Skip Navigation
Expert Brief

“Citizens Without Proof” Stands Strong

A 2006 national survey commissioned by the Brennan Center found that as many as 11% of voting-age adults, millions of Americans did not have a current and valid government-issued photo ID. This is a troubling statistic for advocates of strict photo ID policies. Hans von Spakovsky and Alex Ingram of the Heritage Foundation recently published a memo attacking our study and this widely-cited, independently confirmed statistic. This document contains details on our study and explains why their attacks are baseless.

Published: September 8, 2011

A Response to Von Spakovsky and Ingram

Wendy Weiser, Keesha Gaskins, & Sundeep Iyer

In Novem­ber 2006, the Bren­nan Center for Justice published Citizens Without Proof, a report docu­ment­ing the find­ings of a survey conduc­ted by the Opin­ion Research Corpor­a­tion, a prom­in­ent inde­pend­ent research firm, about the extent to which Amer­ican citizens possess govern­ment-issued photo ID and docu­ment­ary proof of citizen­ship.[1]  The report detailed the study’s meth­od­o­logy, its prin­cipal find­ings, and the asso­ci­ated margins of error.  Most prom­in­ently, the study found that 11% of voting-age Amer­ican citizens—and an even greater percent­age of African Amer­ican, low-income, and older citizens—do not have current and valid govern­ment-issued photo IDs.

Since its public­a­tion, Citizens Without Proof has been widely cited by schol­ars, legal experts, and the media, and its find­ings have been widely accep­ted.  What is more, its prin­cipal find­ings have been repeatedly confirmed by multiple inde­pend­ent stud­ies.  For example:

  • The 2001 Carter-Ford Commis­sion on Elec­tion Reform found that between 6 and 11 percent of voting-age citizens lack driver’s licenses or altern­ate state-issued photo IDs.[2]
  • A 2007 Indi­ana survey found that roughly 13 percent of registered Indi­ana voters lack an Indi­ana driver’s license or an altern­ate Indi­ana-issued photo ID. [3]
  • A 2009 study in Indi­ana found that of the citizen adult popu­la­tion, 81.4% of all white eligible adults had access to a driver’s license, compared to only 55.2% of black eligible adults.  It also found that strict photo ID require­ments have the greatest impact on the elderly, racial and ethnic minor­it­ies, immig­rants, those with less educa­tional attain­ment and lower incomes.   [4]
  • A 2007 report based on exit polls from the 2006 elec­tions in Cali­for­nia, New Mexico, and Wash­ing­ton State found that 12% of actual voters did not have a valid driver’s license.[5]
  • A prom­in­ent national survey conduc­ted after the Novem­ber 2008 elec­tion found that 95% of respond­ents claimed to have a driver’s license, but that 16% of those respond­ents lacked a license that was both current and valid.[6]

Recently, however, Citizens Without Proof has come under attack by polit­ical oper­at­ives support­ive of strict photo ID require­ments for voting.  On August 24, 2011, Hans von Spakovsky and Alex Ingram, through the Herit­age Found­a­tion, published a memor­andum[7] seek­ing to discredit the study, criti­ciz­ing the study’s meth­od­o­logy and the Bren­nan Center’s report­ing of its results.  This docu­ment responds to their base­less criti­cisms.

