Hesitation Media

Reading Tuesday's county counts against a tied race

Mon, 02 Nov 2026 00:00:00 GMT

Kent County, Michigan, which includes Grand Rapids, gave Gretchen Whitmer a 10.2-point margin in the 2022 governor’s race, within a tenth of a point of her statewide margin. If every Michigan county moves by the same number of points between that race and Tuesday’s Senate race, a Kent County result near an even split would mean a tied statewide race. Between 2018 and 2022, however, Kent swung 5.5 points further toward the Democrat than Michigan did. If it does that again, a tied Michigan race would show Kent at about 5.5 points for the Democrat.

Both numbers can be computed for every county in the 2026 battlegrounds. The first is the county’s tie pace: its 2022 margin minus the state’s 2022 margin, the margin the county would post in a tied race if every county moved alike. The second adds the county’s own gap from its state’s swing between the 2018 and 2022 midterms. The table below gives both for the four counties that cast the most votes in each of the 12 Senate battlegrounds analyzed in this series and in Pennsylvania’s governor’s race. The full sheet, every county in all 13 races, is data/baselines/tiepace_2026.csv.

Large counties and the gap

Of the 48 counties in the table with a 2018-to-2022 comparison, 39 swung further toward the Democrat than their state did, by 4.6 points on average, and the other nine swung less, by 2.1. Across all 48 the average gap was 3.3 points toward the Democrat. Hennepin County, Minnesota, ran 8.9 points ahead of its state’s swing and Johnson County, Iowa, 10.5. The first piece in this series found the same tendency across all counties, larger ones swinging further toward the Democrat in 13 of 14 comparisons. If 2026 follows it, a large county that reports at its tie pace would point to a statewide result better for the Republican than a tie, by roughly the county’s own gap, which averaged 3.3 points in this table. The gap is drawn from one cycle and from each state’s governor’s race (Ohio’s Senate race for Ohio), so it describes 2018 to 2022 and is not a forecast range.

Governor’s races as benchmarks

In Maine, Michigan, Minnesota, Nebraska and Texas, the 2022 benchmark is the governor’s race, because none of the five held a Senate race that year. Whitmer, Tim Walz and Janet Mills ran as incumbents in 2022, and the Democrats running against Greg Abbott and Jim Pillen lost by 10.9 and 23.4 points. Seven of the states here held both a Senate and a governor’s race on the 2022 ballot. In six of them, each county’s gap between its Senate margin and its governor’s margin differed from the statewide gap by a vote-weighted root-mean-square of 1.8 to 3.9 points. In Kansas, where Laura Kelly won the governorship by 2.2 points and Jerry Moran won his Senate seat by 23.0, it was 7.3. Maine, Michigan, Minnesota, Nebraska and Texas held no Senate race in 2022, so the same measure cannot be taken for them; in the seven states where it can, a county’s Senate and governor’s margins on one ballot strayed by those amounts.

In Georgia, Iowa, Kansas, New Hampshire, North Carolina, Ohio and South Carolina, the benchmark is the 2022 Senate race. For Georgia it is the November general, in which Raphael Warnock led Herschel Walker by 1.0 point, not the December runoff. North Carolina held no statewide race in 2018, so its counties have no gap.

Partial counts

Tie pace describes a county’s final count. A partial count can sit well away from where the county will finish if the ballots counted first lean differently from the rest, so the comparison with tie pace applies to a county’s complete count.

