ArticlePDF Available

Self‐Reported Understanding of Ranked‐Choice Voting*

Authors:

Abstract and Figures

Objectives Ranked‐choice voting (RCV) is relatively complex compared to plurality voting. We test if some voters find it more challenging. Methods We conducted surveys in RCV cities and plurality cities to assess how voters reported understanding voting instructions, and how they reported understanding election systems. Results Fewer voters reported instructions were easy to understand in RCV cities. Within RCV cities, we found little evidence of race/ethnic differences in reported understanding, but older voters reported less understanding of instructions in RCV cities and less understanding of RCV elections. Across all cities, Asians and women reported less understanding of elections generally, and education correlated with greater reported understanding. Conclusions Our evidence is not consistent with concerns about a racial/ethnic bias specific to RCV, but suggests a need for additional voter education.
Content may be subject to copyright.
Self-Reported Understanding of
Ranked-Choice Voting
Todd Donovan, Western Washington University
Caroline Tolbert, University of Iowa
Kellen Gracey, DeSales University
Objectives. Ranked-choice voting (RCV) is relatively complex compared to plurality voting. We test
if some voters find it more challenging. Methods. We conducted surveys in RCV cities and plurality
cities to assess how voters reported understanding voting instructions, and how they reported
understanding election systems. Results. Fewer voters reported instructions were easy to understand
in RCV cities. Within RCV cities, we found little evidence of race/ethnic differences in reported
understanding, but older voters reported less understanding of instructions in RCV cities and less
understanding of RCV elections. Across all cities, Asians and women reported less understanding of
elections generally, and education correlated with greater reported understanding. Conclusions. Our
evidence is not consistent with concerns about a racial/ethnic bias specific to RCV, but suggests a
need for additional voter education.
Complexity of Preferential Voting
Some cities in the United States have replaced simple plurality elections with ranked-
choice voting (RCV). By providing opportunities for voters to cast multiple preferences,
the act of voting may be more complex under RCV than compared to demands associated
with the single vote required in a plurality system. If this complexity challenges some voters
more than others, adoption of RCV may add to existing biases in political participation
and representation.
One concern with having voters rank preferences is that confusion may cause overvoting,
undervoting, and uninformed ballot order effects (Orr, 2002). Some voters may cope with
the demands of RCV by not ranking. Burnett and Kogan (2015) reported that voters’ failure
to rank sufficient candidates was associated with high rates of ballots being exhausted before
the final count. Neely and McDaniel (2015) report that overvotes were more common
under San Francisco’s RCV elections in minority precincts, and attributed this to demands
placed on minority voters. McDaniel (2016) also attributed demographic differences in
aggregate turnout patterns to the complexity of RCV, and Neely and Cook (2008) also
found racial differences in aggregate overvote patterns. In contrast, Kimball and Anthony
(2016) reported RCV had little to do with ballot completion.
Direct correspondence to Todd Donovan, Department of Political Science, Western Washington Univer-
sity, Bellingham, WA 98225 Todd.Donovan@wwu.edu. Data and code used for this article are available
upon request. Appendices are online at http://faculty.wwu.edu/donovat/Appendix_rcv_ssq.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License,
which permits use, distribution and reproduction in any medium, provided the original work is properly cited
and is not used for commercial purposes.
SOCIAL SCIENCE QUARTERLY
C2019 The Authors Social Science Quarterly published by Wiley Periodicals, Inc. on behalf of Southwestern
Social Science Association
DOI: 10.1111/ssqu.12651
2 Social Science Quarterly
Where survey data have been available, studies document that voters appeared rather
capable of navigating newly adopted preferential systems in New Mexico (Cole, Taebel,
and Engstrom, 1990), New York (Kimball and Kropf, 2016), South Dakota (Engstrom and
Barrilleaux, 1991), and Texas (Brischetto and Engstrom, 1997). Upwards of 90 percent of
respondents reported they understood the system they voted in, with very similar levels
of understanding reported by white voters and minority voters (Brischetto and Engstrom,
1997:978). High levels of reported understanding were found among voters participating
in San Francisco’s 2004 RCV election, yet African Americans and Latinos were less likely
to say they understood RCV (Neely, Blash, and Cook, 2005). However, the racial disparity
in understanding RCV found in 2004 was not evident in San Francisco in 2005 (Neely,
Cook, and Blash, 2006).
Demographic Differences in Understanding
One concern about replacing plurality systems with RCV is that RCV may accentuate
racial/ethnic biases in voting (e.g., McDaniel, 2016). This would be problematic if vot-
ers who find preferential voting overly complex were not similarly disadvantaged by the
plurality system that preferential voting was replacing. Related concerns are that dispar-
ities in understanding will affect turnout, and completed ballots, and that this may give
disproportionate weight to votes cast by whites.
One difficulty here in assessing how voters respond to RCV is that socioeconomic
biases in political participation are widespread and well documented (e.g., Schlozman,
Verba, and Brady, 2012). We might find biases in understanding associated with any
system. Another challenge is that assumptions about biased understanding of RCV are
generally inferred from precinct data, and/or from studies that rely on data from a single
jurisdiction.
