Counties in focus

The 100 counties a simulated 2028 presidential election turns on most, and how their economies are doing: one economic score, read two ways, for people already here and for people getting started.

A county doesn’t have to switch parties to matter. Each state goes to whoever gets the most votes statewide, so a big county can tip a close state just by giving its usual winner a bigger or smaller lead than last time. That’s why Philadelphia ranks first, even though Democrats won it by about 60 points in 2024.

This week, nationally

Their figures, not part of the score

The newest national figures beside the slower county data below. State and region lines, and each county’s own permits, house-price and personal income lines, are on every county card and profile. How these are chosen →

Which counties could tip a simulated presidential election?

County electoral leverage · 2028 simulation

We simulate 400,000 2028 presidential elections from the 2024 result and ask, for every county: how often would a shift in this county’s vote alone change the Electoral College winner? The county does not have to change sides: in a big county, a bigger or smaller lead for its usual winner can be enough to tip its state.

What it measures: Each county’s share of simulated elections in which its own result tips the winner. The simulation adds swings sized from past elections to the 2024 result, with no polls, candidate information or economic forecasts.
What it is not: A comparison, not a forecast. It does not predict who wins in 2028, measure how much one person’s vote counts, or show where campaign spending pays off. Use it to compare counties with each other.
Where the uncertainty is: The ± beside each chance is simulation error, and it is small. The bigger uncertainty is the assumptions: the page reruns the model under 16 alternative sets, and each county’s rank range shows how far its rank moves across them. See the sensitivity tests.

97%
of the simulated chance of being the tipping-point state sits in seven states: Pennsylvania, Michigan, Georgia, North Carolina, Wisconsin, Nevada and Arizona.
39 of 100
highest-leverage counties in the simulation are in Pennsylvania, and all 100 sit in those seven states, mostly big metro and suburban counties.
1.17%
is the largest simulated chance that one county’s own result changes the winner (Philadelphia County, PA; ± 0.01% simulation error). Leverage is concentrated, but no county is likely to decide the presidency on its own.

Seven states hold 97% of the tipping-point chances

The tipping-point state is the one whose electoral votes carry the winner past 270. Pennsylvania holds 3.5% of the electoral votes but is the tipping point in 32.6% of simulated elections, so electoral votes alone mislead. In the 2032 simulation, with no trend extrapolated, North Carolina’s share rises from 14.3% to 16.4% and Pennsylvania’s falls from 32.6% to 27.2%.

Chance each state is the tipping pointPennsylvania is the tipping-point state in 32.6% of simulated elections while holding 3.5% of the electoral votes. Bars show 10 states with the highest chances; the table lists all 55 electoral-vote units.

2028 tipping-point chanceShare of electoral votes2032, no trend extrapolated2032 if recent trends continued (scenario)