Tie pace for the largest counties

Race (2022 benchmark)	County	Share of 2022 vote	2022 margin	Tie pace	2018-22 gap	Tie pace if gap repeats
Georgia Senate (Senate general 2022, state +1.0)	Fulton	10.6%	+49.0	+48.0	-1.2	+46.8
	Cobb	7.9%	+16.3	+15.3	+1.1	+16.4
	Gwinnett	7.6%	+20.4	+19.4	+2.2	+21.6
	Dekalb	7.6%	+70.2	+69.2	+1.1	+70.4
Iowa Senate (Senate 2022, state -12.2)	Polk	15.7%	+12.7	+24.8	+5.2	+30.1
	Linn	7.7%	+9.0	+21.2	+6.7	+27.9
	Johnson	5.2%	+42.6	+54.8	+10.5	+65.3
	Scott	5.2%	-3.6	+8.6	+1.2	+9.8
Kansas Senate (Senate 2022, state -23.0)	Johnson	26.1%	+1.3	+24.3	+6.3	+30.6
	Sedgwick	15.3%	-20.8	+2.2	-0.1	+2.1
	Shawnee	6.5%	-11.4	+11.6	+0.7	+12.3
	Douglas	4.7%	+35.9	+58.8	+2.4	+61.2
Maine Senate (governor 2022, state +13.1)	Cumberland	24.0%	+40.2	+27.1	+8.1	+35.2
	York	15.5%	+18.3	+5.2	+6.0	+11.2
	Penobscot	10.2%	-5.3	-18.4	-6.9	-25.3
	Kennebec	8.9%	+4.1	-9.0	-3.4	-12.3
Michigan Senate (governor 2022, state +10.2)	Oakland	14.2%	+23.1	+12.9	+5.5	+18.4
	Wayne	13.7%	+42.7	+32.5	-0.9	+31.6
	Macomb	8.7%	+5.2	-5.0	+1.1	-3.9
	Kent	6.8%	+10.2	+0.1	+5.5	+5.5
Minnesota Senate (governor 2022, state +7.7)	Hennepin	22.9%	+43.5	+35.9	+8.9	+44.8
	Ramsey	8.7%	+46.4	+38.7	+6.3	+45.0
	Dakota	8.0%	+14.3	+6.7	+6.4	+13.1
	Anoka	6.2%	-1.3	-9.0	+2.2	-6.7
Nebraska Senate (governor 2022, state -23.4)	Douglas	28.6%	+5.0	+28.4	+4.5	+32.9
	Lancaster	17.1%	+4.1	+27.5	-2.0	+25.6
	Sarpy	9.7%	-15.2	+8.3	+8.5	+16.7
	Buffalo	2.4%	-45.8	-22.4	-3.6	-26.0
New Hampshire Senate (Senate 2022, state +9.1)	Hillsborough	27.8%	+8.7	-0.4	+0.8	+0.4
	Rockingham	24.9%	+3.1	-6.0	+1.7	-4.2
	Merrimack	11.4%	+13.3	+4.2	+2.0	+6.2
	Strafford	9.1%	+19.5	+10.4	+2.3	+12.8
Ohio Senate (Senate 2022, state -6.1)	Franklin	10.3%	+32.5	+38.6	+8.0	+46.7
	Cuyahoga	9.9%	+35.6	+41.8	+3.6	+45.3
	Hamilton	7.3%	+15.4	+21.6	+9.6	+31.2
	Summit	4.8%	+13.2	+19.3	+4.3	+23.7
South Carolina Senate (Senate 2022, state -25.9)	Greenville	10.5%	-33.2	-7.3	+2.2	-5.1
	Charleston	9.0%	-4.9	+20.9	+6.8	+27.8
	Horry	7.8%	-45.6	-19.7	-0.1	-19.8
	Richland	7.4%	+25.5	+51.3	+2.8	+54.1
Texas Senate (governor 2022, state -10.9)	Harris	13.6%	+9.6	+20.4	+1.4	+21.9
	Dallas	7.7%	+26.8	+37.7	+4.8	+42.6
	Tarrant	7.3%	-4.1	+6.8	+4.1	+10.9
	Bexar	6.7%	+16.4	+27.3	+7.7	+35.1
North Carolina Senate (Senate 2022, state -3.2)	Wake	12.0%	+26.6	+29.8
	Mecklenburg	9.5%	+32.5	+35.7
	Guilford	5.0%	+18.2	+21.5
	Forsyth	3.6%	+11.3	+14.5
Pennsylvania governor (governor 2022, state +14.8)	Allegheny	10.7%	+39.0	+24.2	+5.1	+29.4
	Philadelphia	9.3%	+72.6	+57.8	-0.8	+57.0
	Montgomery	7.7%	+39.8	+25.0	+6.4	+31.5
	Bucks	5.9%	+19.9	+5.1	+3.7	+8.8

Margins are the Democrat’s share minus the Republican’s, in points of all votes cast. Tie pace is the county’s 2022 margin minus the statewide 2022 margin shown in the first column. The 2018-22 gap is the county’s swing minus its state’s swing between the two midterms in the same office.

Three counties in the table sit near an even split at tie pace: Kent in Michigan at 0.1, Hillsborough in New Hampshire at -0.4 and Sedgwick in Kansas at 2.2. Sedgwick, which includes Wichita, voted for Moran by 20.8 points in 2022, while he won the state by 23.0, and between 2018 and 2022 it moved with its state to within a tenth of a point. A Sedgwick count close to even on Tuesday would, on 2022’s arithmetic, describe a tied Kansas Senate race.

Notes and sources

File names in this section refer to the project’s working data files, which are not yet public. The data and code are available from the author on request.

Passage	Figures	File
Kent County lede	2022 governor margin 10.24, state 10.19, tie pace +0.05; 2018-22 gap +5.47; tie pace if gap repeats +5.52	`data/baselines/tiepace_2026.csv` (MI, SEN)
Large counties and the gap	48 of the 52 table counties have a gap (North Carolina’s four do not); 39 positive, averaging +4.57; nine negative, averaging -2.09; mean of all 48 +3.32; Hennepin +8.91; Johnson, Iowa +10.51	`tiepace_2026.csv`
Earlier finding	13 of 14 comparisons with a positive size-swing correlation	piece 1; `cycle_change.csv`
Senate against governor, 2022	vote-weighted RMS of county (Senate minus governor) gap around the state gap: NH 1.81, IA 1.89, PA 2.38, GA 2.43, SC 3.78, OH 3.88, KS 7.34; Kansas governor +2.21, Senate -22.96	`data/baselines/senate_vs_governor_2022.csv`; `data/clean/statewide_results.csv`
Incumbents and 2022 margins	Texas governor -10.88, Nebraska governor -23.45; Georgia Senate general +0.95 (runoff +2.80)	`statewide_results.csv`
Close-to-even counties	Kent +0.05; Hillsborough 8.65 against 9.06, -0.41; Sedgwick -20.79 against -22.96, +2.17, gap -0.05	`tiepace_2026.csv`
Table	generated, not typed	`data/baselines/tiepace_2026_top.md` from `pipeline/tiepace.py`

Alaska is excluded because its returns here are by state house district, which were renumbered in 2021. Maine’s 2018 Senate race (Angus King, an independent) and Alaska’s 2022 ranked-choice races are excluded from all margin comparisons; see data/MANIFEST.md.