Research Design, Data, Methods
We examine reported understanding of voting instructions and reported understanding
of electoral systems in jurisdictions that used RCV for local elections, and compare this to
reported understanding of voting instructions and election systems in similar jurisdictions
using plurality voting. We isolate exposure to electoral system with an immediate post-
election survey of voters in four California jurisdictions that used RCV for contested local
elections in November 2014, while simultaneously surveying voters in similar California
jurisdictions using plurality local elections. Three RCV jurisdictions (Oakland, Berkeley,
and San Leandro) had previously used RCV twice, and San Francisco used it since 2004.
Case selection was constrained by the small number of RCV places that had contested races.
Demographically similar plurality cases were identified from a limited number of California
cities that also had contested races at the same time. Jurisdictions surveyed are listed in
Online Appendix Table 1. See John, Smith, and Zack (2018) for additional applications of
this design. This matching method allows us to minimize (but not eliminate) nonelectoral
system differences across the cases being compared.
A random digit dial survey targeting registered voters was contacted via land lines and
cellphones, with interviews conducted in English and Spanish. Surveys of voters were
conducted by the Eagleton Poll at Rutgers University. A series of questions screened self-
reported voters from nonvoters (see the Online Appendix). The sample includes 2,189
Self-Reported Understanding of Ranked-Choice Voting 3
voters, with a subsample of 1,220 from places using RCV, and a subsample of 969 from
plurality voting cities. Respondents confirmed their city of residence. The survey, and details
on survey administration and response rates, is in the Online Appendix. Each city was
sampled independently with subjects drawn from each city (not a pool of cities). The two
subsamples are similar in terms of median age, percent with high school education, percent
married, and percent white/nonwhite. Given these local races were held in conjunction
with a statewide general election, voters also cast preferences in plurality-winner contests
for statewide offices and for U.S. Congress.
All voters were first asked how well they understood voting instructions (see the Online
Appendix for survey). We asked: “When you voted in the election Tuesday, how easy was
it to understand the voting instructions?” Second, people were asked about understanding
their local electoral system, with the name of the system stated. Respondents in RCV cities
were asked: “How well do you think you understand ranked-choice voting?” Respondents
plurality cities were asked: “How well do you think you understand plurality voting?” After
this, all respondents were informed that California recently adopted a top-two primary
system, and were asked: “How well do you understand the top-two primary system?”
Following this, all respondents were asked: “How well do you understand winner-take-all
voting rules?”
We do not expect that respondents had nuanced understandings of election systems when
they were asked about them by name. Rather, we examine if there were election-system-
specific race/ethnic differences in reported levels of understanding voting instructions,
and in reported understanding of electoral systems. We assess if race/ethnic differences in
reported understanding of instructions and elections were specific to RCV cities, or if such
differences were found generally.
Hypotheses
This discussion suggests a primary hypothesis: people may find preferential voting in-
structions more difficult to understand than plurality voting instructions. In the cities
using RCV examined here, the RCV ballots, and instructions on the ballot, required some
attention from voters. RCV ballots had all candidates for an office listed in three separate,
parallel columns. In addition to standard information about using a black or blue pen (not
pencil), and on how to properly mark a choice, instructions on RCV ballots also informed
voters they could rank up to three choices and to “vote across in each race.” The ballot
instructed them to mark their first choice in Column 1, mark a second choice “different
from your first choice” in Column 2, and to mark a third choice that “must be different
from your first and second choices” in Column 3. Ballots in plurality cities were standard
office-block style with instructions about how to mark a choice, and the statement “Vote
for One” for each office.
There may also be reasons to expect that understanding voting instructions, if not the act
of voting itself, is more demanding for certain people, even with simple plurality voting.
Sinclair and Alvarez (2004:19) hypothesize that errors in voting punch-card ballots were
related to how well people were connected to the political system, and operationalized
this with aggregated measures of race/ethnicity, education, and other demographics. This
suggests a second hypothesis: variation in reported understanding of elections in general is
explained by race/ethnicity and education. Specifically, members of minority groups, and
those with less education, are expected to be less likely to report a firm understanding of
voting instructions.
4 Social Science Quarterly
TABLE 1
Voters’ Reported Ease of Understanding Voting Instructions (Percent)
By Electoral System
All Plurality City RCV City
Very difficult 2.6 2.2 3.0
Somewhat difficult 7.3 4.7 9.4
Somewhat easy 23.7 20.2 26.6
Very easy 66.3 73.0 61.0
2=38.8, p=0.01
Number of cases 2,177 966 1,211
By Electoral System, and by Race/Ethnicity
Plurality City RCV City
Nonwhite White Nonwhite White
Very difficult 2.7 1.7 2.9 3.0
Somewhat difficult 6.6 3.1 11.1 8.3
Somewhat easy 25.6 25.6 25.9 27.0
Very easy 65.2 79.6 60.1 61.6
2=25.9, p=0.00 2=2.5, p=0.47
Number of cases 966 1,211
Neely and McDaniel (2015), furthermore, infer from aggregate patterns of RCV voting
errors (overvotes) in San Francisco that there may be the potential for political inequality
given that RCV may demand more of Latino/a, African-American, and elderly voters.