Show all 55 electoral-vote units as a table

Tipping-point chance by electoral-vote unit
Unit	Electoral votes	Share of EV	Tipping point 2028	± simulation error	2032	2032 trend scenario	Democratic win share	Expected margin
Pennsylvania	19	3.5%	32.6%	0.06%	27.2%	24.7%	35.1%	−1.8 pts ± 5.5
Michigan	15	2.8%	16.7%	0.05%	15.0%	13.4%	37.7%	−1.6 pts ± 5.9
Georgia	16	3.0%	16.2%	0.06%	16.0%	17.5%	30.4%	−2.5 pts ± 5.6
North Carolina	16	3.0%	14.3%	0.05%	16.4%	18.9%	25.5%	−3.2 pts ± 5.4
Wisconsin	10	1.9%	10.7%	0.04%	10.0%	10.7%	43.3%	−0.9 pts ± 5.9
Nevada	6	1.1%	3.6%	0.03%	3.7%	3.7%	30.3%	−3.0 pts ± 6.7
Arizona	11	2.0%	2.9%	0.03%	4.1%	4.8%	12.5%	−5.7 pts ± 5.5
New Hampshire	4	0.7%	0.7%	0.01%	1.1%	0.7%	70.5%	+2.8 pts ± 5.9
Minnesota	10	1.9%	0.5%	0.01%	1.2%	0.8%	80.4%	+4.2 pts ± 5.5
New Jersey	14	2.6%	0.4%	0.01%	1.0%	1.0%	85.1%	+5.9 pts ± 6.2
Ohio	17	3.2%	0.3%	0.01%	0.8%	0.6%	2.2%	−11.4 pts ± 5.7
Alaska	3	0.6%	0.1%	0.01%	0.3%	0.4%	2.2%	−13.8 pts ± 7.7
Virginia	13	2.4%	0.1%	0.01%	0.5%	0.2%	86.3%	+5.8 pts ± 5.9
New Mexico	5	0.9%	0.1%	0.01%	0.3%	0.4%	88.5%	+6.3 pts ± 5.7
Maine (statewide)	2	0.4%	0.1%	0.00%	0.2%	0.2%	91.1%	+7.2 pts ± 5.8
Iowa	6	1.1%	0.1%	0.01%	0.3%	0.2%	1.7%	−13.2 pts ± 6.1
Texas	40	7.4%	0.1%	0.00%	0.3%	0.5%	1.1%	−13.8 pts ± 5.8
Utah	6	1.1%	0.1%	0.00%	0.2%	0.2%	1.1%	−22.3 pts ± 9.2
Maine 2nd district (county proxy)	1	0.2%	0.1%	0.00%	0.1%	0.1%	3.5%	−9.6 pts ± 5.4
Indiana	11	2.0%	0.0%	0.00%	0.2%	0.1%	0.6%	−19.2 pts ± 7.1
New York	28	5.2%	0.0%	0.00%	0.2%	0.2%	97.9%	+12.7 pts ± 6.2
Nebraska 2nd district (county proxy)	1	0.2%	0.0%	0.00%	0.1%	0.0%	92.4%	+10.3 pts ± 7.7
Hawaii	4	0.7%	0.0%	0.00%	0.1%	0.0%	98.9%	+23.6 pts ± 9.9
Illinois	19	3.5%	0.0%	0.00%	0.1%	0.1%	96.8%	+11.1 pts ± 6.2
Connecticut	7	1.3%	0.0%	0.00%	0.1%	0.1%	98.3%	+14.7 pts ± 6.9
Florida	30	5.6%	0.0%	0.00%	0.1%	0.1%	1.1%	−13.4 pts ± 5.6
Colorado	10	1.9%	0.0%	0.00%	0.1%	0.0%	96.7%	+11.1 pts ± 6.2
Rhode Island	4	0.7%	0.0%	0.00%	0.1%	0.1%	98.0%	+14.2 pts ± 6.8
Montana	4	0.7%	0.0%	0.00%	0.0%	0.0%	0.4%	−20.5 pts ± 6.9
Missouri	10	1.9%	0.0%	0.00%	0.0%	0.0%	0.3%	−19.0 pts ± 6.1
Delaware	3	0.6%	0.0%	0.00%	0.1%	0.0%	98.0%	+14.4 pts ± 7.0
Kansas	6	1.1%	0.0%	0.00%	0.1%	0.0%	0.6%	−16.0 pts ± 5.8
California	54	10.0%	0.0%	0.00%	0.1%	0.0%	99.7%	+20.7 pts ± 6.9
Louisiana	8	1.5%	0.0%	0.00%	0.0%	0.0%	0.0%	−22.6 pts ± 5.4
Nebraska (statewide)	2	0.4%	0.0%	0.00%	0.0%	0.0%	0.3%	−20.3 pts ± 6.7
Oregon	8	1.5%	0.0%	0.00%	0.0%	0.0%	99.2%	+14.9 pts ± 5.8
South Carolina	9	1.7%	0.0%	0.00%	0.0%	0.0%	0.2%	−18.2 pts ± 5.6
Arkansas	6	1.1%	0.0%	0.00%	0.0%	0.0%	0.0%	−31.3 pts ± 6.0
North Dakota	3	0.6%	0.0%	0.00%	0.0%	0.0%	0.0%	−37.3 pts ± 8.5
Massachusetts	11	2.0%	0.0%	0.00%	0.0%	0.0%	99.9%	+25.9 pts ± 6.5
Maine 1st district (county proxy)	1	0.2%	0.0%	0.00%	0.0%	0.0%	99.9%	+24.6 pts ± 7.1
Mississippi	6	1.1%	0.0%	0.00%	0.0%	0.0%	0.0%	−23.3 pts ± 5.4
Alabama	9	1.7%	0.0%	0.00%	0.0%	0.0%	0.0%	−31.1 pts ± 6.0
Tennessee	11	2.0%	0.0%	0.00%	0.0%	0.0%	0.0%	−30.4 pts ± 6.3
District of Columbia	3	0.6%	0.0%	0.00%	0.0%	0.0%	100.0%	+86.6 pts ± 12.2
Idaho	4	0.7%	0.0%	0.00%	0.0%	0.0%	0.0%	−37.5 pts ± 6.9
Kentucky	8	1.5%	0.0%	0.00%	0.0%	0.0%	0.0%	−31.2 pts ± 6.1
Maryland	10	1.9%	0.0%	0.00%	0.0%	0.0%	100.0%	+29.3 pts ± 6.6
Nebraska 1st and 3rd districts (county proxy)	2	0.4%	0.0%	0.00%	0.0%	0.0%	0.0%	−32.9 pts ± 6.9
Oklahoma	7	1.3%	0.0%	0.00%	0.0%	0.0%	0.0%	−34.9 pts ± 6.4
South Dakota	3	0.6%	0.0%	0.00%	0.0%	0.0%	0.0%	−29.7 pts ± 7.1
Vermont	3	0.6%	0.0%	0.00%	0.0%	0.0%	100.0%	+32.8 pts ± 7.3
Washington	12	2.2%	0.0%	0.00%	0.0%	0.0%	99.8%	+18.8 pts ± 5.9
West Virginia	4	0.7%	0.0%	0.00%	0.0%	0.0%	0.0%	−42.6 pts ± 7.2
Wyoming	3	0.6%	0.0%	0.00%	0.0%	0.0%	0.0%	−47.0 pts ± 7.1

How the tipping point is defined

In each simulated election, states are sorted by margin and electoral votes are added up from the winner’s strongest state; the state that pushes the total past 270 is the tipping point. Maine and Nebraska also award district electoral votes, which are modeled here with county proxies rather than official district allocations. The 2032 scenario marker adds each county’s recent structural drift; it is a labelled scenario, not a forecast.

Deciding a state is not the same as signalling the result

Each dot is a county. Left to right is the chance it changes the winner; bottom to top is how much its result would reveal about the national winner. The two diverge: 57 counties rank in the top 150 for information but below 1,000 for leverage (Seminole County, Florida is 9th for information and 1,307th for leverage). The most informative county, Will County, IL, carries 0.44 bits of the roughly 0.9 bits of uncertainty about the winner.

Model-implied.

Information value mostly reflects how closely a county tracks the national swing. It has not been tested against a record of real bellwether counties.

Chance of changing the winner against information about the winner3,102 counties. The top 25 for leverage sit at the right; the 57 highlighted in amber reveal a lot about the national result but almost never decide it. Counties below 1 in 10M share the band at the left.

