Most North Carolinians think it is reasonable for voters to show identification before casting a ballot. A solid, although not overwhelming, majority voted last fall to place a voter-ID requirement in the state constitution, although a vocal minority continues to see it as dangerous and discriminatory.
I’ve long argued that both sides of the voter-ID dispute tend the exaggerate the consequences. Most studies of actual elections held under voter-ID requirements find little effect on the number of votes cast. Some states have implemented the policy and seen big increases in voter turnout. Some states without voter-ID laws haven’t exactly been setting turnout records lately.
You can’t just eyeball election results and draw valid conclusions, of course. Perhaps turnout would have soared even higher if a given state hadn’t implemented an ID rule. Perhaps when critics widely and repeatedly made the argument that voter-ID laws were intended to discriminate against poor or non-white voters, that serve to motivate those very groups to vote in higher numbers.
Researchers have used a variety of means to try to tease out these potential cross-currents. Most of the time, such studies still find negligible effects.
One reason may be that the number of people who both intend to cast ballots and lack ID is minuscule. Consider a recent study by Mark Hoekstra of Texas A&M and Vijetha Koppa of Dubai’s Institute of Management Technology. It examined election results for more than 2,000 races in Florida and Michigan. Hoekstra and Koppa chose those states because they allow ballots to be cast without IDs and then track those ballots separately.
The researchers found that at most the share of votes cast without IDs were in the range of .10 percent to .33 percent. “Thus, even under the extreme assumption that all voters without IDs were either fraudulent or would be disenfranchised by a strict law,” they wrote, “the enactment of such a law would have only a very small effect on turnout.”
The effect was so small, in fact, that it would have been highly unlikely to flip races. “Even if the worst fears of proponents or critics were true,” Hoekstra and Koppa wrote, “strict identification laws are unlikely to have a meaningful impact on turnout or election outcomes.”
The empirical evidence undercuts claims made by both sides. If turnout isn’t much affected by an ID requirement, its usefulness as a “voter-suppression” device is rather unimpressive. On the other hand, if requiring ID doesn’t significantly reduce turnout, there must not be that much illegal voting going on — or at least not the kind that an ID requirement could block or deter.
To my way of thinking, there are three reasons why an apparently low-stakes ID requirement for voting remains a reasonable policy. The first is that it will at least modestly increase public confidence in elections, even if the public is mistaken about the prevalence of impersonation fraud, residency fraud, or other forms of illegal voting for which an ID requirement may be relevant.
The second is that “minuscule” is not “zero.” Occasionally we see races in North Carolina and elsewhere settled by dozens of votes or fewer. Could preventable voter fraud tip the balance in such races? It’s a low-probability scenario, to be sure. But as long as the net cost is also low, why not take out insurance against it? (A similar argument applies to absentee-ballot fraud, which North Carolina is now taking more seriously, as it should.)
The third argument is, indeed, that the net cost is low — because implementing the ID rule has some ancillary benefits. Those who lack photo IDs face other impediments in modern society. It may not be impossible for them to use financial services, public buildings, and certain forms of transportation — advocates sometimes misstate this — but it is certainly more cumbersome than it needs to be.
Voter ID is now the law of the land. Now it’s time to make it work, and to move on to more productive public-policy debates.