At this point, the politics of "voter fraud" are pretty clear. Republicans use claims of voter fraud to justify draconian rules and voter purges that are targeted at traditionally Democratic-leaning demographics and geographic areas.
But how do you explain the so-called voluminous cases of non-citizen voting, double voting, deceased voting, and felon voting? Aren't there numerous studies that show that those sorts of fraudulent voters are commonplace?
The talking point is "fraudulent voting is rare" but what are the real facts? Some juicy tidbits:
- "Dead voters" typically voted before they died, often by absentee ballot
- "Felon voters" are frequently people convicted of more minor crimes, and are actually eligible to vote
- The matching rules used to find voters who voted more than once are statistically useless
- Virtually none of these cases of fraud would be resolved by a more stringent voter ID law
A report by the Brennan Center in 2007 absolutely eviscerates claims of voter fraud.
Matching Problems
One class of allegations of voter fraud centers around finding the same person voting in multiple places. But the assumptions that underly those accusations assume that two people with identical names and birthdates are the same person.
In reality, the chance that two people in a group of just 23 people have the same birthday is 50%. The chance that two people in a group of 57 people have the same birthday is 99%. The Daily News breathlessly reported in 2004 that more than 46,000 people were registered to vote in both NY and CA.
To check for voters registered in both places, the Daily News matched computer records from election officials in New York City and Florida.
Records of the city's 3.7 million registered voters were matched against a statewide Florida database of 10.7 million registrants based on first and last name, gender, birth date and middle initials
.
Which brings us to the next point, being registered more than once.
Multiple registrations
The Daily News hung its hat on the fact that being registered in more than one place is evidence of intent to vote in more than one place. 14% of Americans move each year, and with many New Yorkers moving to Florida, it is not surprising that a large number of New Yorkers who moved to Florida are still registered in New York. In fact, there is no legal requirement for those who move to notify their local election officials.
A small study in New Hampshire found that 67 people were registered to vote in both Dover and Durham, but that none of those voters voted more than once.
Importantly, increased voter ID requirements would not reduce this sort of "voter fraud".
Dead voters
Sensational reports about "dead voting" have a serious methodological flaw. These studies typically match death records after the election with votes. A study of 89 allegedly deceased voters in Maryland in 1995 found zero actual deceased voters. They had all voted before dying.
Additionally, elderly voters frequently vote by absentee ballots, so many of these cases are voters who voted before the election by absentee ballots, and die afterward.
Felon voters
In 2005, the GOP alleged that Dino Rossi lost herhis close election against Christine Gregoire as the result of illegal felon voting. In that case, hundreds of the cases cited were actually people who committed juvenile crimes. And reports of felon voting simply assume that people convicted of a felony are not eligible to vote, and do not investigate the possibility of a restoration of voter rights.
And of course, the matching problems mentioned above are in play here as well (again, simply assuming that somebody with the same name, gender, and birthday is the same person is hysterically irresponsible).
Requiring photo ID would not prevent this, even if it was happening.
Returned mail (aka caging)
A popular technique used recently is to show that people are still on the voter rolls even if they no longer reside in the district in question. There are numerous reasons that could cause mail to be returned undelivered:
- Pure error (in 1990, 4.8 million forms were returned undelivered, but 1.8 million were able to be delivered by hand)
- The person uses a PO box
- The voter has moved but remains eligible to vote without re-registrations (moving within a precinct)
Requiring photo ID would not prevent people voting if they moved outside the eligible district.
Commercial address
Another cited example of fraud is signing up to vote in areas zoned as commercial or in government buildings. Firstly, zoning is not dispositive. Additionally, in many cases, homeless people are registering to vote at the location of their shelter.
Requiring photo ID would not resolve the issue of people fraudulently registering to vote in places where they do not live.
Documented cases of fraud
When all of these maybes are taken into consideration, positively tiny numbers of documented fraud are found. Usually in the single digits in a given state. For instance, in Missouri, a state that was the subject of a frenzied search for voter fraud, only 6 cases of double-voting were found, none of which could have been prevented by increasing photo ID requirements.
In 2005, an attempt to prove that significant double-voting had occurred in New Jersey found just 8 cases of invalid voting--all double-voting. None could have been prevented by more stringent photo ID requirements.
Summary
Claims of voter fraud are based on incredibly flawed metrics for "identical persons". In the case of voter registration problems, the most significant "problems" are due to moving to a new address and re-registering there, something that 14% of Americans do every year.
The celebrated Daily News study took an extremely common case of such moves (New York to Florida), but showed just 400 cases of claimed double-voting (when their headline sensationally reported 40,000 cases of fraud). With more than 7 million votes case in both New York and Florida, it's not at all surprising that 400 cases of people with the same full name, gender, and birthdate voted in both states. In fact, it would be surprising if there were significantly fewer such cases.
And here's the kicker: these claims, completely unsubstantiated, lead to significant voter purges (the recent dramatic lawsuit in Ohio is an example of such a case). So instead of solving a serious problem, Republicans are looking to purge common statistical side-effects (same-name, same-gender, same-birthdate registered twice or deceased), juvenile "felons", people who moved in the last year.
And in many cases, the screen for "same person" is even less stringent than same-name, same-gender, same-birthdate. If you see a study claiming to prove significant voter fraud, take a look at how they determined same-personhood.
And finally, increased voter identification requirements resolves virtually none of these problems. I'd strongly recommend taking a look at the full study. We need to be armed with the facts about how common problems and statistical anomalies are being twisted into claims of widespread fraud.