# Your Vote Doesn't Matter: full text digest

This document is the complete content of yourvotedoesntmatter.com in plain text, generated from the same datasets the site serves, for readers and AI assistants that cannot run the interactive app.

## What the site computes

For a voter in each state, the probability that their single vote decides the presidency. That probability factors into two events, both of which must happen: (1) the voter's state finishes in an exact tie without them, so their vote breaks it, and (2) the state's electoral votes are exactly the ones that carry the winner to 270. The site also computes the same per-race odds for a complete South Carolina ballot.

The answer for the most competitive state, Nevada: about 1 in 2,131,055. For South Carolina: about 1 in 26,644,207,035. These are not rhetorical numbers; the model, its inputs, and its validation are documented below.

## Presidential odds by state

Expected margin is the model's expected two-party margin in the state (positive leans Republican). "Decides the presidency" requires both a tied state and a pivotal Electoral College outcome; "ties the state" is the first event alone.

| State | Electoral votes | Expected margin | Chance your vote decides the presidency | Chance your vote ties the state |
|---|---|---|---|---|
| Nevada | 6 | 0.4 pts Rep | 1 in 2,131,055 | 1 in 146,286 |
| Wisconsin | 10 | 0.4 pts Rep | 1 in 2,591,619 | 1 in 365,353 |
| Pennsylvania | 19 | 0.7 pts Rep | 1 in 2,696,870 | 1 in 719,835 |
| Georgia | 16 | 2.2 pts Rep | 1 in 2,739,409 | 1 in 619,586 |
| North Carolina | 16 | 2.8 pts Rep | 1 in 3,163,773 | 1 in 717,480 |
| Michigan | 15 | 0.1 pts Dem | 1 in 3,973,446 | 1 in 690,466 |
| Arizona | 11 | 3.5 pts Rep | 1 in 4,066,097 | 1 in 537,009 |
| New Hampshire | 4 | 3.7 pts Dem | 1 in 4,173,485 | 1 in 135,200 |
| Maine | 4 | 6.6 pts Dem | 1 in 7,530,219 | 1 in 226,660 |
| Minnesota | 10 | 4.6 pts Dem | 1 in 7,921,984 | 1 in 613,914 |
| Alaska | 3 | 12.7 pts Rep | 1 in 12,310,412 | 1 in 510,328 |
| Iowa | 6 | 11.0 pts Rep | 1 in 23,207,793 | 1 in 1,495,742 |
| Texas | 40 | 10.3 pts Rep | 1 in 25,829,930 | 1 in 8,522,198 |
| Virginia | 13 | 6.9 pts Dem | 1 in 34,510,468 | 1 in 1,717,068 |
| Florida | 30 | 7.8 pts Rep | 1 in 36,485,717 | 1 in 7,166,707 |
| Colorado | 10 | 10.1 pts Dem | 1 in 39,904,244 | 1 in 2,257,184 |
| New Mexico | 5 | 7.9 pts Dem | 1 in 44,004,583 | 1 in 649,288 |
| Utah | 6 | 20.6 pts Rep | 1 in 47,211,816 | 1 in 3,082,932 |
| Ohio | 17 | 9.7 pts Rep | 1 in 53,730,356 | 1 in 7,030,886 |
| New Jersey | 14 | 10.4 pts Dem | 1 in 89,433,221 | 1 in 4,500,727 |
| Delaware | 3 | 15.3 pts Dem | 1 in 284,726,562 | 1 in 2,726,257 |
| Rhode Island | 4 | 16.2 pts Dem | 1 in 302,087,522 | 1 in 4,008,701 |
| Montana | 4 | 19.0 pts Rep | 1 in 1,109,935,074 | 1 in 30,112,539 |
| Indiana | 11 | 18.2 pts Rep | 1 in 1,386,676,034 | 1 in 96,041,182 |
| Missouri | 10 | 17.7 pts Rep | 1 in 1,436,684,464 | 1 in 87,458,167 |
| Kansas | 6 | 16.7 pts Rep | 1 in 1,454,595,139 | 1 in 47,659,810 |
| Connecticut | 7 | 15.9 pts Dem | 1 in 2,706,710,280 | 1 in 30,694,095 |
| Illinois | 19 | 13.8 pts Dem | 1 in 6,431,148,327 | 1 in 83,701,395 |
| Oregon | 8 | 14.1 pts Dem | 1 in 12,088,134,941 | 1 in 60,863,759 |
| Nebraska | 5 | 21.3 pts Rep | 1 in 14,128,699,648 | 1 in 306,168,921 |
| South Carolina | 9 | 15.3 pts Rep | 1 in 26,644,207,035 | 1 in 367,690,057 |
| New York | 28 | 17.6 pts Dem | 1 in 38,825,916,979 | 1 in 616,167,302 |
| Mississippi | 6 | 20.1 pts Rep | 1 in 414,813,071,871 | 1 in 3,575,688,680 |
| California | 54 | 24.6 pts Dem | 1 in 712,263,318,870 | 1 in 15,961,820,976 |
| South Dakota | 3 | 28.4 pts Rep | 1 in 883,995,166,949 | 1 in 11,553,816,832 |
| North Dakota | 3 | 35.7 pts Rep | less than 1 in a trillion | 1 in 57,497,449,171 |
| Washington | 12 | 18.2 pts Dem | less than 1 in a trillion | 1 in 12,252,561,781 |
| Massachusetts | 11 | 28.2 pts Dem | less than 1 in a trillion | 1 in 221,048,774,411 |
| Louisiana | 8 | 20.5 pts Rep | less than 1 in a trillion | 1 in 106,631,791,699 |
| West Virginia | 4 | 41.1 pts Rep | less than 1 in a trillion | less than 1 in a trillion |
| Kentucky | 8 | 29.0 pts Rep | less than 1 in a trillion | less than 1 in a trillion |
| Vermont | 3 | 32.2 pts Dem | less than 1 in a trillion | less than 1 in a trillion |
| Tennessee | 11 | 27.1 pts Rep | less than 1 in a trillion | less than 1 in a trillion |
| Maryland | 10 | 29.4 pts Dem | less than 1 in a trillion | less than 1 in a trillion |
| Alabama | 9 | 28.6 pts Rep | less than 1 in a trillion | less than 1 in a trillion |
| Hawaii | 4 | 26.8 pts Dem | less than 1 in a trillion | less than 1 in a trillion |
| Arkansas | 6 | 29.0 pts Rep | less than 1 in a trillion | less than 1 in a trillion |
| Idaho | 4 | 33.8 pts Rep | less than 1 in a trillion | less than 1 in a trillion |
| Wyoming | 3 | 45.6 pts Rep | less than 1 in a trillion | less than 1 in a trillion |
| District of Columbia | 3 | 85.7 pts Dem | less than 1 in a trillion | less than 1 in a trillion |
| Oklahoma | 7 | 34.3 pts Rep | less than 1 in a trillion | less than 1 in a trillion |

