Ohio


Today, Ohio is considered a "former swing state", and now is "safe Republican". This is quite funny.



What makes a swing state? The obvious characteristic is if the state is prone to switching which party it votes for in Presidential elections. If a state doesn’t switch in recent history, then it’s probably not a swing state. More technically is how close the Dem:GOP margins are. If a state, in recent history, has frequently had Dem:GOP margins within 5-6% or so, then it’s "competitive", and thus a likely "battleground" state, liable to "swing" and change which party it voted for since the last election. As computed in the accordion below, if a typical state has about 4.6m people (since this is the population of the two median states, that means means 23 states have more population, 23 have less), and making rough national-level calculations about age, non-citizen immigrants, felonies, and state-level voter turnout, we find this is about 129k people in a "typical state". And finally, a swing state typically has a substantial amount of EVs in the Electoral College. This criteria is a bit softer (Nevada is considered a swing state today, with only 6 EVs; the minimum possible is 3), but important nonetheless.


How many votes is 6% in a typical state? Let’s look at two states with the median population (that is, if we sort states by their population, the two states which would be in the middle), Louisiana and Kentucky, both with approximately 4.6m people. Overall, about 78% of Americans are 18 years or older, and thus eligible to vote. Further, according to Pew, about 47.8m people in the US (in 2023) were foreign-born - 14.3% of the US population - and 23.4m of those (49%) were naturalized citizens. So 7.3% of the population are non-citizens, or by complement, 92.7% are citizens. Thus, we can roughly assume 92.7% of the 78% voting-age-eligible population are eligible to vote - that is, 72.3% of the population. There’s another twist, because felons’ voting rights vary state-to-state. According to the Prison Policy Initiative in 2023 (date found by looking at snapshots in the Wayback Machine), about 19m in the US (about 5.7% of the 2023 population, estimated at 334.9m) have ever been convicted of a felony. It seems in most states felons lose their vote eligiblity only while incarcerated/have-to-finish-sentence (the exceptions being AZ, WY, NE, IA, KY, VA, TN, MS, AL, FL, DE; 6/11 of these south of the Mason-Dixon line). Overall, I’m not an expert, but the vast majority of those currently in prison have a felony conviction, whereas those in jail are either not-yet-convicted, or have a misdemeanor conviction. According to the Prison Policy Initiative, there are 1.9m people incarcerated in the US, and 550k are in local jail, and thus 1.35m in prison or federal jail (they don’t disaggregate "Federal Prisons & Jails" (208k people), so probably slightly over-estimating the prison population (as opposed to the jail+prison system)). Thus, about 0.4% of the US population is in prison; assuming most of them have felony convictions, that means about 72.3% - 0.4% = 71.9% of the population is eligible to vote.


Since most states only rescind the voting eligibity for felons while in prison or serving their sentence, let’s assume our "typical state" is such a one (We are using Kentucky population here for its median (~representative) state population, and while its eligibity restrictions continue beyond incarceration/sentence-served, we aren’t actually looking at Kentucky specifically, just using it to get an idea of the typical state population). Thus, about 0.719*4.6m = 3.31m in our "typical state" are eligible to vote. In terms of state-by-state voter turnout, per University of Florida Election Lab, most states in 2016 had turnout between 60-70%. So let’s say, roughly, a typical state has 65% voter turnout (the national turnout value will differ, because different states have different populations). In other words, about 0.65*3.31m = 2.15m people in our typical state are expected to vote in a given year. If our "swing margin" is 6%, that means the difference between the vote share of the Democrats and GOP is 6%. So we can just compute how many votes that is: 0.06*2.15 = 129k. Thus, in a "typical state", if it’s Dem:GOP margin is within 6% somewhere betwen 0 to 129k voters need to "change their minds" (this is ignoring the fact that people die, non-voters decide to vote, young people age-in to eligibility, and so on; but let’s keep it simple).


Notably, the plurality of stats (n = 24) have a population between 0 and 4.9m; the mode is roughly 1.56m. In this case, 1.12m would be eligible to vote, 728k would vote, and the 6% would be 44k votes. However, such a low-population state would probably have about 4 EVs, and thus probably not a swing state though.


If it’s before Nov 5th 2024 then, let’s say our assessment of the "battleground states" should look at states which have flipped parties within the past four elections (2004 → 2008, 2008 → 2012, 2012 → 2016, 2016 → 2020). These would be:


  • Indiana - flipped twice (2008, 2012)
  • Colorado - flipped once (2008)
  • North Carolina - flipped twice (2008, 2012)
  • New Mexico - flipped once (2008)
  • Michigan - flipped twice (2016, 2020)
  • Wisconsin - flipped twice (2016, 2020)
  • Ohio - flipped twice (2008, 2016)
  • Pennsylvania - flipped twice (2016, 2020)
  • Iowa - flipped twice (2008, 2016)
  • Virginia - flipped once (2008)
  • Georgia - flipped once (2020)
  • Arizona - flipped once (2020)
  • Nevada - flipped once (2008)
  • Florida - flipped twice (2008, 2016)

Of these, actual margins are important. I’ll report these below as x/𝑥 (margins from [0,6%]), a/𝑎 (margins from (6%,12%]), b/𝑏 (margins from (12%,18%]), c/𝑐 (margins from (18%,24%]), and d/𝑑 (margins from (24%,30%]). The italic letter means that year the margin favored Democrats, and non-italic means favored the GOP. I’ll also capitalize the letter, if that year the state "swung" from one party to the other.


  • Indiana - 𝑋Acb
  • Colorado - 𝐴𝑥𝑥𝑏
  • North Carolina - 𝑋Xxx
  • New Mexico - 𝐵𝑎𝑎𝑎
  • Michigan - 𝑏𝑎X𝑋
  • Wisconsin - 𝑏𝑎X𝑋
  • Ohio - 𝑋𝑥Aa
  • Pennsylvania - 𝑎𝑥X𝑋
  • Iowa - 𝐴𝑥Aa
  • Virginia - 𝐴𝑥𝑥𝑎
  • Georgia - xax𝑋
  • Arizona - aax𝑋
  • Nevada - 𝐵𝑥𝑥𝑥
  • Florida - 𝑋𝑥Xx

So we can see that in at least one of 2016 or 2020, Indiana and Colorado is far from swing-margins; New Mexico, Ohio, Iowa, and Virginia were somewhat far. And North Carolina, Michigan, Wisconsin, Pennsylvania, Georgia, Arizona, Nevada, and Florida were within our "swing" margins. These are the states that were largely considered swing states this year, except Florida.


Yet there’s a lot more to elections than just "swing". If we look at the change in margins, 27 states saw the pattern →2008:Dem → 2012:GOP → 2012:GOP → 2020:Dem. Of these, the NE, IN, MT, and ND saw the biggest swing in 2008 of 18%-24%; in no year since was the swing as big. In CO, NC, NM, OR, VT, ID, DE, MI, WI, SD, the 2008 swing was 12%-18%; in no year since was the swing as big. In CT, NH, SC, MN, OH, PA, RI, ME, IA, WY, and MO, the 2008 swing was 6-12%; in MN, OH, PA, and RI, the 2016 swing was in the same range. And in AL and KY, the swing was 0-6%; in KY, the swing towards Trump was even larger, in the 6-12% range.


Four states saw the pattern → 2008:Dem → 2012:Dem → 2016:GOP → 2020:Dem. Of these, NY, NJ, MS saw a 2008 swing of 6-12%; only in MS was the 2016 in the same range. In Alaska, the 2008 swing was in the 0-6%, only outmatched by the swing in 2012, in the 6-12% range.


Seven states saw the pattern → 2008:Dem → 2012:GOP → 2016:Dem → 2020:Dem. Virginia saw the biggest swing towards Obama in 2008, in the 12-18% range. TX, GA, WA, and KS in 2008 were in the 6-12% range. Of these TX saw a later comparable swing, a 6-12% swing towards Clinton in 2016. KS saw a comparable swing towards Romney in 2012. MA, AZ, and KS saw a 0-6% swing towards 2008 Obama; MA saw a comparable swing towards Biden in 2020.


MD saw the pattern → 2008:Dem → 2012:Dem → 2016:Dem → 2020:Dem, with the 2008 swing to Obama in the 6-12% range comparable to the 2020 swing to Biden.


Three states saw the pattern → 2008:Dem → 2012:GOP → 2016:GOP → 2020:GOP. Hawaii saw a huge gain for Obama in 2008, in the 36-42% range. In NV and FL, no comparable swing happened since.


Three saw the pattern → 2008:Dem → 2012:GOP → 2016:Dem → 2020:GOP. Only in Utah was there a larger swing than the 2008 one, in 2012 (it was Mormon Romney running), and in 2016 in the 24-30% range away from Trump (I think this is due to the third party candidate from Utah).


Three states saw the pattern → 2008:GOP → 2012:GOP → 2016:GOP → 2020:Dem. All of these had a 2008 swing in the range of 0-6% to the GOP; only in West Virginia has the swing level exceeded that, in 2012 and 2016 (in the 12-18% range). One state saw the pattern → 2008:GOP → 2012:Dem → 2016:GOP → 2020:Dem, LA, and the swings have all been in the 0-6% range. One state saw the pattern → 2008:GOP → 2012:GOP → 2016:GOP → 2020:GOP, AR; it’s biggest swing was in 2008 in the 6-12% range, away from Obama.


The interesting cases are those where there were each year saw a swing. That would be CA, IL, UT, LA, NE, IN, MT, ND, CO, NC, NM, OR, VT, ID, DE, MI, WI, D, CT, NH, SC, MN, OH, PA, RI, ME, IA, WY, MO, AL, and KY. That’s 31 states. In fact, only two states saw gains for only one party in all four elections (MD and AR). Looking at our 2024 swing states, FL and NV saw DGGG, GA and AZ saw DGDD, and NC, MI, WI, PA, IA saw DGGD.


Considering the huge changes Obama could pull off in 2008 (and this ground generally eroding slow in 2012), it’s worth wondering how useful the "swing state" idea is. In NE, IN, MT, ND, CO, NC, NM, OR, VT, ID, DE, MI, WI, SD, VA, TX, GA, WA, KS, MD, HI, NV, CA, IL, and UT, he saw greater-than 12% increase in margins in favor of him. This translated to wins in states that voted Bush in 2004 - IN, CO, NC, NM, OH, IA, VA, NV, and FL. Certainly he was helped by the utter Bush failures 2005-2008, but we remember him not as "lesser of two evils", but a candidate people really believed in, due to his message about things like healthcare reform. Further, if the 2008 margin gain for Obama was applied to the 2020 margin in Trump-won states, the Dems would win:


  • TX (2004→2008: +11.16%; required: +5.58%)
  • IN (2004 → 2008: +21.71%; required: +16.07%)
  • NC (2004→2008: +12.76%; required: +1.35%)
  • IA (2004→2008: +10.2%; required: +8.2%)
  • FL (2004→2008: +7.83%; required: +3.36%)
  • MT (2004→2008: +18.24%; required: +16.37%)

OH is close, at -1.33% as a result (2004→2008: +6.7%; required: +8.03%).


Of course, there’s a lot of things different about 2004 → 2008, and 2020 → 2024. But this breakdown perhaps shows some of the issues in the "swing state" concept. And that is... if you tap into issues that voters actually care about, rather than simply fortifying the partisan lines, you can make enormous gains in "non swing states", while also winning all the "swing states", as Obama did in 2008. In fact, in multiple states (WA, NM, OR, TX, NH, MN, NV, ME, VA, GA, NC, VT, MD, HI, CA, DE, IL, AK, SC, UT, and ID), the vote substantially and near-permanently shifted more pro-Democrat in 2008 and most/all elections since (in AR, TN and LA, it did the opposite). Maybe if someone kept doing that, we wouldn’t have to worry about this swing state nonsense? And as far as "swing states" goes, it seems there’s something important about 2008 that put former GOP strongholds AZ, NV, GA, and NC in play as "swing states".


It went for Clinton over the Republican candidate in 1992 (40.18% vs 38.35%) and 1996 (47.38% vs 41.02%), but necessary context is that Ross Perot, who ran a right-wing anti-free-trade campaign, got 20.98% of the vote in 1992 (total other third party: 0.50%) and 10.66% in 1996 (total other third party: 0.95%). While these voters wouldn’t necessarily all break for the Republican (and thus, possibly being undecisive for the 1996 result), it is interesting that his campaign was specifically anti free trade. In 2000, it voted for GOP Bush Jr. over Gore (49.97% vs 46.46%; total third party: 3.58% (Nader: 2.51%)), in 2004 it again voted for GOP Bush Jr. over Kerry (50.81% vs 48.71%; total third party: 0.48%). But then in 2008, in the wake of the 2008 financial crisis, Ohio went for Obama over McCain (51.50% vs 46.91%; total third party: 1.82%), and in 2012 again went for Obama over Romney (50.58% vs 47.60%; total third party: 1.82%).


But then in 2016, it swung to Trump (Trump 51.31% vs Clinton 43.24%; total third party: 5.45% (Libertarian/Independent Johnson: 3.15%)). In 2020, it continued with Trump (Trump 53.27% vs Biden 45.24%; total third party: 1.49%), leaning towards him even more. This wasn’t the result of declining turnout either, as both candidates made gains, and overall actual turnout rose 3.5 pp. And then in 2024, it swung even more towards Trump (Trump 55.14% vs Harris 43.93%; total third party: 0.94%).


