Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

Advance Voter Registration: The King of Behavioral Interventions?

May 21st, 2016, 7:56am by Sam Wang

Pro-voting activists are constantly trying to increase the rate of voting. They often get interested in behavioral interventions such as voter contact. Successful interventions typically boost turnout by a few percentage points. More generally, the smallness of any get-out-the-vote’s effort means that I don’t have to account for it in any of the polling analysis I do. However, now there’s a game-changer: automatic voter registration.

This Tuesday, the Brennan Center for Justice at NYU held a workshop on Automatic Voter Registration: Why And How. It was a practically-focused but also public event: former Attorney General Eric Holder spoke, as well as California Secretary of State Alex Padilla. Legislators, elections administrators, and activists showed up. I was invited to analyze the brain mechanisms and political consequences of automatic registration – the only time I can recall being asked to do anything in both domains at once!

Automatically registering citizens to vote unless they actively opt out is a big deal. It could potentially increase voter turnout by close to twenty percentage points. Here I’ll publish some notes from what I said.

Approximately 76% of the voting-eligible population is registered to vote. The prize is to register the remaining 24%, which in 2012 was estimated to be 53 million people. Automatic voter registration is a powerful way to achieve this goal – while also using validation mechanisms to reduce whatever levels of error and fraud are in the system now.

Automatic voter registration has recently been enacted or implemented in a few states (Oregon, California, West Virginia, Vermont, and Connecticut). There is very little data yet on its effects. To predict what might happen, we can draw upon principles from behavioral and cognitive science, where research can reveal what AVR could accomplish, and why it works. The two reasons are (1) the power of the default option, and (2) the difficulty of planning ahead.

1) The power of the default option. Human beings don’t like to think too hard about decisions. We are “cognitive misers” and therefore tend to make the convenient choice. Consequence: default options can influence outcomes. The foremost advantage of opt-out registration (that is, you have to take an active step to avoid registering) as opposed to opt-in has several advantages. Foremost is inertia: If you do nothing — which is what most people specialize in — then you’re registered in an opt-out system, and not registered in an opt-in. With near certainty, if other requirements for voting are kept the same, AVR would increase the registration rate.

We can estimate what removing this barrier would achieve. Two examples:

Example #1: Savings plans. In 2001, Brigitte Madrian and Dennis Shea showed that participation rates in an opt-in approach were 20% after 3 months of employment, rose to 65% after 36 months. With automatic enrollment, jumped immediately to 90%, and then 98% in 36 months.

Example #2: Organ donation. This is a famous example. In 2003, Johnson  and Goldstein (Do defaults save lives? Science 302:1338) examined commitments to organ donation registration by drivers in European countries. In some countries, drivers were donors by default unless they opted out; in others they weren’t until they opted in.  Rates of willing donors were 94% in opt-out countries, and only 14% in opt-in countries.

So in these examples, we have an endpoint compliance rate of 94% to 98% for an outcome generally agreed to be good, as long as it is made the opt-out option. In the case of voter registration, approximately 76% of the voting-eligible population is registered, so such a gain would be large increase. Going to 94% would get 40 million more people registered nationwide.

2) Planning ahead and the prefrontal cortex. Planning ahead is difficult – our brains are not so good at looking ahead for months or years. I told workshop attendees “if you had a pastry this morning instead of fruit, that is potentially an example of the difficulty of planning far ahead.” Such planning requires executive function (a technical term meaning the brain’s ability to plan and execute), a function that uses brain regions such as the prefrontal cortex.

In states where advance registration is required, voting relies on being able to imagine what one might want to do on the first Tuesday of November, as far as two years off. AVR takes care of the problem of advance registration. Activists in this area presume that registering will increase voting. We can test whether this is true using data from another rule, same-day registration. How many more people vote if they don’t have to register in advance?


Eleven states and the District of Columbia allow registration on Election Day itself. This removes the barrier of registering in advance. All twelve of those same-day jurisdictions have above-average voter turnout (an average of 66.4% compared with 58.6% nationally). If this 7.8% difference were applied nationwide, 14.7 million more people would vote.