## How the model works

Expected margins blend the state's last three presidential results (weights 0.5, 0.3, 0.2 for 2024, 2020, 2016), computed from county returns. The margin's spread combines the state's historical cycle-to-cycle volatility with a floor, clamped between 4 and 10 percentage points. The probability of a one-vote margin is the density of that normal distribution at zero, divided by expected turnout. The pivot probability comes from a 200,000-run Monte Carlo: given the state is tied, a shared national swing (standard deviation 2.5 points) is drawn conditional on the tie, and the state is pivotal when the rest of the map lands within the state's electoral votes of 270. Maine and Nebraska are treated as winner-take-all, a documented simplification that moves nothing at these magnitudes.

Down-ballot races use the same construction per office class, with margins blended over contested cycles only and volatility pooled across races of the same class (state legislative districts were redrawn in 2022; only same-map cycles are paired). Races whose most recent cycle was uncontested are reported as uncontested and assigned no probability.

## South Carolina ballot

185 races are covered. 94 had no opposition in their most recent cycle and are listed as uncontested. The most decisive contested races:

| Race | Chance your vote decides it |
|---|---|
| State House District 55 | 1 in 1,982 |
| State House District 102 | 1 in 2,052 |
| State House District 75 | 1 in 2,455 |
| State House District 116 | 1 in 3,321 |
| State House District 54 | 1 in 3,729 |
| State House District 115 | 1 in 5,044 |
| State House District 15 | 1 in 6,320 |
| State Senate District 29 | 1 in 8,094 |
| State Senate District 17 | 1 in 8,332 |
| State House District 12 | 1 in 8,702 |
| State Senate District 35 | 1 in 8,812 |
| State House District 64 | 1 in 8,846 |
| State Senate District 40 | 1 in 9,363 |
| State House District 122 | 1 in 9,368 |
| State Senate District 39 | 1 in 12,733 |

A full Berkeley County ballot (President, Governor, U.S. House, State Senate, State House, school board) aggregates to roughly a 1 in 1,760 chance that at least one race on it is decided by the voter, computed live by the ballot tool.

