Monthly Archives: August 2012

Changing a Law to Win an Election?

The Main Line Times’ “Notes from Narberth” columnist wrote a piece on August 7 (“Narberth zoning meeting; more on voter-ID need“) using Al Schmidt’s recent report on voting irregularities in Philadelphia as proof of need for a Voter ID law. You may read that here.

My response (“Changing a law to win an election?“) was published on the Main Line Times’ website on August 11. You may read that on the Main Line Times’ site here, or below.


One of the most interesting things about Philadelphia City Commissioner Al Schmidt’s report on “voting irregularities” is not what it says, but how it is being used. Proponents of Pennsylvania’s new Voter ID have praised the report as damning evidence of pervasive voter fraud – which is ironic, considering that a reading of the report gives no such impression.

First of all, Schmidt’s report is not a scientific study, and says nothing about how extensively voter fraud occurs – it only identifies seven types of voting irregularities. And of those seven kinds, the Voter ID law would stop only one. Whether the other six constitute legitimate problems or not is largely irrelevant to the Voter ID debate.

The problem Voter ID actually addresses can be boiled down to one question: is the person at the polling place who he says he is? Under current law, voters must prove who they are the first time they vote at a new polling place, and may use a variety of means to do so. Voters can face harsh punishments for not complying with election law ($10,000 in fines and 5 years in prison for voter impersonation) while they get little in return (one extra vote in an election decided by hundreds of thousands).

Faced with this information, you might conclude that voter impersonation rarely happens. All available evidence suggests you would be right. Numerous studies and investigations – including by the Bush Justice Department – have all reached this conclusion. The attorneys defending the Voter ID law in court recently agreed as well. So it is not surprising that Voter ID supporters have relied upon a single case: “the mysterious case of Joseph Cheeseboro” that the August 5 “Notes from Narberth” column referenced.

In any given election, Mr. Cheeseboro is little more than a rounding error. The same cannot be said of the more than 758,000 Pennsylvanians who may be disenfranchised by this law – a group disproportionately made up of elderly, minority, student and handicapped voters. That figure is larger than President Obama’s 2008 margin of victory, and to date new Voter IDs have been issued to less than 0.5 percent of that number.

Of course, even one illegal vote is one too many, but any voting system will leave room for error. The fundamental question at the Voter ID debate’s core is how to resolve the natural tension that arises from wanting to both keep elections accessible, while preventing any potential fraud.

Opponents of Voter ID do not, as the “Notes from Narberth” column contends, “think black people are too dumb to figure out how to get a photo ID.” Rather, we are concerned because the process of obtaining the underlying documentation and standing in line for several hours at a PennDOT center makes voting less accessible to legitimate voters, especially elderly and handicapped ones. It also opens the door to further chicanery, by providing another venue for vote challenging at the polls.

But perhaps most concerning of all is the nagging suspicion that the law was aimed to, as House Majority Leader Mike Turzai boasted, “allow Governor Romney to win the state of Pennsylvania.” When you have to resort to changing the law in order to win an election, there is something wrong.


Voter ID Proponents Taking Back Their Own Arguments

Since Pennsylvania’s Voter ID law passed last March, proponents have been steadily walking back each of the arguments they used to support it.

Early on, PA Secretary of the Commonwealth Carol Aichele wrote that:

A Pennsylvania Department of State analysis shows 99 percent of eligible voters currently have acceptable photo IDs, and proposals under discussion will likely expand the list of photo IDs that can be used.

Governor Tom Corbett’s administration stopped using the 99 percent figure after it was revealed that Aichele’s office had reached it by simply dividing the number of photo IDs the PA Department of Transportation has issued by the voting age population — a horribly imprecise measure, considering many people are issued photo IDs that are not eligible to vote, thus inflating the percentage.

Then, this past July, the PA Department of State released new numbers, saying that around 758,000 Commonwealth voters may lack the proper ID needed to vote this November. That’s more than 9 percent of all voters in the Keystone State.

