RecordWell Data

Ranked Choice Virginia's Report on its Surveying and Evaluation Efforts

Arlington County · Arlington County · Other

Open the original document More from Arlington County

Text

Extracted automatically from the original. Tables and formatting may not survive; the original document is authoritative.

Report to Arlington County Board Ranked Choice Voting Evaluation Prepared by: FairVote Ranked Choice Voting Resource Center Ranked Choice Virginia Updated: February 23, 2026 On November 4, 2025, Arlington County conducted its fourth election using ranked choice voting (RCV). In preparation for the election, the County Board awarded funding to two grantees to provide voter education and evaluate voters' experience using RCV: 1. Challenging Racism led the voter outreach work. 2. FairVote led the evaluation work in partnership with the Ranked Choice Voting Resource Center (RCVRC) and Ranked Choice Virginia (RCVa). The report documents the work completed by the evaluation-focused team under that grant agreement. The content is organized in three sections: 1. Outreach describes our collaboration with Challenging Racism, the main grantee for voter outreach work. 2. Evaluation details the design and results of the post-election survey of Arlington voters. Key findings include: ● 96% of voters were confident marking their ballots: 80% “very confident” and 16% “mostly confident” ● 87% of voters understood the ballot counting process: 62% “confident” and 25% “pretty sure” ● 66% of voters support using RCV in future general elections for Arlington County Board: 36% “strongly” and 30% “somewhat” 3. Recommendations identifies opportunities for future investment in RCV outreach, especially for Latino voters and targeted age groups. Page 1 of 12 (1) Outreach Though Challenging Racism (CR) was principally responsible for voter outreach, our team offered guidance based on RCVa's experience providing voter education in Charlottesville and RCVRC's work in more than 30 jurisdictions nationwide. ー We designed printed literature in both English and Amharic for distribution at community events. CR arranged the Amharic translation through their partnership with the Ethiopian Community Development Council. ー We planned and delivered presentations at two events: ● a panel discussion hosted by Advance Arlington at Lubber Run Community Center on September 10, 2025 ● a longer workshop hosted by CR at Central Library on October 13, 2025 ー Producing extensive web content was not within the scope of our work. Still, more than 3,700 unique devices visited RCVa's Arlington web page during the election, which described the County Board's decision to adopt RCV, the voting process, and links to more content from the candidates themselves. As we detail in the Recommendations section, the web traffic may indicate unmet demand from voters for neutral, reliable information about the election, beyond the content already provided by local news outlets and the County's Department of Voter Registration and Elections. Page 2 of 12 (2) Evaluation This section details the design of the post-election survey, including the pilot, contact methods, sampling, and weighting. Survey results begin on page 8. Design Pilot Survey In October 2025, we conducted a pilot survey to inform the design of the post-election survey. The pilot used Multimedia Messaging Service (MMS) messages to contact a random sample of more than 33,000 registered voters in Arlington using data from the research firm Murmuration. We used the pilot to refine the wording and sequence of questions. The pilot also measured how response rates varied across demographic groups. Consistent with common patterns in public opinion surveys, the pilot identified lower response rates among men, younger voters, and respondents whose primary language is not English. These groups were oversampled in the post-election survey so the results could reflect the views of a representative sample of voters. Contact Methods We contracted with SurveyUSA to administer the post-election survey. SurveyUSA is a non-partisan research firm rated among the top 15 polling agencies in America by FiveThirtyEight, a former project of ABC News. In addition to its national reputation, SurveyUSA also has specific experience conducting polling in jurisdictions that use RCV, including New York City, San Francisco, and the state of Maine. SurveyUSA used four contact methods to reach voters: 1. live phone call-center operators 2. recorded voice messages from a professional announcer 3. MMS messages with links to complete the survey online via mobile phone or tablet 4. online messages to Arlington residents in a commercial survey panel maintained by Cint USA Page 3 of 12 All methods informed respondents that the survey was conducted “on behalf of the Arlington County Board” to convey credibility and encourage strong response rates. The MMS and web panel visuals also included the County Board logo to affirm authenticity. Invitations asked voters to share their "experience in last week's election" without specifically mentioning RCV to encourage participation by all voters – not just those who were already familiar with RCV or opinionated about its use. Invitations also disclosed that respondents would be asked about the voting process but not who they voted for to ensure confidentiality. SurveyUSA started fielding responses on Thursday, November 6, 2025 – two days after voting closed and unofficial results were announced. Collection ended one week later on November 13. Fielding responses during this one-week window allowed voters to learn the election results and share reactions while their memory of the voting experience was still fresh. The survey