The Study’s Data Collec­tion Meth­ods Follow Best Prac­tices

Citizens Without Proof reports the results of a national survey conduc­ted by the well-respec­ted inde­pend­ent research firm, Opin­ion Research Corpor­a­tion (ORC), in Novem­ber 2006.  The Citizens Without Proof survey was part of a broader tele­phone survey conduc­ted by ORC that month, for which ORC followed stand­ard industry prac­tice in terms of survey design, select­ing the appro­pri­ate number of survey parti­cipants for stat­ist­ic­ally signi­fic­ant results, random selec­tion of survey parti­cipants, and method of ques­tion­ing survey parti­cipants.  ORC used its stand­ard demo­graphic screens—i.e., ques­tions to determ­ine demo­graphic char­ac­ter­ist­ics of survey parti­cipants, such as race, citizen­ship, and age—­for the entire survey.  In other words, the survey meth­od­o­logy used for Citizens Without Proof was the same widely respec­ted meth­od­o­logy typic­ally used by ORC and similar survey research entit­ies.  With respect to the survey ques­tions relat­ing to photo ID and citizen­ship docu­ment­a­tion, before conduct­ing the survey ORC analyzed and revised the proposed survey ques­tions and correc­ted for any poten­tial bias.

Notwith­stand­ing these facts, von Spakovsky and Ingram criti­cize the survey because it “could have included illegal and legal aliens.”  This is base­less.  As Citizens Without Proof clearly reports, ORC specific­ally ques­tioned survey parti­cipants as to whether they were U.S. citizens, using ques­tions gener­ally accep­ted in the industry.  The survey results were limited to U.S. citizens of voting age and did not include illegal or legal aliens.

Von Spakovsky and Ingram also try to criti­cize the survey on the ground that it was not limited to “actual or likely voters, registered voters, or even eligible to vote at all” [sic].  This too is base­less.  First, Citizens Without Proof does not purport to present find­ings of how many actual, likely, or registered voters do not have the docu­ments stud­ied.  Despite von Spakovsky and Ingram’s insinu­ation, the report does not in any way misrep­res­ent its survey pool.  To the contrary, it clearly says that the survey’s find­ings relate to all voting-age Amer­ican citizens, and not to the survey parti­cipants’ likely parti­cip­a­tion in any elec­tion.  That state­ment remains true. 

Second, contrary to von Spakovsky and Ingram’s asser­tion, the survey was, in fact, essen­tially limited to eligible voters, since it focused exclus­ively on U.S. citizens over the age of eight­een, the main determ­in­ants of voter eligib­il­ity across the coun­try.  (While it is theor­et­ic­ally possible that the survey could have captured a small number of indi­vidu­als rendered ineligible to vote because of disqual­i­fy­ing crim­inal convic­tions, based on the national rate of those disqual­i­fic­a­tions, that number would be minis­cule and would have a stat­ist­ic­ally insig­ni­fic­ant effect on the study’s results.) 

Third, von Spakovsky and Ingram are wrong that a more appro­pri­ate survey pool for assess­ing the fair­ness of photo ID or proof of citizen­ship require­ments for voting would have been actual, likely, or registered voters.  While it is true that citizens in those groups are more likely to vote in any given elec­tion, they are not the only citizens who have the right to vote.  It is certainly relev­ant to assess how many people of those entitled to vote would be preven­ted from doing so if they tried because of a photo ID or proof of citizen­ship require­ment.  The fair­ness of photo ID and proof of citizen­ship require­ments is not solely a factor of their effect on over­all turnout in run-of-the-mill elec­tions; it is also a factor of their effect on the abil­ity of every eligible Amer­ic­an—whether or not she has voted recently—to parti­cip­ate in future elec­tions.  Indeed, it is often people who do not frequently parti­cip­ate in elec­tions who peri­od­ic­ally become excited by a new candid­ate, mobil­ize to vote, and change the outcomes of elec­tions.

The Study’s Survey Ques­tions Are Valid and Trans­par­ent

As noted above, the survey ques­tions used in Citizens Without Proof were analyzed and revised by the inde­pend­ent Opin­ion Research Corpor­a­tion to ensure that they did not reflect any bias.  And indeed they did not.  To enable peer-review­ers and other read­ers to assess the ques­tions for them­selves, Citizens Without Proof reprin­ted them in full.  But von Spakovsky and Ingram appar­ently neglected to read them.

