What the Data Say About Voter ID Laws

There’s a lot of debate in our politics about whether we should have stricter voter ID laws. But both sides are having an argument based almost entirely on assumptions because data on the real effect of these laws are scarce. Not anymore.

In a brand new paper, Stanford Political Scientist Justin Grimmer gives us a fresh look at whether stricter voter ID laws decrease turnout during elections. The numbers may surprise you.

Listen on Apple Podcasts or wherever you enjoy podcasts.

Transcript

William Howell:

I'm William Howell.

Anthony Fowler:

I'm Anthony Fowler.

Wioletta Dziuda:

I'm Viola Dziuda. And this is another politics podcast. So in the aftermath of 2016 election, there has been a lot of movement in the country to restrict voting, or at least to change the way we vote.

Tape:

Republican majority state legislatures across the country are moving rapidly to pass new voting laws amid former President Donald Trump's continued false claims of election fraud.

Wioletta Dziuda:

And there's a lot of pushback from the democratic side.

Tape:

Senate Judiciary and rules committees have started their work, responding to the concerted nationwide attack on voting rights of millions of Americans that we haven't seen since the era of Jim Crow.

Wioletta Dziuda:

So in light of all this mess, it would be good to get some sense of what do we know about the impact of certain laws related to voting on actual voting.

William Howell:

Yeah, so I mean, can we tease out some basic facts about what's at stake, caught up in these debates about vote restriction and efforts primarily waged on the part of the republican party and efforts on the Democratic Party to expand the franchise, our deep concerns about polarization and more immediately, frankly the lies being told about the 2020 election this Trump's claims that it was all fraudulent through and through and so you see lots of folks trying to attend to that lie, either by betting it or finally putting it to rest. So yeah, so the question is, can we cut through this a bit and get a sense of what actually is at stake. And Anthony, you talk to somebody who wrote a paper that tries to do exactly that?

Anthony Fowler:

I did. I spoke with Justin Grimmer at Stanford University. He wrote a paper with Jesse Yoder, called the Durable Differential Deterrence, there's like three D's in there, the Durable Differential Deterrence effects of strict photo identification laws. It's a recently published in political science research and methods. And it's a really interesting paper where he's trying to understand what are the effects of voter id requirements. He focuses on North Carolina which is a nice case for reasons that we discussed in the interview. And I think he does a good job of quantifying just how big the extent of the problem is. How many people are we potentially determined for voting by requiring ID at the polls. So in this paper, you guys study North Carolina? Why is North Carolina a nice opportunity to learn about the effects of voter ID laws.

Justin Grimmer:

North Carolina is a nice opportunity because it has the law in place for 2016 primary election, the law was put in place. And then there was a letter sent out to everyone who the state identified as a registered voter, but did not match up to identification in their system. And the state informed those individuals that they lacked identification. So this then provides people with this big piece of information, I don't have identification that's needed to vote. And then after that election, and appellate court judge comes in and says, "Actually, this law is unconstitutional and can't be in place for the general election."

After the appellate court judge's ruling, there was no subsequent letter to inform people that this requirement was no longer in place. So one of the thing that North Carolina enabled us to do is both to study what we call this mechanical effect I showed to the polls, I don't have identification, and therefore my ballots rejected. And a different thing that we call this differential deterrent effect. It's this idea that because I lack identification, there may be some confusion about the requirements. And even when the laws removed that confusion could persist.

Anthony Fowler:

Okay, so you have this great data. You've got data on every registered voter in North Carolina. And also data on which of those registered voters match up with the DMV records so that you know they have a state ID. I think it's they could potentially have other forms of ID. They could have a passport and you wouldn't know about it. They could have a military ID or something like that. But for the most common forms of ID that people have, you know whether they have it or not. So how many people don't necessarily have the right ID to vote and who are they? Are they systematically representative of the population?

Justin Grimmer:

Yes, great question. So it's about the 3% fail to match up. And our findings are consistent with what basically every other studies found that this group is disproportionately black, Latino. They vote at lower rates at a baseline, which presents some causal like interesting causal inference problems, but they definitely look different than the population as a whole.

Anthony Fowler:

And you mentioned what you call a mechanical effect, which is someone actually showing up to the polls wanting to vote and not being allowed to because they don't have the right ID. How do you measure that?