A good way to define voter turnout is to calculate how many people cast ballots, divided by the voting-eligible population. This statistic is studied and made public by Michael P. McDonald of the University of Florida. I examined state-by-state voter turnout for the 2012 presidential election, compared it with voter registration rules, and found the following:

Of the top four turnout states, three have same-day registration: Minnesota (76.0%), Wisconsin (72.9%), and New Hampshire (70.9%). The fourth state, Iowa, requires registration by 10 days before the election.

Of the bottom-four turnout states, all below 50%, require advance registration by at least 21 to 30 days. The second-lowest is West Virginia 46.3% – 21 days before election. The passage of AVR in West Virginia suggests that this is a key state to watch in the coming years.

The benefit could be even greater than the 14.7 million estimate. Data from same-day registration states doesn’t answer the question of whether even more people would be willing to vote if they were already registered in advance, and received communications about their polling station. There is an area of research called “implementation intention” manipulations: where in addition to asking whether you plan to vote, I also ask how you plan to get there, with whom, at what time, and so on. Guiding people through this planning increases the chances that they’ll turn up. Think of these interventions as an assist to the prefrontal cortex.

However, such behavioral interventions are small potatoes compared with AVR. Generally, behavioral interventions increase voting rates by a few percentage points at best. But AVR can potentially increase voting rates by over 20 percentage points. So even though it sounds technical, it is potentially 10 times as powerful as the best voter outreach efforts.


Additional comment: Afterward, I thought about the fact that AVR generally appeals to Democrats more than it does to Republicans. However, there are several ways in which AVR might appeal to Republican legislators and activists.

First, AVR can reduce fraud because it takes registration to places where documents are required, such as the Department of Motor Vehicles or on state tax forms. Questions such as “Are you a U.S. citizen?” could be asked.

Second, as AVR starts to be passed, there may be snowball effects – and perhaps a feeling of competition among states. For example, in the 2000 Bush v. Gore election, Gore won the popular vote by half a million votes. If all Bush states had AVR in place, I have calculated how many votes Bush and Gore would win. Bush’s popular vote total would grow by approximately 1.2 million more votes than Gore’s. That would have given Bush a popular vote win. Conversely, if all Gore states had AVR in place, Gore’s popular-vote lead would have increased to about 1.7 million votes. In anticipation of a close election in the future, both major parties may be motivated to implement AVR.

One final note. It is tempting to be pessimistic about passing AVR in some states. Considering that the odds of passing AVR in red and purple states increase with getting supporters on both sides, I challenge PEC readers to come up with reasons that would persuade at least a few Republican lawmakers to vote for legislation.

Tags: Politics

38 Comments so far ↓

  • Amitabh Lath

    A bit of the chicken and egg: do the three states you mention have high voter turnout because they have AVR, or do they have AVR because voters there are inclined to be engaged?

    I agree West Virginia will be an interesting test case. But any increase in turnout there will have to be normalized to increase in turnout overall. This election has high polarization and interest which could boost turnout overall.

    I wholeheartedly support AVR, but having daily contact with undergraduates disabuses one of any miracle-pill solution to voter turnout. A few years ago I asked the undergrads working in our lab if they were registered and were going to vote. Most frequent answer: I don’t know, I guess.

    • Ed Wittens Cat

      One thing AVR would do is extend the campaign window allowing for last minute deciders to become motivated. Also extend the window for Black Swan events.
      Do candidates negatives generally rise over the duration of the campaign?
      I find this astonishing–

    • Matt McIrvin

      With college students, there’s an additional problem: many of them are living outside of their home districts and will have to either re-register or vote absentee (in local elections where they may well be unfamiliar with much of what’s going on below the national level). And the locals in college towns are often actively hostile to the idea of letting them register and vote locally, and oppose any effort to make it easy.

      I suspect this is a significant factor in low youth turnout. People get registered to vote at 18 and then go off to college almost immediately thereafter, where it’s unusually hard for them to vote.

  • Ed Wittens Cat

    wow…this would reshape the whole political landscape.
    I cannot imagine that this would ever pass a Republican congress though…by 2044 hispanics will be the largest demographic group in America and beginning in 2008 non-hispanic white children have been the minority in the under five demographic.

