Evan McMullin wins UtahDonald Trump/Mike Pence 45.94% 298 EVs
Hillary Clinton/Tim Kaine: 48.17% 227 EVs
Evan McMullin/Mindy Finn: 0.72% 6 EVs
Gary Johnson/Bill Weld: 3.27% 0 EVs
Jill Stein/Ajamu Baraka: 1.07% 0 EVs
Others: 0.84% 0 EVs
Methods: Here is my McMullin wins Utah map. I wanted him to win by about a percent (in this case, 0.98%). In this scenario, McMullin makes it on the ballot in Wyoming, Nevada, and Arizona in addition to the states that he was in in real life. I figure that adding him in other states isn’t worth the effort due to the small numbers of Mormons in other states. A while back, I found that the correlation of Mormons times the % of people with a college degree in the county tracked McMullin’s results pretty closely (r^2 equals about 0.78). I used that model times 1.6446715 to get the expected percent that McMullin received in a county in Nevada, Wyoming, and Arizona. For Idaho and Utah, I used an average of the model (Mormon * College) and the actual results and then multiplied by that 1.64 number to get the expected percentage by county. I picked this number (1.6446715) because it approximated the amount of votes that McMullin needed to gain in order to win Utah by about a percent.
Once I found how many votes I expected McMullin to get, I subtracted those from the other candidates (85% from Trump, 10% from Johnson, and 5% from Clinton). In the case of Nevada, which has a None of the Above option, I guesstimated 40% of those would’ve voted McMullin and prioritized using those votes for McMullin’s total. In Nevada, I subtracted the difference (modeled-40% of NotA) and then subtracted the result from Trump, Johnson, and Clinton in the proportions described earlier. For Wyoming, I used a similar method, but I assumed that 50% of the unlisted write-ins were for McMullin. [Side note: One of my bigger small regrets is that I didn’t fork out $100 or so to the Wyoming SoS right after the election for them to count the write-ins for McMullin. Oh well.] Arizona was cool in that they listed McMullin write-ins for all counties, so I took those out of the number of votes that I needed to take away from other candidates.
To my friends on Atlas, I apologize for the flipped colors. Blame Wikipedia. Speaking of which, is there a good county template to use? There’s some artifacts on the current map due to the way the colors changed slightly around the edges of the counties; these are especially noticeable on the counties that flipped.
Analysis: McMullin obviously does a lot better in Utah (35.05%) and Idaho (10.44%), where he manages to win some counties. I am actually kind of surprised that McMullin only won 4 counties in Utah. Granted, 3 of them are in the top 6 by population. I was slightly disappointed that he only won a single county in Idaho, but it’s better than nothing (he didn’t even win any precincts there in the real results).
Looking at Nevada, he got 2.06%, which wasn’t enough to do more than shift some Trump 70% counties down to the 60% range. Notably, McMullin got 12.85% in Lincoln County and 5.09% in White Pine County.
Wyoming doesn’t see that much of a change. The most notable change is that Johnson drops under 5%, coming in at 4.96%. McMullin’s 3.62% is concentrated in the counties with large Mormon populations (big surprise there). Only the most Mormon counties had any change (in this case, they all dropped from Trump 70% to Trump 60%). McMullin broke 5% in 5 counties and 10% in 3 of those counties.
Arizona also saw little real change. Only one county even changed percentage bins (Navajo County went from Trump 50% to Trump 40%). The biggest thing here is that the relatively narrow margin between Trump and Clinton shrinks from 3.50% to 2.40%. McMullin polled 2.04% in Arizona.
Overall, a whopping 7 counties flipped. Two from Trump to Clinton and five from Trump to McMullin.
Issues: My data was missing Mormon percentages for a few counties. I looked at surrounding counties and just kinda made a number that seemed consistent with neighboring counties. In Wayne County, UT, my model showed that Johnson would have received -5 votes, which just isn’t possible. I fudged the numbers and added in 3 votes each from Stein and Other. A few counties in Nevada had more votes from the 40% of NotA, so I went with the higher number, which resulted in McMullin getting about 36 extra votes than he would’ve from my base model (big whoop). I also lost a total of 7 votes due to rounding in Idaho that I didn’t really feel like dealing with finding.
It’s also comical how hard it is to get consistent numbers even on Wikipedia. I’m not sure that the numbers on their table add up to 100%. I also probably missed taking some votes off when I added them to the huge clunky national popular vote table and I may have screwed something up. Not sure and it’s just after midnight when I’m writing this after spending most of the evening on this. Some of the calculations for the nationwide popular vote may be slightly off because of those issues.
Data Sources: Slate's Mormon mapUnsure about college education for Utah and Idaho; I had that info a long time before and I can’t find where I got it from. The numbers are fairly close to the census bureau’s though.
https://www.census.gov/censusexplorer/censusexplorer.htmlFor election results, I used Wikipedia for Nevada, Arizona, and Utah because it was easy to copy/paste into excel, for Idaho and Wyoming I used their SoS websites, because Wiki doesn’t have it by county. I also used the AZ SoS for McMullin write ins. I used USElectionAtlas for the “other” votes in Arizona (non-McMullin write-ins).
https://sos.idaho.gov/elect/results/2016/General/president_by_county.htmlhttp://soswy.state.wy.us/Elections/Docs/2016/2016GeneralResults.aspxhttp://apps.azsos.gov/election/2016/Info/ElectionInformation.htm