## The cost of voting

About $30 and 40 minutes: 19 minutes at the polling place on average (GPS smartphone study of 10 million voters), about 13 minutes waiting in line (Stewart/CCES estimates), about 10 minutes of travel each way (stated assumption), time valued at about $36 per hour (Bureau of Labor Statistics average hourly earnings) for roughly $24, plus about 8 miles of driving at the IRS rate of $0.70 per mile for roughly $6. Tranquility is not priced.

## Validation

Across eight perturbations of the model (different volatility floors, weighting schemes, a fatter-tailed error distribution), competitive states move by less than a factor of two, while safe states swing widely in the far tail without ever becoming plausible:

| State | Published | Range across all variants |
|---|---|---|
| Nevada | 1 in 2,131,055 | 1 in 1,673,869 to 1 in 2,917,805 |
| Wisconsin | 1 in 2,591,619 | 1 in 2,089,702 to 1 in 3,893,664 |
| Pennsylvania | 1 in 2,696,870 | 1 in 2,197,985 to 1 in 4,174,650 |
| Georgia | 1 in 2,739,409 | 1 in 2,531,725 to 1 in 3,789,226 |
| North Carolina | 1 in 3,163,773 | 1 in 2,850,051 to 1 in 4,313,763 |
| South Carolina | 1 in 26,644,207,035 | 1 in 110,553,574 to 1 in 565,677,010,890 |
| California | 1 in 712,263,318,870 | 1 in 29,497,973,316 to less than 1 in a trillion |

Walk-forward calibration: the model is fitted using only data from before the target election, then graded on what actually happened. Columns show how many states' true margins fell inside the model's 50/80/95 percent intervals (targets 50/80/95), the standard deviation of standardized errors (1.00 is perfectly sized uncertainty), and predicted vs actual shares of states decided by two points or less. The 2016 test rests on only two prior cycles and a thinner source file, and grades worst, as the methodology page discloses:

| Test | Interval coverage | Error sd | Close races |
|---|---|---|---|
| Fitted blind, graded on 2020 (51 states) | 43/82/96 | 0.96 | 6.4% predicted vs 9.8% actual |
| Fitted blind, graded on 2024 (51 states) | 55/82/92 | 0.73 | 7.1% predicted vs 5.9% actual |
| Fitted blind, graded on 2016 (50 states) | 18/40/64 | 1.79 | 4.5% predicted vs 12.0% actual |

Empirical check, following Mulligan and Hunter (Public Choice, 2003): the same model was fitted walk-forward to 6,675 contested US House races, 1978 to 2024 (MIT Election Lab returns), and its predicted number of photo-finishes was compared with the returns. Decided by 1 vote or fewer: 0.08 predicted, 0 occurred (a naive coin-flip model predicts 35.31). By 10 votes: 0.57 vs 2. By 25: 1.38 vs 3. By 100: 5.44 vs 8 (naive model: 2282.16). No US House race in 48 years was decided by one vote; the closest were Indiana's 8th district in 1984 (4 votes) and Iowa's 2nd in 2020 (6 votes). A follow-up recalibration, fitted on 1978-1999 and tested on 2000-2024, found the uncorrected model nearly exact on the holdout (1.97 races predicted within 100 votes, 2 occurred), so no correction was adopted.

## The usual objections, priced

"Local races are decided by a handful of votes." Sometimes, and the site counts them: that is what the ballot tool is for. Take the strongest case in the data: South Carolina House District 55, the most decisive legislative race in the state, gives one of its voters about a 1 in 1,982 chance per election. That is the maximum, picked from 185 races after the fact. The same ballot carries 94 races with exactly one candidate, a complete real ballot with every race combined comes to about 1 in 1,760, and even District 55 asks for roughly 2,700 years of voting in it for a coin flip's chance of ever mattering. The famous squeakers underline it: a race decided by six votes (Iowa, 2020) was not decided by any one of its voters, since removing any single ballot leaves the winner ahead by five. The only finish at which one vote is the whole story is a one-vote margin, and in 48 years of US House elections it has never happened. Local is smaller. The closest local race in the state is still not close, and the typical one is not a race at all.