Justin Grimmer:

So in North Carolina, when someone shows up, and they don't have identification, their their ballot is potentially rejected, but they're able to cast a provisional ballot. And so we have to make an assumption that once the poll worker says you don't have the identification and someone then goes ahead and cast that provisional ballot. This provisional ballot can then be counted if someone shows up within the next two weeks and says, I have the proper for identification, shows identification, then that ballot is counted. If they fail to do that, then the ballot would be rejected. So we count the number of rejected ballots.

Anthony Fowler:

How many people end up falling into that category where it's very clear that they were deterred from voting because of voter ID in a mechanical sense.

Justin Grimmer:

In the primary election, there's about 1169 voters are mechanically deterred from voting. And so I think that number is an important number because it presents a really interesting challenge and evaluating these laws. But 1169 voters doesn't sound like a lot of voters. And I think that could be true. It could be that you have voter identification laws that have a deterrent effect, but not a huge deterrent effect is on huge numbers of people are being deterred. But it's still the case that that's 1169 people who wanted to turn out to vote wanted to exercise their franchise and were unable to. And so that is certainly I think cause for concern. Even if this isn't a huge effect, it's a massive deterrent, or not hugely moving the turnout rate.

Anthony Fowler:

And presumably, there are a lot more people who are deterred by voter ID other than just these people who tried to cast a provisional ballot and got rejected. Presumably there are people who showed up and didn't cast a provisional ballot. And far more people who just didn't show up at all, because they knew about the voter ID law. So how do you go about trying to estimate the effective voter ID law for other people who didn't cast these provisional ballots?

Justin Grimmer:

Right, so we want to make this sort of counterfactual comparison. So we're in the realm of causal inference now. And what we'd like to know is what the turnout rate would have been for folks who don't have an ID in the absence of this law? Of course, that's hard to know. And in particular, it's hard to make this comparison, because the people who don't have identification already turned out a really low rate. I think that's a big part of the story for why ID laws don't have this massive effect, is that the people who lack identification turn out to not be very participatory at a baseline. And so a standard technique that gets used a lot with these sorts of analyses is differences analysis.

We're interested in comparing the change in turnout rate for individuals who hold identification to the change in turnout rate for individuals who don't cross these elections. And then we'll make a comparison of those differences. Of course, there's a bunch of issues that come up that could make that analysis not a useful comparison. And in particular, the standard difference in differences assumption is that you have parallel trends. But we might expect that individuals who lack identification that their turnout trajectory over time, is qualitatively different than people who hold identification.

And so one of the things that we do here that I think is effective at removing a lot of the confounding, is we make an exact match on turnout history. So we say in the prior elections, we want to find bins of people who hold the exact turnout profile as someone without identification and with identification. And then within that bin, we make the comparison. So that helps us eliminate concerns. Things about people dropping off of the voter file is one thing that can help us with or if people are becoming disengaged or re engaged for a variety of reasons, we are going to capture that there. I think that's the most helpful comparison. We also attempt to deal with obviously, differences in like age and race as well within the specification.

Anthony Fowler:

Okay. You get this estimate that seems pretty small 0.7 percentage points. So if you assume that the voter ID law only affects those people who don't have a legal ID, then your inference, presumably correct me if I'm wrong is that this voter ID law essentially prevented 0.7% of these people from voting, who would have otherwise voted?

Justin Grimmer:

Again, I think it's a reasonable thing to say this is not a huge effect. This is not the sort of massive effect that is going to systematically eliminate the voice of certain groups of people. So I think that there's a lot of rhetoric that goes along with voter ID that perhaps doesn't align with the evidence. But I would still say it is the case that that's people who wanted to turn out to vote, who have the right to turn out to vote, and who are then deterred from participating in the electoral system. And so I think that that's an important number of people. Certainly, if I was someone who wanted to cast my ballot, but was unable to do that, because I lacked identification, I'd be very frustrated, particularly because I believe this is a core component of what it means to be a citizen, with well established the intent of these policies are oftentimes discriminatory intent.

And can be passed to address a problem that I believe has not been effectively demonstrated. There's no conclusive evidence that there's massive amounts of voter fraud. And yet, they could still have a relatively small but still determined to fact. And that's a thing that we try to reckon with in the paper. We explained that these aren't small effects, but still meaningful effects in the sense that there's real people who aren't going to be able to vote because of them.