What 2018 and 2022 say to watch on Nov. 3, county by county

Mon, 02 Nov 2026 00:00:00 GMT

Tim Walz’s margin in the Minnesota governor’s race fell from 11.4 points in 2018 to 7.7 in 2022. In 78 of the state’s 87 counties it fell by more than it did statewide, and in the median county it fell 13.1 points. Hennepin County, which cast 575,529 votes for governor in 2022, 23 percent of the state’s total, moved 5.2 points in his direction.

The MIT Election Data and Science Lab publishes precinct returns for both midterms. Set against each other for the 13 states with 2026 Senate battlegrounds and Pennsylvania, they show a similar pattern. Each county’s 2022 vote total, on a log scale, was compared with its swing between the two midterms in 14 comparisons: the governor’s races in 12 states plus the Ohio and Pennsylvania Senate races. In 13 of the 14, larger counties tended to swing further toward the Democrat. The correlation ran from 0.85 in South Carolina and 0.79 in Minnesota down to 0.14 in the Ohio governor’s race, and in Nebraska it was slightly negative, at -0.05.

Every one of these comparisons pairs different candidates, so a county’s swing mixes the national climate with whatever the 2018 and 2022 nominees brought to it. Ohio and Pennsylvania held Senate races in both years, for different seats, and the other 12 comparisons are governor’s races. North Carolina is absent from the 14 because it held no statewide race in 2018, and Alaska is left out because its 2022 governor’s race had two Republicans in the first-round count and its house districts, the only sub-state unit in its returns, were renumbered in 2021.

Texas

Greg Abbott beat Lupe Valdez by 13.3 points in 2018 and Beto O’Rourke by 10.9 in 2022, a swing of 2.4 points toward the Democrat. Of Texas’ 254 counties, 225 swung less toward O’Rourke than the state did or swung away from him, and the median county moved 3.6 points toward Abbott. In Bexar County, which cast 541,071 votes, the margin went from 6.3 points for Valdez to 16.5 for O’Rourke, and the county’s net Democratic vote grew by 54,663, or 24 percent of the statewide change. Tarrant, Travis and Harris supplied another 53 percent between them.

Ohio and Pennsylvania Senate races

Sherrod Brown won Ohio’s 2018 Senate race by 6.8 points. Tim Ryan lost the 2022 race to J.D. Vance by 6.1. None of Ohio’s 88 counties moved toward the Democrat between the two races. Cuyahoga County cast 483,310 votes in 2018 and 408,543 in 2022, a drop of 15.5 percent, and the Democratic margin there fell from 45.0 points to 35.6. The county’s net Democratic vote shrank by 71,802, which is 13 percent of the statewide shift of 553,574 net votes and the largest share of any county. In Mahoning County, which includes Youngstown, the margin went from 21.0 points for Brown to 3.5 points for Vance.

Bob Casey won Pennsylvania’s 2018 Senate race by 13.1 points, and John Fetterman won the 2022 race by 4.9. Statewide, votes cast rose 7.1 percent, from 5,009,890 to 5,366,638. Philadelphia’s fell 10.1 percent, from 554,507 to 498,531, and its margin fell from 74.8 points to 67.0. Philadelphia accounts for 80,954 of the 394,240 net votes the Democratic margin lost between the two races, about a fifth. Cumberland was the one county of 67 that moved toward the Democrat, by 1 point.

Pennsylvania elects only a governor on Tuesday. In the governor’s race, Philadelphia gave Tom Wolf a 75.7-point margin in 2018 and Josh Shapiro a 72.6-point margin in 2022, and its share of the statewide vote fell from 11.1 percent to 9.3.

Michigan

Gretchen Whitmer won the governorship by 9.6 points in 2018 and by 10.2 in 2022, and her statewide net vote grew by 44,884. In Oakland County her margin went from 17.0 points to 23.1, adding 42,784 to her net. Wayne County, which includes Detroit, cast 53,715 fewer votes than in 2018 and subtracted 25,033. The state’s other 81 counties added 27,133 between them. MEDSL’s 2022 Michigan file sums to 4,431,811 votes for governor and gives Whitmer 54.24 percent, against 54.5 percent in RealClearPolitics’ final result, so the Michigan figures here may run slightly short.

Minnesota precincts

Of 4,074 Minnesota precincts in 2022, 3,776 carry the same county and name as a 2018 precinct, and together they cast 87 percent of the 2022 vote for governor. Among the 476 of those precincts in Hennepin and Ramsey counties, the median swing was 5.2 points toward Walz; weighted by votes, the swing was 3.6 points, and 27 percent of the precincts moved toward Scott Jensen. Among the other 3,268 precincts with at least 10 votes, the median swing was 11.8 points away from Walz, and 84 percent moved toward Jensen.

A matching name is not proof of an unchanged boundary, and these precinct figures have not yet been checked against precinct maps.

Counties to watch

The table lists, for each 2026 Senate battleground, the two counties that cast the most votes in 2022 and their margins in the closest comparable race of each cycle. Where the same office was on the ballot in both years, it also names the county whose change in net votes moved the statewide result most. The full sheet, every county in all 14 states with its share of the statewide vote, is data/baselines/benchmarks_2026.csv. The community type is the American Communities Project’s 2023 classification of each county. Alaska is left out for the reasons above.

