Voter ID, Day Two: Statistically Speaking

A full transcript of the hearing is available here.


Day two of our voter ID trial in Harrisburg was about statistics - both the actual statistical impact of new voter ID requirements on Pennsylvania voters, and on the methodology and science behind statistical study.

ACLU-PA's Statewide Survey

Most of the day was taken up with questioning of Matt Baretto, an associate professor in Political Science from the University of Washington in Seattle and expert in survey science and barriers to elections. ACLU-PA and our co-counsel at Arnold and Porter contracted Baretto to conduct a statewide phone survey in late June and early July, to measure the likely impact of PA's new regulations on voters.

Under questioning from ACLU-PA Legal Director Witold "Vic" Walczak, Baretto provided an intensive crash-course in statistical science, first explaining the rigorous scientific standards by which his survey was designed and conducted, even delving briefly into the history of public opinion surveys, and then reviewing the (occasionally startling) results of his PA survey.

The full version of Baretto's study is available to the public here. The short version is that Pennsylvania's voter ID requirements mean serious trouble.

Firstly, Baretto's study revealed significant problems with awareness. As of June and July, even after the primary elections, fully a third of voters knew nothing about the new ID requirements. Even worse: Asked whether they believed they had ID that would qualify them to vote, 99% of those questioned said yes. When probed with more detailed questions to verify their ID, it turned out 12% of them were wrong. When will those people learn that their ID is not valid? Most likely, when they show up to vote.

The problem isn't that people lack any form of photo ID - less than two percent of Pennsylvanians can say that - it's that Pennsylvania's stringent requirements render many forms of photo ID invalid. Less than 86% of eligible voters in the state hold a form of ID that will qualify them to vote. Of the six million eligible voters who cast ballots in PA in 2008, three quarters of a million of them (about 13%) don't have the ID they will need to vote in November.

And, what's probably worst of all, more than 1 in 4 Pennsylvanians who lack the required ID also lack the backup documents required to obtain one - like social security cards, raised-seal birth certificates, marriage certificates, etc. According to Baretto's study, roughly 379 thousand Pennsylvanians currently have no valid ID for voting, and don't have the required documents to obtain one. 174 thousand of those individuals voted in 2008.

Disproportionate Impacts

Baretto's study attempted to identify groups that were particularly hard-hit by the new requirements. Voters were grouped according to several criteria: Ethnicity, age, education level, annual income, access to transportation, and region. While no subsegment had an especially high rate of ID possession (those earning $60-80 thousand a year did the best, with only 7.1% lacking valid ID) several groups were especially hard-hit:

• 17.2% of women lack valid ID, compared with 11.5% of men
• 18% of Latino voters, compared to roughly 14% of black or white voters
• 18% of voters ages 18-34, or age 75+
• 18.5% of voters with less than a high school education
• 22% of voters earning less than $20 thousand per year

The most alarming numbers, however, had to come from access to transportation. While voters with regular access to a car saw only 11% lacking valid ID, 30% of those whose primary transportation mode was bicycle or public transit did not possess ID that would permit them to vote - and a whopping 42% of voters who said they had no access to regular transportation also had no ID that would permit them to vote. Remember, please, that polling places are frequently within walking distance of an individual's home - PennDOT licensing centers are generally not.

Also notable was the geographic breakdown, with markedly lower rates of voting eligibility (roughly 18% lack valid ID) in Philadelphia and Allegheny Counties, Pennsylvania's most urban and densely populated. Only around 13% of voters in other counties were likely to lack ID.

On cross-examination, attorneys for the Commonwealth attempted to poke holes in Baretto's methodology, but Baretto met most of their barbed questions with explanations of statistical science. At one point a Commonwealth attorney asked Baretto whether he had ever "written on the opposite side," and found that voter ID was "more beneficial than harmful." Baretto explained that he does not apply judgements about benefit or harm, but only reports on data.