Anthony Fowler:

So this law was changed between the 2016 primary and the 2016 general and you still go ahead and you test to see whether or not there's a deterrent effect in the 2016 general, even when the voter ID law is not supposed to be in place. What's the motivation for doing that?

Justin Grimmer:

One of the things that we were thinking about was folks may not be aware that the law has changed. And in general, we can imagine that a lot of what happens when there's big changes to election administration. So this can sow confusion in the electorate. We could suspect that people may not understand exactly what's required of them when they have to turn out to vote. Obviously, political campaigns and candidates can go out of their way to give people that information to help them understand what's necessary to vote. But we were particularly interested in this group of people who were given information about what was required to vote when the law was passed, but then not updated when the law was changed. And so it would be quite reasonable for them, unless they were being very attentive, or they've had a political candidate come help them out, that they would still believe that identification would be needed.

Anthony Fowler:

And what you find there is actually a bigger effect for the 2016 General Election than you find for the 2016 primary election, Does that surprise you that you'd get a bigger effect once the laws no longer in place?

Justin Grimmer:

So we get, of course, the mechanical effect goes to zero. And so we get a bigger estimate of this deterrent effect. And the reason that we think that the deterrent effect would be bigger is that there's much higher participation among the people who don't have identification. So those without identification tend to be occasional voters are relatively rare voters. These are not people who participate in a lot of primary elections. And so at a baseline, you would expect that if someone's just not going to turn out to vote, the law cannot make them any less likely to vote, it's already a low turnout rate. But if you go to the general election, that's the setting whether it be potentially much higher turnout for these individuals. And as a result of that increase in the baseline, if this confusion matters, it's perhaps not surprising that we would get a bigger effect of the law among those without identification.

Anthony Fowler:

So you also do these interesting analysis, to the extent that there appears to be this deterrent effect among people who don't have an ID, how does it vary across groups? So we've already talked about how the kinds of people who don't have an ID are not necessarily representative of the general population. But within that subset, are there heterogeneous effects? What kinds of things do you find there?

Justin Grimmer:

Yeah. I think it's interesting, there are heterogeneous effects of the law. We basically fail to reject the norm that the deterrent effect is constant across different groups. But because we talked about there are differences in who holds the identification and who doesn't. That means that the law has a differential effect. But it's not because of some heterogeneity in the fact it's not like the law caused black voters to be more confused or to have a bigger differential effect. It's just that black voters tended to hold identification at a lower rate. And so you end up with a bigger effect of the law for that group.

Anthony Fowler:

Okay, interesting. We're having lots of public debates about election laws in general. And, of course, the controversial Georgia election reforms. One of the pieces of that is that when you submit your mail in ballot, you now have to write a social security number or driver's license number on your ballot. Is there anything to say? I mean, I know, that's different from what you're setting here in North Carolina. Is there anything to say about that? Does research have anything to say about what the effects of that might be?

Justin Grimmer:

I think that the one thing that we could say is that the aggregate effect of that having to write a social security number or a driver's license number is relatively small. But it's still likely to be the case that this is going to create confusion for some people make it harder for them to complete their ballot. And could result in valid ballots being canceled. And so I think has this interesting dynamic that I think holds true in a lot of election ministration debates that both sides get wrong. So on one side, Democrats say that this is massive amounts of voter suppression, it's going to cause people to be systematically disenfranchised across the board. And that's just a complete overstatement about what's going on with, say that component of the mail and ballot.

However, it is the case that if you're an older person, and you're filling out the mail in ballot might already be difficult. Now, there's just one additional hurdle or one additional thing that you could mess up. And as a result of that, it's almost certainly to be the case that this is going to result in more people having their ballots validated because of having this requirement. And so my guess is that small number of ballots will end up being canceled because people fail to provide the proper identification. Given what we know so far about the rates of fraud in mail and elections, it seems quite likely that the number of ballots that would be canceled because people improperly fill out this information, would be much higher than the number of fraudulent ballots that are prevented from being submitted.

Anthony Fowler:

Is there anything else you'd like to share with our listeners about your paper about this topic in general?

