In the summer of 2001, the National Commission on Federal Election Reform issued its report, To Ensure Pride and Confidence in American Elections. Seven years later, the U.S. Supreme Court decided in Crawford v. Marion County Board of Elections that strict photo voter ID laws were constitutional. Justice John Paul Stevens wrote that “public confidence in the integrity of the electoral process has independent significance because it encourages citizen participation in the democratic process.” And in 2017, President Trump signed an executive order establishing the Presidential Advisory Commission on Election Integrity. The group was charged with studying “laws, rules, policies, activities, strategies, and practices that enhance the American people’s confidence in the integrity of the voting process” and factors that undermine voter confidence.
Viewing voter confidence as a measure of the quality of American elections and assuming that voter confidence is linked to the willingness of Americans to vote are intuitively appealing. However, research indicates only a weak causal connection between voter confidence and voter turnout, and it does not show clear causal links between certain high-profile election administration practices, such as voter ID laws, and voter confidence.
Surveys have been measuring public opinion about voter confidence for nearly two decades. How the question about confidence is asked generally determines whether voters are deemed to have high or low confidence in elections. The strongest influence on levels of voter confidence, regardless of how the question is asked, is whether one’s candidate has won or lost an election. Beyond the “losers’ regret” phenomenon, voter confidence is influenced in smaller ways by the intensity of partisan competition and the experience of casting ballots.
In measuring any concept, it's important to know what’s being measured. When we try to see if voters are confident, we need to ask exactly what they're confident about—that the election was fair? That the votes were counted as they were cast?
When political scientists study voter confidence, they ask voters some variant of the question, “Do you believe that votes in the most recent election were counted as cast?” Note that this question focuses on the mechanics of marking a ballot and having it counted accurately, which is narrower than asking whether the last election was fair—and asking whether the last election was fair is narrower than asking whether “elections in America are usually fair.”
Answers about voter confidence are fairly consistent regardless of the precise wording of the question. However, there is one important exception to this pattern of consistency. When survey respondents are asked about their confidence in the parts of the electoral process with which they have direct contact, such as their own vote, they are much more confident than when they are asked about parts of the electoral process they have indirect contact with, such as the process in the nation as a whole. One consequence of this is that in political debates involving voter confidence, it is possible to cherry-pick survey research in support of arguments that voters either have high levels of confidence in American politics or do not.
Here's an example of how degrees of voter confidence vary with the degree of direct voter contact with the process. In the MIT module to the 2016 Cooperative Congressional Election Study, respondents were asked the following question before the election: “How confident are you that your vote in the General Election will be as you intend?” They were also asked, “How confident are you that votes in [your county/your state/nationwide] will be counted as voters intend?” The response categories were very confident, somewhat confident, not too confident, not at all confident, and don’t know.
In looking at summaries of the answers to these questions (Figure 1), two patterns immediately jump out. First, regardless of the target of the question (“your vote,” the “local vote,” etc.), confidence was greater after the election than before. For instance, the percentage of voters who said they were very confident their vote would be counted as they intended was 40% before the election, but this jumped to 59% when they were asked after the election.
Second, confidence was greatest—both before and after the election—as the target of the question got closer to the voter. For instance, in the pre-election question, while 40% of respondents overall were very confident their own vote would be counted as intended, 35% were very confident that votes would be counted as intended in their own county or community. These percentages dropped to 27% and 19% when the question asked about the state and the nation, respectively.
Questions about the mechanical aspects of voting tend to elicit more optimistic responses in public-opinion surveys than vague questions about the honesty of elections. For instance, immediately before the 2016 election, the Gallup organization asked respondents, “How confident are you that, across the country, the votes will be accurately cast and counted in this year’s election?” To this question, 69% responded that they were either very or somewhat confident. However, the poll also asked about how much confidence they had in the “honesty of elections.” Here, only 30% answered they were confident.
Confidence and outcomes
We've already seen that answers to voter-confidence survey questions vary as the target of the question changes. What else affects voter confidence? The main answer points us in two directions related to political outcomes—who wins, and how close elections are.
The “winner’s effect” is illustrated by the changing answers to voter confidence questions since they were first asked in 2000. Figure 2 shows the percentage of respondents who reported they were very confident their vote was counted as intended in the Survey of the Performance of American Elections, which was conducted immediately after the 2008, 2012, and 2016 elections. The data for prior elections shows the percentages taken from commercial public opinion polls and reported in research by Michael Sances and Charles Stewart.
In the first two elections, Republicans were more likely by a margin of 20 percentage points to say they were confident that their vote was counted as intended. The relative opinions of Democrats and Republicans switched in 2008 and became even more entrenched in 2012. Republican confidence rebounded in 2016 while Democratic confidence sagged. All the major changes in confidence correspond with changes in the parties’ electoral fortunes. Also notice that the overall average level of voter confidence did not change very much across the last five elections, even though the confidence of the partisan contingents changed to correspond with the electoral fortunes of the parties.
Another way elections influence voter confidence is through the sniping that goes on between candidates in the heat of a campaign. In a close contest, it's common for both sides to accuse the other of dirty tricks. The effects of such sniping are evident when we asked respondents whether votes in their own state were counted as intended. Figure 3 shows the percentage of voters who were very confident that their own states’ votes were counted as intended in the 2016 presidential election on the y-axis, and the percentage margin for the winning candidate on the x-axis. Note that the states up against the y-axis—those won by the smallest margins—on average had voters with lower confidence than states in which the winning margin was substantially higher.
Confidence and administration
It's common to justify election reforms by arguing they will increase voter confidence in the electoral system. However, there's little evidence that election administration has a direct effect on voter confidence. The major exception to this statement is that voters who experience problems at polling places tend to be less confident than voters who don’t.
The issue of voter confidence and election reforms has been front and center in justifications for stricter voter ID laws. As noted above, Justice Stevens’s decision justifying the constitutionality of strict voter ID laws in Crawford v. Marion County Board of Elections credited Indiana’s argument that strict voter ID laws could increase voter confidence. However, subsequent research on this question reveals no correlation between the adoption of strict voter ID laws and increases in voter confidence. Indeed, if anything, the political climate created by debates about strict ID laws could actually be reducing confidence and further polarizing opinions along partisan lines.
Where election administration has some influence on voter confidence is on how voters experience the process and whether that experience is positive. Research by scholars such as Lonna Atkeson, Mike Alvarez, Thad Hall, and Paul Gronke tells us that voters tend to be more confident when they don’t wait a long time to vote, when they encounter polling place officials who seem competent, and when they vote in person rather than by mail. Some of these factors certainly can be affected by state policies, but more often, they are influenced by local administrators' decisions about how to allocate resources to polling places and how rigorously they train poll workers.