Commonwealth attorneys questioned the impact of the general lack of knowledge about voter ID, pointing out that the state government is planning to invest more than a million dollars in an education campaign around voter ID. Baretto expressed doubt, citing the remarkably short time frame between the proposed start of this awareness campaign and the November election. He cited a study in which voters were quizzed about the facts of a California ballot initiative, shortly after a multi-million dollar awareness campaign. Fourteen percent answered every single question wrong, while another 19 percent got only one or two questions correct. As Baretto pointed out, that's 33 percent total - almost exactly the number of people in Pennsylvania who are presently unaware of PA's requirements.

Baretto also stressed, repeatedly, how difficult it would be to raise awareness when fully 99 percent of Pennsylvanians say they believe their ID is already valid. As Vic explained it following court, "If people think their ID is already good, when that commercial comes on, they're going to go get a beer."

Undermining Integrity

Baretto's approach contrasted sharply with that revealed by Rebecca Oyler, of the Pennsylvania Department of State. Ms. Oyler testified about internal estimates of voters without PennDOT ID, and the related tax cost estimates supplied by the Department to the State Legislature when the voter ID law was being debated. The Department of State famously (and frequently) touted a figure of 1% of eligible voters in PA (or about 90 thousand individuals) who did not have photo ID from PennDOT. They revised that figure in June, following an internal audit, to roughly three-quarters of a million individuals - 9% of the population - and the revised cost to taxpayers of implementing new voter ID requirements scaled proportionately.

On the witness stand, Ms. Oyler testified that she herself put together the 1% figure in less than 24 hours, at the request of her superiors, relying on 2010 Census figures and statistics from an unknown source that she was told came from PennDOT. When Ms. Oyler asked PennDOT follow-up questions, including what portion of their ID database came from non-citizens, PennDOT did not reply.

Ms. Oyler also testified that, following that initial 24-hour calculation, no further effort was made to verify those numbers until after the voter ID law was passed. She testified that the Department of State believes the real figure is lower than 9%, but they are not presently able to present any evidence to verify that belief. 

In her testimony, Ms. Oyler stated that the Department of State seeks to encourage every eligible citizen to cast a ballot, and agreed that, if PA's voter ID requirement prevents eligible people from voting, then it "undermines the integrity of the election process."

Substantial Conformity, Unprecedented Authority

A buzz-word of the day, "substantial conformity" is the term PA's voter ID statue applies to the similarity between a voter's name on his or her photo ID and the name that appears on state election rolls. The legislature included no definition nor criteria for this term, and Ms. Oyler testified that, while the Department of State may issue recommendations to the county boards of election, those recommendations would be non-binding. Ultimately, the question of substantial conformity, and the decision as to whether two names match - say, for example "James Smith" and "Jim Smith" - will be left to those individual boards of elections, and ultimately to the individual poll workers. 

Leaving such a subjective determination in the hands of so many individuals raises significant questions. Substantive differences in name are not uncommon - particularly for recently-married women, who are likely to have obtained a new driver's license, but highly unlikely to have updated the election rolls. Voters whose ID is rejected would have the opportunity to cast a provisional ballot, but as they will have only six days to order and obtain a corrected ID card, the odds that their vote will be counted are slim. In his testimony, Baretto remarked that any voter who has an ID with a name that is not an exact match with his or her name on the voting rolls is "at risk" come election day.

A similar problem confronts the voter ID law's provision for "indigent" voters. According to the law, voters who are "indigent" are permitted to bypass ID requirements and instead complete a special form, which must be submitted to the county board of elections to accompany their provisional ballot. Once again, however, lawmakers failed to define "indigent," and so it is left to the county boards of elections, and ultimately to the discretion of individual poll workers, to decide who is indigent and who is not - as well as to decide whether to provide the "indigent voter" paperwork on-site at the polling place or require indigent voters to make a separate trip to county election headquarters.

In short, under PA's new law, when you hand that poll worker your photo ID card, you're handing over unprecedented authority over whether or not you can vote. Try to smile.

We'll be uploading video of Vic's post-trial comments later tonight or early tomorrow. Remember to follow ACLU-PA on Twitter and Facebook for regular updates throughout the day.