    • Amitabh Lath

      Has any study been done of people who do not vote and asked them why? If indeed they claim difficulty in registration, then yes I can see AVR making a dent. If however, it is apathy, AVR isn’t going to cure that. As for those who lack access to proper documentation, it will just move that problem from one venue to another.

    • Sam Wang

      Doesn’t the same-day-registration voting data address that issue?

      Also, if you’re not registered you can’t vote. Your concern is quantitative – some of the new registrants will surely vote.

    • truedson

      Having AVR is one less hurdle.

    • Ed Wittens Cat

      oops, i meant non-hispanic caucasian (aka “white”).
      Perhaps AVR could be sold to GOP as a means of reaching Sean Trendes “missing white voters” …

      or is that too 2012? :)

    • MPP

      I imagine an issue is the motivation to get a driver’s license or the like is much stronger than the motivation to vote.

      I might be willing to go through the effort to get/find all the proper documents for a driver’s license, but when I need those to register to vote (even same day) I might decide it’s not worth my time.

      But if I’m at the DMV anyway, and all it requires is a couple extra questions, I’ll end up registered and I’ll have everything I need to vote from that trip.

      Because the personal benefit of voting is infinitesimal, even relatively small barriers to voting make apathy much more likely to block someone from voting.

    • Matt McIrvin

      Has any study been done of people who do not vote and asked them why?

      Yes: here’s a Census Bureau survey from 2014, a midterm in which turnout was very low:

      There’s an Excel spreadsheet linked from this Census page that gives reasons for 2012, and the results are pretty similar, though “not interested” is not quite as high relative to “illness/disability”:

      There’s no overarching reason that gets a majority, but the most common one people give is that they were too busy, which suggests holiday/weekend/early-voting approaches. Registration issues are not often mentioned at all. But this assumes that stated reasons are actual reasons, which I suspect Sam is not assuming.

    • Matt McIrvin

      …”too busy”, “illness/disability”, and “out of town” together make up nearly half of the sample, and sufficiently convenient early-voting or vote-by-mail arrangements ought to help with all of those things, and maybe some of the others as well.

  • Doctor Science

    However, there are several ways in which AVR might appeal to Republican legislators and activists

    You’re joking, right?

    There is *no way* it can appeal to the GOP, because they’re fundamentally opposed to expanding the franchise. “Vote fraud” was always a cover for the real thing they object to, voting by the wrong people.

    Trump’s popularity with GOP voters proves that, no matter how much the party leadership may want to expand their appeal, the base aren’t having any of that. It’s a gut thing for which GOP “leadership” comes up with covering, post hoc explanations — but responding to the explanations will do no good.

    • Commentor27

      If AVR were proposed, it would certainly put this to the test. I suspect we’d find that the GOP will oppose any reform that would make it easier to vote and base their opposition on voter fraud, even if it could be objectively shown to reduce voter fraud.

    • Sam Wang

      Considering that the odds of passing AVR in red and purple states increase with getting supporters on both sides, I challenge PEC readers to stop with the negative statements and come up with reasons that would persuade at least a few Republican lawmakers to vote for legislation.

      It is possible – West Virginia did it.

    • 538 Refugee

      Every American has the the right and duty to participate in the electoral system. Registration is only required to keep order in the voter system and shouldn’t be separate from any other valid system of tracking citizens. We all need Social Security numbers, drivers licenses, tax forms, etc…. If you want to serve the will of the people, you need to know the will of ALL the people and we need less to be disenfranchised.

      The Republicans are in desperate need of more moderate votes at the moment. You could argue that people that don’t vote now are content to go along with whatever happens during the election so those who don’t vote now are default middle of the road.

      In Ohio they used to use voter registration lists as the sole source for jury pool selection. They have added drivers license lists but there is still probably a perception from years of past practice that you can avoid jury duty by not registering to vote.

      Bottom line is, they will look at the demographics and resist. Everything I know looks like this would push the system a little to the left. You have to make it look like only an unpatriotic (whatever) would oppose it.

    • Doctor Science


      Yes, WV did it. WV is also 94% non-Hispanic white. This is not coincidence.

      AVR is politically doable in blue states, and in red states with small non-white populations — using arguments similar to what succeeded in WV.

      It should be even easier to argue for it in large, sparsely-populated states like Wyoming and Montana, where the burden of getting to official offices is particularly high. Rural whites are under-registered, in part for this reason.