Justin Grimmer:

I just think that there's this like real opportunity to think about election administration now, particularly after the last election. If we think about it, carefully. Think about the places where we can do better. And there's real risk that it's going to get captured by people who either want to make a political argument that there's a lot of fraud, and therefore we need some dramatic changes that really needed or by people who say that any change to election administration is going to systematically ruin the electorate or keep people from voting. Like a lot of things in life, I think the truth lies somewhere in the middle between those two extremes. There's an opportunity to do elections better if we sort of take the reform opportunity seriously.

Anthony Fowler:

Great. It's good to talk to you Justin, I really appreciate it.

Justin Grimmer:

My pleasure.

William Howell:

So terrifically interesting, Anthony, the paper is looking at the effects of a voter ID law in North Carolina in 2016, that's in place for the primaries, but then is lifted before the general election. Before we dive into kind of, there are interpretations of the findings. Can you restate for us what the core findings actually are?

Anthony Fowler:

There are several interesting findings, the first interesting analysis is just kind of matching the voter file to the DMV records and showing that about 97% of registered voters have a legal form of ID. And about 3% of them don't match the DMV records. So they may not necessarily have the kind of ID they would need to have voted under the voter ID law. Within that 3%, we can look to see, are they representative or not? It turns out, they are more likely to be black, they're more likely to be Hispanic, they're more likely to be Democratic. So they're not representative. But we are only talking about 3%.

The next part of the paper is a kind of accounting exercise where they just asked, in a mechanical sense how many people showed up to the polls? Didn't have an ID? Cast a provisional ballot? And then that provisional ballot was later rejected because they didn't have it? And then the third part of the paper, potentially the most interesting points that are pulling all this together is to ask, what's this deterrent effect of the law? How many people potentially didn't even show up to the polls at all, because of the voter ID law? And so they're going to do some kind of difference in differences design, but they're going to try to match people on this alternate history. So find people who are of the same race, the same age. Who voted and didn't vote in all the same elections in the past.

And then let's see in the 2016 primary, were the people who didn't have an ID less likely to vote, the people who did have an ID conditional on all of those things? And it turns out, they were less likely to vote. But the effect is pretty small. It's on the order of 0.7%. And we're already talking about 3%. 3.3% of the population has this turnout effect of 0.7 percentage points. So in total, this is a pretty small effect. We talked about that in the interview that the one other big pieces, they do the same thing for the 2016 General, where the law shouldn't even apply. And you'll end up getting similar if not larger determine effects even in 2016 General.

William Howell:

Yeah, so in the implementation of the law, people were expressly informed that if they didn't have an ID before the primaries, they weren't going to be able to vote. When the law was overturned in the courts, they weren't subsequently informed that in fact, okay, our previous letter is null and void, you can actually turn out to vote so. And then you see an effect, which is I mean, the point estimates are pretty small. But it's four times the size from a baseline, it's about 2.7% I think, in the general election, as opposed to 0.7 effect in the primary. And then the effects carry over into the 2018 elections. Into the general election there too, you see this differential effect, but it's smaller. It's smaller than the general effect it's observed, excuse me, the effect that's observed in the General of 2016.

Wioletta Dziuda:

I was very surprised by those results. There's a lot of discussion about all these different restrictions that various state legislatures are putting on voting. And there have been a lot of discussion since I remember about the voters ID. So just the fact that in North Carolina, only 3% of the registered voters don't have an ID is already surprising to me. So we are talking about a very small group of people. And I want to emphasize that what's nice about Justin's study is that he looks at the state that before didn't have any voter ID laws. So it's not that there's some people who don't have a voter ID would like to vote and they didn't register. Now, presumably, we see the universe of people who want to vote, and only 3% of them don't have a voter ID, which is already a small fraction. And then the numbers are extremely small. The deterrent effect is in 1000 to 2000 people. Why is this such a big deal? Why do we talk about it so much?

Anthony Fowler:

Yeah, I think it's a miscalculation. I think that both parties think that these laws will have bigger effects than they do. And of course, if you're a politician and if you're a partisan and you believe something is going to help you or something's going to hurt you, you're going to fight really hard for or against it. But given the kind of the level of partisan rancor we have in legislatures and given the level of, sometimes the kind of partisan polarized tenor of media coverage. You would think that these things have huge effects? And I think the answer is, nobody really knows for sure. And the people who have studied these things find that these effects are actually quite small. So I think some of it is they're actually just making a mistake. They're assuming that these effects must be much bigger than they really are.

