Background:
In December 2020, Author Alison Greene published an article in DCReports.org regarding Mitch McConnell re-election win in Kentucky and some questions about voting machine performance in Kentucky (Why The Numbers Behind Mitch McConnell’s Re-Election Don’t Add Up. How Does an 18% Approval Rating Result in a 58% Win? here — www.dcreport.org/...).
Ms. Greene goes on to report on a pattern of unexpected voting results that she links specifically to the use of Election Systems and Software (ES&S) voting machines.
The idea that electronic voting machines are not accurately tabulating votes is a heated topic of debate right now, and has far-reaching significance for American democracy. These claims should not be made lightly, and when made, should be given careful and thoughtful scrutiny. Because Ms. Greene’s reporting relies in large part on data that is publicly available — county-level Kentucky election results — I thought it should be possible to review the data and see if I can at least confirm Ms. Greene’s reported numbers. And maybe dig around some more in those numbers to see what else I might find. Along the way, I want to include the source documents I used so that interested readers can know where the data is coming from, judge the quality of that data, and if interested, check it for themselves.
Data:
In 2020, Breathitt county reported 11,475 registered voters — 9,484 registered Dems — 1,596 registered Republicans (elect.ky.gov/...). In this county, there are six Dem voters for every one Repub voter. For the 2020 election, the county reported voter turn-out of 43.6% (elect.ky.gov/...). This was apparently on the low side of turn-out of all Kentucky counties (www.wkyt.com/...).
Breathitt county uses ES&S voting machines (elect.ky.gov/...)
There were 1,206 straight Repub party votes; 535 straight Dem party votes.
Pres — Trump = 4,265 votes; Biden = 1,301 votes.
Senate — M McConell (R) = 3,738 votes; McGrath (D) 1,652 votes.
Rep. — H Rogers (R) = 4,428 votes; M Best (D) = 1,145
State Rep — B Wesley (R) = 3,022 votes; B Clemons-Combs (D) = 2,542 votes
State SC Justice — R Conley (R) = 2,248 votes; C Harris (D) = 2,413 votes
Source for election result numbers — elect.ky.gov/...
So I can confirm that Ms. Greene accurately reported the voting tallies in Breathitt county. I can also confirm the anomalous nature of the results as Ms. Greene noted: in a county with six Dem voters to every one Repub voter, we would not expect to see Trump get at least three times as many votes as Biden, nor would we expect McConnell to receive twice as many votes as McGrath. Indeed, in this heavily Dem-majority county, the only Dem candidate to poll better than his/her Repub rival was the candidate for SC justice.
Is this evidence that ES&S voting machines are not accurately tabulating votes? No — not in and of itself.
How would we know if ES&S voting machines are incorrectly counting votes? The best way to do this is to count the votes using a “gold standard” (a method proved and trusted to be highly accurate) and the ES&S voting machines, and then compare the results to see how close to the “gold standard” the ES&S voting machines are. This best practice is not available to us after the voting is done. Perhaps another (and admittedly inferior) way to check is to compare results in Breathitt county with another county in the same election that uses a different type of voting machines. So I looked for another county in KY where Dem voters vastly outnumber Repub voters, AND that did not use ES&S voting machines.
Counties that have > 2x more Dems than Repubs and voting machines used in those counties
Bath — 6,682 D — 2,552 R — Hart InterCivic eScan
Breathitt — 9,484 D — 1,596 R — ES&S Invotronic
Carroll — 5,389 D — 2,233 R — Hart InterCivic eScan
Elliot — 4,245 D — 901 R — ES&S Invotronic
Floyd — 22,171 R — 6,358 R — ES&S Invotronic
Franklin — 25,670 D — 11,617 R — Hart InterCivic eScan
Fulton — 2,659 D — 1,467 R — Hart InterCivic eScan
Knott — 8,886 D — 1,936 R — ES&S Invotronic
Marion — 9,274 D — 3,894 R — Hart InterCivic eScan
Menifee — 3,633 D — 1,719 R — ES&S Invotronic
Morgan — 6,807 D — 2,171 R — ES&S Invotronic
Nicholas — 3,796 D — 1,495 R — Hart InterCivic eScan
Perry — 13,163 D — 6,561 R — Hart InterCivic eScan
Robertson — 1,240 D — 476 R — Hart InterCivic eScan
Union — 6,845 D — 3,257 R — Hart InterCivic eScan
Wolfe — 4,528 D — 949 R — ES&S Invotronic
(Source for voting machine info — elect.ky.gov/...)(Source for party registrations by county — elect.ky.gov/...)