      Possibly the need to make registration easier for rural whites can be used to promote AVR even in reddish states with substantial minority populations (e.g. TX). Maybe. But you’d have to persuade state Republican parties that they wouldn’t be committing political suicide.

    • Phil

      While that may be the motive for many GOP legislators I doubt all of them go home and think of themselves as good people who like to disenfranchise voters. The question isn’t how to persuade all of them, it’s how to convince enough that it is the right thing to do.

    • Matt McIrvin

      While they may not think of themselves as people who disenfranchise voters, many of them do insist that voting should not be too easy, because requiring people to jump through some hoops filters for seriousness.

      They also express concern that too many ignorant, stupid or irrational people vote–which I often hear from liberals as well.

  • Ken

    The logical extension of this idea is compulsory voting. Australia has compulsory voting and the voter turnout there is among the highest in the world at about 93-94%.

  • Ashbel green

    There is data available from Oregon’s 2016 primary. I don’t recall the exact numbers but something 6 % have opted out since the law was implemented Jan. 1. Turnout was up in the primary although not as high as 2008. Also, Oregon doesn’t just automatically register, it also mails you a ballot. Our constitution requires being registered 21 days in advance. Auto registration will all but eliminate that barrier. The Oregon a Secretary of State website has monthly registration numbers since Jan. 1.

    • Phil

      The primary election may not be the best to test this. Oregon has a closed primary and if I recall, the state mails you the registration conformation but to be affiliated with a party you need to check a box and mail it back.

  • Amit

    As Sam alluded to above, a possible strategy would be to implement AVR in blue states as a D version of fraud prevention mechanism. While blue states would be eager to implement this anyway, positioning this as fraud prevention gives D’s a viable (and from a voter inclusion perspective, superior) alternative to the current plan in red states.

  • Kevin

    This could be quite powerful combined with universal absentee voting, as we have in Washington state.

    The incentive to increase raw turnout would be higher if national popular vote actually counted for anything.

  • Lorem

    I’m very confused about where the “It could potentially increase voter turnout by over twenty percentage points.” claim is coming from.

    You say that 76% of the population is already registered to vote, so in the best case AVR registers the remaining 24%. Then, of those 24%, 20% would need to turn out? That would imply an 83% turnout among this previously-unregistered group – far higher than among people who actually bother to register on their own. This seems to stretch what I normally imagine when I read the word “could”.

    More plausibly – still using optimistic assumptions – I’d imagine e.g. 97% final registration, gaining 21%. Then, of those, below-average registered voter turnout, so perhaps 50%, for an increase of 10%.

    Which is by no means insignificant – just not 20%.

    • dk

      You’re right to be confused. It’s difficult to be clear when talking about percentage changes of percentages. As I read this, Sam is saying that the voter participation rate (which he tells us is 58.6% nationally) might increase by a factor of 1.2, to 70.3%. That would be a 20% increase in the number of voters.

      But perhaps he’s thinking that the national voter participation rate could rise to about the current level in Minnesota, 76%.

      Either of these outcomes seems unlikely to me, because those who are currently unregistered are the least motivated voters. However, the first one – a rise to 70.3% — would be reached if just half of the currently unregistered voter-eligible population were to vote and the participation rate for everyone else stayed the same.

      Perhaps Sam can clarify.

    • Sam Wang

      Actually I had attempted to make all quantities percentage points, meaning fraction of the Voting-Eligible Population. I agree that the improvement would probably be less than 20% of the VEP. I have made small wording changes to reflect this.

  • Jeff Alworth

    Instead of thinking how to persuade GOP states to adopt automatic registration, think of ways automatic registration (and other parallel forces) might transform because of it.

    The GOP is at the end of a cycle of idea dominance dating back to 1980. It’s been a long time since the party has come up with new ideas. In the era before that, 1932-68, the Dems controlled the idea market following the roaring 20s. we think of demographics, not ideas, as determinative, but there’s nothing special or permanent about a winning bloc. (Southern whites were a part of the two previous ruling coalition.)

    It’s the GOP more than Democrats who seem bent on competing on demographics. Forced to compete on ideas, they might find a winning strategy–and then voter registration would snowball in their favor.