State	County	Community type	2018 margin	2022 margin	Share of 2022 vote	Why it is listed
Georgia (governor 2018, Senate 2022)	Fulton	Big Cities	+45.7	+49.0	10.6%	most votes cast, 2022
	Cobb	Urban Suburbs	+9.6	+16.3	7.9%	most votes cast, 2022
Iowa (governor 2018, Senate 2022)	Polk	Urban Suburbs	+18.3	+12.7	15.7%	most votes cast, 2022
	Linn	Middle Suburbs	+14.0	+9.0	7.7%	most votes cast, 2022
Kansas (governor 2018, Senate 2022)	Johnson	Exurbs	+16.7	+1.3	26.1%	most votes cast, 2022
	Sedgwick	Big Cities	+6.0	-20.8	15.3%	most votes cast, 2022
Maine (governor 2018, governor 2022)	Cumberland	Exurbs	+26.4	+40.2	24.0%	most votes cast, 2022; moved the statewide net most (+24,430 votes)
	York	Graying America	+6.7	+18.3	15.5%	most votes cast, 2022
Michigan (Senate 2018, governor 2022)	Oakland	Urban Suburbs	+13.7	+23.1	14.2%	most votes cast, 2022
	Wayne	Big Cities	+40.1	+42.7	13.7%	most votes cast, 2022
Minnesota (Senate 2018, governor 2022)	Hennepin	Big Cities	+49.3	+43.5	22.9%	most votes cast, 2022
	Ramsey	Big Cities	+52.7	+46.4	8.7%	most votes cast, 2022
Nebraska (Senate 2018, governor 2022)	Douglas	Big Cities	+4.0	+5.0	28.6%	most votes cast, 2022
	Lancaster	College Towns	+11.7	+4.1	17.1%	most votes cast, 2022
New Hampshire (governor 2018, Senate 2022)	Hillsborough	Exurbs	-9.7	+8.7	27.8%	most votes cast, 2022
	Rockingham	Exurbs	-14.2	+3.1	24.9%	most votes cast, 2022
Ohio (Senate 2018, Senate 2022)	Franklin	Big Cities	+37.4	+32.5	10.3%	most votes cast, 2022
	Cuyahoga	Big Cities	+45.0	+35.6	9.9%	most votes cast, 2022; moved the statewide net most (-71,802 votes)
South Carolina (governor 2018, Senate 2022)	Greenville	Urban Suburbs	-16.0	-33.2	10.5%	most votes cast, 2022
	Charleston	College Towns	+14.9	-4.9	9.0%	most votes cast, 2022
Texas (Senate 2018, governor 2022)	Harris	Big Cities	+16.7	+9.6	13.6%	most votes cast, 2022
	Dallas	Big Cities	+33.0	+26.8	7.7%	most votes cast, 2022
North Carolina (no 2018 statewide race; Senate 2022)	Wake	Big Cities		+26.6	12.0%	most votes cast, 2022
	Mecklenburg	Big Cities		+32.5	9.5%	most votes cast, 2022

Margins are the Democrat’s share minus the Republican’s, in points of all votes cast in the race. In Michigan, Minnesota, Nebraska and Texas the 2018 column is the state’s Senate race that year and the 2022 column is the governor’s race, because none of the four held a Senate race in 2022. In Georgia, Iowa, Kansas, New Hampshire and South Carolina it runs the other way: none held a Senate race in 2018.

Polls and forecasts

FiveThirtyEight’s election-eve forecasts in 2018 missed the final margin by 3.1 points on average across 20 Senate and governor’s races in these states. The Democrat beat the forecast in 12 races and fell short in eight, and the signed misses averaged out to -0.04 points. Where the forecasts overstated the Democrat, the largest misses came in Michigan, Ohio and Iowa. FiveThirtyEight forecast a 14.3-point win for Debbie Stabenow in Michigan’s Senate race; she won by 6.5. It forecast 13.2 points for Brown in Ohio; he won by 6.8. It forecast a 1.9-point win for Richard Cordray in Ohio’s governor’s race, which Mike DeWine won by 3.7, and a 2.0-point win for Fred Hubbell in Iowa’s, which Kim Reynolds won by 2.7. In the governor’s races in Texas, Kansas and South Carolina, the Democrats beat their forecasts by between 5.1 and 7.3 points.

In 2022 the polls overstated the Republican. RealClearPolitics’ final averages understated the Democrat in 10 of the 13 races where the average contained polls, and across all 13 the average miss was 2.9 points in the Democrat’s favor. Its average had Whitmer ahead by 1.0 point; she won by 10.2. It had Maggie Hassan ahead by 1.4 in New Hampshire’s Senate race; she won by 9.1. It had Mehmet Oz ahead of Fetterman by 0.4; Fetterman won by 4.9. FiveThirtyEight’s forecasts, which combine polls with other inputs, understated the Democrat in 14 of 20 races, and across all 20 the average miss was 1.1 points in the Democrat’s favor. Of the six races that broke toward the Republican, the largest misses came in two governor’s races: DeWine beat FiveThirtyEight’s forecast by 3.6 points in Ohio, and Jim Pillen beat it by 6.4 in Nebraska.

Before the 2022 vote, RealClearPolitics’ average had Walz ahead by 4.3 points, and FiveThirtyEight forecast an 8.9-point win. He won by 7.7, with Hennepin County adding 10,130 to his net vote while casting 51,133 fewer votes for governor than it had in 2018, and with the state’s other 86 counties, taken together, subtracting 113,113.

