How would Americans vote if there were a popular vote?

Logo of the NPV Interstate Compact

There's been a recent surge in talk about how to switch US presidential elections to be based on a popular vote rather than the electoral college. As I have pointed out before, there is no such thing as the US popular vote even though the press likes to add the real popular votes in the swing states to the nobody-cares surveys of the safe states and call it "the popular vote."

The National Popular Vote Interstate Compact got excited by adding Connecticut's 7 votes this year, but I think that's actually a bad sign for the compact, not a positive one. The compact hopes to get 270 electoral votes so it can be activated, but the 12 states which have joined it are all "blue states" which reliably support the Democratic Party. Adding another bright blue state just makes it look more and more partisan, and that makes it more and more difficult to get a "red" state to consider joining it, and it can never get to 270 with just blue states. (On the other hand, the compact has passed in one house in some red states, notably Arkansas and Oklahoma, so it is not entirely without hope, but one house is just a beginning.)

Ten years ago I pointed out a better way to do a compact needs just 3 or 4 safe states -- if they are the right states, a bipartisan collection representing about 50 blue votes and 50 red votes.

But if we did either of those things, the unanswered question is "what would the popular vote look like, if there were a real popular vote, and every voter's vote counted the same?" It might be quite different from what we see today in the safe states.

In order to figure out what it would be, we would need to use polling and statistical methods to figure out how all the factors in safe states alter the results. Or we might be better off in just doing polling. Most polls try to survey "likely voters" but whether somebody is a likely voter or not depends on whether they think their vote will count for anything.

Let's consider the factors we need to undo:

Lower turnout

As noted, when voters don't think their vote counts, they are less likely to vote, and vote for different reasons. Of course, the votes in down-ticket races still count, but sadly turnout in Presidential elections is 50% higher than in off-year elections. A large fraction of voters really only turn out when they can vote for President.

Many still turn out even when they don't feel their vote counts, but not as many. My analysis of 2016 suggests the average turnout was 3% less (out of the total) in safe states than swing states. It's actually impressive that so many do come out to vote. We need to understand why they are voting, since they know their vote will not change anything. In essence, all voting in a safe state does is add your number to the total printed in the paper for your state, and in the sum the newspapers call the popular vote.

That's hard to read. In general, polls which measure how much you care about registering your view when the poll is publish are considered highly unscientific and invalid. Voters are what is called self-selected. Whether you vote is a measure of how much you care about the survey as much as it measures your opinion. In a scientific survey, subjects must be selected through a carefully designed process that avoids bias of this sort.

On the other hand, these voters do care, since voting takes some work (compared to filling out a survey) so we might be able to find correlations between how they vote and how they would vote if it actually counted.

Inverse lower turnout

Perversely, we might see voter engagement and turnout decrease in some of the swing states. Voters in those states see tremendous attention, money and "bribes" from candidates. Small swing states that used to get candidate visits and heavy campaigning might find themselves deserted, with some GOTV effort, but much less. Candidates would be very unlikely to personally visit low population areas and motivate voters.

Likewise, more people, nationwide, might feel "there is no chance my vote will change the result." While the chance of that is always low, in a very close swing state, voters can get very motivated because that chance is more real. With a popular vote, it would be the same, very low, probability everywhere in the nation.

Third party votes

In safe states, voters can vote for third parties without any risk of altering the national result in a way they don't like. In theory, this should allow third parties to more easily get support, which can demonstrate their growing strength. In fact, this is the only thing third parties get from elections, since they won't win. While there are still many 3rd party voters in swing states, those voters get told they are "spoiling" or "throwing away" their vote. It will become very difficult for a third party to rise from minor status -- though it already is very difficult. The use of a multi-candidate ballot, such as Approval voting or ranked choices is the best answer for this.


Perhaps the biggest difference is money. Presidential campaigns raise about $6B -- and spend almost all of it in the swing states. If you live in those states you are bombarded with ads and get out the vote efforts. A lot. In the safe states, it's crickets. You get exposed to a few national ads, and today, you will see ads on sites like YouTube and your social network feed, but nothing like what you see in a swing state.