Some say Ohio "used to be a swing state", but no longer is. But if we look at other midwest "swing states" - Wisconsin, Michigan, and Pennsylvania - we see something interesting. In 1992, Michigan went (Dem:GOP:Perot) 43.77%: 36.38%: 19.30%, 1996:

Table SSTrend - Vote share of parties from 1992 → 2024. Dem:GOP:Perot:Nader:Libertarian:Green:OtherSum
State 1992 1996 2000 2004 2008 2012 2016 2020 2024
Michigan (7.39) 43.77: 36.38: 19.30:x: 0.24: x: 0.31 (13.21) 51.69: 38.48: 8.75: 0.72: 0.06: 0.29 (5.2) 51.3: 46.1: 0.1‡: 2.0*: 0.4:*:0.2 (3.4) 51.2: 47.8:x: 0.5: 0.2: 0.1: 0.1 (16.44) 57.33: 40.89:x: 0.66: 0.47: 0.18: 0.46 (9.46) 54.04: 44.58:x:x: 0.16: 0.46: 0.45 (-0.23) 47.27: 47.50:x:x: 3.59: 1.07: 0.56 (2.78) 50.62: 47.84:x:x: 1.09: 0.25: 0.2 (-1.42) 48.31: 49.73:x:x: 0.40: 0.79: 0.77
Ohio (1.83) 40.18: 38.35: 20.98:x:x:x: 0.5 (6.36) 47.38: 41.02: 10.66: 0.07*:x:*: 0.88 (-3.51) 46.46: 49.97: 0.57‡: 2.51*: 0.29:x: 0.21 (-2.1) 48.71: 50.81:x:x: x:x: 0.48 (4.59) 51.50: 46.91:x: 0.74: 0.15: 0.35: 0.58 (2.98) 50.58: 47.60:x:x: 0.89: 0.33: 0.42 (-8.1) 43.24: 51.34:x:x: 3.15: 0.84: 1.46 (-8.03) 45.24: 53.27:x:x: 1.14: 0.32: 0.03 (-11.21) 43.93: 55.14:x:x: 0.49:NA†:0.45
Pennsylvania (9.03) 45.15: 36.12: 18.20:x: 0.43:x: 0.09 (9.2) 49.17: 39.97: 9.56:x: 0.62:x: 0.56 (4.17) 50.60: 46.43: 0.33‡: 2.10*: 0.23:*: 0.31 (2.5) 50.92: 48.42:x: 0.05: 0.37: 0.11: 0.13 (10.32) 54.47: 44.15:x: 0.71: 0.33: 0.0: 0.33 (5.38) 51.97: 46.59:x:x: 0.87: 0.37: 0.20 (-0.72) 47.46: 48.18:x:x: 2.38: 0.81: 0.55 (1.16) 49.85: 48.69:x:x: 1.14: 0.02: 0.30 (-1.71) 48.66: 50.37:x:x: 0.47: 0.49:x
Wisconsin (4.35) 41.13: 36.78: 21.51:x: 0.11: 0.47 (10.33) 48.81: 38.48: 10.35: 1.31: 0.36:x: 0.69 (0.22) 47.83: 47.61: 0.44‡: 3.62*: 0.26:*:x (0.4) 49.7: 49.3:x: 0.6: 0.2: 0.1: 0.1 (13.91) 56.22: 42.31:x: 0.59: 0.30: 0.14: 0.45 (6.94) 52.83: 45.89:x:x: 0.67: 0.25: 0.37 (-0.77) 46.45: 47.22:x:x: 3.58: 1.04: 1.7 (0.63) 49.45: 48.82:x:x: 1.17: 0.03: 0.33 (-0.86) 48.74: 49.60:x:x: 0.31: 0.36: 0.8
United States (5.56) 43.01: 37.45: 18.91:x: 0.28: 0.34 (8.53) 49.24: 40.71: 8.40: 0.71*: 0.50:*: 0.43 (0.52) 48.38: 47.86: 0.43‡: 2.74*: 0.36:*: 0.22 (-2.46) 48.27: 50.73:x: 0.38: 0.32: 0.10: 0.20 (7.28) 52.93: 45.65:x: 0.56: 0.40: 0.12: 0.34 (3.86) 51.06: 47.20:x:x: 0.99: 0.36: 0.39 (2.09) 48.18: 46.09:x:x: 3.28: 1.07: 1.38 (4.46) 51.31: 46.85:x:x: 1.18: 0.26: 0.41 (-1.47) 48.27: 49.74:x:x: 0.42: 0.55: 1.03

* This year, Nader was the Green Party candidate. So I didn’t put anything in the "Green Party" section, and only in the Nader section.

† Wiki says "votes not counted"

‡ This year, Pat Buchanan ran for the "Reform" party, which Perot ran under in 1996.


Table SSData - Data on the midwest swing states, circa 2018-2022
State Rural % Suburban % Urban % W:B:L:A:AI Median Income Poverty Unemployment 2019
Michigan 18.5% 58.2% 23.3% 73.7: 14.1: 6.0: 3.6: 0.8 $68,505 13.5% 3.7%
Ohio 19.7% 51.6% 28.7% 76.7: 13.4: 4.8: 2.8: 0.3 $66,990 13.3% 4.3%
Pennsylvania 14.3% 64.2% 21.5% 74.1: 12.3: 8.9: 4.2: 0.5 $73,170 12.0% 4.4%
Wisconsin 26.4% 58.6% 15.1% 79.5: 6.6: 8.1: 3.3: 1.2 $72,458 10.7% 3.2%

While clearly Ohio leans more Republican in general, the 1990s margins for the Democrats were probably in large part due to Perot. In 2000 and 2004, the margins became much less favorable for them. But under Obama, these states voted Democrat with strong margins (and not because of a strong right-leaning third party). But in 2016, this flips: not only do these states return to the lower Democrat turnout, they swing for Trump. In 2020, Biden barely ekes out a victory in all but Ohio, but nothing like the margins even in 2000 and 2004. And then in 2024, all of them swing towards Trump even more than in 2016.


Perhaps it’s correct to say "Ohio is no longer a swing state". But the fact that the overall trend among these states is the same (strong 90s performance likely due to Perot), weaker Dem turnout in the 2000 and 2004 elections, big turnout for Obama, and a strengthening of the Trump vote, indicates a deeper problem.


Ohio, being the most GOP-leaning state - yet with one of the most left-wing Senators in the country, Sherrod Brown, is an interesting case study here.


So far, we’ve looked at overall turnout. But what about party turnout? Let’s look at Ohio, it’s pretty interesting.


For this, I will compute "actual {Dem/GOP} turnout {2020/2024}". I compute this as the total votes for either party in a given year, divided by the total voter eligible population (VEP). Why this calculation? Suppose in year 0, the Democrats win 50% of the vote and the GOP 50% of the vote, and turnout was an incredible 100% (for simplicity, I won’t consider third parties in this mini-example). Then say in year 4, once again the Dems and GOP go 50:50, but turnout falls to 70%. It might appear the state is just as tight, but actually it’s just by coincidence that both parties lost an equivalent number of votes. And it’s this actual gain or loss of votes which is interesting. First, the overall results for the county.


Table C.I - Correlation of county income metrics with change in county turnout, for several states
Item Type 2020 TO 2024 TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini) %Δ turnout†
Ohio All 64.5% 63.1% 0.2842 -0.3426 -0.3310 -0.4792 -1.4%
Rural 62.4% 62.0% 0.4384 -0.2987 -0.3473 -0.2356 -0.3%
Rural+ Suburban 64.9% 64.1% 0.3101 -0.3283 -0.3353 -0.3746 -0.8%
Suburban 65.8% 64.9% 0.3860 -0.4460 -0.4457 -0.4541 -0.9%
Urban+ Suburban 65.0% 63.3% 0.3822 -0.4908 -0.4742 -0.6798 -1.7%
Urban 63.5% 60.4% 0.9268 -0.9998 -0.9976 -0.2929 -3.1%

Table C.II - Correlation of county racial composition with change in county turnout, for several states
Item Type 2020 TO 2024 TO W:B:L:A:AI Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN) %Δ turnout†
Ohio All 64.5% 63.1% 0.4201 -0.4997 0.0206 -0.1456 -0.0285 -1.4%
Rural 62.4% 62.0% 92.8: 2.1: 3.0: 0.6: 0.1 -0.1401 -0.0428 0.2211 0.0994 0.0462 -0.3%
Rural+ Suburban 64.9% 64.1% 85.6: 6.8: 3.6: 1.8: 0.1 0.2490 -0.3628 0.1010 -0.0142 -0.0038 -0.8%
Suburban 65.8% 64.9% 82.8: 8.7: 3.8: 2.2: 0.1 0.4530 -0.5526 -0.0519 0.0285 -0.1272 -0.9%
Urban+ Suburban 65.0% 63.3% 75.3: 14.9: 4.5: 2.9: 0.2 0.6234 -0.6827 -0.1778 -0.1296 -0.1871 -1.7%
Urban 63.5% 60.4% 62.0: 25.9: 5.6: 4.0: 0.2 0.8819 -0.8595 -0.7360 0.2745 -0.6123 -3.1%

How about per party? Here are the results.


Table OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 20 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.34286135 0.34603252 0.003171170.08180942 0.06474629 0.03358156 -0.23044145
Rural (50) 0.44139369 0.44924858 0.0078549 0.22747511 -0.05599367 -0.12439025 -0.09824478
Rural+ Suburban 0.39200607 0.39711255 0.005106470.08064685 0.08902347 0.04507984 -0.14253889
Suburban (35) 0.37289054 0.37718262 0.004292080.2601464 0.26944116 0.24119407 -0.16940764
Suburban+ Urban 0.31849272 0.32068609 0.002193370.21755156 0.18242506 0.17029364 -0.32898418
Urban (3) 0.22221449 0.21919692 -0.003017570.04510121 0.31399833 0.26765612 -0.8038064

Table OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 20 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.34286135 0.34603252 0.003171170.21523286 -0.25373264 0.14827186 -0.27466178 0.13790771
Rural (50) 0.44139369 0.44924858 0.0078549 0.25278178 0.03993723 0.24617079 0.06987298 0.13725459
Rural+ Suburban 0.39200607 0.39711255 0.005106470.09861556 -0.14230139 0.19247288 -0.22591053 0.15097962
Suburban (35) 0.37289054 0.37718262 0.004292080.24663965 -0.19857508 0.12758518 -0.37054321 0.15339979
Suburban+ Urban 0.31849272 0.32068609 0.002193370.37496973 -0.35301812 0.04814166 -0.41037559 0.11144304
Urban (3) 0.22221449 0.21919692 -0.003017570.7384539 -0.19546516 0.88357098 0.81507944 0.94949987

Table OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 20 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.29116655 0.27569505 -0.015471490.50002835 -0.6017019 -0.54824892 -0.55240987
Rural (50) 0.1719355 0.16379758 -0.008137920.58494975 -0.54469499 -0.56954426 -0.37165022
Rural+ Suburban 0.24587506 0.23557857 -0.010296490.55170903 -0.61476583 -0.57642447 -0.47005712
Suburban (35) 0.27449345 0.26301813 -0.011475320.65704386 -0.77590287 -0.72879348 -0.61126774
Suburban+ Urban 0.3206543 0.30317336 -0.017480940.62376334 -0.7751904 -0.72175934 -0.67244971
Urban (3) 0.40235397 0.37530734 -0.027046630.42569407 -0.72030835 -0.68587462 0.43390362

Table OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 20 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.29116655 0.27569505 -0.015471490.47453318 -0.5625468 -0.11079675 0.01632532 -0.17567719
Rural (50) 0.1719355 0.16379758 -0.008137920.0531475 -0.15166045 0.15663524 0.1137194 -0.06936396
Rural+ Suburban 0.24587506 0.23557857 -0.010296490.30694729 -0.44601451 -0.03102266 0.1997225 -0.1560976
Suburban (35) 0.27449345 0.26301813 -0.011475320.40703571 -0.56739262 -0.18947009 0.30591551 -0.3009684
Suburban+ Urban 0.3206543 0.30317336 -0.017480940.57032364 -0.65902168 -0.29604467 0.10188936 -0.34834442
Urban (3) 0.40235397 0.37530734 -0.027046630.96761878 -0.28414406 -0.99999683 -0.45112099 -0.98630527


Just from party turnout, we see that the Dems, county-by-county, registered losses, and Trump saw a very marginal increase (EXPLAIN DIFFERENCES WITH ALL-STATE TABLE). Keep in mind: if the correlation is weak for change in turnout, it means that the group was neither "turned off" or "on" by the given party. If the correlation is significant (ie magnitude ≥ 0.30-ish), that means that a group was "turned off" or "on" by a given party.


For the GOP, the income trend is ambiguous. Certainly, the corr(Gini) remains negative, yet rarely a significant value (ie absolute-value(corr) ≥ 0.3-ish). We do see turnout for the party maintain a positive correlation with both income and poverty (except for Rural poverty), although the correlations are nearly all weak. At very least, it’s a mixed picture.


Interestingly, rural Ohio low-income voters saw a negative correlation, albeit weak, with GOP turnout. One might expect that the so-called "working class" Ohioan JD Vance would be a boon for low-income Ohio turnout for the Trump ticket. If all the propaganda about him is correct (despite how out of touch he is (here’s him attempting to... order donuts)), that’s what we would expect. Home state advantage, plus Mr Hillbilly Elegy on the job! But the results are pretty anemic.


The results for the Democrats are dire, however. We see moderate-to-strong negative correlations of Dem turnout changes from 2020→2024 with poverty and the Gini coefficient. The one variable in Ohio counties that correlated with increased Dem turnout was median income.


In terms of race, the GOP’s only significant trends were: worstening with black voters in suburban+urban, strong correlation with urban white voters (only three urban counties; and the trend weakens from suburban → urban+suburban for Latino, Asian and AIAN voters). There are positive but weak correlations with Latino population and generally a bad showing with Asian voters. The correlation with rural and suburban white voters is stronger than for Asian and Latino, but still weak.


Dems on the other hand, saw strong gains with suburban, and likely urban, white voters. They did horribly with black and Latino voters, and saw weak to moderate correlation with Asians.


The Sherrod Brown Factor


Image 2

Caption: Sherrod Brown (in blue jacket and mustard shirt) with American Federation of State, County and Municipal Employees (AFSCME) members. (source)



How does Sherrod Brown compare?


Image 2
Figurex 1

Figure OHP. Polling performance of candidates for (1) Senate and (2) President in Ohio, 2024. On slide (3), the Senatorial trends are compared with the Presidential trends. Around the Nov 5th vertical line, small colored dots indicate the actual vote share of candidates in their races. Dot color code, in order of vote share value from highest to least: Light Orange: Trump (55.14%); Dark Orange/brownish - Moreno (50.09%); Dark Purple - Brown (46.47%); Light Purple - Harris (43.93%) (I put the values highest to least so you can identify who the dots belong to, even if my color description doesn’t line up with your color perception; so the "top dot" is Trump, "bottom dot" is Harris). Note: I’ve edited these graphs so their x/y axes align, making them easier to visually compare. The data/plots were not themselves manipulated, however.


We should be very careful with polls (note for both GOP candidates in figure OHP, the polls underestimated their vote share by ~3% (outside of the given error interval even), though got the Democrat vote share right), as the sample they select tends not to represent actual voters (generally, under-sampling higher-income voters). Yet we can at least get an idea of trends.


The "cost" of this Senate race was was $483.4m, according to AdImpact, making it the most expensive "for any non-Presidential election on record". $251.3m of this was GOP, edging out the Democrats at $232.4m. For reference, nationally the presidential campaign in total cost $5.5b, with total spending at $14.8b (per PBS). The Ohio Senate race cost nearly 10% of the Presidential campaign itself, with the next closest race being Pennsylvania’s Senate seat, costing $121m less than Ohio, per ABC5 News 5 Cleveland. They also report on the confusion on a lack of a Ohio Senatorial debate, and who was responsible. For the record, Brown has participated in Senatorial debates in each of his elections. Notably, the overall spending in Ohio was $714m, per NPR; although the messaging for the Presidential and Senatorial races aren’t independent of each other, that means at most $230.6m was spent on the Presidential campaign here, less than half of the total for the Senatorial race.


AdImpact reports that while the Dems spent more early on, this shifted in September, when Republican groups (not directly linked with Bernie Moreno’s campaign) poured in money, outspending the Dems by around $10m a week throughout September, and maintaing this spending lead into late October; crypto was a big part of this surge, per Axios, though much of this was from untrackable "dark money" groups, per OpenSecrets (via WFIN) (see OpenSecrets’ PAC reporting here). Dems were outspent in virtually all radio and TV markets, especially in central and northern Ohio.


Per AdImpact, immigration was mentioned in 33% of all ads, but the Republicans focused on it more. The Dems mentioned it 19.3k times, and the GOP 54k times. Democratic messaging was more focused than the GOP on the topics of "Character" (39.7k, vs 10.9k), "Abortion" (27k vs 8.8k), "Jobs" (17.3k vs 13.4k), "Crime" (15.4k vs 13.8k), and "Illegal Drugs" (27.1k vs 0?). Other than Immigration, the GOP focused more on "Donald Trump" (27.5k vs 5.8k), "Inflation" (23.8k vs 4.8k), "Transgender Sports" (20.0k vs 6.8k), and "Joe Biden" (21.3k vs 2.9k). Messaging was overall negative, with the GOP ads 77% "negative", vs Democrats at 66%. Though Brown’s campaign had more money than Moreno’s, it was outside money which tipped spending in favor of the GOP here.


