Ranked-choice voting — the grand experiment in elections that Minneapolis tried out again last fall — may salve some of what ails our democracy. But, for now, it leaves open the well-documented voting gap that favors white voters and the affluent. That populist conclusion arises from careful statistical analyses of votes in the Minneapolis election and the evidence it yields of differences in participation between communities of color and the poor vs. their white and affluent counterparts.

We take our hats off to the enthusiasts of ranked-choice voting (or RCV) for sizing up problems in our democracy and then rolling up their sleeves to do the hard work of actually reforming elections in Minneapolis. We also salute another accomplishment: The RCV process — in which voters rank up to three preferred candidates, and then the weakest vote-getters are dropped until one of those remaining achieves a majority — accomplished something that truly astounds us. Negative campaigning became bad politics.

Slamming an opponent risked alienating his or her supporters and losing any chance of securing their second- or third-place rankings. Here's something we've never seen: The media and voters struggled to detect candidate criticisms of one another during the Minneapolis mayor's election. Did anyone following the 2012 presidential election struggle to identify Barack Obama's and Mitt Romney's differences?

But fans of RCV promised more. FairVote Minnesota, its champion, announced in its 2013 press kit that the reform ensures that a "larger, more diverse swath of the electorate gets to participate in the political process." Important promise — equal voice for each citizen is a fundamental democratic principle. What does the evidence show?

We carefully examined voting in the 13 wards in Minneapolis to see if RCV contributed to broader participation based on actual vote returns obtained from the Minneapolis city clerk. Unfortunately, the evidence shows a clear pattern. Voters who were more affluent and white turned out at a higher rate, completed their ballots more accurately and were more likely to use all three opportunities to rank their most preferred candidates compared with voters living in low-income neighborhoods and in communities of color.

Specifically, we compared the three wards that stood out as the most affluent (11, 12 and 13) with those that were least affluent (2, 3, and 5), as well as those that had the highest percentage of white voters (10, 11, 12 and 13) with those with greatest proportion of people from communities of color (4, 5 and 9). We did not include the Sixth Ward, because its voting participation was an outlier; the disparities we describe below are stronger when the Sixth Ward is included.

RCV did not close the well-documented turnout gap that favors affluent and white voters. Among registered voters in the most affluent wards, 42 percent turned out in the Minneapolis election, compared with 28 percent in the poorest areas. In wards with the greatest concentration of whites, 39 percent of voters turned out, compared with 26 percent in wards with the most people of color. These differences are statistically significant, justifying a high level of confidence in them.

The complicated RCV ballot — including its new rules and rows of candidates — raised another obstacle. In economically better off sections of town, 3.36 percent of the ballots were "spoiled" — the legal term for voter stumbles in selecting candidates according to strict guidelines — compared with 4.92 percent in the poorer parts. The comparable proportions for communities with smaller and larger proportions of racial minorities were 3.37 percent and 5.77 percent, respectively. These differences are highly significant in statistical terms.

Advocates for RCV also champion its pioneering process for recording a fuller range of voter preferences by allowing citizens to rank up to three candidates. Unfortunately, the better-off were advantaged here, too. In affluent wards, 21 percent of voters failed to fill out all three preferences, compared with 24 percent in poorer neighborhoods. The comparable proportions for white and minority wards were 22 percent and 26 percent, respectively. Although these differences may appear small, they are highly significant in statistical terms, indicating a robust and important pattern.

How important are these results? Racial and income disparities in Minneapolis triggered public alarm when they produced one of the largest "education gaps" in the country. What we are pointing to represents another disparity — a "democracy gap."

Passions run hot on RCV. Our research is not an "interpretation" or "opinion." It is a sober, objective analysis of voting results from the Minneapolis city clerk's data.

Let's turn down the passion around RCV and put on our thinking caps. RCV contributes to improving our democracy in certain respects but still falls short of its promise to improve participation by all parts of our community.

What can be done? Plans to decrease the number of candidates and possibly revise the ballot to reduce its complexity might help. We also would recommend new thought about how the city and media can improve the quality of information that is distributed to all communities.

Candidates have a role, too. Playing hide-and-seek on differences to avoid alienating opponents' supporters deliberately trips up voters who are trying to accurately sort the campaigns. We need to set expectations and scrutinize candidates to come clean on where they stand and how they differ from their rivals.

Lawrence R. Jacobs is director of the Center for the Study of Politics and Governance at the University of Minnesota's Humphrey School. Joanne M. Miller is an associate professor in the university's Department of Political Science.