Many people think that money has a serious and dangerous effect on politics. I've seen many people decry the Citizens United decision, which allowed corporate super-PACs to spend heavily in elections, and declare it the greatest threat to democracy out there. They're even ready to weaken the 1st amendment to stop it! But if so, none of that money is spent in the safe states. How can we calculate how they would vote if they got exposed to all that money. Well, not all of it, because now it would not be concentrated in a small number of states, so everybody would see the ads, but see fewer of them.


Related to money is the non-monetary part of a campaign. Visits. Calls. Door-knocking, and most importantly "get out the vote" efforts which focus on getting weak supporters motivated to show up. Nobody bothers to do this in the safe states, and it may account for some large fraction of that lower turnout.

One portion of campaigning does take place in the safe states -- fundraising -- to get money to spend in the swing states. But to fundraise, candidates do visit and sometimes do rallies, but mostly they do fundraising dinners for bigger donors.

With both of these, we might imagine very little connection between safe state voting and voting in a real election. Today, thanks to the internet, a lot more messages do go out nationally, and of course there are the traditional national messages that come from national media and things like debates. Voters in safe states see a much less hyped campaign, and probably vote more on a reasoned look at the issues, because nobody is spending billions to sway them. The more a voter cares, the more they are likely to see, because they have to go seek out election information rather than have it thrust at them.

Voter suppression

There is less motive for parties to do various political tricks, including voter suppression, in safe states. Especially the ultra-safe states like California, where even the down-ticket races are largely decided in advance. Why take the serious effort of voter suppression if it's not likely to win you much?

Where is the real data?

In order to figure out the real values, there are some tricks we might do

  • Expensively, perform surveys just before the election in both safe and swing states with different methodologies, and compare the survey results to actual results. In particular, in safe states, ask voters about what they would do if they lived in a swing state, and try to find how much that differs from their safe state intentions. (Honesty may be hard to measure here.)
  • Study states which switched between safe and swing, possibly on a county by county basis to control for other changes, and look for patterns.
  • Look in safe states for regions with highly contentious house and local races which bring in more serious voters and have a resulting large boost in turnout (ideally 50% boost.) Their vote for President probably mirrors what they cast in these polls.
  • Look at voters who believe their vote counts, even though the campaigns have judged the state safe and are not acting in the states.
  • Get a group of voters in a safe state and have them consume all media from swing states, loaded with Presidential advertising. Observe their difference from a control group. Possibly use people who moved from swing states who won't mind watching those TV stations and reading those newspapers and other media. Take them to rallies via video feed.
  • Calculate effect of voter suppression in swing states, and apply it to surveys in safe states, or to actual votes from polling stations with similar demographics.

Some people have the intuition that all these effects will affect the two parties equally, and thus a real popular vote would be similar to past numbers. ie. that we can still look at California's numbers, even though no campaigning was done, because neither party spent any resources there. We don't have enough information to say that yet, but perhaps we could test that hypothesis in some fashion. Money is distributed far from equally. One side usually raises more money than the other, and they decide to spend it differently. The highly local ad becomes less effective (and doesn't scale to the whole country) so more national advertising would be done.

One difference might be that physical campaigning (meetings, GOTV) is more expensive in low density areas and scales better in cities. Democrats have most of their base in cities. If the amount you can spend per voter drops, this might alter things.

But overall it is very difficult to come up with an accurate methodology. It's hard to get true answers from voters about a hypothetical. After all, every voter answering a poll thinks their answer matters, so they are not going to admit (even to themselves) that their voting behaviour would change if they knew their vote counted.


By and large, voter suppression is a Republican phenomenon. That, plus the strong gerrymandering that took place in the wake of the 2010 GOP victories, is a result of the fact that the Republican party is losing the demographic battle year by year. You may sneer at the popular vote as meaningless, but the fact is that Republican presidential candidates haven't won it since 2004, and haven't won it convincingly since 1988. This is not a healthy thing for a supposedly national party.

Would an actual popular vote tally be different from the popular vote we see now under the electoral college system? As you point out, that's really hard to say. But I think we have to assume it wouldn't be hugely different. Overall, today's safe states would see higher voting participation, and swing states lower. But the effects would likely balance, both along the safe vs. swing axis and the Republican vs. Democrat axis.

WRT money, particularly when spent on TV, I think its effect is overrated. At campaign time, we see massive numbers of TV ads, generally using the same language and attack style, often one after the other attacking either side of a contested race. After a certain point, these become just noise. So long as both sides can air ads in some quantity, neither achieves much with them. At one time, TV might have been an effective campaign tool. By now, TV attack ads are so generic and overused, they have little to no effect.