  • Jinchi

    If all Bush states had AVR in place, I have calculated how many votes Bush and Gore would win.

    How did you calculate this? It seems like this involves huge assumptions about the political preferences of the additional voters.

  • FiveToolPlayer

    “I challenge PEC readers to come up with reasons that would persuade at least a few Republican lawmakers to vote for legislation.”

    I will take you up on that challenge. I had a college friend a couple of decades ago who shared a piece of information he learned from a business course on marketing. The message is simple and obvious: when trying to sell someone something, convince THEM why it’s in their best interest to buy it.

    People in the Republican camp get all worked up about preventing voter fraud. At the same time, they claim that they strongly encourage all eligible citizens to exercise their right to vote.

    State legislatures should market automatic voter registration bills as a means to satisfy both of these concerns. All instances of automatic voter registration would require a photo ID, so the pool of newly registered voters would be “fraud free” — a fact which should make Republican lawmakers happy. Also, more people would now be able to to express their constitutional right to vote. Republicans love the word “constitutional,” so they should be strongly in favor of granting more people this “constitutional” right.

    I am not naive. I do realize that most people in the Republican party will find some reason to oppose automatic voter registration. After all, in order for candidates in their party to win elections, it is not in their best interest to increase voter turnout.

    Should Republican lawmakers refuse to embrace automatic voter registration, however, the Democratic party would be able to more effectively and clearly expose the true intent in Republican backed voter ID laws — voter suppression.

  • Kevin

    I’d love to see some analysis of the *quality* of voting, and how that’s impacted by a requirement to signal an intent to participate in advance. In politics “quality of vote” is difficult (impossible?) to define, but I can imagine having my math students vote on statements, some of which are objectively better than others. I can’t imagine getting them to register to do so, however.

  • xian

    a bit off topic but will differences in GOTV efforts have a negligible effect if Clinton has an obama style microtargeted effort and trump reloes entirely on the party and superpacs to supplement his cult of personality approach?

  • A New Jersey Farmer

    Interesting discussion. In the wake of the certified ignorance and gullibility that led to the Great Recession, many states required financial literacy courses in the public schools. Maybe it’s time to pair AVR with a concerted effort at civics education as a graduation requirement. The voter registration card and AVR could become more valued than the actual diploma.

  • anonymous

    Some ways to get Republicans on board for AVR:

    1. Follow Obamacare Medicaid expansion / Race to the top examples, i.e. introduce strong federal financial incentives, get low-hanging fruit first (implementation in blue/purple states), then slowly get the others to change based on the incentives. Use the courts to get the stragglers based on evidence from early implementations.
    2. Argue voting ease for populations that tend to be Republican, e.g. older white voters.
    3. Argue cost-saving, if any.
    4. Appeal to fairness (make it easy for every voter to vote).

    The Republican political establishment can probably run numbers as well as anyone, and will figure out who AVR is likely to help politically most. So the hope for ways 2-4 is to convince independents leaning Republican, who are not making cynical political calculations. The most effective way is likely to be 1.

  • Jay Sheckley

    The GOP surely wants to know where everyone’s home address! Having this done as essentially a data block between elections allows out of season checking for voter fraud. true, there is no voter fraud, but some say theyre concerned

  • Phil

    It would mean that GOP legislators wouldn’t be as likely to be primaried (assuming they make voting easier too). This pew study compares politically active voters to the rest of the public. Would make for less ideologically extreme candidates.

  • Disillusioned

    While I realize these are back-of-the-envelope calculations to illustrate a point, it seems analytically inappropriate to compare states cross-sectionally. I doubt AVR would improve turnout in a consistent way among all states; there are too many other factors at play. My personal hypothesis is that some combination of the electoral college and gerrymandering makes a lot of states and districts very uncompetitive, depressing turnout.

    That said, while I am in disagreement about the magnitude of the benefits, I agree that it would have a sizable effect on turnout. I would also intuit that anything that presents obstacle to voting (arcane registration process, early registration deadlines, lack of early or weekend voting, having election day be on a weekday, long lines, needing to actually make an effort to look up when important deadlines are, etc.) has the effect, however modest, of reducing turnout.

Leave a Comment