Conservatives say that voters should be required to show ID as a measure to prevent fraud. Liberals say that ID requirements disenfranchise the disproportionately poor and minority voters who don’t have IDs and might have trouble getting them — even if efforts are made to provide IDs for free. The liberals’ panic has always struck me as overblown, though it’s also difficult to say how much fraud exists that’s preventable by requiring IDs.
A new study argues that voter ID probably doesn’t matter much either way, and also suggests a middle road: “non-strict” ID laws that ask voters to provide identification but allow them to vote, subject to some restrictions, even if they don’t. It focuses on two states that already have such laws: In Michigan, those without IDs must sign an affidavit certifying their identity; in Florida, their signatures at the polls are checked against the signatures on their registration forms.
For scientific purposes, the neat thing about these states is that they keep track of how many people vote without ID, which tells us how many voters might be disenfranchised if these alternative voting processes weren’t available. The answer is not many:
Our best guess as to the fraction of ballots cast without identification in local elections in Florida is only 0.03%. Our upper bound estimate is that 0.10% of ballots cast are without ID. Similarly . . . in state and national elections in Florida, our best estimate and upper bound of the fraction of votes cast without identification are only 0.016% and 0.064%, respectively. The rate of voting without ID is somewhat higher in Michigan, where we estimate that 0.3% (from 2004–2016) and 0.31% (2012–2016) were cast without IDs.
The question, of course, is how much we can generalize from these states to other states with different laws. The results do suggest that stricter states could switch to non-strict rules without opening the floodgates to massive numbers of ID-less, potentially fraudulent voters. It’s slightly less clear what the numbers mean for lax states: It’s possible that some ID-less voters, fraudulent or legitimate, might vote when there’s no ID law at all but would be discouraged by an affidavit requirement or signature check.
To me, the takeaway is that these middle-of-the-road laws are a good approach. With these policies in place, no claim of “disenfranchisement” can pass the laugh test, there are systems in place to discourage ID-less voters from committing fraud, and rates of ID-free voting are low enough that any such fraud will virtually never affect an election.