The ques­tion that led to the bulk of the report’s find­ings was the follow­ing ques­tion: “Do you have a current, unex­pired govern­ment-issued ID with your picture on it, like a driver’s license or a milit­ary ID?”  Eleven percent of all survey respond­ents said that they did not, and even higher percent­ages of African-Amer­ican, low-income, and older citizens said that they did not.  It is hard to imagine how such a straight-forward ques­tion can be inter­preted as biased in any way. 

Nonethe­less, von Spakovsky and Ingram try to claim that the ques­tion was indeed biased.  They do so by completely misquot­ing the ques­tion, claim­ing that the survey “did not ask respond­ents whether they had govern­ment-issued IDs but rather asked whether respond­ents had ‘read­ily avail­able iden­ti­fic­a­tion.’”  That is completely untrue.  Although the term “read­ily avail­able iden­ti­fic­a­tion” might indeed be confus­ing if used in a survey, it was not used in this survey at all. 

Von Spakovsky and Ingram further criti­cize this ques­tion because it does not include a more compre­hens­ive list of all govern­ment-issued photo IDs, includ­ing milit­ary IDs and student IDs, as examples.  Again, they misread the ques­tion, which expressly includes milit­ary ID as an example.  And while the ques­tion does not list state-issued student photo IDs as an example, those IDs are clearly covered as a form of “govern­ment-issued ID with your picture on it.”  Moreover, posses­sion of student photo IDs is not espe­cially relev­ant to the ques­tion of the fair­ness of voter ID laws; only some states that require photo ID to vote accept student photo IDs.

In addi­tion to their failed attempt to discredit the photo ID ques­tion, von Spakovsky and Ingram criti­cize the study’s ques­tion relat­ing to docu­ment­ary proof of citizen­ship.  That ques­tion was: “Do you have any of the follow­ing citizen­ship docu­ments (U.S. birth certi­fic­ate/U.S. pass­port/U.S. natur­al­iz­a­tion papers) in a place where you can quickly find it if you had to show it tomor­row?” Seven percent of survey parti­cipants answered no.  Von Spakovsky and Ingram claim that the inclu­sion of the clause, “in a place where you can quickly find it if you had to show it tomor­row” biased the results.  That clause was included to differ­en­ti­ate between people who actu­ally had posses­sion of their birth certi­fic­ates and those who believed that at some point in their lives their parents obtained a birth certi­fic­ate for them and thus that the docu­ment must exist some­where.  Absent that clause, test survey parti­cipants who did not have birth certi­fic­ates but who assumed they should have access to their birth certi­fic­ates were likely to erro­neously answer yes.  What is more, Citizens Without Proof accur­ately repor­ted its results.  Specific­ally, it repor­ted that 7% of respond­ents “do not have ready access to citizen­ship docu­ments.”  In other words, unlike with the photo ID ques­tion, it did not purport to reflect find­ings of how many people have citizen­ship docu­ments some­where, but rather only those who have those docu­ments read­ily access­ible to them.  Even if the survey could have accur­ately determ­ined how many people have citizen­ship docu­ments some­where—and we concluded that it could not because of the confu­sion relat­ing to birth certi­fic­ates—the number of people with ready access to those docu­ments is argu­ably more relev­ant to the ques­tion of the fair­ness of proof of citizen­ship require­ments for voting.

The Study Accur­ately and Ethic­ally Discloses Its Results and Limit­a­tions

Without any basis, von Spakovsky and Ingram repeatedly suggest that the Bren­nan Center misrep­res­en­ted the find­ings of its survey.  To the contrary, Citizens Without Proof very scru­pu­lously and accur­ately reports each of the survey’s find­ings.  Indeed, before the Bren­nan Center publish­ing the find­ings, the report was reviewed in full by the inde­pend­ent Opin­ion Research Corpor­a­tion to make abso­lutely sure that the results were accur­ately described.  ORC also provided the Bren­nan Center with the margins of error for each of the find­ings repor­ted, and all such margins of error were fully included.