asked each respondent 14 questions and took roughly three minutes to complete. Sampling The survey sample was drawn from public records of registered voters in Arlington County who were matched to contact information through the commercial data firm Aristotle. Respondents were then asked to self-report whether they voted in the recent election as an initial screening question. To encourage sufficient response rates across demographic groups – and in light of the pilot survey results – men and younger voters were oversampled in the post-election survey. Phone operators asked to speak with the youngest male in roughly 30% of households and the youngest adult in approximately 70% of households. In total, 584 respondents completed the survey. Table 1 above reports response rates by contact method. Nearly two-thirds of responses were collected by MMS. Roughly one in four respondents completed the survey by phone, and one in ten were drawn from the web panel. Page 4 of 12 Table 1 Response Rates Contact Method Response Rate # of Respondents Share of Sample MMS 1.55% 377 64.6% Phone - Recorded Voice 0.86% 90 15.4% Phone - Live Operator 0.81% 50 8.6% Web Panel -- 67 11.5% Total 584 100% Notes: The response rate is the number of completed surveys divided by the number of respondents contacted. The commercial web panel subcontractor did not report a response rate. Weighting SurveyUSA weighted the raw respondent sample to mirror Arlington's 2025 electorate on age, gender, race and ethnicity. Weighting allows the survey results to better reflect the views of all voters who participated in the election, not just those most likely to respond to surveys. Since ballots do not identify the demographics of individual voters, there are no data that perfectly capture the demographic traits of the electorate. To estimate the composition of Arlington's electorate, SurveyUSA combined data from the US Census of Arlington residents with CNN's post-election polling of Virginia's 2025 voters. To illustrate the weighting procedure, consider the representation of Latino voters in the survey sample. CNN found that Latinos were 5% of all Virginia voters in 2025, while the US Census estimates that Latinos were 9% of Virginia residents, so Latinos voted at lower rates than other ethnic groups. Under the assumption that turnout rates for Latino voters in Arlington mirror the statewide rate, SurveyUSA can estimate the Latino share of Arlington voters from Census data on the Latino share of Arlington residents. 1 Though voters' demographics are not perfectly captured by any data, there are some voter traits that are accurately measured and publicly reported – including voters' chosen ballot method. The County Department of Voter Registration and Elections reports the share of ballots cast by mail, early in-person, and in-person on Election Day. Respondents also self-reported their ballot method on the survey, so we can compare the weighted survey sample to Arlington's actual electorate as an external check on the weighting procedure. Table 2 presents the results of that validation exercise. Early in-person voters were overrepresented in the raw respondent sample by roughly 8 percentage points. Sample weighting closes that gap to 4 points. The remaining differences are within the margin of error for the given sample size, though early in-person 1 This assumption would fail if the turnout rate of Latino voters in Arlington differed dramatically from the turnout rate of Latino voters statewide. An alternate weighting procedure would use individual-level data from public records on Arlington voters and impute gender and ethnicity from voter names and precincts. Since sample weighting did not substantially alter the survey results (as shown in Appendix A), we did not invest in this additional analysis. The County Board could explore that work in future research. Page 5 of 12 Table 2 Survey Sample vs. Arlington Electorate Survey Sample Arlington County Voting Method Unweighted Weighted Board Voters Difference Mail 12.3% 11.1% 13.2% -2.1 pp In-Person - Early 40.8% 37.1% 32.8% 4.3 pp In-Person - Election Day 46.9% 51.8% 54.0% -2.2 pp Notes: The difference column reports the gap in sample shares between the weighted survey sample and the sample of all 2025 Arlington County Board voters. Early in-person voters are slightly over-represented, though the gap is within the weighted sample's margin of error. voters may remain slightly over-represented in the weighted survey results. If early voters are also more likely to be highly-engaged and informed about local elections, their views may also be slightly over-represented in the weighted results. Table 3 illustrates how weighting changes the demographic composition of the survey sample. The largest adjustments were for age. Voters age 45 and older were overrepresented by roughly 9 percentage points in the unweighted sample, along with homeowners (8 points) and White voters (8 points). The weighted sample amplifies the views of voters younger than 45, renters, and Latino and Asian voters. Weighting did not substantially alter the gender composition of the sample or the share of Black voters. Page 6 of 12 Table 3 Unweighted vs. Weighted Survey Samples Survey Sample Demographic # of Respondents Unweighted Weighted Difference Gender Male 284 49% 48% 0.3 pp Female 282 48% 49% -0.6 pp Age 18-29 years 22 4% 13% -9.1 pp 30-44 years 87 15% 24% -9.2 pp 45-64 years 250 43% 36% 7.3 pp 65+ years 225 39% 28% 11 pp Race & Ethnicity White 444 76% 68% 7.9 pp Black 49 8% 8% 0.1 pp Hispanic 37 6% 10% -4.1 pp Asian 23 4% 7% -2.6 pp Homeownership Homeowners 441 76% 68% 7.7 pp Renters 120 21% 28% -7.9 pp Notes: The difference column reports the gap in