From this list, let us choose Marion county as a comparator: it has a large majority of D voters, and it uses Hart voting machines
In 2020, Marion county reported 13,871 total registered voters — 9,274 registered Dem voters, and 3,894 registered Repub voters (elect.ky.gov/...). In Marion county, there are 2.3 Dem voters for every 1 Repub voter. For the 2020 election, Marion county reported a county-wide voter turn-out of 64.9% (elect.ky.gov/...). This was apparently on the higher side of turn-out of all Kentucky counties (www.wkyt.com/...).
There were 2569 straight-line Repub votes; 1504 straight-line Dem votes
Pres — Trump = 6,113 votes; Biden = 2,722.
Senator — M McConnell = 6,274 votes; A McGrath = 3,144 votes.
US Rep. — J Comer (R) = 5,885 votes; J Rhodes (D) = 2,719 votes
State Rep. — B Reed = 4,034 votes; T Mills (D) = 4,825 votes
Source for election result numbers — elect.ky.gov/…
As with Breathitt county, we see conservative candidates doing much better than county voter demographics would suggest. If all registered Republicans made it to the polls and voted for Trump, that would mean 2219 Dem voters would need to switch parties to also cast a vote for Trump: if all registered Republicans made it to the polls and voted for McConnell, that would mean 2380 Dem voters would need to switch parties to also cast a vote for McConnell. It is hard for me to imagine that at least 1/3 of Trump’s votes and McConnell’s votes came from Dem voters. That just doesn’t seem likely to me among our divided electorate.
Conclusions:
In an attempt to put some verifiable facts around the reporting of Ms. Greene, I have examined voting results from two Kentucky counties.
1) I can confirm Ms. Greene’s reporting on vote totals in Breathitt county. Ms. Green’s reporting in this regard is accurate.
2) I join Ms. Greene in concluding that the voting results in Breathitt county are unexpected and unusual in this heavily Dem-majority county. This unexpected and unusual result is mirrored in the heavily Dem-majority Marion county.
3) Given that we see unexpected and unusual voting results in two different Dem-majority counties wherein two different types of voting machines were used, I did not find evidence that supports the idea that ES&S voting machines cause the anomalous voting results in Breathitt county.
In her reporting, Ms. Greene identifies three Dem-majority counties in Kentucky (Breathitt, Elliot, and Wolfe) where Repub candidates did notably better than expected given the make-up of voters in that county. All three of those counties use ES&S voting machines. We now see that there is at least one more Dem-majority county in Kentucky (Marion) where Repub candidates did very well. Unlike the other three counties, Marion does not use the ES&S voting machines.
Other than voting machines, how else can we explain these better-than-expected voting results for Repubs in Dem-majority counties? One possibility is that our expectations are off: perhaps a lot of Kentucky Dems vote for Republicans. Another possibility is that Repub candidates do a better job of campaigning and advertising than do Dem candidates in this very red state: perhaps the Dem voters never hear about the Dem candidates and so don’t vote for them. Another possibility is we are seeing random chance variation. There are 120 counties in Kentucky and we have so far only surveyed four of them (3.3% of the total number of counties): what appears to be a trend is in fact simply random.
What should happen next? It makes sense to me to review election results in all the Dem-majority counties in Kentucky to see it the apparent trend holds or disappears. It would likewise be a good idea to review results in a few Repub-majority counties as well to see if those counties have expected or unexpected voting results. It seems to me that the Democratic leadership in Breathitt county or perhaps Kentucky itself should be notified of these anomalous results in Dem-heavy regions. I myself wonder how the Democratic party in Kentucky would explain these voting results — they might have a reasonable explanation that is not apparent to me. (Full disclosure: I am not a member of the Democratic party — for registration purposes I am “unaffiliated” — and I do not live or vote in Kentucky.)
