Voter Identification - The Pew Charitable Trusts

Voter Identification Michael J. Pitts 1 At first blush, the data generated by the Pew Center on the State’s research related to provisional balloting and voter identification would seem to indicate, as at least one commentator has surmised, that the level of passion and rhetoric related to the highly partisan voter identification debate might be a bit overblown. 2 After all, according to the data, only 5,431 persons had a provisional ballot rejected because of insufficient identification. When viewed in comparison to the overall total ballots cast for president, these persons represent just .0045% of the electorate. However, there are several legitimate reasons to think that the number of persons prevented from voting because of insufficient identification is much higher. For starters, 11 states did not report data for the reasons why provisional ballots were rejected, and some of these states, such as Texas and Illinois, have quite large electorates. 3 Perhaps just as importantly, among those not reporting data are three of the states (Indiana, Georgia, Arizona) with the most stringent identification requirements in the country. 4 Indiana and Georgia both require almost every prospective election‐day voter to provide government‐
issued photo identification. While no comprehensive study on photo identification’s impact in Georgia has been conducted, previous research from the 2008 primary election in Indiana might lead one to conclude that Indiana’s numbers alone could increase by about 900 persons (approximately 15 percent of the 2008 national totals) the number of provisional ballots rejected for insufficient identification. 5 Moreover, a recent judicial opinion, Gonzalez v. Arizona, involving Arizona’s voter identification requirement (where voters are required to present either photo identification or two pieces of non‐photo identification) 6 could lead one to surmise that nearly three thousand provisional ballots went uncounted in that State due to insufficient identification. 7 Admittedly, these estimates are rough and speculative. However, 1
Associate Professor, Indiana University School of Law – Indianapolis.
Stephen Ansolabehere, Effects of Identification Requirements on Voting: Evidence from the Experiences of Voters
on Election Day, 42 PS: Pol. Sci. 127, 130 (2009) (“The experience of individuals at the polls on Election Day
suggests that there is much less to the voter-identification controversy than appears in the pages of the court
decisions or debates in public forums.”).
3
The States that did not report reasons for rejecting provisional ballots are: Arizona, Connecticut, Georgia, Indiana,
Illinois, Mississippi, Rhode Island, South Dakota, Texas, Virginia and Wisconsin.
4
In addition, South Dakota, which also has a somewhat controversial photo identification law, did not report reasons
why provisional ballots were rejected. See Spencer Overton, Stealing Democracy 148-50 (2006) (describing the
controversy over South Dakota’s photo identification law). South Dakota law, however, is a bit less restrictive than
Indiana and Georgia law because South Dakota has an affidavit fail-safe. Id. at 163.
5
Michael J. Pitts, Empirically Assessing the Impact of Photo Identification Through An Examination of Provisional
Balloting, 24 J. L. & Pol. 475 (2008). That study determined that 16.1% of provisional ballots cast in the 2008
Indiana primary election were cast due to problems with identification and that 21.1% of those ballots ultimately
were counted. Id. at 499. The current 2008 general election data shows Indiana with 7,239 total provisional ballots,
so one could estimate that 1,165 provisional ballots were cast due to identification issues and that only 246 of those
provisional ballots ultimately were counted.
6
Gonzalez v. Arizona, No. CV-06-1268-PHX-ROS, slip op. at 5 (D. Ariz. Aug. 20, 2008).