Neeley and Cook (2008) also found RCV overvotes disproportionately in San Francisco
precincts with more African-American voters. However, Kimball and Kropf (2005) and
others (Bullock and Hood, 2002) have documented that overvoting is generally more
common among African Americans in standard plurality election systems. This is attributed
to the qualities of elections and voting equipment—not to qualities of African-American
voters. Nonetheless, this suggests a third hypothesis: racial/ethnic and age differences in
how people report understanding voting instructions would be particularly pronounced
under RCV, with RCV instructions being understood less well by minority voters and the
elderly.
Next, we examine if voters’ reported understanding of voting instructions, and their
reported understanding of electoral systems, were structured by race/ethnicity, age, and
education. Further, we examine if residence in an RCV city was associated with heightened
race/ethnic or age differences in how voters reported understanding voting instructions.
We present bivariate results initially, and then report predicted probabilities of reporting
high understanding estimated from multivariate ordered logistic regression models.
Results: Understanding of Voting Instructions
Our measure of understanding voting instructions was a question that asked: “When you
voted in the election last Tuesday, how easy it was to understand the voting instructions?”
The top panel of Table 1 illustrates that more people in plurality cities (73 percent) said it
was “very easy” to understand voting instructions than in RCV cities (61 percent). However,
Self-Reported Understanding of Ranked-Choice Voting 5
TABLE 2
Predicted Probability of Saying Voting Instructions Were “Very Easy” to Understand
All Respondents Plurality Cities RCV Cities
Plurality city 0.76 (0.02)
RCV city 0.63 (0.02)
White 0.76 (0.02) 0.79 (0.02) 0.61 (0.02)
African American 0.73 (0.03) 0.67 (0.05)0.60 (0.04)
Latinx 0.69 (0.03)0.69 (0.04)0.60 (0.05)
Asian 0.67 (0.04)0.60 (0.06)0.60 (0.06)
Age (75) 0.75 (0.02) 0.81 (0.02) 0.57 (0.02)
Education (highest) 0.78 (0.02) 0.83 (0.02)0.61 (0.03)
at p<0.05 in multivariate models. Reference category for race/ethnicity is whites. Age and education
set to mean values, gender set as female, interest in the local election set as interested. Post-estimation
simulations from ordered logit models reported in Online Appendix Table 3.
regardless of whether a respondent was voting in a city that used RCV or simple plurality,
large majorities in both groups claimed their instructions were “somewhat easy” or “very
easy” to understand, with slightly higher proportions saying this in plurality cities (93
percent) than in RCV cities (88 percent). The bivariate difference is statistically significant
(2=38.8, p=0.01).
This difference remains significant in our multivariate estimates (see Table 2, and Online
Appendix Table 3) that control for race/ethnicity, age, gender, education, and interest in
the election. This does not establish that respondents in RCV places had more difficulty
when they actually voted, but it is consistent with the idea that RCV may be more
demanding. The bottom panel of Table 1 illustrates bivariate race/ethnic differences in
reported understanding in plurality cities; this is not evident in RCV cities.
Table 2 reports the post-estimation predicted probability a respondent said voting in-
structions were very easy to understand (with SEs in parentheses). Models used to generate
predictions in Table 2 (Online Appendix Table 3) were estimated using ordered logistic
regression, with SEs clustered by city. A binary variable distinguishes between people who
voted in an RCV city or plurality city. We also estimate models for respondents voting
in plurality and RCV jurisdictions, respectively. Models include dichotomous measures
representing, respectively, African-American, Latinx, and Asian respondents, and those
who identified as “other” or mixed race. Non-Hispanic whites are the reference category.
Age, gender, education (measured in seven categories), and a dichotomous measure of the
respondent’s report of how interesting the race are also included. The latter item serves as
a proxy for voter interest in elections.
The first column in Table 2 confirms that the difference in reporting that instructions
were “very easy” to understand between plurality respondents (prob. =0.76, SE =0.02)
and RCV respondents (0.63, SE =0.02) remains in multivariate analysis. As for differences
across racial and ethnic groups, Latinx (prob. =0.69, SE =03) and Asian respondents
(0.67, SE =0.04) were less likely than whites (0.76, SE =0.02) to report instructions
were very easy to understand across all respondents. However, results from respondents in
RCV (Column 3, Table 2) are inconsistent with the idea that RCV voting instructions
have demands that disproportionately affect people of color. There was no difference in the
predicted probability of white (prob. =0.61, SE =0.02), black (0.60, SE =0.04), Latinx
(0.60, SE =0.05), or Asian voters (0.60, SE =0.06) reporting that voting instructions were
very easy to understand. Female respondents (included in our reference group in Table 2),
6 Social Science Quarterly
TABLE 3
Voters’ Reported Understanding Named Election Systems (Percent)
Winner-Take-All Top-Two PluralityRCV
Not at all well 9.8 11.2 14.1 12.9
Somewhat well 30.6 40.1 22.9 32.9
Very well 29.9 28.6 44.8 32.1
Extremely well 29.6 20.1 18.1 22.0
Number of cases 2,084 2,129 871 1,213
Only asked of voters in jurisdictions using the electoral system.