Notes and sources

File names in this section refer to the project’s working data files, which are not yet public. The data and code are available from the author on request.

All results come from the MIT Election Data and Science Lab’s precinct returns: Precinct-Level Returns 2018 by Individual State (doi:10.7910/DVN/NVQYMG) and Precinct-Level Returns 2022 by Individual State (doi:10.7910/DVN/UYQIEP), both CC0, aggregated to counties by pipeline/clean.py. The comparisons, correlations and benchmarks are the author’s analysis of those returns. Margins count every vote cast in the race, third-party votes included. File-level hashes, coverage and known gaps are in data/MANIFEST.md; definitions are in docs/METHODS.md. An independent review of draft 1 recomputed 30 figures from these files: docs/REVIEW-draft1.md.

Passage	Figures	File
Minnesota lede and close	statewide 11.42 and 7.66; 78 of 87 counties; median county -13.1; Hennepin votes 626,662 and 575,529, margin 38.38 to 43.54, net change +10,130; statewide net change -102,983	`data/baselines/cycle_change.csv` (state MN, office GOV)
Correlations	log 2022 votes against swing: SC 0.85, MN 0.79, OH governor 0.14, NE -0.05; positive in 13 of 14	`cycle_change.csv`; printed on each panel of fig. 1
Texas	statewide -13.30 and -10.88; 225 of 254 counties; median -3.6; Bexar 6.30 to 16.45 on 541,071 votes, 24.0 percent of net change; Tarrant 18.7, Travis 18.3, Harris 16.0 percent	`cycle_change.csv` (TX, GOV)
Ohio	statewide +6.81 and -6.11; 0 of 88 counties toward D; Cuyahoga 483,310 and 408,543 votes, 44.98 to 35.64, net -71,802 (13.0 percent of -553,574); Mahoning 20.98 to -3.46	`cycle_change.csv` (OH, SEN)
Pennsylvania Senate	statewide 13.12 and 4.90; votes 5,009,890 and 5,366,638; Philadelphia 554,507 and 498,531, 74.81 to 66.97, net -80,954 of -394,240; Cumberland +0.99	`cycle_change.csv` (PA, SEN)
Pennsylvania governor	Philadelphia 75.72 and 72.58; vote share 11.055 and 9.286 percent	`data/baselines/benchmarks_2026.csv` (PA, GOV)
Michigan	statewide 9.57 and 10.19; net change +44,884; Oakland +42,784, 16.96 to 23.08; Wayne 659,859 and 606,144 votes, net -25,033; other 81 counties +27,133; MEDSL total 4,431,811 and 54.24 percent against RCP’s 54.5	`cycle_change.csv` (MI, GOV); `data/clean/statewide_results.csv`; `data/raw/expectations/expectations.csv`
Minnesota precincts	4,074 precincts in 2022, 3,776 matched, 87.4 percent of votes; 476 Hennepin and Ramsey precincts, median +5.2, vote-weighted +3.6, 27.3 percent toward R; 3,268 others with 10 or more votes, median -11.8, 84.1 percent toward R	`data/clean/precinct_match_rates.csv`; `data/clean/precinct_matched/MN.csv.gz`; fig. 4
2018 forecasts	538 mean absolute miss 3.14 over 20 races, mean signed -0.04, 12 above and 8 below zero; largest misses toward R: MI Senate -7.8, OH Senate -6.3, OH governor -5.6, IA governor -4.8; Stabenow forecast 14.33, result 6.50; Brown 13.16 and 6.82; Cordray 1.85 and -3.74; Hubbell 2.02 and -2.73; Kansas +5.1, Texas governor +5.3, South Carolina +7.3	`data/baselines/expectations_vs_results.csv`
2022 polls and forecasts	RCP: 10 of 13 understated D, mean signed +2.89 over all 13; Whitmer avg +1.0, Hassan +1.4, Oz +0.4; 538: 14 of 20, mean signed +1.06 over all 20; DeWine -3.6, Pillen -6.4 against 538; Walz RCP +4.3, 538 +8.91	`expectations_vs_results.csv`
Watch-list table	generated, not typed	`data/baselines/watchlist_2026_senate.md` from `pipeline/watchlist.py`

Pre-election expectations: FiveThirtyEight’s Deluxe forecast files for 2018 and 2022 (CC BY 4.0), read from Wayback Machine copies because the live site now redirects to abcnews.com, using the last forecast dated on or before election day; RealClearPolitics’ final averages and final results through the RCP data connector. The exact URLs are in data/raw/expectations/SOURCES.md. Six RCP averages reported with no polls in them are excluded. Races where a Democrat-minus-Republican margin does not describe the contest (Maine’s 2018 Senate race, won by Angus King, an independent; Alaska’s 2022 ranked-choice races) are excluded from every margin comparison.

Not yet done: a boundary check of the precinct matches against VEST shapefiles; county results for 2010 and 2014; prediction-market prices; a precinct crosswalk for Pennsylvania and Michigan, whose precinct names do not match across the two cycles.