William Howell:

So I have a different take. What we're getting from this paper is an estimate of one particular effort to impose one particular kind of impediment to voting. And it's an impediment that yields effects that while small, differentially target Democrats and differentially target Hispanics and African Americans. In that sense, as an effort to chip away at the kind of support that Democrats are going to have in the upcoming election. This isn't such a bad thing, if you're from a Republican standpoint. At the same time, I'll say this is like part of a much larger effort as we know, by Republicans all across the country to do not just voter ID laws, but to roll out all kinds of different kinds of impediments.

And so you can imagine, we should think about it. We should talk about it. What are these things? Are the effects additive? Or the cumulative, when you think about this in combination with other kinds of efforts. And I would be reticent to say that Republicans all across the country have kind of made a mistake, that they have miscalculated. That this might actually improve their chances of holding office. And I'd also say that on the Democratic side, it might be good politics for them to say, this is a huge deal. They're coming after the franchise, this sacred feature of our democracy.

It's not so much I guess, I would say it's not so much a miscalculation as it's a there are two things in play. One is that this isn't a standalone effort. It's part of something much larger. And there's a politics associated with them that we have to take account of. In the same way, it's good politics for Democrats to scream and holler over these laws. It's part of the reason why it's good politics, is because it can stimulate turnout. You can rally the base in order to turnout at higher levels. And it is rough telling of the history of Georgia in the last four years. I suggest that's what's happened.

Anthony Fowler:

There are other studies that have tried to look at this. And one good example is this Paper by Ken Tony and Pons. They look at a bunch of different states, as states are changing their voter ID laws. What happens to turnout and on net they find essentially zero. They find no effect on turnout. One of their some findings is actually that, there seems to be an increase in Democratic mobilization efforts in states with voter ID laws. And so it could be that there's a very small deterrent effect of the voter ID law that's more or less counteracted by mobilization efforts that happened in those places. And it's not to say that there's something important about the laws, because it turns out on average they have little effect. But I do think these accounts are overblown. I do think it's a mistake to make a big deal about them when they turn out to have small effects.

William Howell:

It's a mistake for who?

Anthony Fowler:

Well, I think it's a mistake for everybody. I think if you're just a journalist, for example, who's trying to report honestly on what's going on, it would be a mistake to say this is like some major effort to disenfranchise people when in fact affects such a small number of people. The other mistake, so why is it a mistake for the Democrats, I think? One reason I think it's a mistake, is it's just a missed opportunity. It turns out that these effects are very small. It turns out Republicans seem to care about these things. So Democrats could go to the Republicans and say you guys seem to care a lot about election security and mitigating voter fraud. And you think that having a voter ID law is one way to do that? Great. Let's implement a voter ID law. And then you guys can stop complaining about voter fraud.

And so that's a win-win. And the health of our democracy will be improved. People will trust the results of election. And let's combine that with something that we want. So we'll give you guys voter ID laws and give us something that we just some grand horse trade, and then they'll get some policy concessions that they wouldn't otherwise get could be a policy concession. If it was me, I'd say why don't we have compulsory voting? We've had our compulsory voting debate on this podcast before, let's have compulsory voting, but with voter ID laws. That would be great. That's okay, that's interesting idea. They would either have to say, great, I'm happy about this, or they would have to admit that they actually were fighting for voter ID laws in bad faith to begin. So either way, that's kind of a win for the Democrats.

William Howell:

Okay, but maybe part of the reason why these effects individually, again, let's hold out the possibility that cumulatively, these things add up to something reasonably large. I mean, cumulatively across multiple efforts to impede people's ability to vote. So there's voter ID. There also is, let's get rid of early voting. Let's reduce the number of places where you can cast ballots. And on and on and on. Let's require you to write your social security number when you mail in a ballot. These things may add up to something substantial and in a world like the one that we inhabit, where in the outcome of presidential elections hangs out upon 10s of 1000s of votes in a handful of states. This could be really consequential.

