Voter confidence–a citizen’s perception whether their ballot will be counted as intended–came under scrutiny in the recent meeting of the Midwest Political Science Association.  Two papers examined the validity of voter confidence directly, my own paper with James Hicks, which can be downloaded here,  and a paper by Lonna Atkeson, Michael Alvarez, and Thad Hall. This last group of authors is particularly noteworthy because, as far as I can tell, Alvarez and Hall invented the concept of voter confidence when they included it in a survey conducted before the 2002 election.

Is voter confidence a key metric by which we can measure and monitor the performance of American elections?  Heather Gerken seems to think so–she suggests it as one potential entry into the Democracy Index sweepstakes.  Yet, as Michael Kang notes in review of Gerken’s book, we have to pay very close attention to the quality of the input into any such index.  Garbage in/garbage out is not a way to move the debate over election reform from “shouting to data driven arguments.”

In my recent paper with James Hicks, we started from the assumption that voter confidence was not a good metric of electoral performance, and was instead reflected broader assessments of the state of the nation and the performance of the political system.  Marc Hetherington, for example, argues that the collapse in public support for activist government was caused by declining trust in government.

What if we similarly conclude that “voter confidence” is similarly driven by the public mood, the popularity of the sitting president, or the state of the economy rather than breakdowns in election technology or well-publicized failures in ballot counting?

I haven’t read Atkeson’s paper yet, but I hope to put a review up here in the next week or two, but I spoke to her at the conference, and their conclusions seem very much in line with our own.  Voter confidence responds pretty much as we’d hope it would–while it is (not surprisingly) related to broader measures of trust and confidence in government, is is a) not at all responsive to short term policy evaluations and b) is most strongly related to features of the election, including:

• Did you vote for the winner or the loser?
• The quality of poll workers
• Concerns about voter fraud
• Overall evaluations of the voting experience

We did not find, however, contrary to some previous work, that voter confidence levels were significantly different across different modes of balloting (early in person, absentee, and at the precinct on election day).

We can debate whether survey measures belong in a democracy index, but these results leave me  confident that voter confidence measures what we think it measures.