NOTE: Respondents were asked “how well do you think you understand” each system.
and older respondents were significantly less likely to say voting instructions were easy to
understand in RCV places. Those who reported their local election was interesting were
more likely to see RCV voting instructions as easy to understand (see Online Appendix
Tab le 3 ).
We also predict a significantly higher probability of white respondents (0.79, SE 0.02) in
plurality cities reporting their voting instructions were very easy to understand, compared
to African-American (0.67, SE =0.05), Latinx (0.69, SE =0.04), and Asian respondents
(0.60, SE =0.06). Race/ethnic differences in reported understanding of instructions are
evident among voters in these plurality cities, but not in cities using RCV.
Results: Reported Understanding of Different Electoral Systems
Our survey also included items that gauged voters’ reported understanding of various
election systems. We use these to assess if there was a systematic difference between
voters in RCV places and voters in plurality places in terms of how they responded to
questions about understanding elections. Voters in plurality cities were asked how well
they understood “plurality voting,” while voters in RCV cities were asked how well they
understood “ranked-choice voting.” We also asked all respondents how well they understood
“winner-take-all voting rules,” and the state’s “top-two primary.” The point was not to assess
how well people navigated voting under these systems, but, first, to assess if self-reported
understanding was generally lower among people we sampled in RCV places. If not, we can
have more confidence that differences reported in the top panel of Table 1 were associated
with RCV, rather than with the people we sampled in RCV cities being less likely to say they
understood things about elections generally. Second, examining the patterns of responses
to these items allows us to assess if racial/ethnic, education level, and age differences in how
people report understanding elections is particular to RCV elections, or common to how
they report understanding elections generally.
Table 3 reports responses to these questions. Few voters said “not at all well” when
describing how much they understood winner-take-all elections (9.8 percent), the top-two
primary (11.2 percent), plurality voting (14.1 percent), and ranked-choice voting (12.9
percent). As for the issue of whether or not respondents from RCV jurisdictions may have
been less prone to self-report they understood voting and elections generally, chi-square
tests comparing responses in Table 3 across the subsamples demonstrate that there was no
significant difference in how people in RCV versus plurality places reported understanding
the top-two primary (chi-square p=0.19) or winner-take-all voting (p=0.64). This
Self-Reported Understanding of Ranked-Choice Voting 7
TABLE 4
Voters’ Reported Understanding Named Election Systems (Percent)
Winner-Take-All Top-Two
White Nonwhite White Nonwhite
Not at all well 8.9 11.0 11.1 11.2
Somewhat well 25.6 37.2 35.3 46.7
Very well 32.6 27.1 30.1 26.6
Extremely well 33.2 24.7 23.5 15.5
N=2,084, 2=41.6 (p<0.00) N=2,129, 2=35.8 (p<0.00)
Plurality RCV
White Nonwhite White Nonwhite
Not at all well 14.6 13.6 11.3 15.3
Somewhat well 19.8 26.5 32.2 34.1
Very well 44.6 45.0 33.3 30.3
Extremely well 21.1 14.8 23.1 20.3
N=871, 2=9.1 (p<0.03) N=1213, 2=5.7 (p=0.12)
By race/ethnicity.
suggests that lower levels of reported understanding of voting instructions among people
in RCV places (Table 1) may be something associated with the demands of RCV, rather
than with people in RCV places having lower propensities of reporting they understand
elections. That is, since people from RCV places were not less likely than people from
plurality places to claim to understand various other elections, we can be more confident
that modest differences between these places in reported levels of understanding voting
instructions could be associated with the complexity of RCV. These results are consistent
with our hypothesis about preferential voting being more demanding.
But what of our other hypothesis? Does race/ethnicity or age structure differences in
who reports understanding these elections, and are minorities more likely to say they do
not understand RCV? In Table 4 we cross-tabulated responses for questions about electoral
systems to compare responses from non-Hispanic whites to racial/ethnic minorities. When
asked about three of the four electoral systems, nonwhite respondents were significantly less
likely to report understanding it, and the results are unchanged (not reported here) regard-
less of how we categorize race/ethnicity. Again, we find no apparent effects of race/ethnicity
that are specific to self-reported understanding of RCV.
We used ordered logit models (Online Appendix Table 4) to estimate responses to the
four questions about electoral systems. Those models generated the predicted probability
a respondent said he or she understood these four election systems “very well.” These
are reported in Table 5 (with SEs in parentheses). Regardless of the election asked about,
education corresponded with greater reported understanding, and those who found their
local election interesting were generally more likely to say they understood elections (see
the Online Appendix). Conversely, Asian respondents (on all four items), women (on three
of the four items, see the Online Appendix), and African Americans (on two) were less
likely to say they understood a named election system. These patterns appear to reflect
something general to self-reported understanding of elections, not something specific about
RCV. There is, however, one demographic result that seems unique to RCV places. Age
had no association with reported understanding of elections, apart from RCV. Older voters
8 Social Science Quarterly
TABLE 5
Predicted Probability of Reporting Understanding Election System “Extremely Well”
Winner-Take-All Top-Two Primary Plurality RCV
White 0.28 (0.02) 0.19 (0.01) 0.19 (0.02) 0.22 (0.02)
African American 0.24 (0.02) 0.14 (0.01)0.14 (0.02)0.19 (0.02)
Latinx 0.22 (0.02)0.19 (0.02) 0.21 (0.04) 0.27 (0.04)
Asian 0.15 (0.02)0.12 (0.02)0.13 (0.03)0.16 (0.03)
Age (75) 0.28 (0.02) 0.20 (0.01) 0.20 (0.03) 0.19 (0.02)
Education (highest) 0.34 (0.02)0.23 (0.01)0.25 (0.02)0.25 (0.02)
p<0.05 in multivariate models. Reference category for race/ethnicity is whites. Age and education set
to mean values, gender set as female, interest in the local election set as interested. Post-estimation
simulations from ordered logit models reported in Online Appendix Table 4.