"The margin is made of votes. Collectively, they matter." Collectively, yes. That is a statement about millions of people, not about any one of them. Take the closest case in modern memory: Arizona's 2020 presidential race was decided by 10,457 votes. Not one of those 10,457 voters decided it; remove any single ballot and the outcome is identical, 10,456 times over. Closeness is not a rebuttal to this arithmetic, it is an input to it: it is exactly why Nevada's number is 1 in 2.1 million instead of 1 in 26 billion. A crowd can be decisive while every member of it is replaceable, the way an army wins a war without any single rifle being the reason.

"Voting sends a signal. Campaigns pay attention to turnout." Campaigns pay attention to blocs: demographics, donors, and turnout operations that move thousands of votes at once. Take the state they watch most closely: in Nevada, one ballot shifts the expected margin by about 0.00007 percentage points, below the rounding error of any spreadsheet a campaign has ever opened. The signal theory inherits the same arithmetic as the tie theory: a vote too small to break a tie in the closest state in the country is too small to be the part of the signal anyone reacts to in any state. Turnout operations matter precisely because they are not one vote.

"At least you get a say." The say is the ballot, and the site is the measurement of how much say it is. Take the maximum on offer anywhere in America: a Nevada vote is one 2.1-millionth of a presidential decision, the largest say any voter in the country gets. Calling the act itself the benefit is the expressive theory of voting, and it may even be true. It is just a consumer good, priced like one: about $30 and 40 minutes for a sticker, a feeling, and a story to tell. Those are real. They are not influence, and relabeling them does not change the exchange rate.

"I'm voting for other people. It's altruistic." The altruist's arithmetic is the strongest version: a tiny chance multiplied by millions of affected people can beat $30. Grant it the best ballot in America, a Nevada voter's 1 in 2.1 million instead of the average 1 in 60 million. The formula still quietly assumes your side winning is a benefit to those millions, and on standard policy-knowledge surveys most voters score poorly enough that the sign of that benefit is close to a guess. Altruism toward an outcome you cannot evaluate is loyalty with a halo. And if helping people at scale is really the goal, voting is dominated on its own terms: the same $30 given to a cause with a measurable return buys orders of magnitude more expected good, at odds that are not millions to one against. The genuinely altruistic version of voting costs far more than a ballot: years spent getting actually informed on economics and policy, for a return you will never see. Almost nobody pays that price. The people who call voting altruism stop at the cheap part.

"If everyone thought that way, nobody would vote." It would not get there. The objection imagines a cliff where the math has a thermostat: as turnout falls, each remaining vote gets stronger, the same density arithmetic divided by fewer people, so long before participation reached zero it would cross the line where voting is worth it for the marginal voter, and settle there. Collapse requires the incentive to shrink as the crowd does. It grows. The argument also grades a rule, not your decision: you are not choosing whether millions of people vote, only whether you do, and they are not waiting for your example. Universalized, it obliges you to do literally everything: what if nobody farmed, what if nobody enlisted, what if nobody fixed teeth? And the collapse has had its chance regardless: this arithmetic has been in textbooks since 1968, turnout tracks habit and mobilization rather than political science syllabi, and the republic is still standing in line.

"It expresses who I am." It does, and the mechanism has a name: coalitional psychology. Humans are team animals, and a ballot is a jersey you wear once every two years; the warm feeling afterward is the feeling of allegiance, which is why it is comforting and why it is not evidence about policy. The enforcement arm gives the game away: defectors get punished. "You didn't vote?" arrives as a moral accusation rather than a scheduling observation, because the ritual is partly the tribe collecting dues. None of this makes the feeling fake. It makes it a feeling, priced at about $30 and 40 minutes, which is what team spirit costs when the stadium is a polling place.

## Sources

Presidential and House returns: MIT Election Lab and Open Elections; House history 1976-2024 via Harvard Dataverse (doi:10.7910/DVN/IG0UN2). Address-to-district lookup: US Census Geocoder, with an OpenStreetMap fallback disclosed on the privacy page. Academic anchors: Gelman, Silver and Edlin, "What Is the Probability Your Vote Will Make a Difference?", Economic Inquiry 50(2), 2012 (average voter about 1 in 60 million in 2008); Mulligan and Hunter, "The Empirical Frequency of a Pivotal Vote", Public Choice 116, 2003.

## Machine-readable endpoints

- https://api.yourvotedoesntmatter.com/api/v1/states (all states, JSON)
- https://api.yourvotedoesntmatter.com/api/v1/states/{abbr} (one state, JSON)
- https://api.yourvotedoesntmatter.com/api/v1/states/SC/races (South Carolina races, JSON)
- POST https://api.yourvotedoesntmatter.com/api/v1/ballot (address in the request body only, never stored; returns the voter's races and odds)