While I’m skeptical of the positive impact ads have (ie convincing a voter to vote for you), there’s a more subtle impact worth considering. Sherrod Brown has managed to keep hold in Ohio, and with substantial margins, with his labor-oriented, anti-free-trade brand of politics. Yet here "Jobs" were only 10.4% of Democratic messaging; "Inflation" was 2.9% - both together at 13.3%. Without a debate, and his core message drowned out by issues he’s weaker on (and the Democratic Party itself is evidently weaker on nationally), he was going into election day with a big handicap, even setting aside the issue of Harris’ campaign itself (advertising aside) dragging down turnout throughout the country (nearly every state but three to five saw a drop in turnout; and fewer saw a rise in Democrat turnout).


Per AdImpact, GOP total spending in the Ohio Senatorial race was $251.3m, the most for any national Senatorial race. Considering PBS reporting that the GOP spent $543m in Pennsylvania overall, and that AdImpact reports they spent $173.1m on the Pennsylvania Senatorial race, that indicates they spent $369.9m in the most important swing state. For "National" spending, the GOP spent $286m, per PBS. This means that the Ohio Senatorial race was 68% of the GOP Presidential budget in Pennsylvania, and 88% of their "National" spending. The third highest priority race in the country! The huge sums of money the GOP put into this race - more than any swing state Presidential campaign other than Pennsylvania (MI: $201.2m, WI: $178m, AZ at most $160m, NC at most $196m, Nevada at most $111m, Georgia at most $139m) - is indicative of the enormous bounty required to take down a labor-populist in a "deep red state", in pursuit of a Senate majority. Mind you, this "deep red state" quite recently swung consistently for Obama (in both 2008 and 2012), back when the Dems had the auspices of working-class outreach, a shattered illusion which has since been backfiring on them in the form of Trump.


Even with all of this, amidst a dwindling Democrat voter pool, he vastly outperformed Harris here, pulling votes off from Trump away from opponent Bernie Moreno (note that 99.1% of Ohioans who voted for a Presidential candidate, also voted for a Senatorial, with Moreno losing votes to both the Libertarian and Brown). Maybe there’s something to Brown’s politics that makes him so formidable? Perhaps an orientation to working-class needs, and constant effort to foster relations with organized labor, is a powerful boon. One wonders if there really is such a thing as a "deep GOP state", or if the Democrats just betrayed a huge swathe of voters, and left them behind.


With that basic summary, let’s get into the details. First, what was the change in turnout from 2018, Brown’s last election?


Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.64472171 0.63057404 -0.01414767 0.2842143 -0.34263724 -0.3310191 -0.47917912
Rural (50) .62363143 0.62046389 -0.00316754 0.43844055 -0.2986767 -0.34730813 -0.23559121
Rural+ Suburban .64876885 0.64119488 -0.00757397 0.31006914 -0.3283222 -0.33533927 -0.37463772
Suburban (35) .65849832 0.64911968 -0.00937864 0.38596247 -0.44595059 -0.44565683 -0.56415331
Suburban+ Urban .64993768 0.63305676 -0.01688091 0.38223971 -0.49083334 -0.47416283 -0.67982325
Urban (3) .63478627 0.60420168 -0.03058458 0.92681144 -0.99979096 -0.99762807 -0.29292169]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.64472171 0.63057404 -0.01414767 0.42014213 -0.4997177 0.02060952 -0.14562637 -0.02849184
Rural (50) .62363143 0.62046389 -0.00316754 -0.14014623 -0.04283768 0.22112433 0.09937891 0.04621907
Rural+ Suburban .64876885 0.64119488 -0.00757397 0.24896438 -0.36281322 0.10103737 -0.01423618 -0.00378476
Suburban (35) .65849832 0.64911968 -0.00937864 0.45303584 -0.55257658 -0.05193496 0.02852723 -0.12722751
Suburban+ Urban .64993768 0.63305676 -0.01688091 0.62339273 -0.68269561 -0.17777002 -0.12957008 -0.18710995
Urban (3) .63478627 0.60420168 -0.03058458 0.88189681 -0.8594617 -0.73604664 0.27451103 -0.61233218]

Table S.OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.34286135 0.34603252 0.00317117 -0.33997458 0.13418305 0.1641966 -0.58224085
Rural (50) .44139369 0.44924858 0.0078549 0.08907606 0.00212118 -0.10885797 -0.38184212
Rural+ Suburban .39200607 0.39711255 0.00510647 -0.37917927 0.21055207 0.21510162 -0.46259196
Suburban (35) .37289054 0.37718262 0.00429208 -0.4328562 0.29629465 0.33486533 -0.50818
Suburban+ Urban .31849272 0.32068609 0.00219337 -0.30416028 0.11035786 0.16653589 -0.67588148
Urban (3) .22221449 0.21919692 -0.00301757 -0.91296754 0.99844472 0.99457385 0.25896344]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.34286135 0.34603252 0.00317117 0.79498299 -0.73084618 -0.27412186 -0.73058916 -0.0550509
Rural (50) .44139369 0.44924858 0.0078549 0.44156621 -0.45027117 -0.14571343 -0.53082132 -0.03523846
Rural+ Suburban .39200607 0.39711255 0.00510647 0.72846125 -0.63635941 -0.21715296 -0.68669897 -0.03025512
Suburban (35) .37289054 0.37718262 0.00429208 0.77759126 -0.61946291 -0.29991947 -0.71674986 -0.17867
Suburban+ Urban .31849272 0.32068609 0.00219337 0.85100931 -0.75926381 -0.38196112 -0.73278163 -0.22556143
Urban (3) .22221449 0.21919692 -0.00301757 -0.89800006 0.8408672 0.75949978 -0.24037184 0.6398779 ]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.29116655 0.27569505 -0.01547149 0.68501944 -0.46970734 -0.52596463 0.21686283
Rural (50) .1719355 0.16379758 -0.00813792 0.30894152 -0.27674276 -0.26277726 0.02750052
Rural+ Suburban .24587506 0.23557857 -0.01029649 0.70718819 -0.51923068 -0.56102911 0.10507775
Suburban (35) .27449345 0.26301813 -0.01147532 0.80051575 -0.73690869 -0.76880214 -0.02154681
Suburban+ Urban .3206543 0.30317336 -0.01748094 0.75431308 -0.64615869 -0.69198688 0.13337993
Urban (3) .40235397 0.37530734 -0.02704663 0.87954267 -0.9913823 -0.9838759 -0.18527373]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.29116655 0.27569505 -0.01547149 -0.53663682 0.41753612 0.22664974 0.71883155 -0.04775023
Rural (50) .1719355 0.16379758 -0.00813792 -0.3169895 0.20931166 0.19267121 0.40538814 -0.04753441
Rural+ Suburban .24587506 0.23557857 -0.01029649 -0.50121762 0.33789477 0.19078302 0.69965754 -0.06487459
Suburban (35) .27449345 0.26301813 -0.01147532 -0.32434508 0.1032873 0.18674662 0.7129947 0.01466497
Suburban+ Urban .3206543 0.30317336 -0.01748094 -0.38420414 0.23782425 0.22721271 0.72113551 0.0421908
Urban (3) .40235397 0.37530734 -0.02704663 0.92866649 -0.79758744 -0.80646104 0.16637317 -0.69609062]

Table S.OH.Dem.H.Income - Correlation of county income metrics with county ΔActualDem, if using 2018 values of Brown, and 2024 values of Harris
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.29116655 0.27569505 -0.01547149 0.65264573 -0.44998775 -0.49820902 0.26259669
Rural (50) .1719355 0.16379758 -0.00813792 0.32443243 -0.30947251 -0.27540535 0.05384612
Rural+ Suburban .24587506 0.23557857 -0.01029649 0.68256852 -0.51088242 -0.54238978 0.13592746
Suburban (35) .27449345 0.26301813 -0.01147532 0.76231205 -0.68453235 -0.71781705 0.04208387
Suburban+ Urban .3206543 0.30317336 -0.01748094 0.69441754 -0.56535074 -0.61423552 0.22750576
Urban (3) .40235397 0.37530734 -0.02704663 0.86899696 -0.98830204 -0.97975774 -0.16387926]

Table S.OH.Dem.H.Racial - Correlation of county racial composition with county ΔActualDem, if using 2018 values of Brown, and 2024 values of Harris
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.29116655 0.27569505 -0.01547149 -0.5758657 0.45626567 0.27549256 0.73020049 -0.01348656
Rural (50) .1719355 0.16379758 -0.00813792 -0.29830594 0.14829382 0.24701791 0.42080854 -0.03155368
Rural+ Suburban .24587506 0.23557857 -0.01029649 -0.52147975 0.3523889 0.23787494 0.70407192 -0.03201069
Suburban (35) .27449345 0.26301813 -0.01147532 -0.3953259 0.17105328 0.23746815 0.73573083 0.08157799
Suburban+ Urban .3206543 0.30317336 -0.01748094 -0.47980397 0.33490469 0.2879974 0.75093882 0.11405979
Urban (3) .40235397 0.37530734 -0.02704663 0.93650612 -0.78429343 -0.8191175 0.14490982 -0.71152521]

This is a tricky comparison in the first place. Election dynamics in midterms are different than presidential elections; even with depressed turnout, presidential elections will usually register higher turnout than midterms. However, (1) people don’t all vote for the same party when they hit the ballot box (so the change in votes from {year} → 2024 is still informative), (2) of Ohioans who voted for a presidential candidate, 99.1% of them voted for a Senatorial candidate (51k short). GOP candidate Benny Moreno got only 89.9% of the total votes that Trump did (323k votes short); by contrast, Sherrod Brown got 104.6% of the votes that Harris did (117k more). Notably, the Libertarian Senatorial candidate, Don Kissick, got 477.1% of the vote that all third party presidential candidates got (155k more). Even assuming all the 50k who didn’t vote Senatorial were Trump voters, that means Moreno was 272k votes short, and Brown won 43% of that. Thus, even with drastically falling Democratic turnout in the 2024 election, Sherrod Brown was still much more competitive (especially considering there were nearly as many votes for Senate as for President).


However, we must be cautious comparing the correlations. If we plug in the values for Harris’s votes in 2024 and Brown’s for 2018, we find virtually the same picture as using Brown’s for both. We should be careful then, in drawing conclusions about the different correlation’s for the 2020 → 2024 Presidential trends, and the 2018 → 2024 Senatorial trends, from the data alone. This makes sense; while Brown did get 104.6% of Harris’s votes, that’s just 4.6% more. It’s largely the same voters for the Dem in 2024.


Yet even within this limited, more GOP-oriented voter pool, Brown was still able to pry off 2% of Trump’s voters. That’s not magnificent, but for an election, it’s pretty significant. It’s funny to think, Harris’s whole campaign was supposedly right-wing to "win over the moderate Republican". But Sherrod Brown - perhaps one of the most left-wing Senators at the moment, even if lame-duck, was able to outperform, even in a turnout pool that Harris’s bad campaign had managed to bleed 1.5 pp off from 2020 Democrat turnout, and pro-Trump turnout increased by 0.3 pp.


Here’s Brown’s interview with Politico. The interviewer seems confused that "working class" can be a reference to non-white straight people? He also seems confused that working class didn’t so much switch to Trump (setting aside those already pro-Trump), but didn’t turn out. Brown also doesn’t clarify this latter point.


How about for 2008 → 2012, 2016, 2020, 2024?


Here are the tables, I’ll give some comments after.


Tables: Trends for Presidential Elections, {2008, 2016} → {2012, 2016, 2020, 2024}


-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 08 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.63662541 0.63057404 -0.006051370.25414166 -0.26089786 -0.29370982 -0.64584725
Rural (50) 0.5948224 0.62046389 0.025641480.27117962 -0.10744124 -0.25909963 -0.41488514
Rural+ Suburban 0.62909 0.64119488 0.012104890.28191818 -0.24010095 -0.29736687 -0.57263408
Suburban (35) 0.64299549 0.64911968 0.006124180.43936179 -0.47291374 -0.47761978 -0.777033
Suburban+ Urban 0.64758173 0.63305676 -0.014524970.42877259 -0.51323782 -0.50157987 -0.81516575
Urban (3) 0.65594669 0.60420168 -0.051745010.88454599 -0.99271754 -0.98571972 -0.19569879]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 08 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.63662541 0.63057404 -0.006051370.65152319 -0.66714996 -0.32476078 -0.2719164 -0.23942917
Rural (50) 0.5948224 0.62046389 0.025641480.34992998 -0.33339761 -0.17547705 -0.4343418 -0.1461145
Rural+ Suburban 0.62909 0.64119488 0.012104890.56967806 -0.60563826 -0.27744824 -0.15197411 -0.23325198
Suburban (35) 0.64299549 0.64911968 0.006124180.73815954 -0.78996182 -0.39361813 -0.01694783 -0.42808402
Suburban+ Urban 0.64758173 0.63305676 -0.014524970.79279791 -0.81381294 -0.46047302 -0.16542594 -0.44365365
Urban (3) 0.65594669 0.60420168 -0.051745010.92467524 -0.8039481 -0.80013647 0.17683515 -0.68842715]

Table OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 08 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.29794104 0.34603252 0.04809148 -0.48822967 0.2930713 0.31757705 -0.47873031
Rural (50) 0.32635868 0.44924858 0.12288991 -0.2073154 0.25152934 0.13326372 -0.20961712
Rural+ Suburban 0.31919743 0.39711255 0.07791512 -0.53855815 0.38251708 0.38194725 -0.34255793
Suburban (35) 0.31629145 0.37718262 0.06089117 -0.54753377 0.43590058 0.46829309 -0.40256376
Suburban+ Urban 0.29049295 0.32068609 0.03019314 -0.43521515 0.26390923 0.31410544 -0.56786469
Urban (3) 0.24343834 0.21919692 -0.02424142 -0.94641734 0.99936364 0.99991853 0.34609505]

Table OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 08 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.29794104 0.34603252 0.048091480.75219211 -0.66211226 -0.2888995 -0.78356961 -0.06965053
Rural (50) 0.32635868 0.44924858 0.122889910.36289592 -0.26984795 -0.2149339 -0.61771454 -0.03741359
Rural+ Suburban 0.31919743 0.39711255 0.077915120.68551973 -0.55405611 -0.24136918 -0.7516623 -0.05127087
Suburban (35) 0.31629145 0.37718262 0.060891170.68731312 -0.50071037 -0.27630421 -0.75516365 -0.20436393
Suburban+ Urban 0.29049295 0.32068609 0.030193140.75896853 -0.64303908 -0.34658913 -0.77231709 -0.23673185
Urban (3) 0.24343834 0.21919692 -0.02424142-0.85406258 0.88678425 0.69691517 -0.32801895 0.56701919]

Table OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.32711674 0.27569505 -0.05142169 0.77930491 -0.5505495 -0.60063156 0.07238745
Rural (50) 0.25441566 0.16379758 -0.09061809 0.47932503 -0.37614537 -0.38560566 -0.14959708
Rural+ Suburban 0.29720169 0.23557857 -0.06162311 0.79873148 -0.59493068 -0.63215698 -0.0432311
Suburban (35) 0.31456389 0.26301813 -0.05154575 0.87238535 -0.78639785 -0.81681229 -0.14062344
Suburban+ Urban 0.34617125 0.30317336 -0.0429979 0.83366056 -0.70754873 -0.75022691 0.01396601
Urban (3) 0.4038208 0.37530734 -0.02851346 0.92810698 -0.9998558 -0.99786058 -0.29623271]