in RCV cities were significantly less likely to report understanding voting instructions very
well, and RCV was the only election type that older voters were significantly less likely to
report understanding very well (Table 5).
Discussion
This article documents reasons to expect that preferential voting systems may be more
confusing to voters than simple plurality, particularly for race/ethnic minorities. We build
on previous literature by using survey (rather aggregate) data, survey data from several
jurisdictions that adopted RCV, and comparable survey data from jurisdictions that did not
use RCV. Our results suggest widespread self-reported understanding of voting instructions
in these cities, with high proportions of voters reporting voting instructions were very easy to
understand. We do find lower levels reporting this in cities using RCV elections. However,
when people in RCV cities were asked how well they understood “ranked-choice voting”
only 13 percent said “not at all”—a level similar to what voters reported when asked about
other election systems.
Previous studies found patterns that raised concerns about a potential racial/ethnic bias
in who understands RCV. Our evidence is not consistent with this. We found no differences
within RCV cities between whites and people of color in reports of understanding voting
instructions. Furthermore, we found no differences in RCV cities in how whites, African
Americans, and Latinx respondents reported understanding RCV. We did find women
and Asians in RCV cities less likely to report they understood RCV elections, but we also
found women and Asians less likely to report they understood a range of different elections
(Online Appendix Table 2 contains discussion of a potential language bias problem in
interviews of Asians). It is difficult, then, to conclude that results for women and Asians
reflect something particular to understanding RCV voting instructions or RCV elections.
However, we do find one sign of potential bias associated with RCV. Older people in RCV
cities were less likely to report understanding voting instructions, and less likely to report
understanding RCV elections.
The adoption of new, relatively complex voting systems presents the risk of confusion
that may lead to increased voting errors. Our results speak mainly to how voters recalled
ballot instructions a day or so after voting, and should not be read as dismissing concerns
that information demands associated with RCV may cause some people to fail to cast
valid ballots, or express a full range of preferences. Scholars of elections must continue to
Self-Reported Understanding of Ranked-Choice Voting 9
search for evidence of bias—particularly racial/ethnic bias—associated with such systems.
Our models suggest reported understanding of voting instructions and election systems is,
not surprisingly, associated with education. This suggests that voter education campaigns
might play a role in increasing understanding of RCV and reducing potential voting errors.
It must be recognized that our method of measuring reported understanding is limited by
the inability to observe actual voted ballots and to link those to actual voters. It may very well
be that some of our survey respondents cast invalid RCV ballots, but then reported having
a good understanding of voting instructions and of RCV. If the propensity to commit a
voting error (perhaps unknowingly), and subsequently offer a survey response that reported
understanding that act of voting is concentrated among racial/ethnic minorities, we could
underestimate the scale of possible racial/ethnic bias in understanding RCV. We await
future research—potentially experimental—that might sort those forces.
REFERENCES
Brischetto, Robert, and Richard Engstrom. 1997. “Cumulative Voting and Latino Representation: Exit Surveys
in Fifteen Texas Communities.” Social Science Quarterly 78(4):973–91.
Bullock, Charles. S. III, and M. V. Hood III. 2002. “One Person—No Vote; One Vote; Two Votes: Voting
Methods, Ballot Types, and Undervote Frequency in the 2000 Presidential Election.” Social Science Quarterly
83(4):981–93.
Burnett, Craig, and Vladimir Kogan. 2015. “Ballot (and Voter) ‘Exhaustion’ Under Instant Runoff Voting:
An Examination of Four Ranked-Choice Elections.” Electoral Studies 37:41–49.
Cole, Richard L., Delbert. A. Taebel, and Richard L. Engstrom. 1990. “Cumulative Voting in a Municipal
Election: A Note on Voter Reactions and Electoral Consequences.” Western Political Quarterly 43 (1):191–99.
Engstrom, Richard. L., and Charles J. Barrilleaux. 1991. “Native Americans and Cumulative Voting: The
Sisseton-Wahpeton Sioux.” Social Science Quarterly 72 (2):388–93.
John, Sarah, H. Smith, and E. Zack. 2018. “The Alternative Vote: Do Chances in Single-Member Voting
Systems Affect Descriptive Representation of Women and Minorities?” Electoral Studies 54:90–102.
Kimball, David, and Joseph Anthony. 2016. “Voter Participation with Ranked Choice Voting in the United
States.” Paper presented at the American Political Science Association Meeting.
Kimball, David. C., and Martha Kropf. 2005. “Ballot Design and Unrecorded Votes on Paper-Based Ballots.”