So there's that. But let's put that aside for a second, which is to say, for any one intervention, part of the reason why the effect may be small is because what accompanies them is efforts by the Democratic Party to step in and to say, this is outrageous what they're doing. And they're whole bunch of people who are willing to volunteer, and to rally to get out the vote and then go get their ideas. Because they're so offended by this. And if they simply said, well, not such a big deal, all is good here. I'll let this slide if you've given me something else. That may actually alter what kind of an impact these laws have on the electorate.

Wioletta Dziuda:

So you might be right. Well, but I think if we look at a couple of our podcasts before, I think the little evidence that we have sort of points against your story. So first of all, we had a podcast on mailing ballots. And it turns out that if you allow for mailing ballots in a more unrestricted way, you actually don't have a partisan effect. Yes, you have an increase in turnout, but you don't have a partisan effect. So now, if you scale back mailing voting and Democrats are the ones who get upset about this, and sort of lobby that people go and vote nevertheless, if anything, this is going to backfire. This is going to backfire against Republicans. So I think I'm with Anthony a little bit here that I don't think those efforts on the part of Republicans are completely thought through, unless they are just virtue signaling for certain type of voters that they are wink win that we don't want certain people to vote. But in that case, the Anthony strategy would work that calling their bluff would actually work.

William Howell:

You think this is a bluff? Because really, this is a bluff? What's happening all across the country. These efforts to go guns blazing, to raise all kinds of impediments? That is a miscalculation on the part of Republicans that those kinds of efforts might both, because for the reasons you just pointed out Viola, which it's a way to rally the base. Their own base. And to chip away, they think at the ability of people who are Democrat, black and brown to turn out to vote. In here, these effects are small, but again, they may be cumulative.

Anthony Fowler:

But either way, I mean, I don't really care what their motivations are. If it turns out that the policies don't actually have much effect, who cares? One way it's possible that they really do, just that they're motivated by partisan considerations and they really think this is going to be a big boon for Republicans. Voter ID laws are going to really help Republicans. That turns out to not be true, but they maybe they think that. It's also possible that they genuinely believe that there's lots of fraud and the voter ID is going to reduce fraud.

Either way, I think it would be perfectly fine to say, okay, let's talk to you, and let's come up with some kind of compromise, and let's implement your preferred thing. And then let's bundle it with something that I really want. And then either you get a better deal out of it, because they're making too big of a deal about the small thing. Or they have to turn and say, actually, I don't actually like this deal. And it turns out I wasn't so worried about voter fraud after all, I was actually just trying to do this partisan thing. I mean, either way that seems good for Democrats.

William Howell:

Hey. If you're getting a lot out of the research that we discussed on this show, there's another University of Chicago Podcast Network Show that you should check out. It's called Capitalisn't. Capitalisn't uses the latest economic thinking to zero in on the ways that capitalism is, and more often than not isn't working today. From the debate over how to distribute a vaccine to the morality of a wealth tax, Capitalisn't clearly explains how capitalism can go wrong and what we can do about it. Listen to Capitalisn't part of the University of Chicago Podcast Network.

William Howell:

So do you think there's a deal to be heard here? I guess I'd be skeptical that I don't think there's a deal to be heard with Republicans. Wherein they would say, sure, as long as we have voter ID laws, then let's make voting compulsory. Like it's a terrifically interesting proposition—

Anthony Fowler:

What if it's automatic registration. What if we're going to exchange automatic voter registration for voter ID laws? Is that a deal that they would take? If they really care about voter ID. If they really think that like fraud and impersonating people at the polls is a problem, then that should be a great deal for them.

Wioletta Dziuda:

Yeah, I don't think there's a deal to be made exactly. Because I don't think that they truly believe that this is super beneficial to the per se and just as a law. And because we have all this evidence to be looked at, and there's no evidence to that effect. So I think it would be strange for them to truly believe that. But if they are offered this kind of deal, and they rejected this sort of shows that they don't actually care so much about this voter ID.

William Howell:

Yet, but in the meantime, it is clear, but what Justin's paper shows is that a set of people who by all rights ought to be able to vote are deterred from doing so. Or having their ballots cast out because of the introduction of a particular law. And in that sense, that's terrifically consequential.