Likewise, I think a lot of the sophisticated campaign targeting techniques pioneered during the Obama campaigns may not be as effective as anyone thinks. Having been on the ground implementing high-tech campaigns, I think they are overrated. They tend to send volunteers to nearly random doors, and as often irritate voters as motivate them. The 2016 Clinton campaign was highly data-driven, using much of the same tech as Obama had used previously. I visited their national offices early on, and they were large and sophisticated. Yet in the end all that tech was drowned out by events, especially the Comey emails debacle.

Despite what many think, modern US presidential elections are massive, stupid beasts. Their outcomes swing on what are often the least-informed voters, and despite all the expertise and money poured into them, run in many ways on a kind of catatonic autopilot. Witness, for example, the fact that with just a couple of exceptions, every presidential election since 1952 has run on a two-terms-and-switch-parties pattern. Eisenhower-Eisenhower, then Kennedy-Johnson, then Nixon-Nixon, then Carter (only instance of single party term), then Reagan-Reagan-Bush (only instance of a third term for one party), then Clinton-Clinton, Bush-Bush, Obama-Obama, and now Trump. Through massive societal changes, the pattern stays. I don't think that's a coincidence.

WRT any popular vote scheme, Republicans, having benefited twice from an Electoral College minority vote win, will fight it tooth and nail. So the only way it will come about will be if Democrats manage such giant victories in the presidency, legislative branch, and statehouses that they can push through a constitutional amendment. That doesn't seem likely, but OTOH I didn't think Trump would make it to the Oval Office either.

The effect of money is indeed debatable, and possibly overrated. Yet at the same time, I know that in my circle, people are so intensely upset about Citizens United that they have been willing to talk about weakening the first amendment for the first time in history to undo it, which is pretty dramatic. I am not of that view, even remotely.

If money is indeed not that important, it would be great if the candidates could realize that, and then they would stop raising all the money, and stop having to owe favours do donors.

Yes, I have pointed out the switch pattern. It's worse than you thought for the Democrats. Ever since the Republican party was born in the 1850s, not once ever has a Democrat beat the Republican when trying to replace a retiring Democrat president. The closest they came was the very first Republican, Fremont, who lost to a Democrat who had usurped the sitting President at the convention. After that, zip, nada. Clinton was doomed to any Republican under that formula.

And perhaps even the popular vote would not have changed that. While it seems unlikely for Trump, Republicans like Schwartzenegger and Reagan and others have won California. It can happen.

I think money is less effective in high-profile contests, like the presidency and Senate/House campaigns, and for governors. Generally those feature well-enough funded opponents that TV ads reach saturation and everyone tunes them out. But for smaller races, money can do more. Of course, so can a good ground game (which isn't necessarily mutually exclusive with campaign funding). I think 2016 is a good example of how dark money can work on a subterranean level, where people aren't paying attention. The general expectation that Clinton would win turned out to be highly damaging to Democrats in multiple ways. And years of Citizens United funded down-ballot races had produced a hollowed-out Democratic party, so when they lost the presidency, their situation was dire.

For the upcoming midterms, we see the usual calculus turned on its head. Rather than money causing political support, the white-hot Democratic anger level is pushing lots of small-donor money. As a result, most Democrats have plenty of money, despite Citizens United. Of course we can't count on this for every election. Generally, elections awash in money will tend to favor the wealthy and their proxies.

I doubt that anything other than direct legislation will reduce the amounts of money going into politics. And I don't think we should be Pollyannas about the baleful effects of that much money. It forces legislators to spend massive amounts of their time raising money, and political discourse based on television commercials is impoverished and useless. We need to stop thinking of money as a First Amendment expression, and work explicitly to limit it.

Yes, the effect of money on small elections, particularly with candidates who don't have high name recognition, does seem demonstrated. Operation Redmap relied on it, for example.

It does make one wonder, however, why $6B gets donated to Presidential campaigns, and it spent by them. It's much harder to buy influence with the President. Surely only the very largest donors to a President get anything for their money, and not even that much. Yet they donate like crazy, and spend like crazy.