Table OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 08 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.32711674 0.27569505 -0.05142169-0.38109162 0.26449566 0.0652215 0.72806764 -0.11986137
Rural (50) 0.25441566 0.16379758 -0.09061809-0.03680417 -0.02601814 0.02642601 0.28110786 -0.12467044
Rural+ Suburban 0.29720169 0.23557857 -0.06162311-0.31899598 0.15217438 0.03120565 0.71839128 -0.13387975
Suburban (35) 0.31456389 0.26301813 -0.05154575-0.17770515 -0.04931559 -0.01111963 0.76406219 -0.10899959
Suburban+ Urban 0.34617125 0.30317336 -0.0429979 -0.2495419 0.10097769 0.02408284 0.75847863 -0.08885553
Urban (3) 0.4038208 0.37530734 -0.028513460.88025809 -0.86122773 -0.73369682 0.27784103 -0.60958927]

-----------------------------------------------


-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 08 A-Total TO 20 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.63662541 0.64472171 0.0080963 0.1862159 -0.16953612 -0.21412221 -0.57659766
Rural (50) 0.5948224 0.62363143 0.02880902 0.1086055 0.01022406 -0.13427997 -0.35161149
Rural+ Suburban 0.62909 0.64876885 0.01967885 0.20112357 -0.14451554 -0.20869457 -0.5148711
Suburban (35) 0.64299549 0.65849832 0.01550283 0.39465438 -0.41242652 -0.41844805 -0.74596333
Suburban+ Urban 0.64758173 0.64993768 0.00235595 0.39531742 -0.45911746 -0.45102291 -0.77137789
Urban (3) 0.65594669 0.63478627 -0.02116042 0.84038153 -0.97838562 -0.96722459 -0.10921866]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 08 A-Total TO 20 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.63662541 0.64472171 0.0080963 0.60855994 -0.59360351 -0.40116266 -0.26662412 -0.27717968
Rural (50) 0.5948224 0.62363143 0.02880902 0.44096183 -0.3449732 -0.28540799 -0.51555369 -0.17888385
Rural+ Suburban 0.62909 0.64876885 0.01967885 0.56378294 -0.55847296 -0.36721865 -0.17214823 -0.27174433
Suburban (35) 0.64299549 0.65849832 0.01550283 0.74223676 -0.76687059 -0.47264457 -0.0328172 -0.48536093
Suburban+ Urban 0.64758173 0.64993768 0.00235595 0.76618073 -0.76840659 -0.52292581 -0.16008656 -0.49713214
Urban (3) 0.65594669 0.63478627 -0.02116042 0.95442129 -0.74888076 -0.84950994 0.09011458 -0.74919858]

Table OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 20 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.29794104 0.34286135 0.04492031-0.49597342 0.29526509 0.32569565 -0.46274429
Rural (50) 0.32635868 0.44139369 0.11503501-0.26201586 0.26391782 0.16306878 -0.18438665
Rural+ Suburban 0.31919743 0.39200607 0.07280865-0.54277599 0.38007897 0.38701245 -0.32962164
Suburban (35) 0.31629145 0.37289054 0.05659909-0.53816477 0.4193036 0.45753906 -0.39888004
Suburban+ Urban 0.29049295 0.31849272 0.02799977-0.43198507 0.25514337 0.3099841 -0.557568
Urban (3) 0.24343834 0.22221449 -0.02122386-0.97885738 0.98746196 0.9939475 0.45844859]

Table OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 20 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.29794104 0.34286135 0.044920310.75012281 -0.65020213 -0.3243527 -0.77349529 -0.09423309
Rural (50) 0.32635868 0.44139369 0.115035010.42297052 -0.2781985 -0.27418064 -0.63154291 -0.07088953
Rural+ Suburban 0.31919743 0.39200607 0.072808650.69157979 -0.5482384 -0.28239479 -0.73814515 -0.07883179
Suburban (35) 0.31629145 0.37289054 0.056599090.68726841 -0.4979075 -0.30952553 -0.74051976 -0.2375898
Suburban+ Urban 0.29049295 0.31849272 0.027999770.75393092 -0.63400991 -0.3738568 -0.76330763 -0.26594527
Urban (3) 0.24343834 0.22221449 -0.02122386-0.7838879 0.93673504 0.60378749 -0.44130172 0.46181422]

Table OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 20 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.32711674 0.29116655 -0.0359502 0.73329034 -0.46802392 -0.53237926 0.18867481
Rural (50) 0.25441566 0.1719355 -0.08248017 0.40960195 -0.30140211 -0.30637557 -0.08362542
Rural+ Suburban 0.29720169 0.24587506 -0.05132663 0.77141359 -0.53451996 -0.58306323 0.04476945
Suburban (35) 0.31456389 0.27449345 -0.04007043 0.84616642 -0.72345992 -0.76794695 -0.02822787
Suburban+ Urban 0.34617125 0.3206543 -0.02551696 0.77335538 -0.60092665 -0.65977221 0.16699022
Urban (3) 0.4038208 0.40235397 -0.00146683 0.97255982 -0.99159901 -0.99669929 -0.4326607 ]

Table 0H.08t20.Dem.Racial - Correlation of county racial composition with county ΔActualDem, 2008 → 2012
Type A-Dem TO 08 A-Dem TO 20 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.32711674 0.29116655 -0.0359502 -0.50334963 0.39629592 0.09210091 0.77583885 -0.09289971
Rural (50) 0.25441566 0.1719355 -0.08248017 -0.05437872 0.00580272 -0.00650096 0.2938388 -0.12589055
Rural+ Suburban 0.29720169 0.24587506 -0.05132663 -0.41172596 0.25529542 0.04047262 0.75209783 -0.11679561
Suburban (35) 0.31456389 0.27449345 -0.04007043 -0.28813641 0.06566426 0.02797721 0.79879353 -0.05887456
Suburban+ Urban 0.34617125 0.3206543 -0.02551696 -0.40224739 0.25936369 0.09318087 0.80863726 -0.01885154
Urban (3) 0.4038208 0.40235397 -0.00146683 0.80144643 -0.92626427 -0.62649704 0.41527214 -0.48717007]

-----------------------------------------------


-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.63662541 0.61000538 -0.02662004 0.14078262 -0.16481638 -0.15857381 -0.48745669
Rural (50) .5948224 0.58120695 -0.01361546 0.25268128 -0.12383833 -0.19212013 -0.27003311
Rural+ Suburban .62909 0.60731771 -0.02177229 0.14988369 -0.14460911 -0.1509794 -0.44071794
Suburban (35) .64299549 0.6176401 -0.0253554 0.28809793 -0.29717185 -0.27224268 -0.67937423
Suburban+ Urban .64758173 0.61729768 -0.03028405 0.29876917 -0.34603917 -0.31215184 -0.68167896
Urban (3) .65594669 0.61668754 -0.03925914 0.01979388 -0.37492246 -0.32959187 0.7635253 ]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.63662541 0.61000538 -0.02662004 0.5649291 -0.55530717 -0.33927642 -0.23462586 -0.2524323
Rural (50) .5948224 0.58120695 -0.01361546 0.4410958 -0.45841102 -0.2150723 -0.28974017 -0.20463824
Rural+ Suburban .62909 0.60731771 -0.02177229 0.55766369 -0.56835493 -0.30770733 -0.15844549 -0.24431853
Suburban (35) .64299549 0.6176401 -0.0253554 0.70048456 -0.726589 -0.44139096 -0.0159167 -0.40959698
Suburban+ Urban .64758173 0.61729768 -0.03028405 0.69354763 -0.69223453 -0.48786461 -0.12330134 -0.43203045
Urban (3) .65594669 0.61668754 -0.03925914 0.78063817 0.13143862 -0.91208678 -0.77578273 -0.96785307]

Table GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.29794104 0.31301621 0.01507517 -0.52870306 0.30821536 0.35178206 -0.39277077
Rural (50) .32635868 0.389047 0.06268832 -0.21740256 0.18330705 0.11882843 -0.10720435
Rural+ Suburban .31919743 0.35283099 0.03363357 -0.57178966 0.38619713 0.4090141 -0.25337506
Suburban (35) .31629145 0.3385137 0.02222224 -0.62278267 0.51651107 0.56218929 -0.29572999
Suburban+ Urban .29049295 0.29376379 0.00327084 -0.51601532 0.3499231 0.41154542 -0.47493011
Urban (3) .24343834 0.2140272 -0.02941115 -0.99625711 0.9616143 0.97377373 0.56108761]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.29794104 0.31301621 0.01507517 0.70686141 -0.60807996 -0.2912063 -0.74716265 -0.09059323
Rural (50) .32635868 0.389047 0.06268832 0.39164644 -0.29618512 -0.21455829 -0.51086051 -0.09221729
Rural+ Suburban .31919743 0.35283099 0.03363357 0.64314501 -0.50368519 -0.24557795 -0.70655663 -0.07461229
Suburban (35) .31629145 0.3385137 0.02222224 0.61726193 -0.42075749 -0.28319855 -0.73082868 -0.20166812
Suburban+ Urban .29049295 0.29376379 0.00327084 0.69953891 -0.57399155 -0.35052662 -0.75837512 -0.23611923
Urban (3) .24343834 0.2140272 -0.02941115 -0.70431135 0.97177309 0.50449202 -0.54509227 0.35282552]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.32711674 0.26378462 -0.06333212 0.60069738 -0.33487767 -0.38489402 0.23420663
Rural (50) .25441566 0.16021754 -0.09419813 0.27515973 -0.12731869 -0.10984921 0.02818637
Rural+ Suburban .29720169 0.22076691 -0.07643478 0.63683868 -0.39968621 -0.43187042 0.08009429
Suburban (35) .31456389 0.24470392 -0.06985996 0.74681889 -0.61878219 -0.66107067 -0.01455619
Suburban+ Urban .34617125 0.29000975 -0.05616151 0.65095075 -0.46892481 -0.52717097 0.23457309
Urban (3) .4038208 0.37073687 -0.03308393 0.94184128 -0.99976262 -0.99964485 -0.33303277]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.32711674 0.26378462 -0.06333212 -0.47975271 0.40982518 0.00940191 0.69153871 -0.11294608
Rural (50) .25441566 0.16021754 -0.09419813 0.0564329 -0.02807959 -0.18435123 0.27820548 -0.16467954
Rural+ Suburban .29720169 0.22076691 -0.07643478 -0.34841206 0.23915281 -0.05814822 0.6501118 -0.14246211
Suburban (35) .31456389 0.24470392 -0.06985996 -0.31362509 0.12123619 0.0125619 0.73448683 -0.06205805
Suburban+ Urban .34617125 0.29000975 -0.05616151 -0.47398159 0.35737612 0.10346386 0.75687672 -0.00623147
Urban (3) .4038208 0.37073687 -0.03308393 0.8612039 -0.88028033 -0.70680717 0.31486855 -0.57840341]

-----------------------------------------------


-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.63662541 0.62039664 -0.01622877 0.40495702 -0.36041524 -0.44216611 -0.38638418
Rural (50) .5948224 0.57618835 -0.01863405 0.48948316 -0.32821992 -0.43639329 -0.55450215
Rural+ Suburban .62909 0.61180664 -0.01728335 0.41042308 -0.37102889 -0.4488915 -0.48065101
Suburban (35) .64299549 0.62609095 -0.01690454 0.46262925 -0.47355324 -0.48451754 -0.43168333
Suburban+ Urban .64758173 0.6318179 -0.01576383 0.4361373 -0.43399401 -0.45125321 -0.2550795
Urban (3) .65594669 0.64218715 -0.01375953 -0.88276897 0.65728143 0.69300409 0.92126443]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.63662541 0.62039664 -0.01622877 0.0617885 -0.0650378 -0.04667237 0.01983308 -0.14817326
Rural (50) .5948224 0.57618835 -0.01863405 0.32306716 -0.36293735 -0.03325133 -0.29196612 -0.15779533
Rural+ Suburban .62909 0.61180664 -0.01728335 0.13218629 -0.15703343 -0.05721428 0.01121727 -0.15144018
Suburban (35) .64299549 0.62609095 -0.01690454 0.24137178 -0.29761285 -0.15431751 0.09706922 -0.09604086
Suburban+ Urban .64758173 0.6318179 -0.01576383 0.12076206 -0.13418997 -0.13012812 0.08371192 -0.08860131
Urban (3) .65594669 0.64218715 -0.01375953 -0.20438234 0.94359957 -0.04680927 -0.91362751 -0.21337752]

Table S.OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.29794104 0.29532571 -0.00261532 0.26391494 -0.47329984 -0.44444957 -0.41016631
Rural (50) .32635868 0.3274849 0.00112623 0.55742154 -0.57914103 -0.58510662 -0.52416781
Rural+ Suburban .31919743 0.31892339 -0.00027404 0.27112413 -0.46910095 -0.44405302 -0.4190467
Suburban (35) .31629145 0.31548989 -0.00080156 0.21366033 -0.35496763 -0.30981433 -0.088769
Suburban+ Urban .29049295 0.28701736 -0.00347559 0.21644113 -0.37851749 -0.32983689 -0.17625675
Urban (3) .24343834 0.23546478 -0.00797357 -0.65374681 0.34102531 0.38614543 0.99954163]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.29794104 0.29532571 -0.00261532 0.27901712 -0.27013098 -0.03719953 -0.14260366 -0.13225792
Rural (50) .32635868 0.3274849 0.00112623 0.38303434 -0.41749402 0.04040409 -0.21970324 -0.14394251
Rural+ Suburban .31919743 0.31892339 -0.00027404 0.27572035 -0.28091744 -0.01420849 -0.10089352 -0.12425371
Suburban (35) .31629145 0.31548989 -0.00080156 0.29025198 -0.33303856 -0.14274191 -0.0007379 -0.11098693
Suburban+ Urban .29049295 0.28701736 -0.00347559 0.34950606 -0.36147601 -0.19609566 -0.09365484 -0.14850579
Urban (3) .24343834 0.23546478 -0.00797357 0.16264008 0.76053751 -0.40412756 -0.99877617 -0.55155664]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.32711674 0.31377605 -0.0133407 0.00159828 0.28561861 0.17214339 0.12877412
Rural (50) .25441566 0.23521207 -0.01920359 -0.28416996 0.4710621 0.36328802 0.14172312
Rural+ Suburban .29720169 0.28066132 -0.01654037 0.00090289 0.2727 0.16518341 0.05927788
Suburban (35) .31456389 0.29888822 -0.01567566 0.05517883 0.07868687 0.02650666 -0.23970413
Suburban+ Urban .34617125 0.33407313 -0.01209812 0.03063099 0.12550233 0.06742521 -0.04507153
Urban (3) .4038208 0.39777917 -0.00604163 -0.6985338 0.90773962 0.88635971 -0.11494868]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.32711674 0.31377605 -0.0133407 -0.23946718 0.25014975 -0.07006267 0.1307624 0.00079392
Rural (50) .25441566 0.23521207 -0.01920359 -0.12329953 0.18859169 -0.17289599 -0.05827511 0.01601167
Rural+ Suburban .29720169 0.28066132 -0.01654037 -0.1729057 0.19010923 -0.10862675 0.07609968 -0.01293994
Suburban (35) .31456389 0.29888822 -0.01567566 -0.04732727 0.08218707 -0.04108946 -0.01846879 -0.00970572
Suburban+ Urban .34617125 0.33407313 -0.01209812 -0.18501991 0.2114284 0.02775642 0.05217012 0.03282229
Urban (3) .4038208 0.39777917 -0.00604163 -0.99692506 0.58244332 0.94568522 0.13399903 0.87791568]