Von Spakovsky and Ingram try to cast asper­sions on the accur­acy of the survey’s report­ing by criti­ciz­ing the study’s weight­ing of survey responses to account for the under­rep­res­ent­a­tion of minor­ity respond­ents.  This again displays their ignor­ance of proper survey meth­od­o­lo­gies.  In fact, this type of weight­ing of survey responses is stand­ard prac­tice in the field.  Weight­ing removes sample bias from a survey sample so that the results better reflect the target popu­la­tion.[8]  For example, imagine a random survey of 100 Amer­ic­ans ages 15 to 64, where respond­ents include 60 males and 40 females.  In that age range, the general popu­la­tion is 50% male and 50% female. To correct for the demo­graphic discrep­ancy between the random sample and the popu­la­tion, a researcher would weight each male respond­ent as 0.83 and each female respond­ent as 1.25.  Consist­ent with accur­ate, ethical and respons­ible survey prac­tices, ORC weighted the survey results to accur­ately reflect the rate of photo ID posses­sion among all Amer­ican citizens.

Von Spakovsky and Ingram also find it omin­ous that 135 survey respond­ents indic­ated that they have both a U.S. birth certi­fic­ate and U.S. natur­al­iz­a­tion papers, suggest­ing that this means that results were not fully repor­ted.  While this does indic­ate some misun­der­stand­ing among respond­ents about the nature of the docu­ments described in the ques­tion, the misun­der­stand­ing in no way biased the survey’s results in favor of lack of docu­ment­a­tion.  All 135 of those respond­ents were included in the repor­ted results as indi­vidu­als who have citizen­ship docu­ments.  Had they been excluded from the results, the study would have found an even higher percent­age of Amer­ic­ans without citizen­ship docu­ments.

Finally, von Spakovsky and Ingram raise the base­less criti­cism that the report improp­erly relies on 2000 Census data.  Again, this is highly mislead­ing.  Noth­ing in the survey find­ings depends on Census data.  For example, the find­ing that 11% of all voting-age citizen survey respond­ents said they did not have govern­ment-issued photo ID has noth­ing to do with Census data.  Rather, the Census data was used only narrat­ively, to provide read­ers with an under­stand­ing of the total number of people who are “11% of all voting-age citizens.”  Accord­ing to 2000 Census data, that number was around 21 million people; the number is almost certainly higher in 2011.  (Although it is completely irrel­ev­ant, von Spakovsky and Ingram are also wrong that the Census uses non-citizen popu­la­tion numbers to estim­ate the number of voting-age citizens.)

The Study’s Find­ings Have Been Repeatedly Confirmed, Not Under­mined

As noted above, the find­ings of Citizens Without Proof have been repeatedly confirmed in subsequent stud­ies.  Nonethe­less, von Spakovsky and Ingram try to under­mine the Citizens Without Proof by citing stud­ies that supposedly came to differ­ent results.  Again, their criti­cism is base­less.  First, unlike Citizens Without Proof, none of the stud­ies they cite attempt to survey the number of voting-age Amer­ic­ans who have current state-issued photo IDs.  And second, those stud­ies each have seri­ous prob­lems.

The 2006 survey they cite for the propos­i­tion that only 23 people out of 36,000 nation­wide were unable to vote because of an ID require­ment is both irrel­ev­ant and mislead­ing.[9]  First, in 2006, only one state (Indi­ana) required voters to present govern­ment-issued photo ID to vote.  Thus, the lack of photo ID would not have preven­ted voters in other states from voting.  Second, and more import­antly, the 2006 survey did not ask whether people voted by regu­lar ballot or by provi­sional ballot; it merely asked whether they voted.  But under federal law, anyone whose vote will not count because they cannot meet a photo ID require­ment is still entitled to vote a provi­sional ballot.  In other words, it is possible that thou­sands of people who voted a provi­sional ballot in 2006 and thus told survey ques­tion­ers that they voted in fact did not have their ballots count because they did not present photo IDs.