Now, there has been some debate over the validity of these numbers as well. Namely, the state’s computers appear to have returned ‘false hits‘ for people whose names contained spaces (such as “Mary Sue”), were hyphenated (“Smith-Jones”), or who used nicknames on one document, but not the other (such as using “Frank” on your voter registration, but “Francis” on your ID).

Yet the 758,000 figure is still significant, especially when one considers that the Department of Transportation has (to date) only issued between 3,000 and 3,500 new Voter IDs. That is less than 0.5 percent of the number of people the Department of State says could potentially be missing the appropriate form of identification now needed to vote — it might as well be a rounding error.

Even if one thinks that the 758,000 figure overstates the problem, the gap between ID-less voters and new IDs issued is large enough to dwarf any concerns of inaccuracy. Imagine for a moment that 95 percent of the 758,000 people identified by Aichele’s office actually do have IDs. Under this scenario, there are still 37,900 people that do not have the proper identification — and PennDOT has so far only issued new IDs to about 9 percent of that new number.

Then there are the 906,000 people that the Department of State says have a driver’s license that will have been expired for more than a year by Election Day. As the Voter ID law requires the identification to be valid with a year, these people may face some difficulty.

Also, University of Washington political scientist Matt Barreto estimated in his testimony to the PA Commonwealth Court that more than 1.3 million voting-age Pennsylvanians could lack the necessary identification. It is important to note the difference between Barreto and the Department of States’ figures. Barreto looks at the whole voting-age population (which includes individuals that can register, but have not yet), while the state’s figures refer only to those currently registered to vote.

Furthermore, the issue of people’s names being different on voter registration when compared to IDs raises another concern: vote challenging. Poll watchers (or other voters) can challenge someone’s vote if, for instance, the name on an ID does not match the name listed an individual’s voter registration. Perhaps it is the difference between your legal name, and the name you go by in everyday life.

Barreto testified that more than 4 percent of eligible voters could have a “name listed on their ID [that] did not match that which would appear on the voter registration records.” This could possibly leave them open to a vote challenge.

The Committee of Seventy describes the process that follows:

The challenged Voter is put under oath by the Judge of Elections, after which they must sign a Challenge affidavit and produce one qualified Voter of the Division willing to sign the Affidavit alleging that the challenged voter is who they claim to be. After this process the challenged Voter may vote on a machine.

This process has a history of being used by political parties looking to gum up the works for opposition voters. The ultimate goal here is the same as that of Voter ID: make it as difficult as possible for people to vote.

With Voter ID, the hope is that the cost of underlying documentation (birth certificate, Social Security card), transportation, and long lines at the DMV will prevent numerous people from being able to comply with the law.

Barreto estimates that around 27 percent of eligible voters that lack a photo ID also lack some of the documentation needed to get a new Voter ID card. And, as the Brennan Center notes, 24 percent of Pennsylvania’s voting-age population live “more than 10 miles from nearest ID-issuing office that is open more than two days a week,” and 10 percent do not have access to a vehicle.

Indeed, as the Philadelphia Inquirer has documented multiple times, logistical problems abound:

Nine counties in Pennsylvania lack full-service PennDot photo and licensing centers, and officials are sometimes misinformed about where their closest full-service photo center is.

Witnesses that testified in the court case against the Voter ID law described “long lines, short hours, and misinformed clerks” at PennDOT offices.

These barriers do not need to prevent all of the opposition from voting, just enough to deliver 51 percent of the vote. In a tight election, such chicanery could make or break a race, and the Voter ID law only exacerbates this.