List the 57 high-information, low-leverage counties

Counties in the top 150 for information and below 1,000 for leverage
County	Information rank	Information	Leverage rank
Seminole County, FL	9	0.42 bits	1,307
Peoria County, IL	12	0.42 bits	1,370
Frederick County, MD	16	0.41 bits	2,108
Clay County, FL	18	0.41 bits	1,606
Okaloosa County, FL	21	0.41 bits	1,823
Duval County, FL	22	0.41 bits	1,206
Escambia County, FL	25	0.41 bits	1,499
Roanoke city, VA	26	0.41 bits	1,008
Dorchester County, SC	29	0.41 bits	1,799
Pinellas County, FL	30	0.41 bits	1,386
Baltimore County, MD	34	0.41 bits	2,109
York County, SC	36	0.41 bits	1,687
St. Clair County, IL	39	0.41 bits	1,056
Berkeley County, SC	47	0.40 bits	1,721
Santa Rosa County, FL	49	0.40 bits	1,740
Prince George's County, MD	52	0.40 bits	2,110
Franklin County, WA	56	0.40 bits	2,111
Alachua County, FL	61	0.40 bits	1,652
St. Johns County, FL	62	0.40 bits	1,520
Baltimore city, MD	64	0.39 bits	2,112
York County, VA	65	0.39 bits	1,121
Greenville County, SC	69	0.39 bits	1,523
Charleston County, SC	70	0.39 bits	1,572
Adams County, CO	71	0.39 bits	1,093
Guadalupe County, TX	77	0.39 bits	1,349
Coryell County, TX	82	0.39 bits	1,545
Ogle County, IL	83	0.39 bits	1,358
Richland County, SC	84	0.39 bits	1,373
Madison County, MS	87	0.39 bits	2,113
Anne Arundel County, MD	94	0.38 bits	2,114
Ada County, ID	95	0.38 bits	2,090
Snohomish County, WA	96	0.38 bits	2,045
Clark County, WA	98	0.38 bits	2,115
Lexington County, SC	102	0.38 bits	1,685
Pierce County, WA	103	0.38 bits	2,018
Yamhill County, OR	104	0.38 bits	1,622
Walker County, TX	109	0.38 bits	1,615
Spokane County, WA	110	0.38 bits	2,092
Skagit County, WA	112	0.38 bits	2,116
San Benito County, CA	113	0.38 bits	2,117
Rock Island County, IL	118	0.38 bits	1,625
Yuba County, CA	120	0.37 bits	2,118
Carroll County, MD	121	0.37 bits	2,119
Wilson County, TX	122	0.37 bits	1,483
St. Mary's County, MD	124	0.37 bits	2,120
DeKalb County, IL	125	0.37 bits	1,656
Lancaster County, SC	128	0.37 bits	2,008
Wyandotte County, KS	134	0.37 bits	1,396
Taylor County, TX	135	0.37 bits	1,092
Benton County, WA	137	0.37 bits	2,121
Bonneville County, ID	139	0.37 bits	2,122
Fresno County, CA	142	0.37 bits	1,467
Clackamas County, OR	143	0.37 bits	1,127
Medina County, TX	146	0.37 bits	1,807
Polk County, OR	148	0.37 bits	1,492
Marion County, OR	149	0.37 bits	1,476
Tom Green County, TX	150	0.37 bits	1,460

Who is each county working for?

Each dot is a county. Further right, the county works better for homeowners and longer-standing, older residents, including those on fixed incomes; higher up, for households headed by people under 35, and newcomers. The lines at 50 mark the middle of all counties. Ringed: the counties in focus, named where there is room, and the one you pick.

For people getting started: higher ↑

Open door 554 counties

Works for both 966 counties

Struggling for both 1,012 counties

Works for owners, hard to get into 560 counties

For people already here: higher → No score (Connecticut, and counties with too few inputs)

The same four groups on the map

Loading the county map…

For people getting started: higher ↑

Open door 554 counties

Works for both 966 counties

Struggling for both 1,012 counties

Works for owners, hard to get into 560 counties

For people already here: higher → No score (Connecticut, and counties with too few inputs)

Pick a county

Click a dot or a county on the map, or find one in the table below.

Every county, as a table

County

Group

3,112 counties

Group	For people already here	For people getting started	Economic conditions (90% range)
Works for both	69	60	57 to 68
Works for owners, hard to get into	60	41	48 to 54
Struggling for both	46	50	42 to 53
Works for owners, hard to get into	57	48	51 to 54
Works for both	63	65	57 to 70
Open door	50	54	44 to 58
Works for owners, hard to get into	58	46	46 to 54
Works for both	56	54	49 to 58
Struggling for both	33	34	33 to 35
Open door	48	72	52 to 64
Struggling for both	44	39	33 to 49
Works for both	65	61	59 to 66
Works for both	61	51	51 to 63
Struggling for both	29	37	30 to 38
Open door	42	66	48 to 61
Works for both	56	63	54 to 62
Struggling for both	37	46	35 to 48
Works for both	55	58	54 to 60
Struggling for both	48	36	35 to 45
Struggling for both	48	46	42 to 52
Struggling for both	32	45	34 to 42
Works for both	66	55	57 to 63
Struggling for both	29	21	24 to 31
Struggling for both	38	30	33 to 38
Works for owners, hard to get into	54	44	46 to 52

Page 1 of 125

Explore every county

Pick a measure, then pin a county to see why it ranks where it does. Gray: Alaska and Connecticut, each simulated as one state-level unit. Kansas City, Missouri is one composite of three counties.

Loading the county map…

1 in 10M
1 in 1M
1 in 100K
1 in 10K
1 in 1K
1 in 100

Alaska and Connecticut: simulated as one state-level unit each, not ranked

Simulated and conditional on the no-poll prior: a comparison between counties, not a forecast. Counties below one in ten million share the darkest colour.

In the simulation, could a different result in this county alone change who wins?

The chance that a fresh draw of this county's own result changes the Electoral College winner. It needs a swing bigger than the state margin and a state that is pivotal in that same simulated election, so values are small; the ranking is the useful part.

Click or tap a county to pin it. Hold Ctrl or ⌘ and scroll, or pinch, to zoom; drag to pan when zoomed; double-click a state to zoom into it. Keyboard: focus the map, then use the arrow keys to move between neighbouring counties, Enter to zoom into the selected county's state, + and − to zoom, and Escape or 0 to reset.