sample shares between the unweighted and weighted survey samples. Green shading highlights demographic groups that are overrepresented in the unweighted sample. SurveyUSA's sample improves on the County's past efforts to collect resident feedback about RCV. In 2023 and 2024, the County invited residents to complete forms posted online about their RCV experience. Since those surveys did not involve proactive, unsolicited outreach to all registered voters, they captured the views of residents who were connected to the County's public information streams and motivated to share their views about RCV specifically rather than the voting experience broadly. Table 4 compares the demographic composition of the weighted survey sample to the respondents of the County's past feedback forms. Older, White, and homeowning residents were dramatically overrepresented among 2023 respondents. The 2024 form was markedly more representative of the Arlington electorate, though likely still overrepresented the views of the County's most engaged residents. Though weighting is an important methodological consideration, in practice weighting has little impact on the substantive results in this sample. All of the results reported in the next section use the weighted sample. Appendix A compares the weighted results to responses from the unweighted sample and finds few differences larger in magnitude than the sample's margin of error. Page 7 of 12 Table 4 Past County RCV Survey Samples Demographic 2023 2024 2025 White 89% 72% 68% Homeowners 83% 72% 68% 60+ years old 54% 40% -- Notes: The 2023 and 2024 surveys reported results for ages 60+. The 2025 survey grouped respondents ages 65+ together (27.5% of the weighted sample). The different age groups prevent direct comparison across all three years. Results Survey questions focused on four dimensions of the voting experience: 1. Did voters anticipate using RCV in this election? 2. Were voters comfortable marking their ballots? 3. Did voters understand the ballot counting process? 4. Do voters support using RCV in future general elections for Arlington County Board? All of the results that follow are reported for the weighted survey sample. The margin of error for full-sample results is +/- 5 percentage points, with larger bounds for subsample results. Anticipation The 2025 election was Arlington's second use of RCV in a general election for County Board, and most voters anticipated their ranked choice ballots. More than 70% of voters knew they would be using RCV before they received their ballots —either by mail or in-person at their voting location. Another 26% learned about RCV when they received their ballots. The remaining 3% were unsure. Anticipation rates were largely consistent across racial and ethnic groups except for Latino voters. Less than half of all Latino voters (45%) knew they would be using RCV before they received their ballots. As we detail in the Recommendations section, additional outreach to Latino voters, including Spanish-language content, may be needed to narrow the awareness gap. Ballot Marking Though 71% of voters anticipated using RCV, fully 80% were "very confident" marking their ballots , which indicates that some voters found the RCV ballot easy to use without advance instruction. Another 16% of voters said they were "mostly confident" marking their ballots. Just 4% were "not very confident" or "did not understand" the ballot marking process at all. Comfort with ballot marking was consistently high across demographic groups, except for the youngest voters. Roughly 12% of voters ages 18-29 were "not very confident" about ballot marking, compared to just 3% of voters ages 30 and older. Additional research would be needed to understand whether young voters' discomfort reflects RCV-specific concerns or general unfamiliarity with the voting process. While Latino voters were less likely to anticipate using RCV than the broader electorate, most Latino voters were still comfortable marking their ballots: 73% were "very confident" and Page 8 of 12 another 21% said they were "mostly confident". Just 6% of Latino voters expressed low confidence with ballot completion. Respondents were also asked how many candidates they chose to rank. 2 More than half (56%) ranked more than one candidate and another 35% ranked exactly one candidate. Exploring why voters ranked only one candidate is important for assessing how well they understand RCV. Among the 35% who ranked one candidate: 3 ー 64% said they only liked one candidate ー 29% said they didn't know enough about the other candidates ー 25% expressed concern ranking more options could hurt their favorite candidate The first two reasons do not necessarily imply that voters are confused about RCV. Voters may choose to rank only one candidate if they genuinely don't know or like any of the other options. As we discuss further in the Recommendations section, RCV may increase voters' demand for information about all the candidates, and the County could choose to play a role in amplifying credible, neutral voter guides. The last of these reasons is more concerning to voter educators. It indicates that roughly 9% of voters may hold a misconception about RCV – that ranking more options could make it harder for a voter's favorite candidate to win. Though the share of voters who hold this belief is relatively small, the County should continue to address this concern in voter outreach. Ballot Counting Respondents were somewhat less confident in their understanding of the ballot counting process: 62% were "confident" they knew how their ballot was counted, and another 25% were "pretty sure." Just 13% said they were "not sure," or "did not understand" how ballots were