Wioletta Dziuda:

So I'm very sympathetic to this argument, but I think I have this knee jerk reaction when people dump everything into the same bucket and get a trade of anything because it can get outrage. So there are certain laws that obviously are aimed at minimizing the voting of minorities. For example, if you restrict how long can you vote, or you say you can only vote on Tuesday from nine to five. A lot of people have jobs, and they just cannot show up. So there are certain laws that we should be outraged at. Because it's obviously the differentiator we have. I mean, it's maybe not obvious we should know, we are academics, we should search for data.

But it's sort of it seems reasonable that they were differentially affect minorities. But when I think about the voter ID law per se, I think this is somewhere where Democrats should say, well, it's not such a crazy idea. I don't know, maybe this is my background. If I come from a country where if you go to vote, you have to have an ID, otherwise there's no way to verify that you are who you are. And getting an ID is not such a difficult thing to do. And maybe we should make it easier for people to get an ID.

Maybe it should be for free or maybe it should be much more streamlined process. But for me, it is not a crazy idea to have a voter ID. And especially in a situation which people believe in this slide that there's a lot of voter fraud. This is like a cheap way to say, let's make it a little bit tighter. Let's sort of like make sure that these 1000 votes that Justin found in his paper, are truly people who just forgot their ID and not people who are trying to vote pretending being someone else. So I feel like, yes, there's room for outrage, but voter ID per se is not the place where you should place this outreach.

William Howell:

So let's talk about this. So at bay, these effects are small. But at baseline, Hispanics, African Americans, and Democrats are much less likely to have identification, then are people who are white and Republican in the state. Now what you might say is, well, let's fix that. I think that's what you're suggesting, Viola. Let's fix that fact. But we live with that fact right now. And so, look, that which appears to be at a race neutral is not in fact, race neutral or neutral with regards to partisanship is in fact, not neutral, because of these baseline differences. And then that's likely to persist as long as you have one party that disproportionately represents people who are marginalized and vulnerable.

Because those people have a very different relationship to the state than those people who are less vulnerable and kind of center of the poverty. And so in that sense, I don't know how exactly we get around this. We can certainly... The rhetoric can be inflated, we ought to pay attention to facts. I think that that's true. But the distill is something to worry about in this paper. Because you see these differential effects and because when we were on election night, looking at what was going on in Georgia, where there's an effort to not just do voter ID, but a whole lot of other stuff, it came down to a few 1000 votes.

And that's a very big deal. So we live in a world in which these margins are terrifically narrow in highly consequential elections. And so the small effects matter a bunch and let me say this last thing they'll back off, which is, we've talked about in previous episodes how the effective mobilization campaigns tend to be very small. Money spent on ads. But the returns are pretty small. These are kind of comparable to those. And so this, I don't know, this strikes me as something that we should be keeping an eye on and we should be worrying about.

Wioletta Dziuda:

Yeah, but I don't think we should live in a world in which we make decisions based on what kind of voters affected, Now because in this world we're going to have a continuous fight, and bobbing heads are going to try to push in one direction and Democrats in the other direction to just find this one vote that's going to tip the scale. I think we should live in the world, and I know that what I'm saying is completely idealistic. But we should live in a world in which we have some principle. If you think about the lectures. If you take us back you can think about which laws are prior seem to be good and sort of reasonable and which laws are unreasonable. And I think voters ID law strikes me as a law that's in principle extremely reasonable as we see from Justin's paper.

Yes, it affects differentially different groups. But it's a very small group of people. And at the end of the day, in his paper, a lot of people who are turnout from voting were people who just simply forgot their ID and I don't know whether these are minorities in a disproportionate way. While there are certain laws that strike me as wrong. Restricting times at which you can vote to a very narrow window. That seems like I don't even care at home, it affects disproportionately, it seems like that's disenfranchising the entire population, because that's making it harder to vote. So I wish again that's not the climate we're living in, but I wish there was a little bit more of this sort of rational discussion about what are the good laws? Whether the acts under reasonable laws and whether they are not.

Anthony Fowler:

But I do want to emphasize again. I've been pushed back a little bit against Will saying these effects are comparable to like get out the vote effort or something like that. These are very small effects. So in the paper, if you kind of take all their estimates at face value, there are estimates are that 0.03% of white people are deterred from voting. Of white registered voters are deterred from voting because of voter ID laws. And something like 0.06, or 0.07 of black and Hispanic voters are deterred from voting, so that's twice as much. So if I do the kind of percent comparison, it seems like it's twice as much deterrent. And then the other hand is almost no people that we're talking about. And so we're talking about two numbers that are extremely close to zero.