7
The district court noted that: “In the 2006 primary, 2006 general, and the 2008 Presidential preference elections [in
Arizona], 3,135,951 ballots were cast. Of these, 4,194 ballots, or 0.13%, were uncounted due to a lack of proof of
identification.” Gonzalez v. Arizona, No. CV-06-1268-PHX-ROS, slip op. at 15 (D. Ariz. Aug. 20, 2008) (internal
citations omitted). Applying this rate to the 2,293,475 ballots cast for president in Arizona at the 2008 general
2
1
the fact remains that data are lacking from several of the states that employ the most stringent voter identification laws. There would also seem to be a lack of data among the states actually reporting reasons for rejecting provisional ballots because 11 8 states, including a couple with very large electorates (New York and Pennsylvania), reported that absolutely no provisional ballots had been rejected for insufficient identification. It’s true that six 9 of these eleven States do not have an identification requirement aside from the Help America Vote Act’s (HAVA) mandate for first‐
time voters who register by mail. 10 Yet certainly it would seem that a few voters in these States would be excluded by the HAVA requirement. 11 Even more startlingly, five of the States that reported no provisional ballots cast due to insufficient identification have an identification requirement that extends beyond the minimum requirements of HAVA. For instance, Pennsylvania requires all first‐time voters (not just those who register by mail) to present identification. However, Pennsylvania reported that not a single provisional ballot was rejected due to insufficient identification despite almost 33,000 provisional ballots cast. Contrast this with Kansas which has a requirement similar to Pennsylvania’s and which reported 1,152 provisional ballots rejected due to insufficient identification with about 40,000 provisional ballots cast. 12 For these reasons, it seems likely that some of the uncounted provisional ballots in these States were voter identification‐related and were either misclassified or would be found among the more than 22,000 uncounted provisional ballots nationwide that were lumped into the catch‐all category of “other” as the reason for rejection. Furthermore, a comparison with data from the 2006 general election would also point toward a possible undercount of the number of provisional ballots rejected for insufficient identification. In 2006, there were 791,831 provisional ballots cast nationwide and 5,938 of those ballots were rejected because “no identification was provided.” 13 In 2008, however, more than twice as many provisional ballots were cast (2,037,887), yet the number of ballots rejected because of insufficient identification decreased (5,431). One can see this reduction election, one could conclude that 2,982 ballots went uncounted due to insufficient identification. In addition, it’s
worth noting that the 2008 general election data shows that more than 40,000 total provisional ballots went
uncounted in Arizona, so an estimate that about 3,000 went uncounted due to insufficient identification does not
seem unreasonable.
8
Alaska, Delaware, District of Columbia, Hawaii, Maine, Nebraska, New York, Oregon, Pennsylvania, Tennessee,
and Vermont.
9
District of Columbia, Maine, Nebraska, New York, Oregon and Vermont.
10
42 U.S.C. § 15483(b). It’s worth noting that Maine’s lack of rejections may be caused by the fact that it employs
election-day registration.
11
It’s possible that in some of these States voters were exempted from the HAVA identification requirement
because: (1) the voter registered by mail under Section 6 of the National Voter Registration Act; (2) as part of the
registration application, the voter provided a driver’s license number or at least the last four digits of the voter’s
social security number; and a State or local election official matched the number with an existing State identification
record bearing the same number, name, and date of birth as provided in the registration. 42 U.S.C. §
15483(b)(3)(B).
12
All characterizations of the voter identification laws of the individual States in this paragraph were taken from
Electionline.org’s comprehensive list that is available at
http://www.pewcenteronthestates.org/uploadedFiles/voterID.laws.6.08.pdf (last visited May 4, 2009).
13
U.S. Election Assistance Commission, The 2006 Election Administration and Voting Survey 19, 21 (Dec. 2007).
