Andrew Rubenstein, Intern, AEI-Brookings Election Reform Project
December 17, 2008
In the aftermath of Obama's historic victory last month, the country will soon forget another of its pre-election obsessions: the failed prospect of a catastrophic, Floridian malfunction in the election's administration.
But this success owes itself as much to accident as accomplishment. Rightly, election experts emphasize long lines, intermittent machine failures, and, what some consider
the year's most important administration failure, the exclusion of voter names from mismatched statewide and local registration rolls. According to the National Journal
, the Election Protection Coalition alone received 50,000 calls on November 4th with reports of voting problems. Jonah Goldman, of the Coalition, speculates, "If this election were being decided by 537 votes," as was Florida in 2000, "there were a dozen different areas where you could find [points of controversy]."
With this increased awareness and interest in the prevention of election malfunctions, embodied in 2002's Help America Vote Act (HAVA), researchers have shifted their attention to an important correlate: voter confidence in the electoral process. As Charles Stewart of MIT has written
, "a combination of greater media attention to election-system failures, transitional chaos and razor-thin election margins have increased anxieties" over the management of elections.
Exit polls now reflect the popularization of this concern, as pollsters have asked voters since 2004 and 2008, "How confident are you that votes in your state will be counted accurately?" These numbers, though crude, reveal a striking pattern. In 2008, of the 9 percent of voters who responded "Not very/Not at all confident," 53 percent supported McCain while only 42 percent Obama. In 2004, the inverse pattern appeared in greater disproportion with 27 percent of the "Not Confident" pool voting Republican, versus 69 percent
The evidence of such disproportion, and such a marked shift, indicate the malleability of voter confidence and how it reacts to exogenous variation, raising two related and significant questions: What factors influence voter confidence? And, of equal if not greater consequence, how does voter confidence affect voter turnout?
Recent research speaks to these concerns. In a wide study (PDF)
, "Are American's Confident Their Ballots Are Counted?", the Caltech/MIT Voting Project surveys two voter groups following the 2004 election, identifying three striking determinants of voter confidence: race, political affiliation, and surprisingly, the technology on which voters cast their ballots.
They find with regard to African-American voters, for example, that the number who are confident their vote in 2004 "was counted as intended is significantly lower than the proportion of white voters who are similarly confident." Likewise, though race and party affiliation may reinforce one another, the data show a "direct relationship between party identification and voter confidence," which the 2004 and 2008 exit polls underline. Importantly, neither Republican nor Democratic values breed electoral distrust in and of themselves; rather, the extent to which an election's result pleases one party or the other encourages or discourages confidence. The researchers call this a "winner's effect," in that the success of a partisan's party inspires his or her confidence in an election's outcome. (An earlier study (PDF)
, "Voter Confidence in Context and the Effect of Winning," confirms the finding: "At the national level identification with the winning party/candidate increase[s] voter confidence following the election.")
Evidence of such racially and politically disproportionate voter confidence levels should interest election students and officials for two reasons. The first requires further investigation: the possibility of a causal relationship between voter confidence in elections and voter trust in government. To the extent that voters distrust an election outcome, do they question the legitimacy of that elected government? Though the two trust metrics may vary independently (the Caltech/MIT team notes in their paper, "Voters may believe that the electoral process is fair and accurate but simultaneously hold the belief that all politicians are crooks"), their relationship has profound policy implications. "Extensive evidence" shows that distrust in government disinclines voters to reelect incumbent candidates and can even "shap[e] the limits of [an elected official's] political possibilities (PDF)
," suggesting the extent to which voter trust issues meaningfully impact public opinion.
The second reason has more immediate implications: voter confidence directly affects voter participation. The Caltech/MIT researchers observe "a positive relationship between voter turnout and confidence...[that] more confident voters are more likely to turnout to vote." The team calls its findings "preliminary," but they too suggest something important. As we have seen, lower voter confidence exists among historically losing voters-minorities and out-of-office partisans-who might vote against a party in power. Where losing voters' lack of confidence discourages them from voicing their opposition, the very source of a vital democracy suffers.
With these consequences in mind, election officials should work to understand better their role in shaping voter confidence. They cannot, of course, control for voters' histories, but they can address a third variable that the Caltech/MIT team identifies: that voter confidence fluctuates across districts and technologies. The researchers emphasize technology in their study, concluding, "Paper absentee ballots and precinct-cast electronic ballots appear to have the largest negative effect on confidence [of the various technologies]."
But such explanations, though attractive in their clarity, highlight only proximate determinants of confidence. Recent studies find that voters' confidence varies based on their experiences with a given technology, which is a more malleable variable than the specific technology they use. "Voters' direct experience with the voting process influences their voter confidence," write (PDF)
Lonna Rae Atkeson and Kyle Saunders in a recent paper. "[T]he more a voter enjoyed her method, the more confident she was that her vote counted."
What factors, then, influence a voter's experience at the polling place or with a given technology? The addition of a voter verified paper audit trail (VVPAT), for one, seems to increase the confidence of DRE-users, suggesting that "the effect on voter confidence of an electronic voting device equipped with a VVPAT and a paper ballot are indistinguishable (PDF)
Voters' experiences with technologies depend to an even greater extent, though, on their experiences with the poll workers who administer those systems. As voters encounter only poll workers during an election--not county clerks or other more-senior officials--Thad Hall, J. Quin Monson, and Kelly Patterson find
that "overlooking the recruitment and training of competent poll workers can have a detrimental effect on voter confidence." In a recent comparison
of DRE and paper ballot voters, Stein et al. roundly summarize these findings: "[G]ood policy implementation and administration can provide voters with an effective, efficient, secure, and usable election system, independent of the specific voting technology used
...electoral administration dominates ballot effects on voter confidence" (emphasis mine).
This growing body of research highlights four significant points:
(1) Not only does a considerable group of voters lack confidence in the electoral process, it is a racially and politically disproportionate
group of voters.
(2) Voter confidence impacts voter participation.
(3) Administration decisions--the quality of poll workers, the addition of a paper trail--directly influence voters' experiences on Election Day and their perceptions of voting technologies.
(4) And, lastly, voters' experiences at the polling place directly influence their confidence.
These findings require further investigation, especially the relationship of voter confidence to voter trust in government; administration decisions that increase voter confidence; and the full extent to which voter confidence affects participation, which the Caltech/MIT experts handle only preliminarily in their work.
Still, they add an important shade to the study of elections. The goal of election reform should expand to include not only direct disenfranchisement but indirect disenfranchisement through administration decisions that depress voter confidence. As such, trust in the electoral process may stand as another metric by which to gauge the success of election reform efforts. Through promoting distrust in the very mechanism of democracy, we shoot ourselves in the foot before casting a single vote.
Viewpoint is an occasional feature analyzing various election reform issues.