Public Opinion Quarterly 69(4):508–29.
———. 2016. “Voter Competence with Cumulative Voting.” Social Science Quarterly 97(3):619–35.
McDaniel, Jason. 2016. “Writing the Rules to Rank the Candidates: Examining the Impact of Instant-Runoff
Voting on Racial Group Turnout in San Francisco Mayoral Elections.” Journal of Urban Affairs 38(3):387–408.
Neely, Francis, Lisel Blash, and Corey Cook. 2005. An Assessment of Ranked-Choice Voting in the San Francisco
2004 Election. Daly City, CA: Public Research Institute, San Francisco State University.
Neely, Francis, and Corey Cook. 2008. “Whose Votes Count? Undervotes, Overvotes, and Ranking in San
Francisco’s Instant-Runoff Elections.” American Politics Research 36(4):530–54.
Neely, Francis, Corey Cook, and Lisel Blash. 2006. An Assessment of Ranked-Choice Voting in the San Francisco
2005 Election. Daly City, CA: Public Research Institute, San Francisco State University.
Neely, Francis, and Jason McDaniel. 2015. “Overvoting and the Equality of Voice Under Instant-Runoff
Voting in San Francisco.” California Journal of Politics & Policy 7(4):1–27
Orr, Graeme. 2002. “Ballot Order: Donkey Voting in Australia.” Election Law Journal 1(4):573–78.
Schlozman, Kay. L., Sidney Verba, and Henry E. Brady. 2012. The Unheavenly Chorus: Unequal Political Voice
and the Broken Promise of American Democracy. Princeton, NJ: Princeton University Press.
Sinclair, D. B., and R. Michael Alvarez. 2004. “Who Overvotes, Who Undervotes, Using Punchcards? Evidence
from Los Angeles County.” Political Research Quarterly 57(1):15–25.
... This includes a higher rate of manipulation for the Plurality voting rule [27,28], bias towards the candidate with the highest first place votes [29] and a higher familiarity of the Plurality voting rule [21]. Surveys to analyze the understanding of ranked choice voting have reported that the respondents do not find it as difficult to follow the rules of ranked choice voting as it was perceived by the researchers [55,56]. Naturally, all of these insights can be highly valuable in constructing effective explanations but they are very particular to the domain of ranked voting while our feature-based explanations are general and can be used for any domain. ...
Preprint
Full-text available
Designing and implementing explainable systems is seen as the next step towards increasing user trust in, acceptance of and reliance on Artificial Intelligence (AI) systems. While explaining choices made by black-box algorithms such as machine learning and deep learning has occupied most of the limelight, systems that attempt to explain decisions (even simple ones) in the context of social choice are steadily catching up. In this paper, we provide a comprehensive survey of explainability in mechanism design, a domain characterized by economically motivated agents and often having no single choice that maximizes all individual utility functions. We discuss the main properties and goals of explainability in mechanism design, distinguishing them from those of Explainable AI in general. This discussion is followed by a thorough review of challenges one may face when when working on Explainable Mechanism Design and propose a few solution concepts to those.
... In this method, each voter (QoS) is allowed to vote for one SDaaS provider; i.e. the provider with the highest perceived value is selected. The provider that polls more than any other candidate is the winner [27]. • Instant Runoff: also known as ranked choice voting (RCV). ...
Preprint
We propose a novel framework for the recommendation of swarm-based drone delivery services based on the consumers preferences. We propose a density-based pruning approach that uses the concept of partnerships with charging station providers to reduce the search space of swarm-based drone service delivery providers. A weighted service composition algorithm is proposed that considers the providers capabilities and consumers' preferences in selecting the best next service. We propose a voting-based recommendation algorithm to select the best providers. We conduct a set of experiments to evaluate the efficiency of the framework in terms of consumer satisfaction, run-time, and search space reduction cost.
... This includes a higher rate of manipulation for the Plurality voting rule [27,28], bias towards the candidate with the highest first place votes [29] and a higher familiarity of the Plurality voting rule [21]. Surveys to analyze the understanding of ranked choice voting have reported that the respondents do not find it as difficult to follow the rules of ranked choice voting as it was perceived by the researchers [55,56]. Naturally, all of these insights can be highly valuable in constructing effective explanations but they are very particular to the domain of ranked voting while our feature-based explanations are general and can be used for any domain. ...
Preprint
In many social-choice mechanisms the resulting choice is not the most preferred one for some of the participants, thus the need for methods to justify the choice made in a way that improves the acceptance and satisfaction of said participants. One natural method for providing such explanations is to ask people to provide them, e.g., through crowdsourcing, and choosing the most convincing arguments among those received. In this paper we propose the use of an alternative approach, one that automatically generates explanations based on desirable mechanism features found in theoretical mechanism design literature. We test the effectiveness of both of the methods through a series of extensive experiments conducted with over 600 participants in ranked voting, a classic social choice mechanism. The analysis of the results reveals that explanations indeed affect both average satisfaction from and acceptance of the outcome in such settings. In particular, explanations are shown to have a positive effect on satisfaction and acceptance when the outcome (the winning candidate in our case) is the least desirable choice for the participant. A comparative analysis reveals that the automatically generated explanations result in similar levels of satisfaction from and acceptance of an outcome as with the more costly alternative of crowdsourced explanations, hence eliminating the need to keep humans in the loop. Furthermore, the automatically generated explanations significantly reduce participants' belief that a different winner should have been elected compared to crowdsourced explanations.
