This site asks where a dollar does the most good. Every two years a different dollar competes for the same wallet: the political one. In the run-up to November 2024, OpenSecrets projected total spending on federal races at about $15.9 billion, the most ever. A lot of that came from small donors who believed the money would change an outcome.
The evidence on that belief is unusually clear, and it changes how a political donation should be compared with a charitable one, and with the ballot itself.
What a political dollar buys
Persuasion: almost nothing, in general elections. The largest review of the question, by Joshua Kalla and David Broockman in 2018, pooled 49 field experiments covering more than 150,000 people and found the average persuasive effect of campaign contact and advertising on vote choice in general elections was, in their words, best estimated at zero. Effects appear in two narrow cases: primaries and ballot measures, where voters have weaker priors, and very early contact whose effect then fades. Their follow-up on the 2020 presidential race reached the same conclusion.
Turnout: real but expensive. Getting an existing supporter to actually vote works. Alan Gerber and Donald Green’s decades of randomized experiments put door-to-door canvassing at roughly one extra vote per 14 people reached, or about $19 per additional vote at typical labor costs. Mail and phone banks cost more per vote; TV ads are far worse.
Where it does matter. Money buys ballot access, staff and the ability to be heard at all. A candidate with nothing cannot compete. Past a threshold, though, the marginal million buys ads that the research says do not move general-election voters. In a presidential race the threshold was passed long ago; in a school-board race it may not be.
So a political dollar has a decent return at the bottom (a low-budget local race, a primary, a ballot measure) and a return near zero at the top (a well-funded general election for federal office), which is exactly where most of the $15.9 billion went.
What a charitable dollar buys
For comparison, the best-evidenced charitable programs are documented on this site in detail: about $3,500 to $5,500 per child’s life saved for GiveWell’s top charities, a few dollars per healthy year of life for lead-paint regulation, tens of dollars per year of better mental health. These are marginal estimates with real error bars, but they are measured, published and repeatedly audited. No campaign publishes anything comparable, because there is nothing comparable to publish.
What a vote is worth
A vote is free, so “impact per dollar” is the wrong frame; the right one is impact per hour. Two things decide it.
How likely it is to change the result. Gelman, Silver and Edlin estimated the chance that one vote decides a US presidential election at about 1 in 60 million nationally, rising to around 1 in 10 million in the most competitive states. That sounds like nothing until you multiply by what is at stake for 330 million people: the expected value is not small. In local races the arithmetic is different in kind. City council, school board and county contests are regularly decided by a few dozen votes, and some by one.
How many people each vote speaks for. A state senator represents perhaps 100,000 people; a US senator from California represents 39 million. One vote in the first race carries hundreds of times the representational weight of one vote in the second. Combine the two factors and the races where a single ballot matters most are almost always the ones at the bottom of the ballot that most people skip.
Our sister site, Elections Simplified, rates every race on a voter’s ballot on exactly these two dimensions, how concentrated the representation is and how competitive the race is, and shows what each candidate has actually done. Its mission page explains the method.
Putting them side by side
| Political donation (federal general election) | Political donation (local race, primary, measure) | Charitable donation (top program) | Voting the full ballot | |
|---|---|---|---|---|
| Cost | $ | $ | $ | About an hour, including the research |
| Best evidence of effect | Near zero on persuasion; $19+ per turnout vote | Plausibly meaningful; rarely measured | $3,500 to $5,500 per life saved, published models | Decisive in a measurable share of local races |
| Who measures it | Nobody, publicly | Nobody, publicly | Independent evaluators | Election results |
What this suggests
- If you give politically, give down-ballot and early. A local race or a primary in which your candidate is under-funded is where the research says money can still matter. A federal general election that already has nine figures behind each side is where it cannot.
- Do not treat a political gift as a substitute for a charitable one. They are not measured on the same scale, and the one with public evidence of cost per outcome is the charitable one.
- Spend the hour, not just the dollar. The bottom of the ballot is where one vote carries the most weight and where the least information reaches voters. That hour of reading is the cheapest high-leverage act available to you, and it is not for sale.
Method notes
Spending figures are OpenSecrets’ October 2024 projection for the 2024 federal cycle. Persuasion findings are as stated by Kalla and Broockman (2018) and Broockman and Kalla (2023). The $19 per vote canvassing figure is Gerber and Green’s, as reported by the Chronicle of Philanthropy; it assumes canvassers reach about 12 people an hour at $16 an hour. Vote-probability estimates are from Gelman, Silver and Edlin for the 2008 election. Charitable figures are GiveWell’s 2022 to 2024 averages. Representation figures are rounded population counts. Elections Simplified is published by the same team as this site.
Sources
- OpenSecrets: Total 2024 election spending projected to exceed previous record (Oct 8, 2024)
- Kalla and Broockman: The minimal persuasive effects of campaign contact in general elections, evidence from 49 field experiments (American Political Science Review, 2018)
- Broockman and Kalla: When and why are campaigns' persuasive effects small? Evidence from the 2020 U.S. presidential election (AJPS, 2023)
- Chronicle of Philanthropy: Cost of door-to-door canvassing is $19 per vote, research finds (on Gerber and Green, Get Out the Vote)
- Gelman, Silver, Edlin: What is the probability your vote will make a difference? (Economic Inquiry, 2012)
- GiveWell: Our top charities
- Elections Simplified: what is on your ballot and how much one vote weighs in each race
Figures come from the cited sources as of the dates shown. Cost-effectiveness estimates change as programs and evidence change; check the source before giving. We are not financial advisers and receive nothing from any charity named here.