-----------------------------------------------



-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.61000538 0.64472171 0.03471633 0.16586544 -0.11231352 -0.19421359 -0.45916166
Rural (50) 0.58120695 0.62363143 0.04242448 -0.10081321 0.15292778 -0.00770378 -0.27655568
Rural+ Suburban 0.60731771 0.64876885 0.04145114 0.17560213 -0.0868758 -0.18705323 -0.3888253
Suburban (35) 0.6176401 0.65849832 0.04085822 0.38746116 -0.40867752 -0.44292451 -0.60209643
Suburban+ Urban 0.61729768 0.64993768 0.03264 0.3913951 -0.4554509 -0.47346693 -0.67121189
Urban (3) 0.61668754 0.63478627 0.01809872 0.99609152 -0.89915551 -0.91929202 -0.69662964]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.61000538 0.64472171 0.03471633 0.43133891 -0.41623725 -0.31931738 -0.20258314 -0.20160892
Rural (50) 0.58120695 0.62363143 0.04242448 0.23412514 -0.05847133 -0.22901407 -0.52234234 -0.06670165
Rural+ Suburban 0.60731771 0.64876885 0.04145114 0.34582341 -0.32559035 -0.28437204 -0.11819592 -0.1927486
Suburban (35) 0.6176401 0.65849832 0.04085822 0.57541131 -0.59174854 -0.37092176 -0.03998759 -0.42309698
Suburban+ Urban 0.61729768 0.64993768 0.03264 0.65125922 -0.65627073 -0.43088815 -0.15633037 -0.4394142
Urban (3) 0.61668754 0.63478627 0.01809872 0.56996466 -0.99800526 -0.34646849 0.68272743 -0.18453563]

Table S.OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.31301621 0.34286135 0.02984514 -0.34346115 0.21803921 0.2160058 -0.52915294
Rural (50) 0.389047 0.44139369 0.05234669 -0.25107265 0.32134411 0.18786908 -0.26453948
Rural+ Suburban 0.35283099 0.39200607 0.03917508 -0.36859381 0.28768344 0.2600789 -0.41936438
Suburban (35) 0.3385137 0.37289054 0.03437684 -0.28517401 0.15879831 0.17615948 -0.54752399
Suburban+ Urban 0.29376379 0.31849272 0.02472894 -0.21331427 0.03557675 0.07043921 -0.65800808
Urban (3) 0.2140272 0.22221449 0.00818729 -0.89429693 0.99506182 0.98908821 0.21657781]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.31301621 0.34286135 0.02984514 0.71207451 -0.62682233 -0.33788469 -0.6963111 -0.08572655
Rural (50) 0.389047 0.44139369 0.05234669 0.32722141 -0.14117852 -0.28755372 -0.63039768 -0.00384605
Rural+ Suburban 0.35283099 0.39200607 0.03917508 0.64753704 -0.52614986 -0.29990326 -0.6492535 -0.07110048
Suburban (35) 0.3385137 0.37289054 0.03437684 0.7250684 -0.57886162 -0.31600708 -0.64801667 -0.27441042
Suburban+ Urban 0.29376379 0.31849272 0.02472894 0.77489922 -0.6802354 -0.3769621 -0.68413684 -0.29464796
Urban (3) 0.2140272 0.22221449 0.00818729 -0.91634087 0.81645367 0.78715654 -0.19779477 0.67279523]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.26378462 0.29116655 0.02738192 0.77972215 -0.58763776 -0.66107956 0.0529131
Rural (50) 0.16021754 0.1719355 0.01171796 0.40175713 -0.46770994 -0.52023728 -0.27518844
Rural+ Suburban 0.22076691 0.24587506 0.02510815 0.79142752 -0.62210388 -0.68589373 -0.03053181
Suburban (35) 0.24470392 0.27449345 0.02978953 0.88194807 -0.78704443 -0.82835681 -0.04673471
Suburban+ Urban 0.29000975 0.3206543 0.03064455 0.85916251 -0.73334465 -0.78318851 0.01796894
Urban (3) 0.37073687 0.40235397 0.0316171 0.99136985 -0.97296773 -0.9830092 -0.52334642]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.26378462 0.29116655 0.02738192 -0.41019291 0.26340792 0.22041527 0.72113687 -0.03045338
Rural (50) 0.16021754 0.1719355 0.01171796 -0.26619158 0.07890981 0.40572677 0.10317785 0.05994473
Rural+ Suburban 0.22076691 0.24587506 0.02510815 -0.40731811 0.21158609 0.20382608 0.7198155 -0.03821035
Suburban (35) 0.24470392 0.27449345 0.02978953 -0.20116614 -0.03733806 0.04943723 0.78329619 -0.04448825
Suburban+ Urban 0.29000975 0.3206543 0.03064455 -0.2055693 0.04030395 0.05890912 0.76250603 -0.03811685
Urban (3) 0.37073687 0.40235397 0.0316171 0.73548236 -0.96019738 -0.54275956 0.50689028 -0.39449191

-----------------------------------------------


-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.61000538 0.63057404 0.02056867 0.28756765 -0.27855916 -0.33572771 -0.62435559
Rural (50) 0.58120695 0.62046389 0.03925694 0.19408797 -0.0520634 -0.23704992 -0.41893454
Rural+ Suburban 0.60731771 0.64119488 0.03387717 0.32198425 -0.25935393 -0.34586991 -0.5345207
Suburban (35) 0.6176401 0.64911968 0.03147958 0.46830291 -0.51304532 -0.5375237 -0.71021384
Suburban+ Urban 0.61729768 0.63305676 0.01575908 0.45042617 -0.54650333 -0.55059361 -0.78425424
Urban (3) 0.61668754 0.60420168 -0.01248586 0.98814434 -0.9779515 -0.98691655 -0.50390102]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.61000538 0.63057404 0.02056867 0.56977562 -0.60245912 -0.23667387 -0.23874253 -0.17259839
Rural (50) 0.58120695 0.62046389 0.03925694 0.13004253 -0.08396212 -0.07161849 -0.43120871 -0.03288721
Rural+ Suburban 0.60731771 0.64119488 0.03387717 0.42721854 -0.47567911 -0.17640204 -0.10548476 -0.16090492
Suburban (35) 0.6176401 0.64911968 0.03147958 0.6364059 -0.69708318 -0.29221371 -0.01472702 -0.366738
Suburban+ Urban 0.61729768 0.63305676 0.01575908 0.74334785 -0.7755421 -0.3778288 -0.16868684 -0.38813295
Urban (3) 0.61668754 0.60420168 -0.01248586 0.75064817 -0.95362069 -0.56165396 0.48722496 -0.41521496]

Table S.OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.31301621 0.34603252 0.03301632 -0.32614067 0.21204868 0.19813182 -0.54325515
Rural (50) 0.389047 0.44924858 0.06020159 -0.10125805 0.26662438 0.10246911 -0.30360273
Rural+ Suburban 0.35283099 0.39711255 0.04428156 -0.35442738 0.2877631 0.24503387 -0.42493218
Suburban (35) 0.3385137 0.37718262 0.03866892 -0.32716657 0.22569393 0.23021403 -0.51286579
Suburban+ Urban 0.29376379 0.32068609 0.0269223 -0.24489658 0.08756932 0.11234678 -0.64605318
Urban (3) 0.2140272 0.21919692 0.00516972 -0.72125284 0.92080025 0.90083347 -0.08284543]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.31301621 0.34603252 0.03301632 0.69398881 -0.63600978 -0.23147784 -0.70370446 -0.01954936
Rural (50) 0.389047 0.44924858 0.06020159 0.15742519 -0.10820085 -0.12433684 -0.54634404 0.07639941
Rural+ Suburban 0.35283099 0.39711255 0.04428156 0.60426112 -0.51753513 -0.17816753 -0.6601085 0.00274612
Suburban (35) 0.3385137 0.37718262 0.03866892 0.68695995 -0.54902058 -0.21729181 -0.66638849 -0.17380712
Suburban+ Urban 0.29376379 0.32068609 0.0269223 0.75683795 -0.67200784 -0.29446258 -0.6935534 -0.20656306
Urban (3) 0.2140272 0.21919692 0.00516972 -0.99387856 0.60836121 0.93470678 0.10196296 0.86201354]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.26378462 0.27569505 0.01191043 0.84168437 -0.72897585 -0.76593277 -0.18967
Rural (50) 0.16021754 0.16379758 0.00358004 0.58711077 -0.62034498 -0.67318341 -0.38870291
Rural+ Suburban 0.22076691 0.23557857 0.01481167 0.84464972 -0.73445351 -0.77028868 -0.20732843
Suburban (35) 0.24470392 0.26301813 0.01831421 0.90691215 -0.88427419 -0.89595975 -0.27598435
Suburban+ Urban 0.29000975 0.30317336 0.01316361 0.90275146 -0.87578728 -0.88931219 -0.27826778
Urban (3) 0.37073687 0.37530734 0.00457047 0.91687581 -0.99893775 -0.9955346 -0.26830422]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.26378462 0.27569505 0.01191043 -0.13234436 -0.02353009 0.13180393 0.59066892 -0.09859503
Rural (50) 0.16021754 0.16379758 0.00358004 -0.18479241 -0.00819271 0.39177177 0.13386476 0.01501961
Rural+ Suburban 0.22076691 0.23557857 0.01481167 -0.20462822 -0.0052871 0.15012702 0.65060364 -0.09118946
Suburban (35) 0.24470392 0.26301813 0.01831421 0.0133373 -0.25174159 -0.03864188 0.69569246 -0.15164467
Suburban+ Urban 0.29000975 0.30317336 0.01316361 0.09592017 -0.25600794 -0.08497094 0.60538835 -0.17905509
Urban (3) 0.37073687 0.37530734 0.00457047 0.89369738 -0.84606875 -0.75316515 0.24975991 -0.63240656]

-----------------------------------------------

For 2020 → 2024, see above.
Image 2
Figurex 1
Image 2
Figurex 1
Image 2
Figurex 1

Figure OHBC (OH.PresCompare). Turnout trends, by county, for {2008, 2012, 2016, 2020, 2024} Presidential elections


“It was clear how the Clinton campaign message just wasn’t working here,” said Sappington, who studied at Ohio University and works at the Texas Roadhouse with Mash’s daughter. When Sappington canvassed for Clinton in September, he was taken aback by how many people told him they had settled on voting for Trump. These were registered Democrats.


Mash viewed Trump not as a Republican but as an outsider, just like himself when he ran for council and won. In his opinion, Trump’s talk on trade made him sound like a Democrat should. Mash says he would have taken a serious look at Bernie Sanders had he defeated Clinton in the primary. “He was all about change, too,” he says.


Trump’s focus on jobs and winning let him overlook everything about Trump that Democrats and the media assumed would doom his run. The “Access Hollywood” tape? Didn’t bother Mash ― what annoyed him was the media’s replaying it. Trump’s smear of Mexican immigrants or his plan to ban all Muslims from entering the country? Blown out of proportion. He resented the charge that voting for Trump made him racist.


HuffPost, after the 2016 election



2012. Notice for the Dems, they held on to the lower-income vote from 2008 to 2012 - for example, the correlation of county poverty rate and Democrat turnout change was 0.2856 (positive!) (and the correlation with income was practically zero) - this is even after Obama’s extremely disappointing first term. Still, he was the "change" guy, and the guy who did pass, and fight for, some form of healthcare reform (The Affordable Care Act, aka "Obamacare"). And in both 2008 and 2012, the Democrats won Ohio. While Dem actual turnout fell from 2008 to 2012 (by 1.3 pp), GOP turnout also fell, albeit slower (by 0.3 pp). And their turnout, while increasing weakly with income (0.2639), was moderately negatively correlated with poverty (-0.4733).


2016. But this starts to change in 2016. Part of it could be the Trump appeal, part of it could be disillusionment both with Obama’s performance, and the Democrats-as-stilted-corporate-surrogates in general. From 2008 → 2016, the Dems turnout fell by 6.3 pp, with turnout moderately positively correlated with income (0.6007), and moderately negatively correlated with poverty (-0.3349). The GOP, by contrast, saw a rise in actual turnout by 1.5 pp, moderately negatively correlated with income (-0.5287), and near-moderately positively correlated with poverty (0.3082). Overall though, turnout was weakly negatively correlated with poverty (-0.1648), indicating not all had switched to the Trump camp.


2020. 2020 was a slight recovery overall for the Dems, but it appears more due to higher income Ohio counties. Compared to 2008, actual Dem turnout fell 3.6 pp (an improvement from 2016), strongly positively correlated with county income (0.7333), and moderately negatively correlated with poverty (-0.4680). The GOP kept at the same trend from 2008, though with increasing turnout (+4.5 pp). Now though, while they were still moderately negatively correlated with income (-0.4960) and near-moderately positively correlated with poverty (0.2953), these numbers show improvement with higher-income counties, and worse performance with higher-poverty counties. Overall, turnout from 2008 → 2020 (+0.8 pp) was weakly negatively correlated with poverty (-0.1695), implying that higher poverty counties hadn’t purely flipped to Trump.


From 2016 → 2020, the Dems turnout change correlated strongly with income (+0.7797) and near-strongly negatively with poverty (-0.6611). For the GOP, turnout change correlated moderately negatively with income (-0.3435) and weakly positively with poverty (+0.2180). Overall turnout was near-zero correlated with income (+0.0347) and weakly negatively with poverty (-0.1123), indicating former Clinton low-income voters likely didn’t switch to Trump as a whole.


2024. From 2008 → 2012, Obama eked out gains with higher-poverty Ohio counties. From 2008 → 2016, Clinton saw significant losses here. This trend worstened for Biden in 2020. It was the opposite for income level. And in 2024, those downward and upward trends continued. Overall, Dem turnout from 2008 → 2024 fell by 5.1 pp; not as bad as Clinton 2016, but this is not because Harris recovered the lower-income vote, but strengthened her appeal in higher-income counties. Change in actual Democrat turnout was strongly positively correlated with income (+0.7793), stronger even than in 2020. The correlation with poverty was moderately negatively correlated with poverty (-0.5505), worse than 2020. For the GOP, turnout increased compared to 2008 by 4.8 pp, a slight improvement over 2020, and was again moderately negatively correlated with county income (-0.4882) and near-moderately positively correlated with poverty (+0.2931). Overall, turnout increased by 2.1 pp, and was near-moderately correlated with income (+0.2876) and near-moderately negatively correlated with poverty (-0.2786).


From 2016 → 2024, the Dems turnout increased by 1.2 pp, strongly positively correlated with income (+0.8417), and strongly negatively correlated with poverty level (-0.7290). For the GOP, turnout increased by +3.3 pp, moderately negatively correlated with income (-0.3261) and weakly positively correlated with poverty (+0.2120). From 2020 → 2024, Dem turnout fell by 1.5 pp, moderately positively correlated with income (+0.5000) and moderately negatively correlated with poverty (-0.5482). For the GOP, turnout increased by only by +0.3 pp, and was near-zero correlated with income (+0.0647) and poverty (+0.0336). Overall, turnout fell by 1.4 pp, near-moderately correlated with income (+0.2842) and moderately negatively correlated with poverty (-0.3426).