Good for whom? Every county through two lenses

County economic score · Our research

Rising home prices are good news for people who already own and bad news for people trying to buy. So each county gets two scores from the same published figures: For people already here (homeowners and longer-standing, older residents, including those on fixed incomes) and For people getting started (households headed by people under 35, and newcomers). Economic conditions, at the top of each card below, is the mean of the two, published only as a range.

What goes into each lens

Ten published figures, each weighted out of 100 points per lens. A lens score is where the county sits among all U.S. counties on those figures: 50 is the middle, not a pass mark.

For people already here

Housing40 of 100

Home value change, one yearhigher counts as better25
Owners 65 and older spending 30% or more of income on housinglower counts as better15

Jobs and income40 of 100

Employment change, June-August against a year earlierhigher counts as better15
Unemployment rate, past twelve monthslower counts as better10
Median household income change after inflationhigher counts as better15

Everyday costs20 of 100

Home energy bills as a share of income (modeled)lower counts as better15
Grocery prices, one year (metro area or region)lower counts as better5

For people getting started

Housing60 of 100

Years of income to buy the median home, householders 25 to 44lower counts as better25
Renters 25 to 34 spending 30% or more of income on rentlower counts as better20
Home value change, one yearlower counts as better10
Householders under 35 who own their homehigher counts as better5

Jobs and income35 of 100

Employment change, June-August against a year earlierhigher counts as better15
Unemployment rate, past twelve monthslower counts as better10
Median household income change after inflationhigher counts as better10

Everyday costs5 of 100

Home energy bills as a share of income (modeled)lower counts as better5

In Economic conditions, the mean of the two lenses, housing figures carry 50% of the weight, jobs and income 37.5% and everyday costs 12.5%. One figure counts both ways: “Home value change, one year”. A rise is good for people already here and bad for people getting started.

Four groups

Each county lands in one group, by whether each lens is above or below 50, the middle of all counties. The cards, the chart and the map all use these colours.

Open doorAbove the middle for people getting started, below it for people already here.554 counties
Works for bothAbove the middle on both lenses.966 counties
Struggling for bothBelow the middle on both lenses.1,012 counties
Works for owners, hard to get intoAbove the middle for people already here, below it for people getting started.560 counties

Beside each score, a county’s card also shows young-adult labor-market context: the unemployment rate for ages 20 to 34 next to the all-ages rate, and the change in jobs held by workers aged 25 to 34 at workplaces in the county. It is our research too, and it is not part of the score. How the young-adult context is built →

Our research The score, its two lenses, every weight and every rule here are this project's own research design, not an official statistic. Every input is a published figure from a named source and vintage. Our research: how this score is built →

How the 100 counties in focus are doing

The counties the election simulation ranks highest, in its order. How they are picked is explained near the top of this page. Each card opens with the county’s economic score, then its own jobs, income, housing, costs and population figures, each with its source and period: tap or hover a measure to see them.

The simulation’s rank compares counties and is not a forecast. A measure that reads “Not available” has no published value for that county, with the reason; it is never zero. Costs name the area they are measured for: the county, a metro area or region, or the state.

#1
Philadelphia County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.1%+1,546 people
Loading…
Full county profile
#2
Allegheny County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
−0.2%−2,139 people
Loading…
Full county profile
#3
Fulton County, GA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.8%+8,323 people
Loading…
Full county profile
#4
Wayne County, MI
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.2%+3,504 people
Loading…
Full county profile
#5
Montgomery County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.3%+2,483 people
Loading…
Full county profile
#6
Clark County, NV
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.9%+21,480 people
Loading…
Full county profile
#7
Milwaukee County, WI
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
−0.0%−107 people
Loading…
Full county profile
#8
Mecklenburg County, NC
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+2.2%+26,554 people
Loading…
Full county profile
#9
Gwinnett County, GA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.8%+8,235 people
Loading…
Full county profile
#10
Wake County, NC
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+2.3%+27,760 people
Loading…
Full county profile
#11
DeKalb County, GA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.2%+1,810 people
Loading…
Full county profile
#12
Dane County, WI
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.7%+4,050 people
Loading…
Full county profile
#13
Oakland County, MI
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.5%+5,788 people
Loading…
Full county profile
#14
Chester County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.5%+2,586 people
Loading…
Full county profile
#15
Cobb County, GA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.6%+4,466 people
Loading…
Full county profile
#16
Maricopa County, AZ
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.8%+35,411 people
Loading…
Full county profile
#17
Delaware County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
−0.1%−479 people
Loading…
Full county profile
#18
Bucks County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
−0.1%−588 people
Loading…
Full county profile
#19
Westmoreland County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
−0.3%−1,149 people
Loading…
Full county profile
#20
Macomb County, MI
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.4%+3,724 people
Loading…
Full county profile
#21
Lancaster County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.2%+1,097 people
Loading…
Full county profile
#22
Luzerne County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.4%+1,228 people
Loading…
Full county profile
#23
York County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.7%+3,211 people
Loading…
Full county profile
#24
Berks County, PA
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.4%+1,693 people
Loading…
Full county profile
#25
Kent County, MI
Jobs
Loading…
Loading…
Income & housing strain
Loading…
Loading…
Loading…
Loading…
Loading…
Costs
county
Loading…
county · model estimate
Loading…
county
Loading…
metro or region
Loading…
People
+0.7%+4,820 people
Loading…
Full county profile

If a state is decided by a hair, which counties carry it?

Among simulated elections where a battleground finishes within 1 point, the bar is each county’s share of the statewide swing and the tick is its share across all simulations. How concentrated that is depends on the state: Clark County alone carries 72% of Nevada’s, while Pennsylvania’s largest single share is 8.1% (Allegheny County).