Simil­arly, the Amer­ican Univer­sity survey they cite, in addi­tion to being an outlier, has seri­ous meth­od­o­lo­gical prob­lems.[10]  Although the study concluded that few registered voters in Mary­land, Indi­ana and Missis­sippi lacked photo ID, the way the survey sampling was done assured that the results would dramat­ic­ally under­es­tim­ate the number of voters without ID.  Specific­ally, contrary to common prac­tice, the study did not adjust for under­rep­res­ent­a­tion of minor­ity and poorer popu­la­tions, which are less likely to be included in a random sample. These popu­la­tions have been shown in virtu­ally every other study on the topic to dispro­por­tion­ately lack photo ID.  In fact, the authors of the survey noted, “[s]ome of the limit­a­tions of the study stem from the unanti­cip­ated results. We expec­ted that a much larger number of registered voters would lack a photo ID, and so we did not over-sample any specific popu­la­tion.”  Moreover, even if the results were not skewed, unlike Citizens Without Proof, the Amer­ican Univer­sity survey focused only on registered voters as opposed to eligible voters.

Finally, von Spakovksy and Ingram erro­neously rely upon reports from state and federal agen­cies that more photo IDs are issued than there are registered voters.  Specific­ally, they point to a July 2011 article from the Colum­bus Dispatch report­ing that Ohio has 28,000 more driver’s licenses than voting-age resid­ents.[11]  But a simple compar­ison of the total number of state-issued photo IDs to the number of registered voters is inad­equate to determ­ine whether how many registered or eligible voters there lack current, valid photo ID required by some states to vote.  A closer review of Ohio’s drivers’ license list manage­ment prac­tices shows clear reas­ons why the number of IDs may exceed the number of voting-age citizens without chan­ging the fact that there are many citizens who lack photo ID.  First, the Ohio motor vehicle agency (BMV) only removes people who have moved out of state from the BMV rolls if they apply for a license in another state and that state noti­fies Ohio.  If the BMV does not receive noti­fic­a­tion, the license is marked as expired but remains on the rolls.  The holder of that license, however, is not an eligible voter in the state.  Second, Ohio only removes deceased people from the BMV rolls after a family member sends the BMV a copy of the death certi­fic­ate.[12]  A BMV Drivers License attend­ant estim­ated that “many” of the 105,000 indi­vidu­als over 18 who die in Ohio every year[13] remain in the BMV driver data­base, as local counties do not auto­mat­ic­ally send death certi­fic­ates to BMV.  Family members tend to do so only if they receive a renewal notice for the deceased, which may occur several years after death.  And third, Ohio’s driver’s license list includes signi­fic­ant numbers of perman­ent resid­ents and other non-citizens who cannot vote.  Accord­ing to the Amer­ican Community Survey, there were 191,439 resid­ents of Ohio of voting-age who were not citizens in 2009.  This is equal to 2.17% of Ohio’s voting-age popu­la­tion. (There are no stat­ist­ics avail­able on how many people with driver’s licenses were not citizens.) In other words, the total number of entries on bloated drivers’ license lists does not at all reflect the number of voting-age citizens who have or do not have state-issued photo ID.

*          *          *

In short, in their ill-conceived diatribe, von Spakovsky and Ingram fail to raise any legit­im­ate criti­cism of Citizens Without Proof.  It is for good reason that Citizens Without Proof remains the fore­most study of the number of voting Amer­ic­ans who lack govern­ment-issued photo ID and citizen­ship docu­ment­a­tion.


[1] Citizens Without Proof: A Survey Of Amer­ic­ans’ Posses­sion Of Docu­ment­ary Proof Of Citizen­ship And Photo Iden­ti­fic­a­tion, Bren­nan Center for Justice (2006), avail­able at http://www.bren­nan­cen­ter.org/dynamic/subpages/down­load_file_39242.pdf.