What we see is a steady decline in higher-poverty county support for the Democrats since Obama wasn’t at the top of the ticket. The fall was slower from 2016 → 2020 than 2020 → 2024, but was present. While higher-income counties have seen increased Democrat turnout, this hasn’t compensated for a fall in Democrat Ohio turnout since 2008 (although their turnout increased from 2016 → 2020). By contrast, Trump gained a foothold in higher-poverty counties in 2016 compared to 2008, though that level has hardly budged since. He saw weak positive correlation from 2016 → 2020 here, but no gains from 2020 → 2024. In other words, most of his support in lower-income counties has been there since 2016. At the same time, Trump has seen a steady decline in support in higher-income counties.


What to make of this? Today, it’s easy to see Obama as a false hope. But back then, there was real substance to his appeal. Either for reason of Trump’s appeal (railing against free trade deals, etc), or disillusionment with Obama and the corporate auspices of the Democratic Party with Clinton at the top of the ticket, Trump was able to grab some of the higher-poverty areas and hold on. Yet most higher-poverty areas saw an overall negative trend in turnout, indicating many former Obama voters just stayed home, and this pattern deepened.


It’s this erosion of the Democratic working-class base, and Trump’s smaller toe-hold in this base, which tells us why Ohio is "no longer a swing state". For example, let’s compare the Senatorial races in 2022 (when Republican JD Vance, now Vice President elect, won a seat for Senate) and 2018 (when Democratic Sherrod Brown retained his seat for Senate), and run it in reverse. This is a bit problematic, since the incumbent party tends to do worse in midterms. That is, we can reasonably assume other factors are affecting the election other than the candidates themselves. Yet presumably, this would mean Vance was boosted in 2022 (Dem Biden incumbent), and Brown was boosted in 2018 (GOP Trump incumbent). 2022 didn’t see much enthusiasm for either candidate though, and in 2018, Republican Renacci accidentally called the country company, and he thought since it’s now a Trump state, he could win (Brown: voters dont see liberal or conservative, but if candidate fights for you). Since both would have been "boosted" from this consideration, it’s perhaps less problematic. So what are the trends of 2022 → 2018?


-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.45137185 0.48498452 0.03361267 0.08219533 -0.11290543 -0.15184536 -0.15413653
Rural (50) 0.4391247 0.45058401 0.01145931 -0.07547388 0.16435321 0.15765742 0.30216534
Rural+ Suburban 0.45939807 0.47977176 0.02037369 0.03630039 0.14889853 0.09764488 0.42148622
Suburban (35) 0.46717416 0.49122448 0.02405032 -0.05705474 0.25246349 0.20278139 0.53150129
Suburban+ Urban 0.45438636 0.49362055 0.03923419 -0.10290148 0.33385627 0.27215123 0.70201164
Urban (3) 0.43148179 0.49787647 0.06639467 -0.93731636 0.99996328 0.9992052 0.32053853]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.45137185 0.48498452 0.03361267 0.08017299 -0.10813025 0.02665853 0.02732452 0.03225616
Rural (50) 0.4391247 0.45058401 0.01145931 -0.24542577 0.21706616 0.1082662 0.1572082 0.41637537
Rural+ Suburban 0.45939807 0.47977176 0.02037369 -0.50436209 0.48046576 0.13032197 0.36952566 0.25235366
Suburban (35) 0.46717416 0.49122448 0.02405032 -0.59584854 0.56106911 0.13353064 0.396556 0.10191298
Suburban+ Urban 0.45438636 0.49362055 0.03923419 -0.76090612 0.74532574 0.26640115 0.4842681 0.17519321
Urban (3) 0.43148179 0.49787647 0.06639467 -0.86784773 0.87393123 0.71609717 -0.30229367 0.58913691]

Table S.OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.23907125 0.22591471 -0.01315653 0.49040866 -0.23952421 -0.29414063 0.36461233
Rural (50) 0.30429382 0.26947856 -0.03481526 0.16591322 -0.08628712 -0.06675336 0.21204664
Rural+ Suburban 0.27394177 0.2522457 -0.02169607 0.51936352 -0.29729469 -0.33468929 0.23584655
Suburban (35) 0.2622999 0.24548385 -0.01681604 0.62204859 -0.44687567 -0.50062782 0.14302212
Suburban+ Urban 0.22301738 0.21497828 -0.00803909 0.52806626 -0.30615671 -0.37378911 0.3628695
Urban (3) 0.15265739 0.16079416 0.00813678 0.75121153 -0.93716497 -0.9191715 0.03864454]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.23907125 0.22591471 -0.01315653 -0.53638298 0.46271565 0.1220254 0.703752 0.03700656
Rural (50) 0.30429382 0.26947856 -0.03481526 -0.12879785 0.10909757 0.05349348 0.32734235 0.15519724
Rural+ Suburban 0.27394177 0.2522457 -0.02169607 -0.42886665 0.31651396 0.06557104 0.66813052 0.02036489
Suburban (35) 0.2622999 0.24548385 -0.01681604 -0.40755136 0.22315551 0.03446506 0.7905368 -0.08784592
Suburban+ Urban 0.22301738 0.21497828 -0.00803909 -0.54797381 0.43795098 0.12746461 0.7990624 -0.02701761
Urban (3) 0.15265739 0.16079416 0.00813678 0.98801301 -0.64290153 -0.91805523 -0.05782188 -0.83872761]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.21152951 0.25895814 0.04742863 -0.60372606 0.53632692 0.52109277 0.19618284
Rural (50) .134122 0.18103487 0.04691287 -0.26670297 0.26610926 0.23379269 0.05540458
Rural+ Suburban .18469041 0.22742214 0.04273173 -0.61147762 0.53038542 0.52050548 0.15324506
Suburban (35) .20408651 0.24562362 0.04153711 -0.81506165 0.81702262 0.82657418 0.40980316
Suburban+ Urban .23058258 0.27852029 0.04793771 -0.80268984 0.82087903 0.82558265 0.44975053
Urban (3) .27804041 0.3369515 0.05891109 -0.83258347 0.97534381 0.96351315 0.09506205]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.21152951 0.25895814 0.04742863 -0.10464737 0.18940897 0.09735187 -0.35162237 0.26361144
Rural (50) .134122 0.18103487 0.04691287 -0.11460119 0.10637143 0.05814506 -0.22806682 0.26053161
Rural+ Suburban .18469041 0.22742214 0.04273173 -0.00882508 0.12446384 0.06332504 -0.44425476 0.2523749
Suburban (35) .20408651 0.24562362 0.04153711 -0.16119722 0.34726622 0.10626902 -0.52424204 0.22161832
Suburban+ Urban .23058258 0.27852029 0.04793771 -0.29351722 0.41290665 0.18638563 -0.38583785 0.2659224
Urban (3) .27804041 0.3369515 0.05891109 -0.95857197 0.73937418 0.85693164 -0.07593267 0.75854835]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-IP TO 18 A-IP TO 24 ΔA-IP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.23907125 0.25895814 0.0198869 -0.03612796 0.2202673 0.18949098 0.69401639
Rural (50) 0.30429382 0.18103487 -0.12325895 -0.41541869 0.34105278 0.42370473 0.49842229
Rural+ Suburban 0.27394177 0.22742214 -0.04651963 -0.04906161 0.19444599 0.18943209 0.60857991
Suburban (35) 0.2622999 0.24562362 -0.01667628 -0.1516485 0.25390171 0.24360595 0.74078247
Suburban+ Urban 0.22301738 0.27852029 0.05550292 -0.17604426 0.33620573 0.30403826 0.82112519
Urban (3) 0.15265739 0.3369515 0.18429412 -0.54166558 0.80574675 0.77612258 -0.31089307]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO Interparty (Vance 22 → Brown 18)
Type A-IP TO 18 A-IP TO 24 ΔA-IP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.23907125 0.25895814 0.0198869 -0.79385097 0.76625852 0.37678407 0.46615306 0.25108605
Rural (50) 0.30429382 0.18103487 -0.12325895 -0.53360936 0.5019215 0.20665414 0.35734677 0.29360591
Rural+ Suburban 0.27394177 0.22742214 -0.04651963 -0.70505336 0.67038329 0.32813683 0.3480127 0.2507137
Suburban (35) 0.2622999 0.24562362 -0.01667628 -0.76519665 0.73962804 0.50793296 0.1988518 0.37360918
Suburban+ Urban 0.22301738 0.27852029 0.05550292 -0.86653963 0.84805204 0.55554512 0.35270825 0.39218088
Urban (3) 0.15265739 0.3369515 0.18429412 -0.99250773 0.40849268 0.9915397 0.32908317 0.95581529]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔTO, of Dem22-GOP22 and Dem18-GOP18
Type A-DC TO 18 A-DC TO 24 ΔA-DC TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.02754173 0.0198869 0.04742863 -0.60372606 0.53632692 0.52109277 0.19618284
Rural (50) 0.17017182 -0.12325895 0.04691287 -0.26670297 0.26610926 0.23379269 0.05540458
Rural+ Suburban 0.08925136 -0.04651963 0.04273173 -0.61147762 0.53038542 0.52050548 0.15324506
Suburban (35) 0.05821339 -0.01667628 0.04153711 -0.81506165 0.81702262 0.82657418 0.40980316
Suburban+ Urban 0.00756521 0.05550292 0.04793771 -0.80268984 0.82087903 0.82558265 0.44975053
Urban (3) 0.12538303 0.18429412 0.05891109 -0.83258347 0.97534381 0.96351315 0.09506205]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO, of Dem22-GOP22 and Dem18-GOP18
Type A-DC TO 18 A-DC TO 24 ΔA-DC TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.02754173 0.0198869 0.04742863 -0.10464737 0.18940897 0.09735187 -0.35162237 0.26361144
Rural (50) 0.17017182 -0.12325895 0.04691287 -0.11460119 0.10637143 0.05814506 -0.22806682 0.26053161
Rural+ Suburban 0.08925136 -0.04651963 0.04273173 -0.00882508 0.12446384 0.06332504 -0.44425476 0.2523749
Suburban (35) 0.05821339 -0.01667628 0.04153711 -0.16119722 0.34726622 0.10626902 -0.52424204 0.22161832
Suburban+ Urban 0.00756521 0.05550292 0.04793771 -0.29351722 0.41290665 0.18638563 -0.38583785 0.2659224
Urban (3) 0.12538303 0.18429412 0.05891109 -0.95857197 0.73937418 0.85693164 -0.07593267 0.75854835]

-----------------------------------------------


Image 2
Figurex 1
Image 2
Figurex 1
Image 2
Figurex 1

Figure OHIPD (OH.Sen.Vance2022→Brown2018). Turnout trends, by county poverty-level, from the 2022 Ohio Senate race, to the 2018 Ohio Senate race. In order: (1) overall actual turnout, (2) overall GOP turnout, (3) overall Democrat turnout, (4,5,6) changes in turnout for 2022 Vance → 2018 Brown, correlated with poverty (4), the Gini coefficient (5), and median income (6). (7,8,9) change in turnout interparty, for all regions, colored by income (7), poverty (8), and percent black population (9); (10, 11, 12) change in turnout interparty, for Rural counties, colored by income (10), poverty (11), and percent black population (12). Note that the median poverty rate over all counties is 8.95%, and for Rural counties is 9.0%.


Image 2
Figurex 1
Image 2
Figurex 1

Figure OHSCM (OH.Sen.Compare.Maps). Comparing performance of (1) Δ(Vance 22, Brown 18) (true min: -0.3085), (2) Δ(Vance 22, Ryan 22) (true min: -0.3774) and (3) Δ(Δ(Vance 22, Brown 18), Δ(Vance 22, Ryan 22)) (true min: 0.00867); (4 and beyond) WRITE CAPTION TO DO


Image 2
Figurex 1
Image 2
Figurex 1
Image 2
Figurex 1

Figure OHIPD (OH.Vance.DeWine). Turnout trends, by county poverty-level, comparing Vance Senatorial 2022 to DeWine Gubernatorial {2018, 2022}


We see that Brown sees an improvement in turnout by 4.7 pp (winning 2018 actual turnout by 3.3 pp; among those who voted, 53.40% Brown vs 46.58% Renacci, Δ = 6.82 pp), with turnout moderately negatively correlated with higher-income counties (-0.6037) and moderately positively correlated with poverty rate (+0.5363). The GOP, by contrast, saw a fall in turnout of 1.3 pp, a moderately positive correlation with higher-income counties (+0.4904) and a weakly negative correlation with poverty (-0.2395). Overall, turnout rose 3.4 pp, moderately positively correlated with income (+0.5125) and near-moderately negatively with poverty levels (-0.2936). If we compare trends for Vance 2022 to Brown 2018, we see an increase in party actual turnout (+2.0 pp), near zero correlation with income (+0.0199), and near-moderate correlation with poverty (+0.2203).


Interestingly, the elusive Gini correlation is at +0.6940, indicating in more unequal counties, Brown did significantly better than Vance. Looking at Fig. OHIPD, we can see that as Gini coefficient increases (and with it, Δinterparty-turnout), poverty increases (slides 9,12) and in terms of income, the higher income counties are near the middle, with low income counties at falling and rising turnout extremes; for rural areas, income levels clearly lower with Gini coefficient and turnout. From this, it seems reasonable to infer that, while we can’t see the actual income distributions of counties, that Brown performed better with lower income voters than Vance.


If we look at the difference between how Ryan 22 and Vance 22, and Brown 18 and Vance 22, fared, we see some interesting patterns pop out. Looking at Fig. OHSCM slide 3, we can see that Brown does much better in the southeast part of Ohio (the Appalachian part) than did Ryan. This makes sense - he did well here in 2006, part of his working-class orientation paying off. A similar pattern holds for the western Lake Erie coast region. Overall, he outperforms Ryan in the poverty metrics, had higher turnout across region types; notably while he saw relatively better urban turnout performance, his 2nd best region (comparatively) was rural counties. However, the comparison shows a moderately negative correlation with higher-income counties (see also Slide 7 in Fig. OHSCM; the southwest to northeast diagonal through the middle of the state), with strong negative correlation for suburban and urban counties, and near-moderate negative correlation in rural - generally, it seems doing well with Ohio’s higher-income counties is a red flag for election performance.


It’s not totally fair to compare Ryan and Brown; Ryan was hurt by running during a Democratic incumbent year (and vice versa for Brown). Further, their opponents were different. Still, these patterns seem worth noting.


Something else is worth considering: JD Vance has repeatedly polished his self-described 'low-income Appalachian' background (he is a venture capitalist today, closely tied to Silicon Valley’s largest moguls, Peter Thiel). Did that help him in Ohio? It was widely noticed in the 2022 election that he significantly underperformed deWine’s gubernatorial performance, short hundreds of thousands of votes. This is significant especially since there were only 0.06 pp fewer votes for the Senatorial race than for the gubernatorial (actual 2022 Gubernatorial election turnout: 45.193782%; actual 2022 Senatorial turnout: 45.137185%); that is, many voted for deWine, but against Vance (and for the Democrat Senatorial candidate, Ryan).