Pennsylvaniawithin 1 point in 15% of simulations
1. Allegheny8.1%
2. Montgomery7.1%
3. Philadelphia7.0%
4. Bucks6.1%
5. Lancaster5.1%
Michiganwithin 1 point in 15% of simulations
1. Oakland14.1%
2. Wayne13.0%
3. Macomb9.6%
4. Kent6.6%
5. Ottawa3.7%
Georgiawithin 1 point in 14% of simulations
1. Fulton9.7%
2. Gwinnett8.7%
3. Cobb7.8%
4. DeKalb6.1%
5. Cherokee3.8%
North Carolinawithin 1 point in 13% of simulations
1. Wake12.1%
2. Mecklenburg10.8%
3. Guilford5.1%
4. Forsyth3.8%
5. Union3.0%
Wisconsinwithin 1 point in 15% of simulations
1. Waukesha9.8%
2. Milwaukee8.5%
3. Dane7.8%
4. Brown5.6%
5. Outagamie3.9%
Nevadawithin 1 point in 12% of simulations
1. Clark72.2%
2. Washoe17.3%
3. Douglas2.2%
4. Carson City2.0%
5. Lyon1.9%
Arizonawithin 1 point in 8% of simulations
1. Maricopa67.7%
2. Pima13.5%
3. Pinal5.0%
4. Yavapai3.3%
5. Mohave2.8%

Does it hold up?

The model was re-run as of 2016, 2020 and 2024, each time using only information available before that election. It found where the close races were, but its probabilities are not perfectly calibrated. It ranked which counties could flip their state well (state-flip AUC 0.89 in 2016 and 0.98 in 2020; 2024 had no state flips to test).

Predicted leverage landed in the states that decided the raceFor each past election, the model was re-run using only information available beforehand. The bars compare the share of predicted county leverage inside the closest set of states that would have had to flip the result with that set's share of electoral votes: 2016, 20% against 6.7%; 2020, 19% against 6.9%; 2024, 58% against 8.2%.

Share of predicted county leverageShare of electoral votes

Show the numbers as a table

Leverage concentration in the closest flipping coalition
Election	Closest flipping coalition	Share of leverage	Share of EV	Lift	State-flip AUC
2016	Michigan and Pennsylvania	20.3%	6.7%	3.0×	0.89
2020	Georgia, Arizona and Wisconsin	18.9%	6.9%	2.8×	0.98
2024	Wisconsin, Michigan and Pennsylvania	58.3%	8.2%	7.1×	no state flipped

How much the top 100 depend on modeling assumptionsOf the top 100: 19 low, 74 moderate and 7 high model dependence. Maricopa County, AZ is the most assumption-sensitive of the leading counties (rank confidence 0.16).

Show the numbers as a table

Model dependence of the top 100
Model dependence	Counties
Low model dependence	19
Moderate model dependence	74
High model dependence	7

Maricopa County, AZ (ranked 16th) is the most assumption-sensitive of the top 25: its rank moves with the national-environment assumption.

How many of the top 100 counties survive a change of assumptionsAcross 16 alternative modeling assumptions, between 82 and 89 of the base top 100 stay in the top 100; a different random seed alone keeps 86.

Alternative assumptionSame model, different random seed

Calmer cycles (volatility ×0.6)82

Environment 3 points more Republican83

Turnout variation ×283

State-level correlation ×0.584

Larger national swings (×1.3)85

Recent cycles weighted more85

Different random seed only86

Smaller national swings (×0.7)86

Environment 3 points more Democratic87

No coalition-cluster correlation87

State-level correlation ×1.587

State-based elasticity prior87

More turbulent cycles (volatility ×1.6)87

History from 2008 only87

Turnout variation ×0.588

Same national response everywhere88

Pooled county noise89

7580859095100

Show the numbers as a table

Top-100 overlap under alternative assumptions
Assumption	Counties kept in the top 100	Democratic 270+ share under it
Calmer cycles (volatility ×0.6)	82	31.6%
Environment 3 points more Republican	83	15.5%
Turnout variation ×2	83	31.6%
State-level correlation ×0.5	84	31.6%
Larger national swings (×1.3)	85	35.3%
Recent cycles weighted more	85	31.9%
Different random seed only	86	32.0%
Smaller national swings (×0.7)	86	26.3%
Environment 3 points more Democratic	87	54.1%
No coalition-cluster correlation	87	31.4%
State-level correlation ×1.5	87	32.2%
State-based elasticity prior	87	32.6%
More turbulent cycles (volatility ×1.6)	87	32.3%
History from 2008 only	87	23.9%
Turnout variation ×0.5	88	31.7%
Same national response everywhere	88	33.3%
Pooled county noise	89	31.8%

Calibration is imperfect, and single-county leverage cannot be checked against history

In 2016 the 80% county intervals contained only 60% of results (too narrow); in 2020 and 2024 they contained 98% and 97% (too wide). The model gave the eventual Electoral College winner 26.5% in 2016, 44.7% in 2020 and 41.2% in 2024. County volatility is therefore modeled as a mix of calm and turbulent cycles, and three backtestable cycles cannot confirm the tails.

In 2016, 2020 or 2024, no single state could have changed the winner on its own, so the number of counties that were actually decisive was zero each time. Claims about a single county’s chance rest on the model, not on a record of hits.

Calibration in past elections (model re-run with information available beforehand)
Election	80% county intervals that contained the result	50% intervals	90% intervals	Chance given to the actual winner	Actual tipping-point state	Chance it had
2016	60%	30%	73%	26.5% (Republican)	Pennsylvania	20%
2020	98%	83%	99%	44.7% (Democrat)	Wisconsin	8%
2024	97%	79%	99%	41.2% (Republican)	Pennsylvania	30%

Does economic change predict vote movement?

Not in a way that holds up. Relative economic deterioration had no robust relationship with vote movement after controls in any of the three transitions tested. Population growth, housing-cost deterioration and demographic change (age structure) are the robust within-cycle correlates. Nothing economic carried over to predict a later election: controls fitted on 2016 and 2020 and tested on 2024 did worse than predicting each state’s average (out-of-sample R² -0.75).