Which kinds of places moved between 2018 and 2022, and what to read on Nov. 3

Mon, 02 Nov 2026 00:00:00 GMT

In the 92 Texas counties the American Communities Project classifies as Hispanic Centers, 966,389 people voted for governor in 2022, which is 81,241 fewer than in 2018. Greg Abbott’s margin in those counties grew from 6.6 points over Lupe Valdez to 11.9 points over Beto O’Rourke, while the rest of Texas moved 3.5 points toward the Democrat. Five of the 92 counties moved toward the Democrat by more than the rest of the state did.

The same comparison, run for every community type in 12 states, 11 with 2026 Senate battlegrounds plus Pennsylvania, ranks the types by how they moved against the rest of their own state between the two midterms. The Urban Suburbs ran ahead of the rest of their state in all 10 comparisons where they appear. Working Class Country, Aging Farmlands and Hispanic Centers ran behind in every one of theirs. In the Ohio and Pennsylvania Senate races, every type moved toward the Republican; the ranking describes which moved less.

The American Communities Project sorts each of the nation’s roughly 3,100 counties into one of 15 community types. Its current typology dates from 2023 and was built with Matt Grossman, a Michigan State University political scientist, according to the project’s methodology page. A k-means cluster analysis of about three dozen census and religious-adherence measures guided the grouping, which was then edited by hand to stay consistent with ACP’s original 2012 types. Matched to the MIT Election Data and Science Lab’s precinct returns, the typology covers 1,107 counties in the 14 comparisons used here: the governor’s races in all 12 states, plus the Ohio and Pennsylvania Senate races.

Each comparison pairs different candidates, so the measure here is relative. For each community type in each race, the type’s swing between 2018 and 2022 is set against the swing of every other county in the same state. A positive number means the type moved toward the Democrat more, or away from him less, than the rest of its state. Averages below are weighted by each comparison’s 2022 vote unless stated otherwise.

Suburbs and exurbs

The 30 Urban Suburbs counties in these states ran 5.1 points ahead of the rest of their states, and 36 of 38 county-level comparisons beat the figure for the rest of the state. In Texas, the eight Urban Suburbs counties moved 7.2 points toward O’Rourke while the rest of the state moved 1.4. In the Pennsylvania Senate races, the five moved 4.5 points toward the Republican and the rest of the state moved 9.6.

The 59 Exurbs counties ran ahead in 12 of 14 comparisons, by 4.7 points, and cast 5.7 percent more votes for governor in 2022 than in 2018. Between Sherrod Brown’s 2018 Senate win in Ohio and Tim Ryan’s 2022 loss, the rest of Ohio moved 13.7 points toward the Republican and its five Exurbs counties moved 4.0.

Big cities

The 14 Big Cities counties ran 4.2 points ahead of the rest of their states, but in only five of 10 comparisons, and they cast 8.2 percent fewer votes for governor in 2022 than in 2018. Minneapolis and St. Paul’s two counties ran 10.7 points ahead of the rest of Minnesota, and Texas’ five ran 6.1 ahead of the rest of Texas. In the governor’s races, Atlanta’s Fulton County ran 1.4 points behind the rest of Georgia, Detroit’s Wayne County 1.9 behind the rest of Michigan and Philadelphia 2.2 behind the rest of Pennsylvania, and Wichita’s Sedgwick County matched the rest of Kansas within a tenth of a point.

Hispanic Centers and the rural types

Hispanic Centers ran behind the rest of their states in all six comparisons, by 8.5 points, and their vote for governor fell 8.3 percent. Texas holds 92 of the 108 such counties in these states. Working Class Country ran behind in all 12 of its comparisons, by 4.4 points, with the largest gap in Ohio’s Senate race, 11.9 points. Aging Farmlands ran behind in all seven, and Rural Middle America, which has more counties here than any other type, 313, ran behind in 10 of 12.

The Middle Suburbs

Dante Chinni wrote for ACP on Oct. 14 that the Middle Suburbs “bear close watch” in tight races this year in Michigan, Ohio, Wisconsin and Iowa, and counted seven such counties in Michigan, 10 in Ohio and two in Iowa. Wisconsin is not among the states analyzed here; for the other three, the match finds the same counts. Between 2018 and 2022 the Middle Suburbs stayed close to the rest of their states, within 5.5 points in each of seven comparisons, ahead in four and behind in three. Ohio, with 20 of the 46 county-level comparisons, pulled the weighted average to -1.1 points; unweighted, the seven comparisons average -0.1. In the Ohio governor’s race, its 10 Middle Suburbs counties moved 5.4 points more toward the Republican than the rest of the state did.

What to read on Nov. 3

The table lists, for each 2026 Senate battleground, the three community types that cast the most votes in the closest comparable race of each cycle. In several states the 2018 and 2022 columns are different offices, named in the first column, so the change between them is not a swing. The full sheet, every type in every state for both the Senate and governor’s races on Tuesday’s ballot, is data/baselines/benchmarks_2026_by_type.csv, and the county sheet now carries each county’s type.