[2] The National Commis­sion on Federal Elec­tion Reform, To Assure Pride and Confid­ence in the Elect­oral Process, (2001), avail­able at http://fl1.find­law.com/news.find­law.com/hdocs/docs/elec­tion2000/elec­tion­re­form­rpt0801.pdf.

[3] Matt A. Barreto, et al., WASH­ING­TON INSTI­TUTE FOR THE STUDY OF ETHNI­CITY AND RACE, THE DISPRO­POR­TION­ATE IMPACT OF INDI­ANA VOTER ID REQUIRE­MENTS ON THE ELECT­OR­ATE, (2007) avail­able at http://depts.wash­ing­ton.edu/uwiser/docu­ments/Indi­ana_voter.pdf.

[4] Matt A. Barreto et al., The Dispro­por­tion­ate Impact of Voter-ID Require­ments on the Elect­or­ate—New Evid­ence from Indi­ana, PS: Polit­ical Science and Polit­ics, 111 (Janu­ary 2009), avail­able at http://faculty.wash­ing­ton.edu/mbar­reto/papers/PS_VoterID.pdf.

[5] Matt A. Barreto et al., Voter Id Require­ments And The Disen­fran­chise­ments Of Latino, Black And Asian Voters (2007), avail­able at http://faculty.wash­ing­ton.edu/mbar­reto/research/Voter_ID_APSA.pdf.

[6] R. Michael Alvarez et al., 2008 Survey of the Perform­ance of Amer­ican Elec­tions (2009), avail­able at http://www.pewtrusts.org/uploaded­Files/wwwpewtrust­sorg/Reports/Elec­tion_reform/Final%2520re­port20090218.pdf.

[7] Hans von Spakovsky and Alex Ingram, Herit­age Found­a­tion, Without Proof: The Unper­suas­ive Case Against Voter Iden­ti­fic­a­tion, (2011), avail­able at http://www.herit­age.org/Research/Reports/2011/08/Without-Proof-The-Unper­suas­ive-Case-Against-Voter-Iden­ti­fic­a­tion.

[8] For general illus­tra­tion of this prin­ciple see: Eric L. Dey “Work­ing with Low Survey Response Rates: The Effic­acy of Weight­ing Adjust­ments” Research in Higher Educa­tion Volume 38, Number 2, 215–227. Avail­able at http://www.eric.ed.gov/PDFS/ED387018.pdf.

[9] Stephen Ansol­abehere, The First Big Survey Of Voter ID Require­ment­s—And Its Surpris­ing Find­ings, Slate, March 16, 2007, http://www.slate.com/id/2161928/.

[10] Robert Pastor et al., Center for Demo­cracy and Elec­tion Manage­ment, Amer­ican Univer­sity, Voter IDs Are Not the Prob­lem: A Survey of Three States (2008), avail­able at http://bren­nan.3cdn.net/9dfd902a36ac8a7a27_aqm6b0w84.pdf.

[11] Bill Bush, Ids Exceed Voter-Age Resid­ents, The Colum­bus Dispatch, July 25, 2011, http://thevot­ing­news.com/state/ohio/ohio-ids-exceed-voter-age-resid­ents-the-colum­bus-dispatch/.

[12] In order to have a spouse’s or relat­ive’s Ohio driv­ing record marked as deceased, a copy of the death certi­fic­ate or a letter from the coron­er’s office must be provided. You can mail to the Ohio Bureau of Motor Vehicles, Attn: License Support Veri­fic­a­tion Unit, P.O. Box 16784, Colum­bus, Ohio 43216–6784 or fax to (614)752–7987.

[13] Accord­ing to the Centers for Disease Control, 972,223 people died in Ohio between 1999 and 2007. Less than 3% of deaths are among those below 24. Accord­ingly, if we conser­vat­ively estim­ate that 2.5% of deaths were among those below 18, we can safely estim­ate that between 105,000– 108,000 people in Ohio age 18 and older die every year.