However, we're talking about a world where $6B is spent in 12 states -- $500M per state -- and close to zero is spent on the other states, which is a large enough amount to make a difference. Of course, the money is not spent alone, it is countered with opposite money.

It strikes me that the safe state elections are "purer" elections in some way. The voters are not subject to heavy campaigning. The ones who vote are the ones who care, not people pushed into it by GOTV. The voters who care had to go out and seek to understand each candidate, watch debates, read coverage etc. (Though the news media definitely thrust lots of coverage at them, traditionally that was done in an even handed way.) If it weren't for the counter force, that the more aware voters realized their votes did not count and stayed away.

I am surprised, actually, that safe states still have a much larger turnout in Presidential elections than in off-year. It's clear that voting for President attracts a lot of voters, but it is less clear why casting an impotent vote for President is so attractive. I think people convince themselves the vote is not impotent to some degree, or at least a lot of them do.

I'm from Missouri, which was once considered a swing state, though now it's pretty safe for Republicans. I've both campaigned and voted in presidential elections since 2004 (though the Clinton 2016 campaign didn't really bother much). There are good reasons to do this.

First, elections are not merely about the presidential contest. In every election there will be contested races at many levels, from governorships on down to school boards. Missouri as a whole is Republican, but my area isn't. I've been involved in many races that Democrats ended up winning, even in years when Republicans took the top of the ticket. Further, these local races aren't completely separate from the national contests. I think if the Clinton campaign had bothered to do some organizing in Missouri, they could have elected Jason Kander to Senater rather than Roy Blunt. There's a synergy that's really necessary. It's the reason Claire McCaskill managed to pull out a win in 2012 even as the state went for Romney; the Obama campaign was organized and on the ground, even though they knew there was little chance of winning the state.

Another reason it's important for a presidential campaign to put at least some effort into safe opposing states is that A. You never know if lightning will strike and there may be a chance to win, and B. you want to make your opponent earn their "safe" states. I think it was a mistake for the Clinton campaign to concede Missouri; she could have put in minimal resources, and made the GOP at least try.

But the last reason why campaigns, and voters, vote even if their state is considered safe for the other side, is that it's the right thing to do. One of my proudest votes as a voter was voting for Mondale in '84. I knew he wouldn't win, but it was my chance to officially register my disdain for Reagan, who was an ignorant right winger and whose presidency started the country toward our present catastrophic inequality. And I feel that way every time; beyond the civic duty, it's my way as a citizen to speak my mind. Even in a safe state, it's only considered safe because enough citizens show up and vote. People who give up and stay home are a pestilence, if you asked me.

What is the voter turnout in Missouri in Presidential years vs. off-years? I will bet it's still quite different.

When you voted for Mondale, it wasn't to make him President or win your state. You just wanted to show up in the total printed in the newspaper. Which makes it a very fancy (and better enforced) version of the text message poll on American Idol. Except your text message might alter the result. But you don't feel that way, I know people don't feel that way. They show up, in a way the downticket races don't inspire them, to cast a vote that they should know is impotent and only counts in the newspaper.

If not, the turnouts in safe states would be like the turnouts in off-years.

I think that description really devalues what a vote is, and what it represents. Almost no one's life will be significantly affected by the outcome of American Idol. But politics is about all of us collectively choosing our leadership, and endowing that leadership with massive power, including power of life and death over many of our fellow citizens. And voting isn't just about choice. It also confers a legitimacy on political leadership (or should). Throughout history many societies have had no choice in the selection of their leadership. Democracy here in the US was bought at the price of no small number of lives. Exercising the franchise is so much more than an empty vote for a celebrity.

Does everyone feel this way? Obviously not; witness the huge numbers that don't bother to vote in every election, both presidential and midterm. I think this is a byproduct of living in a rich, peaceful country. We have the luxury of imagining that votes don't matter. Well, they do. And if people don't bother voting, or tell themselves the easy lie that it doesn't matter who wins, they may wake up someday and discover how wrong they were.

But more similar to that than a real election. In the science of measurement, one of the biggest flaws in any survey is what's called the self-selection bias. This occurs when you only survey a subset of the population, and who you survey is their idea rather than the party doing the measurement. Ie. whether you participate depends mostly on whether you care. That's how American Idol and most web polls work.