Associations, not causes. The economic fields on this page are descriptive. No county is shown as sensitive to economic deterioration, and the planned economic-volatility index (p.vol) is not computed.

Net migration uses the site’s one definition: movers in minus movers out, per thousand residents on July first at the end of the period. Here it is IRS full county inflow and outflow headers (people on returns) over the Census population vintage frozen in this analysis; the Movement map shows the same measure over the latest Census vintage.

Estimated effect on a county's margin change, per standard deviation of change relative to its state8 estimates across three election-to-election transitions. 3 pass the robustness rule and structure controls: population growth in 2016→2020 (+0.83 pts), housing-cost deterioration in 2016→2020 (+0.65 pts) and demographic change (age structure) in 2016→2020 (−0.58 pts). Relative economic momentum has no robust effect in any of the three transitions tested.

RobustFragile or explained by compositionNo robust effect95% interval

Show the numbers as a table

Economic and demographic associations with vote movement
Measure	Transition	Effect (points per SD)	95% interval	q-value	Status	Counties
Population growth	2012→2016	+1.59 pts	1.07 to 2.10	<0.001	Explained by composition	2,989
Population growth	2016→2020	+0.83 pts	0.46 to 1.21	<0.001	Robust	2,981
Housing-cost deterioration	2016→2020	+0.65 pts	0.33 to 0.97	0.001	Robust	2,981
Demographic change (age structure)	2016→2020	−0.58 pts	-0.90 to -0.25	0.005	Robust	2,981
Relative economic momentum	2012→2016	+0.49 pts	0.19 to 0.79	0.008	Fragile	2,989
Relative economic momentum	2016→2020	−0.13 pts	-0.40 to 0.14	0.454	No robust effect	2,981
Relative economic momentum	2020→2024	−0.07 pts	-0.23 to 0.10	0.511	No robust effect	2,836
Net migration	2020→2024	+0.13 pts	-0.03 to 0.29	0.174	No robust effect	2,836

The counties that matter most

Ranked by the chance that a shift in the county’s own vote, on top of the national and state swings, changes the winner. The largest is 1.17% for Philadelphia County (± 0.01% simulation error); the top 25 sum to about 10% and the top 100 to about 17%. Values are small because a county must swing by more than its state’s margin in an election where that state is pivotal, so the order matters more than the level. Clark County, NV has the highest chance of flipping its own state (8.8%), but Nevada is the tipping point in only 3.6% of simulations, so it ranks 6th. Philadelphia ranks first even though it is far from close: it is big, its own vote typically moves by about 30,000 votes, and Pennsylvania is the tipping point in 32.6% of simulations, more than any other state.

Counties ranked by the simulated chance that a different result in the county alone changes the presidential winner, top 25
#	County	Chance it changes the winner	Flips its state	State is tipping point	Net-vote movement	Information	Rank range	Into 2032	Model dependence
1	Philadelphia County, PA	1.17% ±0.01%	3.5%	33%	±54,230	0.37 bits	1	Stable	Moderate
2	Allegheny County, PA	0.61% ±0.008%	1.9%	33%	±34,826	0.36 bits	2–3	Stable	Moderate
3	Fulton County, GA	0.51% ±0.008%	3.0%	16%	±43,030	0.35 bits	3–6	Stable	Moderate
4	Wayne County, MI	0.50% ±0.007%	2.8%	17%	±48,910	0.40 bits	3–9	Stable	Moderate
5	Montgomery County, PA	0.47% ±0.006%	1.5%	33%	±29,176	0.41 bits	3–9	Stable	Low
6	Clark County, NV	0.46% ±0.008%	8.8%	4%	±61,067	0.38 bits	3–11	Stable	Moderate
7	Milwaukee County, WI	0.46% ±0.008%	4.2%	11%	±29,692	0.36 bits	4–10	Stable	Low
8	Mecklenburg County, NC	0.43% ±0.005%	2.6%	14%	±40,666	0.37 bits	7–15	Stable	Moderate
9	Gwinnett County, GA	0.42% ±0.007%	2.4%	16%	±28,159	0.30 bits	6–15	Stable	Low
10	Wake County, NC	0.41% ±0.005%	2.5%	14%	±42,376	0.38 bits	8–16	Stable	Moderate
11	DeKalb County, GA	0.40% ±0.008%	2.3%	16%	±33,221	0.38 bits	8–17	Stable	Moderate
12	Dane County, WI	0.40% ±0.008%	3.6%	11%	±27,788	0.38 bits	8–15	Stable	Moderate
13	Oakland County, MI	0.38% ±0.006%	2.2%	17%	±42,492	0.42 bits	9–15	Stable	Moderate
14	Chester County, PA	0.36% ±0.007%	1.2%	33%	±18,999	0.36 bits	9–16	Stable	Low
15	Cobb County, GA	0.34% ±0.008%	2.0%	16%	±24,320	0.32 bits	12–17	Stable	Low
16	Maricopa County, AZ	0.32% ±0.005%	5.3%	3%	±107,386	0.34 bits	5–32	Rising	High
17	Delaware County, PA	0.30% ±0.007%	0.9%	33%	±17,192	0.39 bits	11–19	Stable	Low
18	Bucks County, PA	0.29% ±0.005%	0.9%	33%	±19,823	0.41 bits	15–21	Stable	Moderate
19	Westmoreland County, PA	0.27% ±0.005%	0.8%	33%	±12,315	0.31 bits	15–23	Stable	Moderate
20	Macomb County, MI	0.27% ±0.006%	1.5%	17%	±27,120	0.31 bits	18–23	Stable	Moderate
21	Lancaster County, PA	0.27% ±0.004%	0.9%	33%	±16,894	0.41 bits	17–22	Stable	Low
22	Luzerne County, PA	0.26% ±0.006%	0.8%	33%	±10,038	0.26 bits	21–25	Stable	Low
23	York County, PA	0.25% ±0.005%	0.8%	33%	±15,526	0.41 bits	18–23	Stable	Moderate
24	Berks County, PA	0.21% ±0.005%	0.7%	33%	±12,109	0.39 bits	23–32	Stable	Moderate
25	Kent County, MI	0.20% ±0.005%	1.2%	17%	±20,972	0.40 bits	22–32	Stable	Moderate