Table S.OH.Dem.Income - Correlation of county income metrics with county ΔTO GOP (DeWine Gubernatorial GOP 22 → Vance Senatorial 22)
Type A-DGOP TO 18 A-DGOP TO 24 ΔA-DGOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.28142325 0.23907125 -0.04235201 -0.39611705 0.34878725 0.44115403 -0.03488015
Rural (50) 0.33773005 0.30429382 -0.03343623 -0.36095942 0.35486258 0.45166104 0.30386861
Rural+ Suburban 0.31628694 0.27394177 -0.04234517 -0.39925353 0.36242808 0.44920523 0.0122709
Suburban (35) 0.30806219 0.2622999 -0.04576229 -0.29988577 0.26751538 0.28741063 -0.3107128
Suburban+ Urban 0.26756391 0.22301738 -0.04454654 -0.29091948 0.2525332 0.2752241 -0.25622592
Urban (3) 0.19502635 0.15265739 -0.04236897 0.83805784 -0.58849464 -0.62695428 -0.95188499]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO GOP (DeWine Gubernatorial GOP 22 → Vance Senatorial 22)
Type A-DGOP TO 18 A-DGOP TO 24 ΔA-DGOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.28142325 0.23907125 -0.04235201 0.43512024 -0.3010639 -0.62289365 -0.33594755 -0.35191914
Rural (50) 0.33773005 0.30429382 -0.03343623 0.53308117 -0.2379714 -0.6596392 -0.12601786 -0.34789861
Rural+ Suburban 0.31628694 0.27394177 -0.04234517 0.51325704 -0.34490479 -0.62341578 -0.33768299 -0.35087106
Suburban (35) 0.30806219 0.2622999 -0.04576229 0.42576667 -0.27138899 -0.61004891 -0.33207533 -0.54668133
Suburban+ Urban 0.26756391 0.22301738 -0.04454654 0.31200119 -0.19958031 -0.58445922 -0.30472599 -0.53281644
Urban (3) 0.19502635 0.15265739 -0.04236897 0.11755011 -0.91084588 0.13442834 0.94582597 0.29842398]

Image 2 Image 2

Figure OHDG . Turnout change trends, comparing DeWine and Vance in 2022. Whole state, people-under-150%-poverty-level rate vs ΔTO. Left, color-coded by Income; Right, by region type.


The data tells an interesting story. Overall, Vance saw lost votes in every single county compared to DeWine, with actual turnout falling from -8.50 pp to -0.70 pp. So that’s pretty bad; the best he did was fall by only 0.70 percentage points! However, his votes fell less hard in areas with higher poverty rates and more white people, and faster in areas with non-white people and higher income. The geography of this less-bad loss can be seen in Fig. OHSCM, slide 4. A significant region of better-than-otherwise performance for Vance is in the Appalachian southeast region of Ohio.


Maybe we should still be skeptical of Vance’s "working class" bonafides. However, among the groups of voters that abandoned him (which was nearly everyone it seems) from DeWine, it was whites and higher-poverty counties that did so less harshly.


Yet Vance (2,192,114) was not only short 388,310 votes of DeWine’s 2022 votes (2,580,424), but was even short 43,711 votes of DeWine’s 2018 votes (2,235,825).

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔTO GOP (DeWine Gubernatorial GOP 18 → Vance Senatorial 22)
Type A-DGOP TO 18Gb A-DGOP TO 22S ΔA-DGOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.24548786 0.23907125 -0.00641662 -0.53507786 0.32536475 0.40032325 -0.26430582
Rural (50) 0.29186894 0.30429382 0.01242487 -0.36458753 0.2662292 0.31581512 0.06383612
Rural+ Suburban 0.27302184 0.27394177 0.00091993 -0.56194243 0.38237092 0.44235952 -0.12619359
Suburban (35) 0.2656266 0.2622999 -0.0033267 -0.57768845 0.43707661 0.47367946 -0.22591704
Suburban+ Urban 0.23384418 0.22301738 -0.01082681 -0.50546502 0.31966512 0.37061938 -0.39102495
Urban (3) 0.17739213 0.15265739 -0.02473474 -0.55895148 0.81783505 0.78901056 -0.29115033]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO GOP (DeWine Gubernatorial GOP 18 → Vance Senatorial 22)
Type A-DGOP TO 18Gb A-DGOP TO 22S ΔA-DGOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.24548786 0.23907125 -0.00641662 0.52067775 -0.4204826 -0.26064357 -0.66148873 -0.20240487
Rural (50) 0.29186894 0.30429382 0.01242487 0.19662281 -0.02328039 -0.33102388 -0.23897418 -0.43380838
Rural+ Suburban 0.27302184 0.27394177 0.00091993 0.43161727 -0.28046942 -0.22263551 -0.62720923 -0.2000305
Suburban (35) 0.2656266 0.2622999 -0.0033267 0.3896867 -0.20267327 -0.0915056 -0.72278626 -0.01350694
Suburban+ Urban 0.23384418 0.22301738 -0.01082681 0.50259926 -0.38839571 -0.16261088 -0.73568283 -0.05259023
Urban (3) 0.17739213 0.15265739 -0.02473474 -0.99482443 0.42730113 0.98864 0.30946373 0.94952472]

Comparing these elections, it’s a similar pattern: Vance underperforms as county income rises, and his performance is more favorable as county poverty levels rise; and also, does better with white voters, and worse with non-white voters. Notably, compared to DeWine’s 2018 Gubernatorial race (but not compared to DeWine’s 2022 Gubernatorial race), Vance sees higher actual turnout in Rural counties by 1.2 pp.


Image 2
Figurex 1
Image 2
Figurex 1
Image 2
Figurex 1

Figure OHVDW (OH.Vance vs DeWine Compare). Turnout trends, by county poverty-level, from the 2022 Ohio Senate race, to the 2018 Ohio Senate race. (4) Vance’s 2022 Senatorial actual turnout increase/decrease compared to DeWine’s Gubernatorial 2022 actual turnout



How about 2018 gubernatorial vs 2018 Senatorial?

-----------------------------------------------

Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.48704169 0.48498452 -0.00205716 -0.15251897 0.09093729 0.11902733 0.00966038
Rural (50) 0.4523103 0.45058401 -0.00172629 -0.07516206 0.03026394 0.04814195 -0.00260924
Rural+ Suburban 0.48156858 0.47977176 -0.00179681 -0.15109073 0.09587613 0.12115525 0.03806026
Suburban (35) 0.49304896 0.49122448 -0.00182448 -0.25107545 0.2229222 0.25055919 0.25108012
Suburban+ Urban 0.49576078 0.49362055 -0.00214023 -0.24594486 0.20416576 0.23560387 0.14439535
Urban (3) 0.50057752 0.49787647 -0.00270105 -0.81367646 0.9674537 0.95406547 0.06185944]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.48704169 0.48498452 -0.00205716 0.05038686 0.00190269 -0.18796051 -0.10984077 -0.27677904
Rural (50) 0.4523103 0.45058401 -0.00172629 0.03956772 0.09262017 -0.18560157 -0.20647732 -0.30195712
Rural+ Suburban 0.48156858 0.47977176 -0.00179681 0.02432285 0.05385014 -0.18482991 -0.09965139 -0.27707622
Suburban (35) 0.49304896 0.49122448 -0.00182448 -0.12483674 0.21674007 -0.18376191 -0.01698638 -0.25387995
Suburban+ Urban 0.49576078 0.49362055 -0.00214023 -0.04150784 0.09486494 -0.18553138 -0.03832022 -0.25055603
Urban (3) 0.50057752 0.49787647 -0.00270105 -0.96752605 0.71654293 0.87362054 -0.04268601 0.77982693]

Table S.OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.24548786 0.22591471 -0.01957315 -0.00728924 0.11745615 0.14560601 0.23210492
Rural (50) 0.29186894 0.26947856 -0.02239039 -0.21384335 0.20368374 0.2901587 0.36611586
Rural+ Suburban 0.27302184 0.2522457 -0.02077614 -0.01068492 0.10829948 0.14194129 0.21227735
Suburban (35) 0.2656266 0.24548385 -0.02014274 0.10349109 -0.02127881 -0.06224234 -0.20006583
Suburban+ Urban 0.23384418 0.21497828 -0.0188659 0.08997327 0.01560394 -0.03101927 -0.0494212
Urban (3) 0.17739213 0.16079416 -0.01659797 0.76657432 -0.4872793 -0.52899281 -0.98176875]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.24548786 0.22591471 -0.01957315 -0.10688173 0.14323205 -0.22863926 0.17876048 -0.28879317
Rural (50) 0.29186894 0.26947856 -0.02239039 0.06172268 0.12062942 -0.32659075 0.1578279 -0.31199507
Rural+ Suburban 0.27302184 0.2522457 -0.02077614 -0.04573197 0.09784321 -0.25669068 0.14734299 -0.30004435
Suburban (35) 0.2656266 0.24548385 -0.02014274 -0.04087494 0.04811207 -0.13733966 0.15887581 -0.24327069
Suburban+ Urban 0.23384418 0.21497828 -0.0188659 -0.14918804 0.15307985 -0.0816753 0.21045284 -0.20438139
Urban (3) 0.17739213 0.16079416 -0.01659797 -0.00240172 -0.85476848 0.25230423 0.9779385 0.4107397 ]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.22728575 0.25895814 0.03167239 0.02398108 -0.20071986 -0.21706483 -0.31686375
Rural (50) 0.14491347 0.18103487 0.0361214 0.30142459 -0.31145644 -0.40460308 -0.45428675
Rural+ Suburban 0.1934486 0.22742214 0.03397354 0.02879076 -0.18813226 -0.21277399 -0.2846007
Suburban (35) 0.21249286 0.24562362 0.03313076 -0.13249692 -0.01838147 0.06256958 0.22092308
Suburban+ Urban 0.2479648 0.27852029 0.03055549 -0.10522973 -0.07854336 0.00870211 -0.00260656
Urban (3) 0.3109702 0.3369515 0.0259813 -0.48730841 0.1439846 0.19173404 0.98503114]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.22728575 0.25895814 0.03167239 0.1524514 -0.19108999 0.24758896 -0.23112378 0.22896624
Rural (50) 0.14491347 0.18103487 0.0361214 -0.12461716 -0.0694449 0.38738232 -0.19887768 0.26053058
Rural+ Suburban 0.1934486 0.22742214 0.03397354 0.0556018 -0.11023398 0.29136306 -0.18051147 0.24702222
Suburban (35) 0.21249286 0.24562362 0.03313076 0.08130722 -0.07766361 0.11595726 -0.19132101 0.1706633
Suburban+ Urban 0.2479648 0.27852029 0.03055549 0.25656419 -0.2518443 0.02245831 -0.27860903 0.10548306
Urban (3) 0.3109702 0.3369515 0.0259813 0.35871074 0.61360651 -0.58066512 -0.98815901 -0.7087677 ]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔiTO Interparty (DeWine 18 → Brown 18)
Type A-IP TO DeWine18 A-IP TO Brown18 ΔA-IP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.24548786 0.25895814 0.01347028 -0.14120998 0.29355293 0.27580072 0.67581322
Rural (50) 0.29186894 0.18103487 -0.11083408 -0.47663074 0.38529111 0.47594922 0.50394736
Rural+ Suburban 0.27302184 0.22742214 -0.0455997 -0.16662339 0.27796446 0.28527623 0.59573662
Suburban (35) 0.2656266 0.24562362 -0.02000298 -0.28989932 0.35920821 0.35757109 0.69077854
Suburban+ Urban 0.23384418 0.27852029 0.04467611 -0.28767974 0.41702636 0.39384392 0.77827407
Urban (3) 0.17739213 0.3369515 0.15955937 -0.54393771 0.80734608 0.77782569 -0.30832064]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO Interparty (DeWine 18 → Brown 18)
Type A-IP TO DeWine18 A-IP TO Brown18 ΔA-IP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.24548786 0.25895814 0.01347028 -0.73084602 0.72130389 0.34426914 0.36043011 0.22387111
Rural (50) 0.29186894 0.18103487 -0.11083408 -0.49135562 0.49156673 0.14395552 0.30954395 0.21116114
Rural+ Suburban 0.27302184 0.22742214 -0.0455997 -0.63100795 0.62691158 0.28916528 0.22560962 0.214734
Suburban (35) 0.2656266 0.24562362 -0.02000298 -0.67634202 0.6951499 0.48873473 0.02784937 0.37229149
Suburban+ Urban 0.23384418 0.27852029 0.04467611 -0.80287676 0.8069638 0.5474547 0.21790334 0.3992229
Urban (3) 0.17739213 0.3369515 0.15955937 -0.99283464 0.41096053 0.9911849 0.32652729 0.9550165 ]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔTO, of Dem18Gb-GOP18Gb and Dem18S-GOP18Gb
Type A-DC TO 18Gb/Gb A-DC TO 18S/Gb ΔA-DC TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All -0.01820211 0.01347028 0.03167239 0.02398108 -0.20071986 -0.21706483 -0.31686375
Rural (50) -0.14695548 -0.11083408 0.0361214 0.30142459 -0.31145644 -0.40460308 -0.45428675
Rural+ Suburban -0.07957324 -0.0455997 0.03397354 0.02879076 -0.18813226 -0.21277399 -0.2846007
Suburban (35) -0.05313374 -0.02000298 0.03313076 -0.13249692 -0.01838147 0.06256958 0.22092308
Suburban+ Urban 0.01412061 0.04467611 0.03055549 -0.10522973 -0.07854336 0.00870211 -0.00260656
Urban (3) 0.13357807 0.15955937 0.0259813 -0.48730841 0.1439846 0.19173404 0.98503114]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO, of Dem18Gb-GOP18Gb and Dem18S-GOP18Gb
Type A-DC TO 18Gb/Gb A-DC TO 18S/Gb ΔA-DC TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All -0.01820211 0.01347028 0.03167239 0.1524514 -0.19108999 0.24758896 -0.23112378 0.22896624
Rural (50) -0.14695548 -0.11083408 0.0361214 -0.12461716 -0.0694449 0.38738232 -0.19887768 0.26053058
Rural+ Suburban -0.07957324 -0.0455997 0.03397354 0.0556018 -0.11023398 0.29136306 -0.18051147 0.24702222
Suburban (35) -0.05313374 -0.02000298 0.03313076 0.08130722 -0.07766361 0.11595726 -0.19132101 0.1706633
Suburban+ Urban 0.01412061 0.04467611 0.03055549 0.25656419 -0.2518443 0.02245831 -0.27860903 0.10548306
Urban (3) 0.13357807 0.15955937 0.0259813 0.35871074 0.61360651 -0.58066512 -0.98815901 -0.7087677 ]

-----------------------------------------------


We can see that (1) Brown Senatorial 2018 (Dem) saw higher actual party turnout in all regions compared to Cordray Gubernatorial 2018 (Dem), though his lead was stronger in rural, suburban, and urban (in that order). (2) Overall, it appears that turnout trend was weakly negatively correlated with poverty, and moderately negatively correlated with inequality (that is, the more poverty, the less strong was Brown’s advantage). However, this was due to rural counties; in suburban and urban counties, the pattern flips, though weakly. (3) Brown Senatorial 2018 (Dem) saw higher actual party turnout than DeWine Gubernatorial 2018 (GOP), though with lower turnout in rural (-11.1 pp) and suburban (-2.0 pp) counties, and leading in urban (16.0 pp) counties. This is generally per expectations for a Dem vs GOP. However, there’s also the expected pattern of Brown doing better with the higher poverty and more unequal counties, and worse in higher-income (though the latter is a weak correlation).


There’s a notable comparison for Sherrod Brown: in 2022, DeWine got 2,580,424 votes, but in 2018, Brown only got 2,358,508 votes, 221,916 short of DeWine. Not as bad as Vance was short of DeWine in 2022, but still, hundreds of thousands of votes. What are the patterns of 2022 Gubernatorial → 2018 Senatorial?