Now of course, we are quite used to this in elections. We tell everybody it is their duty to vote, but only 60% vote in Presidential elections, 40% in off-year. We have convinced ourselves that this is right, that we truly wish to register the views of only those who take the effort, and that the views of those who don't should not be counted. This is not necessarily true. But in a real election, like in a swing state, we are satisfied with it.

In a safe state, your vote is impotent. This magnifies the self-selection bias. All your vote gets you is the ability to slightly tweak the number printed in the newspaper. It's amazing that so many people vote -- so amazing that it's clear that many voters treat it like it's real.

There have been proposals for systems of voting that would actually measure the will of the people. For example, you could do an election in a population of a million by picking 5,000 eligible voters at random on election day, and making sure they vote. Like, if they don't show up at the polls, then the police and voting registrars come to their house to force them to, and fine them if they have no valid excuse. This would measure the will of the people far more accurately than what we do today -- unless you define the will of the people to be only the will of those with the desire to vote. (Yes, there are some security issues, as you must not allow the voters to contact with vote buyers etc. but there are ways to deal with this.)

In any event, while there are many who treat a safe state election like a real one, there are also millions who don't. Which means the results from safe state elections may differ quite a bit from if there were a national popular vote.

But if we did either of those things, the unanswered question is "what would the popular vote look like, if there were a real popular vote, and every voter's vote counted the same?"

I fail to see how a country in which every voter's vote doesn't count the same can even call itself democratic with a straight face.

Not a strict democracy. It is very common in democratic republics for voters not to have equal weight, and in parliamentary democracies, it is even stronger. How many of the world's democracies actually have all voters counted the same?

But there is a difference between round-off errors and small-number statistics on the one hand and, on the other, bias built in from the start.

"It is a democratic republic"

So was the German Democratic Republic. :-|

The truth is that constitutional rules are great but not enough. Wish it were otherwise. Another irony is the way that the USA, with a strong separation of church and state in the constitution, has a bigger problem of religious influence in politics than many of the countries that have official state religions.

It will become very difficult for a third party to rise from minor status -- though it already is very difficult. The use of a multi-candidate ballot, such as Approval voting or ranked choices is the best answer for this.

PR is an even better solution. A party with x per cent of the votes gets x per cent of the seats. I don't see how a country in which this is not the case can even call itself democratic with a straight face.

I'm talking about the election of the President here, not the legislature, so PR is more difficult. (I have a future planned article about how it's not as impossible as it seems.)

In the USA, the senate is hard wired as geographical. The districts less so, you could at least do proportional districts within a state, the constitution does not block that. And multi-candidate ballots (within a state) are not prohibited by the constitution either, even for President.

For the senate, PR is not possible because the senators are elected in different elections. So each senator is elected alone, and it's just the one with the most votes. Multi-candidate could be done.

Electing a president by popular vote is not a good idea. What has happened is that people only vote for candidates who promise/appear to support positions that are important to them. What we need instead is a leader who will run the country well, and moderate the various conflicts over issues. I propose that people should vote for electors by name, and that the candidates for president be selected after the electors have been chosen. This would severely curtail campaigning directly to the voters, and so severely reduce the need for candidates to raise money and run ads.

This is of course closer to the original system, and doesn't require constitutional change. But I am unsure how well it would work. Electors would in theory campaign, and would not all of them campaign with, "I promise to vote for this ticket?" And states would make laws making those promises binding, I would suspect. Could an elector truly run with "I will decide after the election who to vote for" or "I will decide after the election which Republican to vote for?" It would all happen so fast. I mean this happens in political conventions, the ones that run with unpledged delegates. But aren't they mostly pledged in the USA?

In practice it would require a change to the Constitution, because that'd be the only way to get all 50 states to agree to it. There might also be an Equal Protection challenge to it without an amendment.

One interesting aspect that is often missed when talking about moving to a popular vote is that the states have a lot of leeway in deciding who is allowed to vote. If we switched to a popular vote, would it make sense for states to allow 14-year-olds to vote? No federal law prohibits it. Felons? Again, right now it's up to the states. Interestingly, this would likely help the Democrats, which could be part of why the blue states are the ones that support it.

There is currently a federal law that restricts voting (in federal races) to citizens. Even there though, it is enforced by the states, and blue states would probably be more lax in enforcing it if there were a popular vote for President.

Add new comment