How to read the columns

Chance it changes the winner: Replace the county’s own (idiosyncratic) outcome with an independent draw and compute whether that flips its state and whether the flip changes the Electoral College winner. The ± figure is Monte Carlo simulation error, not model uncertainty.
Flips its state: The same calculation, stopping at the statewide winner.
Rank range: The 10th to 90th percentile of the county’s rank across 16 alternative modeling assumptions (volatility, correlation, national swing size, turnout, elasticity, environment and history window).
Model dependence: Low model dependence: The rank barely moves across the alternative assumptions. Moderate model dependence: The rank shifts somewhat under alternative assumptions. High model dependence: The rank moves a lot under alternative assumptions. Below simulation resolution: The leverage estimate is smaller than three Monte Carlo standard errors, so its rank among similar counties is not meaningful.
Into 2032: Whether the county’s leverage rises or falls when the same model runs two cycles ahead with zero drift. A change counts only if it exceeds 3 simulation standard errors.

Rising counties and the watchlist

Running the same model two cycles ahead to 2032, with no trend extrapolated, 342 counties gain leverage and 51 lose it. Of the 300 largest risers outside today’s top 100, 63 follow a persistent structural trend, 38 are growing structurally and 199 depend on the scenario.

Rising into 2032

The 25 largest risers outside the top 100
County	Rank now	2032	Type
Hennepin County, MN	135	61	Structural trend
Davidson County, NC	102	71	Structural growth
Pinal County, AZ	144	78	Structural growth
Onslow County, NC	109	81	Structural growth
Hillsborough County, NH	125	88	Structural trend
Carteret County, NC	131	96	Structural growth
Rockingham County, NH	140	111	Scenario-dependent
Salt Lake County, UT	254	115	Scenario-dependent
Lincoln County, NC	180	134	Structural growth
Mohave County, AZ	217	147	Structural growth
Harnett County, NC	195	148	Structural growth
Ocean County, NJ	326	149	Scenario-dependent
Yavapai County, AZ	202	151	Structural growth
Bernalillo County, NM	351	160	Structural trend
Cleveland County, NC	211	164	Scenario-dependent
Wilkes County, NC	224	171	Scenario-dependent
Fairfax County, VA	373	175	Scenario-dependent
Rockingham County, NC	229	181	Scenario-dependent
Rutherford County, NC	248	183	Scenario-dependent
Bergen County, NJ	360	184	Scenario-dependent
Wayne County, NC	251	185	Scenario-dependent
Merrimack County, NH	270	189	Scenario-dependent
Craven County, NC	244	196	Structural growth
Ramsey County, MN	327	201	Structural trend
Middlesex County, NJ	370	218	Scenario-dependent

Watchlist: fast-changing, not yet consequential

Counties outside the top 100, with at least 10,000 expected votes, in a state that is the tipping point at least 1% of the time, and showing an extreme change (at least 2.75 standard deviations) or at least two changes beyond 2.5. Changes are measured against all ranked counties nationally, which is why the list is short (22 counties) and Georgia-heavy (13 of 22).

Watchlist counties
County	Rank	Changing fast in
Burke County, GA	437	employment momentum
Dawson County, GA	205	labor-market momentum, vote growth and migration
St. Joseph County, MI	177	employment momentum
Bibb County, GA	157	structural realignment
Bryan County, GA	266	vote growth
Columbia County, GA	111	vote growth
Effingham County, GA	145	vote growth
Oconee County, GA	260	employment momentum and vote growth
Clarke County, GA	104	structural realignment
Barrow County, GA	108	vote growth and migration
Troup County, GA	279	labor-market momentum
Monroe County, GA	283	migration
Cabarrus County, NC	113	structural realignment
Union County, GA	227	vote growth and migration
Pender County, NC	228	migration
Greene County, GA	410	vote growth and migration
Currituck County, NC	312	vote growth
Santa Cruz County, AZ	433	four-year political trend
Mohave County, AZ	217	vote growth
Yuma County, AZ	277	four-year political trend
Pinal County, AZ	144	vote growth and migration
Apache County, AZ	469	four-year political trend

How to read this, and what it cannot tell you

What we publish about elections

We publish certified election results and documented facts about the 2026 ballot. On two pages, the home page and Counties in focus, we also publish probabilities from a simulation: it is how Counties in focus picks its counties, and the home page shows the same results by state. The simulation starts from the 2024 county results and adds swings whose sizes are estimated from past presidential elections: a national swing, shared swings for regions, states and groups of similar counties, and variation in each county. It uses no polls, candidate information or economic forecasts. For each county it asks one question: across the simulated elections, how often would a different result in this county alone change the Electoral College winner?

These numbers exist to compare places with each other. They are not a forecast of the next presidential election, a measure of how much one person’s vote counts, or a guide to where campaign spending pays off. They change when the assumptions change, and Counties in focus shows by how much. We publish no sentiment scores and no live monitoring feeds.