... 4. Donovan et al. (2019) find evidence of greater understanding of rank vote choice among higher educated voters. 5. Voters may also like or dislike a voting method for its potential consequences, that is, because it contributes to the entry of more (fewer) candidates, the greater (smaller) use of negativity (attacks) in the campaign, or because it better reflects the outcome they would like. ...
Article
Full-text available
We ascertain whether citizens want to have smart ballots, that is, whether they appreciate having the possibility to express some support for more than one option (expression across options) and to indicate different levels of support for these options (expression within options). We conducted two independent yet complementary survey experiments at the time of the Super Tuesday Democratic primaries to examine which voting method citizens prefer, one with the real candidates in the states holding Democratic primaries and one with fictitious candidates in the whole country. In both surveys, respondents were asked to vote using four different voting rules: single, approval, rank, and point (score). After they cast their vote, respondents were asked how satisfied they were using each voting method. The findings are consistent in both studies: the single vote is the most preferred voting method. We show that this is a reflection of a status quo bias, as citizens’ views are strongly correlated with age.
Article
New York City is the largest U.S. jurisdiction to use ranked choice voting (RCV). We examine New York and other U.S. cities using RCV to assess if there were different levels of understanding and utilization of RCV across demographic groups. We placed items on a survey conducted during the 2021 New York City RCV election that had been included in two previous surveys of different U.S. cities using RCV. We find higher levels of reported understanding and rates of ranking multiple candidates in NYC than in other jurisdictions. We find no systematic differences by race/ethnicity in terms of reported understanding of RCV in NYC or the other samples. We also find no systematic association between age and reported understanding of RCV. Respondents with more education were more likely to report understanding RCV in each sample. People of color were less likely to report ranking multiple mayoral candidates in NYC and California, and respondents with more education were more likely to report ranking in two samples. Apart from these important differences in utilization, our search for race/ethnic differences largely produced null results, suggesting RCV may not produce bias in who engages with it.
The ideal of publicity plays an important role in contemporary legal and political philosophy. Yet, to date, it has not been brought to bear on the question of voting method choice. This paper aims to fix this. I argue that voting method publicity is a well-motivated requirement which reveals tradeoffs inherent to democracy between procedural and epistemic equality. I further explore the implications of voting method publicity to the normative status of plurality voting and its possible alternatives.
Article
Proponents of ranked-choice voting (RCV) highlight a number of arguments for why such an approach to elections should be adopted. One major argument is that ranked-choice voting will encourage voters to support more third-party or independent candidates and break the electoral stranglehold of the two main parties in America. Considering approximately two-thirds of Americans want a third major party this argument may prove appealing to American voters, but there is currently no empirical evidence to support such claims. In this project, we explore a theory of why ranked-choice voting may increase voter support for third-party or independent candidates and test the argument that RCV will improve the fortunes of third-party candidates using a survey experiment. We find significant support for the claim that ranked-choice voting increases support for third-party candidates.
Article
Full-text available
When election reforms such as Ranked Choice Voting or the Alternative Vote are proposed to replace plurality voting, they offer lengthier instructions, more opportunities for political expression, and more opportunities for mistakes on the ballot. Observational studies of voting error rely on ecological inference from geographically aggregated data. Here we use an experimental approach instead, to examine the effect of two different ballot conditions at the individual level of analysis: the input rules that the voter must use and the number of ballot options presented for the voter’s choice. This experiment randomly assigned three different input rules (single-mark, ranking, and grading) and two different candidate lists (with six and eight candidates) to over 6,000 online respondents in the USA, during the American presidential primary elections in 2020, simulating a single-winner presidential election. With more expressive input rules (ranking and grading), the distinction between minor mistakes and totally invalid votes—a distinction inapplicable to single‐mark ballots (1MB) voting—assumes new importance. Regression analysis indicates that more complicated input rules and more candidates on the ballot did not raise the probability that a voter would cast a void (uncountable) vote, despite raising the probability of at least one violation of voting instructions.
Article
Full-text available
Theory suggests that ranked choice voting (RCV) may create a more civil campaign environment. As voters must rank candidates, the candidates have an incentive to work with each other more collaboratively. This study uses text analysis software (LIWC) to examine candidate tweets and newspaper articles in RCV versus specifically-chosen plurality cities for evidence of positivity or negativity. In quantitatively comparing the tweets, the results are mixed among the cities. Qualitatively, candidates seem to be more likely to engage each other in RCV cities than in plurality cities. Using LIWC to analyze newspaper articles for campaign tone, one can see that RCV city articles have significantly more positive than negative words. This is the first published study to use direct campaign communication data to study RCV elections and campaign civility. This research validates survey research indicating that citizens perceive RCV campaigns are more civil.