Barreto also noted that a substantial portion of voters are unaware of the new restrictions (34 percent) and almost all (99 percent) believe they have a valid ID. Of course, many of these voters are likely mistaken, given that we have already established that a large number of people do not have the proper identification. Keep in mind that only specific forms of photo ID are permissible under this law, namely:

  • PA government-issued ID (must have an expiration date and be currently valid, or have expired less than a year ago)
  • U.S. passport (must have an expiration date and be currently valid)
  • Military ID
  • Student ID (must be issued by a Pennsylvania college, have an expiration date and be currently valid)
  • Care facility ID (must have an expiration date and be currently valid)
The latter two, it should be noted, often lack expiration dates, and thus would not be considered valid for voting (though many colleges have begun printing expiration dates on their IDs in an attempt to ensure their students do not lose their voting rights). An out-of-state student whose only ID is a student card issued by an out-of-state college is out of luck.

The second major argument for Voter ID that supporters have since retracted is that the law is necessary to prevent voter fraud and ensure the integrity of elections. I have dealt with this fallacy at length in previous blog posts (here and here) and opinion columns (here and here), but the short of it is that numerous studies and investigations (including by the Bush Administration) have found that voter impersonation fraud appears to be relatively rare.

News21, a Carnegie-Knight journalism initiative, recently undertook possibly the most extensive analysis of voter fraud cases in America to date, filing public records requests for such information in every state. In Pennsylvania, they found that third party organizations and election officials were accused of perpetrating voter fraud more often than actual voters, and that voter registration fraud (which, under previous law, would not actually result in any illegitimate votes unless the fakers show up to the polls with identification) was the most prevalent form of voter fraud. Voter impersonation fraud, they contend, is nearly non-existant in the Keystone State and on a national level.

Even the Republican National Lawyers Association, which supports the law, identified only 400 cases of various kinds of voter fraud in the entire country over a decade’s time, and most of these were not voter impersonation, and thus would not be solved by Voter ID. And even those few cases may overrepresent the problem. A News21 analysis of the RNLA’s list reports that their “consisted mainly of newspaper articles about a range of election issues, with little supporting evidence of actual in-person voter fraud.”

The Corbett Administration and PA General Assembly Republicans vehemently maintained that voter impersonation fraud is a real and pressing problem. When he signed the Voter ID law in March, Governor Corbett stated:

Some have argued that there is no evidence of voter fraud. I don’t necessarily agree with that.

So it was surprising when the attorneys defending the Voter ID law in court agreed that they did not know of any voter impersonation fraud in Pennsylvania and “will not offer any evidence or argument that in-person voter fraud is likely to occur in November in the absence of the Photo ID law.”

That voter impersonation fraud is rare should be fairly intuitive; the rewards of voter fraud are very low, while the costs are potentially very high. Being caught trying to perpetrate voter fraud could mean a $10,000 fine and up to five years in prison, while only yielding one extra vote in an election that will likely be decided by hundreds of thousands. University of Essex government professor Sarah Birch noted in a March 2012 Economist article that “most manipulators make only sparing use of blatant election-day frauds,” opting instead to change election laws as a way to deter opponents.

To many, PA House Majority Leader Mike Turzai said as much at a Republican State Committee meeting in June while touting GOP achievements in Harrisburg:

Voter ID, which is gonna allow Governor Romney to win the state of Pennsylvania, done.

Indeed, the Voter ID law disproportionately impacts voters that have tended, in the past, to vote for Democrats.

PoliticsPA reported on an AFL-CIO analysis of state data that broke General Assembly and Congressional voting districts down and ranked them according to which ones would be most affected by Voter ID. The analysis was fairly unsurprising:

With just one exception, Democrats represent each of the districts with the highest rates of voters who lack PennDOT ID: the top 5 in Congress, the top 10 in the state Senate, and 19 of the top 20 in the state House, including each of the 8 House districts where over 50 percent of voters in 8 districts lack PennDOT ID. Republicans represent the 5 seats in Congress, 5 seat in the state Senate, and 10 seats in the state House with the lowest rates of voters who lack ID.

(Click here if you would like to see how your own districts fare.)

A map found on PoliticsPA further specified which areas of the state would be most impacted by Voter ID: cities. Specifically, the strongly Democratic Philadelphia and Pittsburgh would be the most affected.

From PoliticsPA.