What the model cannot tell you

It is not a forecast.
Every probability is conditional on a prior built from the 2024 result and the historical spread of national swings (standard deviation about 5.7 points), with no polls, fundamentals, citizen voting-age population or registration. Under this prior Democrats reach 270 electoral votes in 31.8% of simulations and Republicans in 68.1% (ties 0.1%): a property of the prior, not a prediction.
The national environment matters most.
The assumed national environment is the most influential setting. Under “Environment 3 points more Republican” Democrats reach 270 in 15.5% of simulations; under “Environment 3 points more Democratic”, in 54.1%.
Single-county leverage is small and untested.
No past election in the backtest had a single decisive county, so there is no hit rate against real elections. Leverage of correlated blocs of counties is larger than any one county’s.
Information value is model-implied.
It comes from the simulation, not from an empirical bellwether record.
Elasticity is mostly set by the prior.
For 85% of units (2,629 of 3,104) the estimate is prior-dominated; it is always shown with its 90% interval and that flag.
Turnout and persuasion cannot be separated.
County totals cannot tell them apart, so the related fields in the data file are bounds.
Economic labels are descriptive.
No robust link between economic change and vote movement was found, associations are not causes, and p.vol is not computed.

Definitions, data and sources

What each measure means7 measures

Chance of changing the presidency: The chance that a fresh draw of this county's own result changes the Electoral College winner. It needs a swing bigger than the state margin and a state that is pivotal in that same simulated election, so values are small; the ranking is the useful part.
Chance of flipping its state: The chance that the county's own variation, by itself, changes the statewide winner. It ignores whether that state then matters nationally.
Information about the national winner: How much knowing this county's result would reduce uncertainty about who wins nationally, in bits. Model-implied: it mostly reflects how closely the county tracks the national swing and has not been checked against a record of real bellwethers.
Chance of a large swing: The chance that the county's two-party margin moves five points or more from 2024 in either direction. Small or unusual counties dominate; this is not about the Electoral College.
Expected net-vote movement: The expected absolute change in the county's net votes (Democratic minus Republican) from 2024. Large counties dominate, including many that are safe for one party.
Partisan trend: The change in the county's margin relative to the nation, per presidential cycle, over the last 12 years. Descriptive only: the baseline model does not extrapolate it.
Response to national swings (elasticity): How strongly a county follows national swings, with a 90% interval. Flagged prior-dominated when the model’s prior carries more than 80% of the weight.

Data tiers and qualityQuality scores 0.52–0.96

2020 and earlier results come from audited releases where available and from a research-tier compilation elsewhere; 2024 county results are a secondary compilation. Every county carries a data-quality score (here between 0.52 and 0.96, median 0.72).

2,621 counties: research-tier history, 2024 secondary compilation, audited release
471 counties: 2024 secondary compilation, audited release
10 counties: research-tier history, 2024 secondary compilation

Units that are not ordinary countiesAlaska, Connecticut, Kansas City, Maine, Nebraska

Alaska and Connecticut are each simulated as one state-level unit and not ranked (37 county-level rows are listed with that status).
Kansas City, Missouri is one composite of three counties and the city election board, because the board’s votes cannot be allocated to counties.
Maine and Nebraska congressional districts use county proxies, not official allocations.
1 county has no usable presidential history and is not modeled. In all, 3,102 county-level units are ranked out of 3,145 rows in the data file.

Sources and file hashes24 files, SHA-256

Published from county_consequentiality_v1 (model run 2026-10-06). Publication 35ba9b37fa41eed1…. Every number on this page is read from these files at build time; nothing is typed into the copy.

File	SHA-256
`county_consequentiality_v1/derived/viz_data.json`	`258f8431468ca6a7…`
`county_consequentiality_v1/docs/EXECUTIVE_FINDINGS.md`	`78d61b2bf867e5cb…`
`county_consequentiality_v1/docs/METHODOLOGY.md`	`543000eb76def3b5…`
`county_consequentiality_v1/results/county_metrics.csv.gz`	`22839c7c9960c9e5…`
`county_consequentiality_v1/results/emerging_map_counties.csv`	`38270b75351fa892…`
`county_consequentiality_v1/results/findings.json`	`7895c1bce5ccadec…`
`county_consequentiality_v1/results/s06b_calibration.csv`	`8a2fd0f35b2d06de…`
`county_consequentiality_v1/results/s06b_county_backtest.csv.gz`	`569d70548de3f6e0…`
`county_consequentiality_v1/results/s06b_ranking_validation.csv`	`073288e4252863f9…`
`county_consequentiality_v1/results/s06b_state_backtest.csv`	`96dc0d4eb39482e8…`
`county_consequentiality_v1/results/s07_oos_tests.csv`	`21002ef3bd547af5…`
`county_consequentiality_v1/results/s07_regression_results.csv`	`f5b7ef73e3d046e1…`
`county_consequentiality_v1/results/state_breakdown/AZ.csv`	`87f2dab41a314476…`
`county_consequentiality_v1/results/state_breakdown/GA.csv`	`f69a814e84bbeb80…`
`county_consequentiality_v1/results/state_breakdown/MI.csv`	`4fd2c95b5e44ecb2…`
`county_consequentiality_v1/results/state_breakdown/NC.csv`	`dbbbcdc31602a63e…`
`county_consequentiality_v1/results/state_breakdown/NV.csv`	`ed52c5048398f09e…`
`county_consequentiality_v1/results/state_breakdown/PA.csv`	`bf6229b3c06899cf…`
`county_consequentiality_v1/results/state_breakdown/WI.csv`	`eab02a2e489f8d51…`
`county_consequentiality_v1/results/state_breakdown/_summary.json`	`1600bef27ae7946e…`
`county_consequentiality_v1/results/state_unit_table.csv`	`2bb7c0e987d2f5b0…`
`county_consequentiality_v1/results/watchlist.csv`	`6360c227d2dfcc60…`
`county_consequentiality_v1/s09_outputs.py`	`09b835dd6d234275…`
`web/public/cost/geo/counties-10m.json`	`145aaf5d1433352a…`

This week, nationally

Which counties could tip a simulated presidential election?

Seven states hold 97% of the tipping-point chances

Deciding a state is not the same as signalling the result

Who is each county working for?

The same four groups on the map

Pick a county

Every county, as a table

Explore every county

Good for whom? Every county through two lenses

What goes into each lens

For people already here

For people getting started

Four groups

How the 100 counties in focus are doing

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs

Income & housing strain

Costs

People

Jobs