Article
There exist two, conflicting, conventional wisdoms in the United States about which electoral systems best provide for descriptive representation. On the one hand, single-member districts are understood to be the best way to ensure representation for ethnic and racial minority groups, at least when populations are geographically concentrated. On the other hand, it is well-established that a greater proportion of women are elected in multi-member districts in the United States, especially in state legislatures. At the intersection, minority women are not well served by single-member districts, at least using plurality or majority runoff election, but it is far from clear that multi-member districts are much better. This paper takes a fine-grained approach to electoral systems reform and descriptive representation, exploring the impacts of the reform of the electoral formula in single-member systems in the context of the adoption of the alternative vote (AV) in several California Bay Area cities. Using similar cities in the area that did not adopt AV as controls and a difference-in-differences research design, we show that reform of single-member electoral formulae can have a significant positive effect on descriptive representation.
Article
Objective: This article evaluates the voting experience in the first election using cumulative voting for the Board of Trustees in Port Chester, New York. A growing number of local jurisdictions in the United States are using cumulative voting for multimember elections. While the Port Chester election included some other new features in addition to cumulative voting, the village implemented an extensive voter education program to prepare voters and candidates for the election. Methods: We conducted an exit poll of 1,946 Port Chester voters in June 2010, more than half of the voters in the local election. We used a variety of survey questions to measure voting experience and voting behavior. We also examined election returns for Port Chester, including the 2010 and 2013 elections using cumulative voting. Results: We find that the voter education program helped inform residents about casting a ballot with cumulative voting. Port Chester voters, and Hispanic voters in particular, reported a positive experience in the 2010 election. A large majority of voters also indicated that they understood cumulative voting and cast all of the votes allotted to them. Finally, we find evidence of strategic use of cumulative voting in order to help elect a candidate of one's choice. Conclusions: Our results indicate that voters are capable of effectively participating in elections with cumulative voting. Communities that are weighing the adoption of cumulative voting for local elections should also be prepared to implement a parallel voter education effort.
Article
Politically active individuals and organizations make huge investments of time, energy, and money to influence everything from election outcomes to congressional subcommittee hearings to local school politics, while other groups and individual citizens seem woefully underrepresented in our political system.The Unheavenly Chorusis the most comprehensive and systematic examination of political voice in America ever undertaken--and its findings are sobering. The Unheavenly Chorusis the first book to look at the political participation of individual citizens alongside the political advocacy of thousands of organized interests--membership associations such as unions, professional associations, trade associations, and citizens groups, as well as organizations like corporations, hospitals, and universities. Drawing on numerous in-depth surveys of members of the public as well as the largest database of interest organizations ever created--representing more than thirty-five thousand organizations over a twenty-five-year period--this book conclusively demonstrates that American democracy is marred by deeply ingrained and persistent class-based political inequality. The well educated and affluent are active in many ways to make their voices heard, while the less advantaged are not. This book reveals how the political voices of organized interests are even less representative than those of individuals, how political advantage is handed down across generations, how recruitment to political activity perpetuates and exaggerates existing biases, how political voice on the Internet replicates these inequalities--and more. In a true democracy, the preferences and needs of all citizens deserve equal consideration. Yet equal consideration is only possible with equal citizen voice.The Unheavenly Chorusreveals how far we really are from the democratic ideal and how hard it would be to attain it.
Article
Instant-Runoff Voting (IRV/RCV) is a relatively new reform to the rules of voting and elections that has been adopted by several cities. The impact of IRV/RCV balloting on racial group voter turnout in urban elections has not been subject to rigorous empirical analysis. In order to evaluate the relationship between IRV/RCV and voter turnout, I analyze precinct-level racial group voter turnout rates in five San Francisco mayoral elections from 1995 to 2011. I find significant declines in turnout under IRV/RCV rules, especially among Black voters. Variation in racial group turnout is best explained by factors related to candidate race and racial group competition. In addition, I find that IRV/RCV exacerbates turnout disparities between older and younger groups in the population and between those with high education compared to those with low education.
Article
We examine the first 3 years of San Francisco's instant-runoff voting (IRV) elections to determine whether some voters adapt more easily than others to the more complex ballot and decision task. We draw on studies of uncounted votes to develop hypotheses about tendencies to undervote, overvote, and rank candidates in IRV elections. Individual ballot records and precinct-level census data are used to estimate the relative influences of demographic and election-specific factors. A natural experiment tests whether or not prior experience with IRV makes a difference. The change to IRV appears to have increased the rate of overvotes and decreased tendencies to undervote. Both behaviors are explained by demographics and aspects of the electoral environment. Meanwhile, tendencies to rank candidates were shaped less by demographics and more by contextual factors and prior exposure to IRV. The findings extend the literature on uncounted votes, inform issues of equality in elections, and provide practical insights on this type of electoral reform.
Article
This article examines the use of an election system — cumulative voting — used for the first time in the twentieth century in a municipal election in the United States in July 1987 in Alamogordo, New Mexico. Using survey data collected through exit interviews of voters we examine the results of the cumulative vote in the context of minority representation, voter understanding, voter utilization, and voter evaluation. We find that the system worked as supporters hoped — a minority candidate was elected in an at-large voting context. Further, voters appear to have readily understood the system and to have extensively utilized their cumulative options. We conclude that under certain circumstances cumulative voting can represent a viable alternative to single-member districting as a means of remedying the vote dilution of minority groups in local elections.