-----------------------------------------------


Table S.OH.Total.Income - Correlation of county income metrics with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.45193782 0.48498452 0.0330467 0.01650437 0.21096115 0.14405255 0.56768203
Rural (50) 0.43924671 0.45058401 0.0113373 -0.11821816 0.20305277 0.2107508 0.32724619
Rural+ Suburban 0.45954363 0.47977176 0.02022814 0.013165 0.18351485 0.13550033 0.44013013
Suburban (35) 0.46732874 0.49122448 0.02389574 -0.07410827 0.28685421 0.22910082 0.54486306
Suburban+ Urban 0.45506161 0.49362055 0.03855895 -0.11570579 0.36035823 0.29253637 0.70404276
Urban (3) 0.4330896 0.49787647 0.06478686 -0.88841344 0.99368965 0.98709264 0.20388778]

Table S.OH.Total.Racial - Correlation of county racial composition with county ΔActualTotal
Type A-Total TO 18 A-Total TO 24 ΔA-Total TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.45193782 0.48498452 0.0330467 -0.64359426 0.64412331 0.15713233 0.46043514 0.21821963
Rural (50) 0.43924671 0.45058401 0.0113373 -0.16917089 0.18474386 0.02256113 0.12743796 0.36654123
Rural+ Suburban 0.45954363 0.47977176 0.02022814 -0.48440255 0.47966771 0.07333465 0.35196137 0.21383959
Suburban (35) 0.46732874 0.49122448 0.02389574 -0.59736946 0.57443075 0.10686931 0.37839083 0.07112995
Suburban+ Urban 0.45506161 0.49362055 0.03855895 -0.75657237 0.74567571 0.24432546 0.47131758 0.15080884
Urban (3) 0.4330896 0.49787647 0.06478686 -0.9214608 0.80889018 0.79509559 -0.18505469 0.68234142]

Table S.OH.GOP.Income - Correlation of county income metrics with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.28142325 0.22591471 -0.05550854 0.18972949 -0.00180791 0.00528891 0.29294188
Rural (50) 0.33773005 0.26947856 -0.06825149 -0.09399018 0.14542682 0.21482378 0.32188246
Rural+ Suburban 0.31628694 0.2522457 -0.06404124 0.18909824 -0.02727585 -0.00579713 0.20293494
Suburban (35) 0.30806219 0.24548385 -0.06257834 0.39018953 -0.25121052 -0.28829009 -0.04537097
Suburban+ Urban 0.26756391 0.21497828 -0.05258563 0.33295367 -0.14963223 -0.19981534 0.19964922
Urban (3) 0.19502635 0.16079416 -0.03423219 0.92399096 -0.9996111 -0.99708815 -0.28579766]

Table S.OH.GOP.Racial - Correlation of county racial composition with county ΔActualGOP
Type A-GOP TO 18 A-GOP TO 24 ΔA-GOP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.28142325 0.22591471 -0.05550854 -0.20642304 0.22153571 -0.25952067 0.40827478 -0.17406717
Rural (50) 0.33773005 0.26947856 -0.06825149 0.2190318 -0.06216193 -0.34411228 0.15314077 -0.09384197
Rural+ Suburban 0.31628694 0.2522457 -0.06404124 -0.04513382 0.0538056 -0.32265255 0.34965914 -0.19530673
Suburban (35) 0.30806219 0.24548385 -0.06257834 -0.12752887 0.04845127 -0.30995947 0.52327587 -0.3842212
Suburban+ Urban 0.26756391 0.21497828 -0.05258563 -0.34027115 0.29771244 -0.18576776 0.57394133 -0.30048981
Urban (3) 0.19502635 0.16079416 -0.03423219 0.88538103 -0.85563336 -0.74106422 0.26734697 -0.61819916]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.16855701 0.25895814 0.09040113 -0.22108609 0.19902076 0.13011718 0.1657126
Rural (50) 0.09856961 0.18103487 0.08246526 0.0348956 -0.03716249 -0.11455895 -0.16259286
Rural+ Suburban 0.14100265 0.22742214 0.08641949 -0.22025808 0.1839227 0.12088044 0.11277629
Suburban (35) 0.15727834 0.24562362 0.08834528 -0.51860083 0.54943222 0.54031923 0.53625168
Suburban+ Urban 0.1857837 0.27852029 0.0927366 -0.5266351 0.5685878 0.55614619 0.51219575
Urban (3) 0.23684042 0.3369515 0.10011108 -0.89786607 0.99582776 0.99024085 0.22442078]

Table S.OH.Dem.Racial - Correlation of county racial composition with county ΔActualDem
Type A-Dem TO 18 A-Dem TO 24 ΔA-Dem TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.16855701 0.25895814 0.09040113 -0.34193048 0.32716489 0.45425472 -0.08139197 0.41538455
Rural (50) 0.09856961 0.18103487 0.08246526 -0.38414613 0.20485021 0.4311767 -0.11648688 0.38420669
Rural+ Suburban 0.14100265 0.22742214 0.08641949 -0.33502283 0.3247013 0.4363228 -0.14276601 0.40920084
Suburban (35) 0.15727834 0.24562362 0.08834528 -0.38360182 0.45877115 0.43829387 -0.27000051 0.49667323
Suburban+ Urban 0.1857837 0.27852029 0.0927366 -0.40126006 0.44387456 0.47378177 -0.18907924 0.51509
Urban (3) 0.23684042 0.3369515 0.10011108 -0.91309171 0.82107013 0.78217189 -0.20567033 0.66682466]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔiTO Interparty (DeWine 22 → Brown 18)
Type A-IP TO DeWine22 A-IP TO Brown18 ΔA-IP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.28142325 0.25895814 -0.02246511 -0.09521563 0.28326225 0.26398271 0.7293873
Rural (50) 0.33773005 0.18103487 -0.15669518 -0.49413167 0.41619482 0.51899837 0.56963478
Rural+ Suburban 0.31628694 0.22742214 -0.0888648 -0.12035147 0.26995585 0.27930771 0.65448819
Suburban (35) 0.30806219 0.24562362 -0.06243857 -0.21355828 0.31988506 0.3118507 0.75805635
Suburban+ Urban 0.26756391 0.27852029 0.01095638 -0.220049 0.38403497 0.35288936 0.83363269
Urban (3) 0.19502635 0.3369515 0.14192515 -0.49107414 0.76936106 0.73752529 -0.36648414]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO Interparty (DeWine 22 → Brown 18)
Type A-IP TO DeWine22 A-IP TO Brown18 ΔA-IP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.28142325 0.25895814 -0.02246511 -0.7774535 0.76754056 0.30910214 0.44495192 0.21507128
Rural (50) 0.33773005 0.18103487 -0.15669518 -0.45553925 0.47584699 0.09495216 0.34658597 0.24083755
Rural+ Suburban 0.31628694 0.22742214 -0.0888648 -0.66872239 0.66012645 0.245968 0.3157665 0.20922256
Suburban (35) 0.30806219 0.24562362 -0.06243857 -0.76620775 0.76311228 0.45602231 0.16355476 0.31969478
Suburban+ Urban 0.26756391 0.27852029 0.01095638 -0.87476796 0.86874176 0.51488753 0.33482165 0.3491145
Urban (3) 0.19502635 0.3369515 0.14192515 -0.98355909 0.35387145 0.99747515 0.38427962 0.97150871]

Table S.OH.Dem.Income - Correlation of county income metrics with county ΔTO, of Dem22Gb-GOP22Gb and Dem18S-GOP22Gb
Type A-DC TO 22Gb/Gb A-DC TO 18S/22Gb ΔA-DC TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All -0.11286624 -0.02246511 0.09040113 -0.22108609 0.19902076 0.13011718 0.1657126
Rural (50) -0.23916044 -0.15669518 0.08246526 0.0348956 -0.03716249 -0.11455895 -0.16259286
Rural+ Suburban -0.17528429 -0.0888648 0.08641949 -0.22025808 0.1839227 0.12088044 0.11277629
Suburban (35) -0.15078385 -0.06243857 0.08834528 -0.51860083 0.54943222 0.54031923 0.53625168
Suburban+ Urban -0.08178022 0.01095638 0.0927366 -0.5266351 0.5685878 0.55614619 0.51219575
Urban (3) 0.04181406 0.14192515 0.10011108 -0.89786607 0.99582776 0.99024085 0.22442078]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO, of Dem22Gb-GOP22Gb and Dem18S-GOP22Gb
Type A-DC TO 22Gb/22Gb A-DC TO 18S/22Gb ΔA-DC TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All -0.11286624 -0.02246511 0.09040113 -0.34193048 0.32716489 0.45425472 -0.08139197 0.41538455
Rural (50) -0.23916044 -0.15669518 0.08246526 -0.38414613 0.20485021 0.4311767 -0.11648688 0.38420669
Rural+ Suburban -0.17528429 -0.0888648 0.08641949 -0.33502283 0.3247013 0.4363228 -0.14276601 0.40920084
Suburban (35) -0.15078385 -0.06243857 0.08834528 -0.38360182 0.45877115 0.43829387 -0.27000051 0.49667323
Suburban+ Urban -0.08178022 0.01095638 0.0927366 -0.40126006 0.44387456 0.47378177 -0.18907924 0.51509
Urban (3) 0.04181406 0.14192515 0.10011108 -0.91309171 0.82107013 0.78217189 -0.20567033 0.66682466]

-----------------------------------------------


Here, Brown Senatorial 2018 (Dem) does better than Whaley Gubernatorial 2022 (Dem) in terms of actual party turnout by 9.0 pp - this time, with the lead in urban, suburban, and rural counties, in that order (opposite of 2018 Gubernatorial → 2018 Senatorial). His advantage is better in higher-poverty counties and higher-inequality counties, and worse in higher-income counties, though weakly so for all. This is due to the opposite (but weakly correlated) trend in rural counties; for suburban and urban counties, the overall pattern (better in higher poverty/unequal counties, worse in higher-income) is moderately correlated in suburban counties (and suburban+urban), and strongly correlated in urban counties (except for inequality for urban, which is weakly correlated).


Comparing Brown Senatorial 2018 (Dem) vs DeWine Gubernatorial 2022 (GOP), Brown’s turnout is overall down 2.2 pp, with a dip in rural (-15.7 pp) and suburban (-6.2 pp) counties, and up in urban (+14.2 pp) counties. The expected better performance in higher-poverty/unequal counties, and worse in higher-income counties, holds in all cases (except in urban, where Brown does moderately negatively worse with inequality).


Still, despite being down compared to DeWine 2022, it’s worth noting that DeWine Gubernatorial 2022 (GOP) actual turnout is 28.1%, Whaley Gubernatorial 2022 (Dem) actual turnout is 16.9%, and Brown Senatorial 2018 (Dem) actual turnout was 25.9% (among those who voted, the 2022 Gubernatorial race was 62.41% DeWine vs Whaley 37.38%, Δ = 25.03 pp). Brown 2018 gives a far more competitive comparison than Whaley 2022.


In fact, Brown and DeWine had gone head to head before, with Brown unseating then incumbent Senator DeWine in 2006, with 2,257,369 votes vs DeWine’s 1,761,037.


Table S.OH.Dem.Income - Correlation of county income metrics with county ΔiTO Interparty (DeWine 06 → Brown 06)
Type A-IP TO DeWine06 A-IP TO Brown06 ΔA-IP TO Corr(Median Income) Corr(Poverty) Corr(Poverty150) Corr(Gini)
All 0.19649129 0.25187054 0.05537925 -0.47139813 0.51823419 0.49289764 0.45623001
Rural (50) 0.21426627 0.22586843 0.01160216 -0.57387914 0.50610892 0.52093243 0.45367953
Rural+ Suburban 0.2064758 0.2450162 0.0385404 -0.47983751 0.51259268 0.49381619 0.44938224
Suburban (35) 0.20329863 0.2528252 0.04952657 -0.64135346 0.6364718 0.64369405 0.41530465
Suburban+ Urban 0.19181945 0.25870475 0.0668853 -0.6384075 0.64419389 0.64802046 0.40536836
Urban (3) 0.17101116 0.26936259 0.09835143 -0.77159919 0.94767403 0.93110382 -0.0071984 ]

Table S.OH.Interparty.Racial - Correlation of county racial composition with county ΔTO Interparty (DeWine 06 → Brown 06)
Type A-IP TO DeWine06 A-IP TO Brown06 ΔA-IP TO Corr(white) Corr(black) Corr(Latino) Corr(Asian+) Corr(AIAN)
All 0.19649129 0.25187054 0.05537925 -0.31037325 0.36876959 0.13327462 -0.11439182 0.1783924
Rural (50) 0.21426627 0.22586843 0.01160216 -0.31033809 0.40843034 -0.02498857 0.0319373 0.18688102
Rural+ Suburban 0.2064758 0.2450162 0.0385404 -0.27069112 0.36695426 0.09353269 -0.19501106 0.16195724
Suburban (35) 0.20329863 0.2528252 0.04952657 -0.19684486 0.34882806 0.2236264 -0.45348602 0.18303201
Suburban+ Urban 0.19181945 0.25870475 0.0668853 -0.26902469 0.36209484 0.28144006 -0.34594228 0.22765632
Urban (3) 0.17101116 0.26936259 0.09835143 -0.98266921 0.66667298 0.90513244 0.02639537 0.82118629]

We can see Brown had 5.5 pp higher turnout than DeWine (25.2% vs 19.6%; among those who voted, 56.16% Brown vs 43.82% DeWine, Δ = 12.36 pp), with a lead in urban, suburban, and rural counties, in that order. In all county types, he did better with higher poverty and higher inequality counties (except inequality for urban), and worse in higher-income counties.


It’s worth noting that comparing DeWine 18/22 vs Brown 18, and DeWine 06 vs Brown 06, the gap between rural and urban has widened. While in 2006, Brown’s lead in urban was by 9.8 pp and in rural was 1.2 pp (a gap of 8.6 pp), in 2018 DeWine → 2018 Brown, Brown’s urban lead was 16.0 pp, and DeWine’s rural lead was 11.1 pp (a gap of 27.1 pp), and 2022 DeWine → 2018 Brown, Brown’s urban lead was 14.2 pp and DeWine’s rural lead was 15.7 pp (a gap of 29.9 pp).


Further, Brown’s advantage with higher-poverty counties diminishes over 06 → 06, 18 → 18, and 22 → 18, his advantage with higher inequality counties increases, and his disadvantage with higher-income counties diminishes. Brown’s disadvantage with whiter counties increases; his advantage with blacker counties skyrockets in 22 → 18, his disadvantage with more Asian reverses to advantage, and his advantage with Latino increases (but decreases slightly from 18 → 18 to 22 → 18), and his advantage with American Indian counties barely changes.


Image 2
Figurex 1
Image 2
Figurex 1
Image 2
Figurex 1

Figure OHBC (OH.BrownCompare). Turnout trends, by county poverty-level, for Brown Senatorial {2018, 2006} vs DeWine Gubernatorial {2018, 2022} and DeWine Senatorial {2006}


Overall, Brown won among actual voters in 2006 by 12.36 pp (56.16% vs 43.82%), in 2012 by 6.0 pp (50.70% vs 44.70%), in 2018 by 6.82 pp (53.40% vs 46.58%), and lost in 2024 by -3.62 pp (46.47% vs 50.09%). I haven’t really analyzed 2012 and 2024 however, as these are Presidential election years (except a bit at the start for 2024), so the turnout is driven to a different level, due to the Presidential election.



Overall, given the trends at the beginning, and Brown’s performance (likely brought down primarily by Harris driving down 2024 turnout, given he greatly outperformed her in the 2024 election), indicates Ohio isn’t simply a "GOP state". It was left behind as the Dems hewed closer and closer to big capital.