But that’s just a breakdown of parties and places. At its core, the Voter ID debate is about people — namely, the elderly, minority, young, low income, and handicapped people that Voter ID would disproportionately impact. In this context, it makes sense that citizens living in Philadelphia and Pittsburgh would be hit hardest — both cities have markedly larger minority and low income populations than the state as a whole, and are hubs for large universities.

A recent Azavea study of Philadelphia voters found strong correlations between white, black, and Latino populations and the probability of having a valid ID that can be used to vote this November. That is, white voters were quite likely to have the necessary identification, while black and Latino voters were not. If history is a guide, it would perhaps be better to avoid policies that disproportionately disenfranchise minority voters.

Barriers to getting photo IDs could be especially high for seniors. The PA Department of State’s figures suggested as many as 1 in 4 voters over 80 could lack the necessary identification. Getting such voters to (and having them wait in line at) a PennDOT facility presents certain difficulties that Secretary Aichele herself has acknowledged. Additionally, the Pennsylvania Budget and Policy Center released a report criticizing PennDOT for poorly implementing new policies related to Voter ID. In particular, the report described lack of signage, workers offering incorrect information, and limited PennDOT hours.

Of course, such issues were discussed even before the bill passed into law. If General Assembly Republicans had been worried about such barriers to complying with their new law, they could have passed a number of proposed amendments that would have:

  • Extended DMV hours,
  • Made available more locations at which to secure a Voter ID, and
  • Provided for mobile units to help people register.
They both prevented such amendments from coming up for a vote, and rejected the ones that did.

Currently, the Voter ID law sits in the PA Commonwealth Court, where it waits for Judge Robin Simpson to announce his ruling. As this case deals with the fundamental right to vote, the state will have to prove that it had a compelling case for passing Voter ID legislation, that the law was narrowly tailored to achieve its goal, and that it was done in the least restrictive way.

To me, at least, it seems that the state has failed to prove its case. And if the admission by the lawyers defending the law that they neither know of nor expect pervasive voter impersonation fraud is any indication, it may seem that way to the state as well.



Examining Al Schmidt’s Report on Voting Irregularities

Philadelphia City Commissioner Al Schmidt recently released a report on “voting irregularities” that Voter ID advocates have been touting as irrefutable evidence of rampant voter fraud in Pennsylvania. If true, this would be quite a bombshell, shoring up the case for the Voter ID law and directing contradicting strong evidence that shows voter impersonation fraud just does not occur.

So is it true?

Well, let’s take a look at how the report was produced and what it contains. Schmidt produced the report after non-scientifically examining 14 voting divisions in Philadelphia during the 2012 primary election. By “non-scientifically,” I mean that these divisions were not chosen at random, but were specifically selected because of discrepancies between city and state data.

You see, Pennsylvania keeps election records at both the state and county levels. Your county records that you showed up to vote at your local polling place. After the election, another list is compiled for the state. But there were a lot of people who voted, and thus a lot of room for error.

Accordingly, it is not uncommon for some people to be missed. This does not change the outcome of the election — the voting booths still recorded the proper amount — but it can cause a discrepancy between county and state records.

Schmidt’s report looked specifically at the divisions that had the largest discrepancies between county and state records, and discusses seven types of voting irregularities:

  1. Voting by non-registered individuals
  2. Voting by individuals in the incorrect party’s primary
  3. Divisions with more votes than voters
  4. Voting by individuals in the incorrect legislative district
  5. Individuals voting more than once
  6. Voter impersonation
  7. Voting by non-U.S. citizens

The first thing to notice about this list in the context of the Voter ID debate is that very few of them actually deal with voter impersonation fraud. This specific kind of fraud — which is the only type of voter fraud the Voter ID law would actually prevent — occurs when someone shows up at the poll and says they are someone that they are not in order to receive an additional, illegal vote.

The simple fact is that Voter ID would not prevent the vast majority of irregularities Schmidt’s report identifies. In fact, six of the seven types on this list would not be affected at all by Voter ID. That’s because most of these deal with what poll workers and Election Board officials do and how they do it. Even the single alleged case of a woman double-voting would not have been prevented by Voter ID, because in that instance the woman was who she said she was.