Obviously, anything that differentially affects different groups is something that we should at least care about and pay attention to and empirically examine. But if the differences is tiny, then it might turn out to be politically inconsequential. So I think it's worth at least emphasizing that. And that's certainly the job of the academics, the journalists to at least think about and say, okay, maybe this is a lot of fretting over not much. That's a political impact. It was just Justin's response to that, of course, is that even if our net doesn't matter, it doesn't matter a lot to those few people who wanted to vote and couldn't vote. And I understand that point. But then, I would go back to, there are other ways we can get these people voting. We can get them IDs, if that's really the issue. But it seems like we shouldn't overstate it ourselves. We should point out that we are talking about very small numbers of people who are affected by this.

William Howell:

Yes. But this one law in this one state. But then when there are other laws that have 17 components, not one, is that right? Then it might add up to something more we don't know though. It may be that the same people are affected again, and again. What I take away from this and in some ways, I look at my bottom line on this is the strength of this paper is to hone our attention on precisely who is affected by one particular intervention. It says, but we can lose our way if we just kind of wave our hands at a general problem. And that when you borrow into the voter file, and kind of excavate exactly who's being affected, it tends to be people who are in this case, Democratic, Hispanic, black.

And this is another piece that we haven't said yet, which is that they are kind of the occasional voters, but never voters in that face. They're not affected. They're always voters. They're not affected by this particular intervention. It's the people sort of in between, and that in this instance, the baseline appears to be really quite small. And I'm not willing to give up to say, therefore don't worry about it. Like therefore, it's not consequential, either by reference to the politics that are set into motion, or by the effects that they have on electoral outcomes. I got to get vaccinated, guys. So I'm eager to hear what your bottom lines are.

Anthony Fowler:

I like this paper a lot. I think this was a nice effort to scientifically study a really important and politically contentious question in American politics. And they have some nice data of the fact that they can connect the voter file with the DMV records to actually see who does and doesn't have a legal form of ID, is a nice innovation. The fact that the policy is kind of on and then off again, is also a nice opportunity for learning about the effects of these laws. So I thought that this was a really nice effort. And I think my overall bottom line is that on the whole, we are making too big of a deal about voter ID laws. And if these effects are actually very small. And I would hope that that kind of evidence would change the nature of our political discourse over voter ID, although I don't know if it will. But the actual paper itself, I think is a really compelling paper.

Wioletta Dziuda:

So I also like the paper and I hope and I'm pretty sure that moving forward, we are going to have more similar papers trying to study the effect of all these restrictions that are being passed right now. But I wish that when we talk about how to structure our elections, what kind of flow we should have around them? We should focus less on how particular laws differentially affect particular people and who is going to benefit, who's going to lose? I think we should sort of take a step back and think about just in a reasonable non partisan world, what are the laws that seem to make sense? That seem to minimize fraud and at the same time at least expand it. I'm not targeting particular groups.

William Howell:

Well, I agree with both of you. This is a terrific paper. It does lots of little things then which we didn't actually talk about, but that are really sort of smart moves that they are able to do precisely because they have such high quality data. They can match on people's voting history over multiple elections. Which is a really important innovation that they introduced here, that if you just look at variation, within individuals over time, you see actually much bigger effects. And they say that, but the parallel trends assumptions don't hold. So what we're going to do is in fact, break that down and look at within bins of our passed voter history, and there you see the effects attenuate. There's lots of stuff going on in the paper that it just is careful and illuminating and sobering.

And yet, I'm still not sober. That is, and going to continue to worry. Because I think that these larger politics are playing out. And for reasons that aren't just born of people's mistaken assumptions, that the effects are smaller. That it plays to their self interests. And I'm not especially hopeful that at least in the short term, there is an accommodation to be had so that we can adopt reasonable good election administration laws. That these partisan considerations are going to continue to flow through our politics. But maybe papers like this will push back against it. Maybe it'll put us on a more solid ground. If it did that would be a good thing.

William Howell:

Thanks for listening to Not Another Politics podcast.

Wioletta Dziuda:

Our show is a podcast from the Harris School of Public Policy, and is produced by Matt Hodapp. Thanks for listening.