State (races compared)	Community type	Counties	2018 margin	2022 margin	Share of 2022 vote
Georgia (governor 2018, Senate 2022)	Urban Suburbs	7	+18.5	+23.8	31.2%
	African American South	63	+10.8	+11.9	19.4%
	Exurbs	11	-47.8	-40.9	11.6%
Iowa (governor 2018, Senate 2022)	Rural Middle America	80	-17.8	-30.5	50.2%
	Urban Suburbs	1	+18.3	+12.7	15.7%
	Middle Suburbs	2	+9.8	+3.9	12.9%
Kansas (governor 2018, Senate 2022)	Exurbs	2	+15.4	-0.9	27.1%
	Rural Middle America	38	-14.4	-50.4	25.4%
	College Towns	6	+28.6	-2.2	16.2%
Maine (governor 2018, governor 2022)	Graying America	13	+3.0	+6.8	58.9%
	Exurbs	1	+26.4	+40.2	24.0%
	College Towns	1	-4.0	-5.3	10.2%
Michigan (Senate 2018, governor 2022)	Urban Suburbs	2	+9.5	+18.9	21.0%
	Rural Middle America	29	-15.5	-13.2	20.6%
	Middle Suburbs	7	+3.5	+6.1	20.2%
Minnesota (Senate 2018, governor 2022)	Big Cities	2	+50.2	+44.3	31.6%
	Exurbs	7	+11.8	-3.0	23.3%
	Rural Middle America	47	+2.2	-25.1	19.7%
Nebraska (Senate 2018, governor 2022)	Big Cities	1	+4.0	+5.0	28.6%
	Rural Middle America	27	-42.8	-50.9	24.6%
	College Towns	4	+3.2	-4.0	20.4%
New Hampshire (governor 2018, Senate 2022)	Exurbs	3	-10.5	+7.3	64.0%
	Graying America	5	-5.7	+7.0	21.4%
	College Towns	2	+5.9	+19.7	14.6%
Ohio (Senate 2018, Senate 2022)	Rural Middle America	48	-21.4	-39.0	23.7%
	Big Cities	2	+41.2	+34.0	20.2%
	Middle Suburbs	10	+3.7	-11.4	19.1%
South Carolina (governor 2018, Senate 2022)	Urban Suburbs	3	-18.9	-34.1	22.0%
	Working Class Country	11	-28.3	-45.2	20.0%
	College Towns	3	+18.9	+0.2	18.8%
Texas (Senate 2018, governor 2022)	Big Cities	5	+22.1	+16.7	41.0%
	Urban Suburbs	8	-7.8	-12.6	19.9%
	Hispanic Centers	92	+4.6	-11.9	11.9%
North Carolina (no 2018 statewide race; Senate 2022)	Big Cities	2		+29.2	21.6%
	College Towns	9		+23.0	18.4%
	Working Class Country	18		-31.8	11.3%

Margins are the Democrat’s share minus the Republican’s, in points of all votes cast, summed across the type’s counties.

These are counties, not voters. A type’s movement can come from who turned out as much as from who changed sides, and the Texas Hispanic Centers moved on a smaller electorate: Abbott’s margin there grew by 5.3 points in a year when 81,241 fewer people voted.

Notes and sources

File names in this section refer to the project’s working data files, which are not yet public. The data and code are available from the author on request.

Results: MIT Election Data and Science Lab, Precinct-Level Returns 2018 by Individual State (doi:10.7910/DVN/NVQYMG) and Precinct-Level Returns 2022 by Individual State (doi:10.7910/DVN/UYQIEP), CC0, aggregated to counties by pipeline/clean.py. Community types: American Communities Project, 2023 typology, the “Community Type Sheet” linked from americancommunities.org/methodology (file 2023-Typology-1.xlsx, read 2026-11-02); the sheet spells one type “Urban Burbs,” which ACP’s site and this piece call Urban Suburbs. ACP states no license for the sheet; the raw file stays local and the type labels are cited to ACP. The relative-swing measure (against the rest of the state), the counts and the benchmarks are the author’s analysis, computed by pipeline/communities.py.

Passage	Figures	File
Texas Hispanic Centers lede and close	92 counties; votes 1,047,630 and 966,389; margin -6.62 to -11.91 (swing -5.29); rest of state +3.52; 5 of 92 above the rest-of-state swing	`data/baselines/by_type_comparison.csv` (TX, GOV, Hispanic Centers)
Scope and measure	12 states (GA, IA, KS, ME, MI, MN, NE, NH, OH, SC, TX with 2026 Senate races; PA); 14 comparisons; 1,107 unique counties; relative swing = type swing minus the swing of all other counties in the same race	`pipeline/communities.py`
Urban Suburbs	30 unique counties, 38 county-comparisons, 36 above the rest; weighted +5.14 (unweighted +5.26); 10 of 10; Texas 8 counties, swing +7.23 against +1.38; Pennsylvania Senate 5 counties, -4.47 against -9.60	`by_type_summary.csv`; `by_type_comparison.csv`
Exurbs	59 unique counties; 12 of 14; weighted +4.65 (unweighted +4.35); governor votes 4,194,588 to 4,433,413 (+5.7 percent); Ohio Senate 5 counties, -4.02 against -13.73	same
Big Cities	14 unique counties; weighted +4.17 (unweighted +2.50); 5 of 10; governor votes 7,435,738 to 6,823,228 (-8.2 percent); Minnesota +10.67, Texas +6.14; Georgia -1.44, Michigan -1.94, Pennsylvania governor -2.22, Kansas -0.06	same
Hispanic Centers, rural types	Hispanic Centers 0 of 6, weighted -8.48 (unweighted -10.31), governor votes -8.3 percent, 108 counties (92 in Texas); Working Class Country 0 of 12, -4.44, Ohio Senate -11.87; Aging Farmlands 0 of 7; Rural Middle America 313 counties, 2 of 12	same
Middle Suburbs	Michigan 7, Ohio 10, Iowa 2 counties (matches ACP’s Oct. 14, 2026 article by Dante Chinni, which also names Wisconsin); relative swings -5.43 to +5.15; 4 of 7; Ohio governor -5.43	same; americancommunities.org, “Why the Middle Suburbs Stay So Pivotal in American Politics”
Benchmark table	generated, not typed	`data/baselines/watchlist_2026_senate_by_type.md` from `pipeline/watchlist.py`