There is only one kind of voting irregularity on this list that Voter ID would actually prevent: voter impersonation fraud.

None of this is a reflection upon Schmidt’s report but, rather, how Voter ID supporters have used it to advocate for the new law. Though the report does not seek to quantify voter impersonation fraud (and only contains one example), it has nevertheless been hailed by the Voter ID bill’s sponsor, PA State Rep. Daryl Metcalfe, as “[confirming] what leading Democrat opponents of voter photo ID and those in the mainstream media have been denying all along.”

I will discuss that instance of voter impersonation Schmidt identifies — the infamous “Joe Cheeseboro case” — but first, let us examine another claim often put forward by the Voter ID’s proponents.

Non-citizen voting.

Some of the law’s supporters contend that Voter ID will indeed stop widespread illegal voting by non-citizens. There are two responses to this: that there is little evidence that non-citizen voting is widespread, and that even if it were, this would not stop it.

To be clear, there are two basic types of non-citizens: legal and illegal. The arguments surrounding Voter ID have generally focused on the latter, probably because it plays into pre-existing fears of illegal immigrants as ‘law-breakers.’ The irony is that illegal immigrants are actually less likely to break the law. Indeed, the immigrant population as a whole are less likely to commit crimes, but especially undocumented ones. The reason is simple: illegal immigrants have to worry about being discovered and deported. Keeping out of trouble helps ensure they can remain in the country.

With that in mind, does it make sense that an illegal immigrant would want to risk deportation in exchange for one additional vote in an election that will be decided by hundreds of thousands? A quick look at the potential costs and benefits would suggest not.

Schmidt’s report does highlight an interesting aspect of non-citizen voting, though. Namely, that in the current system, election officials only discover if a non-citizen has voted after they have applied for citizenship. Once they apply, the U.S. Citizenship and Immigration Services (CIS) contacts the Board of Elections to determine if the individual has ever registered to vote. In the months between January 2012 and when the report came out in July, Schmidt’s office has been notified of 19 non-citizens that registered to vote. Of those, seven voted at least once in the last decade.

Again, the report does not seek to measure how often such fraud occurs, but other evidence suggests it is not at the epidemic levels Voter ID supporters have claimed. For instance, despite a vigorous campaign by the Bush Justice Department to uncover voter fraud, there were only 14 instances in which non-citizens were found to have voted illegally between 2002 and 2005.

Here’s the kicker: Voter ID would not have necessarily prevented these non-citizens from casting their ballots. After all, Voter ID does not require poll workers to check citizenship papers, only photo identification. And it is well within a legal non-citizen’s rights to get a driver’s license or other photo ID.

Remember, the seven instances in the Schmidt report were not cases of people showing up at the polls, saying they were someone they were not, and gaining one additional fraudulent vote. They were cases in which a person who was not legally allowed to vote had registered, shown up to the polls, demonstrated they were who they said they were, and then voted. Voter ID would not stop this.

As an interesting historical aside, numerous states (Pennsylvania included) originally allowed non-citizen voting in state elections. Nationalistic sentiment during World War I largely reversed these policies in the states and localities in which they remained, and in 1926 Arkansas became the last state to do away with it. However, even today various localities retain the authority to grant non-citizens a vote in local elections, and some continue to do so. Outside the United States, there are at least 22 countries in which non-citizens have some voting rights.

On the flip side of the argument, being an American citizen has not (and in some cases, still does not) always imply voting rights. Obviously, voting rights are subject to age restrictions, but they were also once subject to gender and race restrictions as well, despite citizenship status. Today, they are subject to one’s criminal record.

None of this is to say that illegal voting is permissible (it is not), nor is it to say that non-citizen voting is a preferable policy (it is not clear to me that it is). This is only to say that the citizenship-voting rights link is much more tenuous than it at first appears to be.

