Upfront disclaimer: Although my degree is in Economics and, in the past, I’ve had heavy involvement in statistical analysis and the like, my chops are very rusty as I haven’t truly picked up economic and statistical analysis in 25 years.
Having said that, there’s something very, very strange with the vote in Pennsylvania. And yes, I’m adding onto the other diaries urging Clinton to challenge the vote in the three states referenced in numerous articles — Wisconsin (filing deadline to challenge is this Friday), Pennsylvania (filing deadline to challenge is this Monday upcoming) and Michigan (filing deadline to challenge is next Wednesday 11/30).
So, referencing my upfront disclaimer, and admitting readily that much of what I’m about to present is open to interpretation, let me explain what I did.
On 11/10/2016, feeling despondent, I started assembling the results from Pennsylvania. This is the site I used as it is the tally of record, and I did my analysis on both the 2012 and 2016 election cycles by PA county. I have referenced this in more than a few comments to other diaries in the past 24 hours, and one reply challenged me that I should have compared 2016 to 2008. I thought about that, and stand by my decision to use 2012 instead of 2008. For one, the demographics have changed dramatically since 2008. For another, the “Obama Effect” on 2008 was massive, and the erosion thereto was evident in his own results in 2012 when compared to 2008. Therefore, I felt that 2012 was a better basis for comparison, and a more conservative one, frankly.
So, I basically went into the Obama-Romney-Johnson-Stein results for 2012 and manually entered them into a spreadsheet, and then did the same for the Clinton-Trump-Johnson-Stein results for 2016. You can find that spreadsheet here. Note that the filename is appended with “corrected” as I found one error where I transposed the Johnson/Stein results. From those 11/10/2016 vote totals, I added the following calculations:
- % of vote by candidate by PA county for both 2012 and 2016
- Turnout, 2016 minus 2012 (total, by PA county)
- Turnout, 2016 minus 2012 (by candidate, by PA county)
- Win/loss margin, 2016 over 2012 (by candidate by PA county)
What I was after here was some comparison as to differences between 2012 and 2016. Hope that makes sense.
The big thing I added to this analysis was the voting method used in each election by PA county, as recorded by Verified Voting’s “Verifier” tool. It’s really an amazing resource that let me choose the voting method by county in a given election. I added that column to both 2012 and 2016.
If you clicked the Verified Voting link above, you can see that PA is uses predominantly DRE (Direct Recording Electronic) voting machines that do not have verified voter paper audit trails (VVPATs). Four PA counties use a blend of paper ballots and DREs without VVPAT, and the handful remainder uses paper ballots.
Now — here is where my upfront disclaimer comes back into play. After 11/10/2016, reviewing all the results I input into the spreadsheet, I had a case of the “feels”. Something feels wrong about these results particularly when compared to 2012. But — being a loud and harsh critic of all the feels with not enough facts, I wanted to continue to try to support my feels with facts.
That’s what brought me to late last night. I thought to myself, “Self, you should revisit the PA results because they will have added vote as a result of absentee and provisional ballots.” So that’s what I did. PA does not have early voting nor no-excuse absentee voting, so those additional votes are likely to be counted after election day. The updated spreadsheet can be found here. Note that there are two tabs. The first, “Sheet 1”, are the pure updated results. “Sheet 2”, however, is my attempt at further analysis.
Basically, I took the vote totals from 11/10/2016 by candidate by county and moved them to Sheet 2. I then took the vote totals from 11/23/2016 by candidate by county and also recorded them on Sheet 2. From there, I calculated vote added by candidate by county between the two referenced dates. And here’s where my feels started aiming more towards facts, though I don’t think the job is done or necessarily dispositive at this point: I calculated the margin of total vote of D (Clinton) over R (Trump) on 11/10/2016, and then the same calculation on the total vote for the figures on 11/23/2016. Finally, I calculated the D over R margin only on the added vote.
Note: When reviewing these, you have to exclude Fayette County as PA took away just over 29K votes from Trump — that’s probably a whole separate subject to explore, because it doesn’t make a lot of sense. You can see the notation of a 29K undervote here. Further, the vote total for Fayette at the link just provided vs. the total on the PA SBE’s page don’t match — don’t know what to make of it yet.
So you can see on my last column on Sheet 2 — the vote difference calculated as D-R for just the added vote is dramatically different than the vote difference on the total vote. Again — I’m back in the feels trying to drive towards facts here — but this seems not insignificant to me. At all. Clinton’s losing margin particularly decreases dramatically when calculating ONLY vote added after election day. It would seem to me that absentee voting would be proportional within some reasonable deviation from the norm. These figures are not that.
So again — I realize this may not mean anything. There are outliers that argue both cases on either side of this question of whether or not the vote was tampered with. There are questions about whether there is enough added vote to draw a reasonable conclusion. There are still many counties that have no vote added to any candidate, meaning either they have yet to add that vote, or there is no vote to be added (which I find unlikely), or that the 11/10/2016 totals already included absentee and provisional votes. But my feels are angling more towards the facts.
I wanted to make this available to whomever to do whatever you like with it. To be clear — I’m not saying voting machines were hacked or were otherwise tampered with, but I’m not saying they weren’t, either. I’m merely saying that the results are weird enough to warrant a closer look. My strongest hope is that Clinton will make that request by the specified dates in the states highlighted at the start of this diary. When you consider the extraordinary lengths to which Russia went to interfere with the election via hacking, it’s not a far leap to have legitimate concern that they went farther than we realize.