Alaska is excluded because ACP types its boroughs while its returns here are by state house district. North Carolina appears in the benchmark table but not in the 2018-to-2022 comparisons, because it held no statewide race in 2018. Maine’s 2018 Senate race and Alaska’s 2022 ranked-choice races are excluded from all margin comparisons; see data/MANIFEST.md.

Pittsburgh Water's 2025 rate case: what it asked for and what it settled

Wed, 28 Oct 2026 00:00:00 GMT

On June 4, 2025, Pittsburgh Water asked the Pennsylvania Public Utility Commission for two years of rate increases, and the notice it filed with the request put a number on what that meant for a household. A typical residential customer using 3,000 gallons a month, paying $100.27, would pay $123.02 in 2026 and $135.49 in 2027. Eight and a half months later, the commission approved a settlement whose own bill table puts the same 3,000-gallon customer at $115.29, with no base-rate increase for 2027.

All three documents behind those figures are published here as one record, with hashes and archived copies.

The typical bill

	Monthly bill	Change from $100.27	Source
Before the case	$100.27		Filing notice, excerpt p. 2
Requested for 2026	$123.02	+$22.75 (22.7%)	Filing notice, excerpt p. 2
Requested for 2027	$135.49	+$35.22	Filing notice, excerpt p. 2
Settled for 2026	$115.29	+$15.02 (15.0%)	Settlement, Appendix F, p. 61
Settled for 2027	No base-rate increase		Settlement, p. 8

The filing notice describes its typical customer as using 3,000 gallons a month on a 5/8-inch meter, charged the base stormwater rate (excerpt pp. 2 and 4). The settlement’s Appendix F row is a 5/8-inch, 3,000-gallon residential account starting from the same $100.27 bill. Customers who use more water see larger changes: the same table puts a 7,000-gallon account at $235.45, up from $197.12.

Pittsburgh Water withdrew its request to set 2027 rates as part of the settlement (p. 8), but 2027 bills will still move. The PENNVEST charge, a separate line on the bill, continues to be updated after January 1, 2027, and on February 1, 2027 the rate structure changes, a change the settlement calls revenue-neutral and whose bill impact is due with the tariff filing for that date (pp. 8 to 9).

Announced: $25 million. In the settlement: $42.6 million.

The commission’s February 19, 2026 press release said Pittsburgh Water’s total annual operating revenues “will increase by approximately $25 million,” about 10.2% over present rates. Pittsburgh Water’s December 17, 2025 release described the settlement as a base rate revenue increase of $25 million. Neither release, nor the authority’s February 19, 2026 release after the vote, states the $17.6 million or a combined total.

The settlement petition does. Its footnote 4, on page 7, says the total base rate and DSIC revenue increase “will be $42.6 million,” made up of the $25.0 million settlement amount plus $17.6 million in PENNVEST surcharges. The PENNVEST charge is the line through which the authority recovers what it owes on loans from PENNVEST, the state’s water-infrastructure lender. The commission’s order repeats the $42.6 million total in its footnote 16 (p. 51), where it corrects another party’s figure of $43.6 million.

A third number sits in the same petition. Pittsburgh Water’s own statement in support of the settlement (p. 141) describes “the Settlement total increase of $43.5 million.” The order quotes that figure (p. 44) and does not reconcile it with the $42.6 million.

The commission’s release describes the original request as about $63.6 million for 2026 and $20.7 million for 2027, roughly $84.4 million over two years.

What this does not settle

These are modeled bills for a typical account, not anyone’s actual statement, and the filing notice’s figures exclude ALCOSAN’s charges for wastewater treatment, which Pittsburgh Water bills on ALCOSAN’s behalf. The commission approved the settlement with modifications: it changed the settlement’s PENNVEST tariff language, set notice and filing requirements for the 2027 rate-structure change, and corrected percentages in one of the settlement’s appendices. The $115.29 figure comes from the settlement’s table, which predates those modifications; Pittsburgh Water’s release after the order continued to cite the $15.02 increase. Whether the compliance tariffs filed in March 2026 change the typical bill is not shown in these documents.

Hesitation Media

Reading Tuesday's county counts against a tied race

Large counties and the gap

Governor’s races as benchmarks

Partial counts

Tie pace for the largest counties

Notes and sources

What 2018 and 2022 say to watch on Nov. 3, county by county

Texas

Ohio and Pennsylvania Senate races

Michigan

Minnesota precincts

Counties to watch

Polls and forecasts

Notes and sources

Which kinds of places moved between 2018 and 2022, and what to read on Nov. 3

Suburbs and exurbs

Big cities

Hispanic Centers and the rural types

The Middle Suburbs

What to read on Nov. 3

Notes and sources

Pittsburgh Water's 2025 rate case: what it asked for and what it settled

The typical bill

Announced: $25 million. In the settlement: $42.6 million.

What this does not settle

Sources