Currently, the first box on a voter registration application asks whether an individual is a citizen of the United States. Beyond that, however, the Election Board may only find out whether a voter is a non-citizen once they are contacted by CIS.

Some states, like Florida, have moved toward a tighter system in which they access a federal database (the Systematic Alien Verification for Entitlements, or SAVE, program) in order to determine an individual’s immigration status, and then compare that to voter rolls. Although voter registration rolls need to be kept up-to-date and accurate, there is some concern that purging voter rolls through comparison with SAVE could accidentally disenfranchise eligible voters. The Brennan Center reports that states’ current voter purging processes are “shrouded in secrecy, prone to error, and vulnerable to manipulation.”

Schmidt himself is described as having reservations about hypothetically using SAVE in Philadelphia, and for good reason. The possibility of tossing eligible voters off the registration rolls is not something that should be taken lightly. Especially when the possible benefits of such a policy are actually fairly small. As described above, available evidence suggests non-citizen voting is fairly rare — something even the Corbett Administration admits.

Obviously steps should be taken to curb it (the Brennan Center has policy suggestions for ways to purge rolls responsibly and transparently), as long as the proper safeguards are in place to ensure that the process is not politicized and legitimate voters are not disenfranchised in the process.

The curious case of Joe Cheeseboro.

There is only one type of voting irregularity in Schmidt’s report that a Voter ID law could actually stop, and that is voter impersonation. Evidence suggests that voter fraud is rare (and that a more usual way to fix elections is changing election law to deter opponents). As with non-citizen voting, the Bush Justice Department vigorously sought out instances of alleged voter fraud, and found little.

Schmidt’s report, though Voter ID supporters have presented it as building a strong case for Voter ID, brings little new to the debate. It only identifies characteristics of voter impersonation fraud through the story of Joseph J. Cheeseborough.

Available evidence suggests Mr. Cheeseborough is the victim of voter impersonation — he is registered at two polling places, and has voted twice in the same election in the past.

It is telling, however, that many Voter ID supporters are still drumming up Mr. Cheeseboro as the damning evidence for pervasive voter fraud. His case proves that voter impersonation fraud has occurred, but not the level of pervasiveness (again, I would point to other studies that say it is rare). Mr. Cheeseboro’s single case, while provocative, is not even a rounding error — it for all intents and purposes does not exist, except on the most excruciatingly technical level.

This is not to make light of potential voter fraud. Obviously, one illegal vote is too many. But it is a foregone conclusion that any democratic voting process will leave some room for error. The question we must answer, as a society, is what safeguards are we willing to enact in order to fulfill the twofold aim of:

  • Ensuring the rights of citizens to vote and
  • Maintaining the integrity of elections.

Currently, there are strong incentives against voter impersonation. In order to successfully impersonate another voter, you must know that he or she is registered at that poll and has not yet voted and does not intend to vote today. If he or she is a new voter at that polling place, you must bring some form of identification. And you must know that the poll workers, who live in that community, will not know that you are not the voter you are impersonating.

Being caught could mean up to a $10,000 fine and 5 years in prison. On the other hand, getting away with it means one lousy extra vote in an election likely to be decided by hundreds of thousands of votes.

In that context, does it really make sense to spend $11 million to disenfranchise more than 9 percent of Pennsylvanians and more than 18 percent of Philadelphians because one man potentially lied about being Joe Cheeseboro? Or would it make more sense simply to enforce existing election law, and make appropriate changes that take care to disturb legitimate voters as little as possible?

Judge Robin Simpson of the Commonwealth Court will likely decide the answers to these questions on August 13, when he issues his ruling regarding the court case against the Voter ID law. And since PA Supreme Court Justice Joan Orie Melvin has recused herself from hearing cases pending the resolution of criminal charges against her, the PA Supreme Court could deliver a tied 3-3 verdict. In that case, Simpson’s decision would stand.

For me, however, it seems fairly clear that the costs of the Voter ID law far outweigh the benefits.