. If you do that plot, it looks *exactly …* *the *… ** same**. This cannot represent a nett movement of votes because both plots are

*the*…

*same*.

I just wanted to get that out very quickly for the people who have already seen Dr Shiva’s video, and may not watch much more of mine.

You are here –

Clearly, Matt the Mathematician doesn’t expect his viewers (702,058 views when last accessed) to be able to read a graph; or to exercise the faculty of reason, for that matter. You might conclude that Matt suffers from the same handicaps, but I’ll rush to his defence here and insist that he certainly knows how to read a graph and how to build a graph, and possesses highly developed rational faculties of the mathematical and self-promotional varieties.

Matt’s Biden X and Y axes are effectively the inverses of those Trump graph. The Y axis range is shifted up relative to the Trump X-range. The votes have to go somewhere. The X-axis is almost exactly the inverse of the Trump X-axis, as it runs from low Democrat voting (i.e. high Trump voting) percentage to high Democrat-voting (i.e low Trump-voting). This simple trick created superficially similar graphs, which was obviously the intention.

When this spurious setup is repaired, the result looks like this.

People will have to make their own assessment of the integrity of Matt Parker, but there is no denying that he has been influential in inducing many to disregard the initial analysis, as a scan through the 10,699 comments will show.

The original video analyses four of the largest counties in Michigan. The data for Kent I found in Matt Parker’s description of his video. For this, thank you Matt. It was a courtesy neglected by Dr Shiva. However, I have not been able to find the corresponding data for the other counties analysed – Macomb, Oakland and Wayne. Wayne, heavily Democratic, does not display the pattern obvious in Kent, Oakland and Macomb. If anyone can point me to this data I would be grateful.