2
playing out on the individual state level as well. Take Ohio, for example. At the 2006 general in Ohio, a total of 129,432 provisional ballots were cast, with 2,726 rejected for lack of identification. 14 In 2008, Ohio had many more provisional ballots cast (206,859) but fewer not counted because of insufficient identification (1,990). Of course, it’s possible that the reduced rate of insufficient identification rejections might reflect, among other things, greater awareness of identification requirements in the marketplace of voters or reflect that those voters lacking identification just stayed home this time (more on this factor in a moment) or reflect that election officials changed counting procedures. At bottom, though, it seems counterintuitive for the total number of provisional ballots cast nationwide to have increased by more than 1.1 million yet the number rejected for insufficient identification to have decreased. A comparison with data from a survey of voters at the 2008 general election also points toward a potential serious undercount of the impact of voter identification. A voter survey indicated that 0.65% of persons who voted cast a provisional ballot after being asked to show picture identification. 15 That would mean more than 784,000 provisional ballots may have been cast because of identification problems. 16 If that many provisional ballots were cast because of a lack of identification and, as the survey of election officials suggests, only 5,431 went uncounted, that would be an extraordinarily high count rate—upwards of 99%. Of course, there could be other explanations for this apparent discrepancy. The voter survey does not necessarily show that voter identification issues were what caused these persons to cast a provisional ballot. A person who was asked to show picture identification may have ended up casting a provisional ballot for a reason unrelated to identification; for instance, because the voter’s name was not on the registration list. Nonetheless, the voter survey provides additional evidence suggesting a significant undercount of the number of provisional ballots rejected due to insufficient identification. Finally, it should be noted that even if the current data was unassailably accurate with regard to provisional ballots rejected due to insufficient voter identification, actual provisional ballots represent only one piece of the puzzle related to gauging voter identification’s impact on the electorate. For instance, many voters who lack identification may not show up at the polling place at all. One estimate is that more than 2 million registered voters stayed away from the polls in 2008 due to a lack of proper identification. 17 Still other persons may have shown up at the polls and not been offered a provisional ballot by poll workers. And still other persons may have refused a provisional ballot even if offered one. That said, it’s worth mentioning the possibility that every person prevented from voting because of insufficient 14
Steven Huefner et. al, From Registration to Recounts 32, 49 (2007).
R. Michael Alvarez, et. al., 2008 Survey of the Performance of American Elections (2009). This rate was
calculated as follows. The survey included 10,000 voters. Of those 10,000 voters, 9,316 indicated that they
definitely voted. Id. at 94. Of the 9,316 persons who voted, 61 (0.65%) indicated that they were asked for picture
identification and then cast a provisional ballot. Id. at 99-100.
16
This number was calculated as follows. At the 2008 general election, 120,698,778 total ballots were cast for
president. Taking 0.65% of the total ballots cast, one can extrapolate that 784,542 persons who were asked for
picture identification cast a provisional ballot.
17
R. Michael Alvarez, et. al., 2008 Survey of the Performance of American Elections 58-59 (2009).
15
3
identification should have been prevented from voting because they were attempting to vote fraudulently. 18 Regardless of which side of the voter identification debate one is on, when it comes to assessing voter identification’s impact, there’s likely more to the story than just the number of provisional ballots rejected due to insufficient identification. In sum, the data related to provisional ballots and voter identification represents a useful starting point for further inquiry and is certainly better than the alternative (no data at all). Nevertheless, like a lot of good data, it raises as many questions related to voter identification as it resolves. 18
One interesting consideration in terms of trying to measure the level of “fraud” are the numbers from Florida. In
Florida, voters are required to show photo identification on election day. If a voter cannot produce photo
identification on election day, the voter then casts a provisional ballot. During the post-election canvassing period,
such a provisional ballot should be counted as long as the signature on the provisional ballot form matches the
signature on the registration form. See Florida Division of Elections, Frequently Asked Questions,
http://election.dos.state.fl.us/gen-faq.shtml (last visited May 20, 2009).
The data indicates that Florida rejected 394 provisional ballots for insufficient identification. Presumably,
this means that canvassing boards found 394 signatures on provisional ballots that did not match the signatures on
voter registration forms. If this is the case, then this may represent 394 instances of voter identity theft.
Of course, there could be alternative explanations other than fraud. Perhaps these provisional ballots were
cast for lack of identification and the prospective voter neglected to sign the provisional ballot. Or perhaps some
canvassing boards lack the expertise to do a proper signature match. Or perhaps some voter’s just displayed poor
penmanship.
4