The real reason no one impersonates dead voters: High risk, little benefit

In an effort to demonstrate that the specter of voting fraud in America is real, the conservative agitator James O'Keefe and his group Project Veritas recently sent a handful of people into a voting center during the New Hampshire primary to obtain ballots on behalf of dead registered voters. (You may remember O'Keefe as the guy who dressed as a "pimp" in an undercover ACORN sting, or who made so much trouble for NPR.) Several were successful, as a selectively edited video from Project Veritas spoon-fed to the Daily Caller demonstrates.

The trend throughout the United States is to enact new laws that will make photo IDs a prerequisite for participating in the democratic process. Proponents of voter ID laws use voting in lieu of dead people as the main example of fraud, while opponents point out that there is no evidence of widespread fraud and significant evidence that such laws make it more difficult for students and those in lower-income brackets to vote. Lawmakers in South Carolina used the accusation that 957 dead people voted in the "recent elections" as proof of the need for voter ID laws—a claim the New York Times' Andrew Rosenthal points out is very poorly supported. (The Justice Department has blocked the measure in South Carolina, so voters on Saturday will not need a photo ID to vote.)

But surely by accident, O'Keefe has actually given use some extremely valuable data about the cost-benefit of trying to vote on behalf of a cadaver. Of the handful of people O'Keefe sent into voting centers to vote as dead people, at least one was recognized as being an imposter. It is unclear how much trouble he or the rest of Project Veritas will end up in since they did not actually cast the ballots they obtained; TPM reports that merely obtaining the ballots fraudulently could violate federal law as well, even if no voting took place. Project Veritas refused to tell me how many people participated in their stunt, but I can be extremely conservative and say that voting for a dead person carries at least a 1 in 100 risk of being recognized and possibly ending up in legal trouble.

Based on the most conservative estimates, then, we can estimate that voter ID laws could disenfranchise between 10,000-500,000 eligible voters for every 1-100 blocked fraudulent votes. Here's how I get there:

It may seem like a government-issued photo ID isn't so much to ask to cast a vote—after all, you need one to drive, get on a plane, or have a beer. The fact is that many eligible voters do not have the right documents under new or proposed laws. The right-wing Heritage Foundation trumpets a paper that claims that only 1.2% registered voters lack valid a photo ID. That may seem low, but nearly 130 million votes were cast in the 2008 presidential election, so that would translate into roughly 1,560,000 voters. The Heritage Foundation's estimate is the lowest I could find. In 2007, the Georgia Secretary of State estimated 198,000 registered voters there did not have government issued photo IDs and in South Carolina, 200,000 registered voters do not have a photo ID that would be valid for voting under the proposed law, according to the state election commission. That translates into roughly 4-5 percent of voters for Georgia and 8-10 percent of voters for South Carolina, based on 2008 registration and vote totals.

Those eligible citizens who do not have a photo ID tend towards the more disenfranchised citizens: 25% of African-Americans have no photo ID, 15% of people earning less than $35,000 have no photo ID, and 18% of the elderly have no photo ID. This represents millions of citizens in each category. Such laws also penalize college students since many of these laws require in-state photo IDs, which prevents college students from voting at their college if they attend from out-of-state.

Voter ID laws do not stop people who have fraudulently registered as themselves. The vast majority of these cases are people who believed themselves to be eligible, notably felons that do not know they are ineligible to vote in a given state. States that bar felons, such as Florida, have traditionally been so vigilant in blocking felons that thousands of eligible voters have been inadvertently purged from the voter rolls in the state's fixation to ensure that felons do not vote. Nor would these laws stop non-citizens from voting as themselves. (Even so, investigations have found voting by non-citizens to be extremely rare; a study of 370,000 votes cast in Milwaukee from 1992-2000 showed 4 votes by non-citizens.)

The main voter fraud that photo IDs would stop, then, is that of a person voting in lieu of another registered voter; this is likely someone who has died, as it is otherwise hard to estimate when a live registered voter will not be voting. Again, studies have shown very few votes by dead people in recent election cycles; this study by the FBI showed that all 89 dead voters in a Maryland election died after they voted. Many other presumed dead voters are caused by clerical errors on death certificates.

Second, the type of voter fraud that voter ID laws could stop is extremely costly to the criminal if he or she is caught, as Veratis' stunt helps us to compute, which is probably why it is so rare.

The cost of impersonating a dead eligible voter is the penalty times the likelihood of being caught plus the fear of being caught (for those who are risk averse). Thus, for a United States citizen or a legal immigrant the cost is, at minimum, 1/100 of up to 5 years in prison plus $10,000.

The benefit of impersonating a dead voter, meanwhile, is one more vote for your chosen candidate. Florida's 2000 Electoral College vote was an anomaly; the average Electoral College vote over the last 10 cycles, 510 elections, was 247,848 votes, with just Florida and New Mexico in 2000 under 1,500.

So here's the question: if the most conservative estimates are correct and 10,000 eligible voters are disenfranchised so that 100 non-eligible votes can be stopped, do you think that that is a fair deal for democracy? What if the more mainstream estimates are true and the number is closer to 100,000 eligible voters being disenfranchised so that 10 fraudulent votes can be stopped? Whichever figures you use, the math comes out squarely against these controversial measures.

David Rothschild is an economist at Yahoo! Research. He has a Ph.D. in applied economics from the Wharton School of Business at the University of Pennsylvania. His dissertation is in creating aggregated forecasts from individual-level information. Follow him on Twitter @DavMicRot and email him at

Want more? Visit The Signal blog or connect with us on Facebook and follow us on Twitter.