counted. The results suggest most Arlington voters have grown comfortable participating in RCV elections, though questions about the tabulation process persist with some residents. Self-reported understanding was lowest among Latino and Asian voters, as well as seniors (ages 65+). In these groups, 23% of voters were “not sure” or did not understand the ballot counting process. These subgroup results motivate the recommendations for targeted voter education outlined in this report's final section. The survey also revealed that a sizable share of voters are confident in ballot marking and counting even when they're not closely tracking the results of a race. Though 87% of voters were "confident" or "pretty sure" they knew how their ballots were counted, just 52% correctly identified Takis Karantonis as the winner of the County Board race, while 47% were unsure. 3 These figures sum to more than 100% because respondents were allowed to choose more than one reason. 2 These self-reported ranking patterns could be externally validated if Arlington County chose to release its cast vote record (CVR), as Charlottesville did after its first RCV election. Page 9 of 12 These findings suggest the survey successfully reached the sizable share of voters who complete the County Board portion of the ballot even when they don't follow the outcome of the election. Overall, voters were confident in the process – whether or not they knew the results. Support for RCV The survey's last section asked voters whether they support using RCV in future general elections for Arlington County Board. Roughly two-thirds expressed support: 36% "strongly" and 30% "somewhat". Roughly one-in-four (25%) opposed continued RCV use: 11% "somewhat" and 15% "strongly", leaving net support at +41 percentage points. Age was the strongest demographic predictor of support for continued RCV use. Though every age group registered positive net support, younger voters were most supportive and older voters were more cautious. Among voters ages 18-29, 27% "strongly support” future RCV elections and 48% "somewhat support." Seniors (ages 65+) had the lowest net support at +26 percentage points. The age group pattern was mirrored in the breakdowns by housing status: net support was stronger among renters (+58 points) than homeowners (+39 points). Support also varied across racial and ethnic groups. More than half of all Asian respondents strongly supported continued RCV use. Both Black and white voters mirrored the full sample results, with roughly +43 points net support. Latino voters expressed more modest but still positive support at +19 points. Taken together, every demographic group recorded net positive support for RCV, with net support highest among renters and Asian voters and lowest among seniors and Latino voters. The breadth of that support – across age, race, gender, and housing status – suggests Arlington voters across the electorate have found value in the RCV system after two general election cycles. Page 10 of 12 (3) Recommendations We conclude by highlighting four potential priorities for Arlington’s future voter education work based on the findings of the post-election survey. Spanish-Language and Latino Voter Outreach Latino voters consistently reported the least familiarity with RCV in the survey sample. Less than half (45%) knew they would be using RCV before they received their ballots, and 23% reported they were not sure or did not understand how ballots were counted. Latino voters also showed the least support for continued RCV use across ethnic groups, though net support was still positive at +19 points. To assist Latino voters in developing their understanding of RCV, the County could invest in earlier, sustained outreach to Latino residents, including Spanish-language education. Organizations like New Mexico-based Democracy Rising have developed comprehensive RCV education guides for Spanish-speaking voters (available in both Spanish and English). Large RCV jurisdictions like New York City and Minneapolis also provide extensive collections of translated RCV content. Effective delivery of Spanish-language content would also require partnerships with Arlington-based organizations that serve Spanish-speaking residents to ensure that quality content reaches the voters who may be less connected to the County’s official public information streams. Age-Tailored Voter Education The survey revealed distinct patterns across age groups that suggest different informational needs. Younger voters (ages 18-29) showed lower confidence marking their ballots: 12% were not very confident, compared to just 3% of voters ages 30 and older. Meanwhile, seniors (ages 65+) reported less confidence in the counting process: 23% were not sure or did not understand, compared to 13% overall. The age patterns observed in Arlington are consistent with national data on RCV education. In 2021, New America surveyed research on RCV jurisdictions and found that age is the most consistent predictor of self-reported RCV understanding among voters – not race, ethnicity, or socioeconomic status. The same studies also find, however, that older voters are often more likely to submit ballots that use all of their rankings despite reporting lower understanding, which makes it difficult for researchers to disentangle factual confusion from subjective concerns about the new voting system. Arlington's Registrar has already prioritized events at senior centers, which is a valuable foundation for reaching older voters. Future education efforts could tailor content and delivery methods to different age groups. Younger voters may benefit from more hands-on practice and digital engagement, while older voters may seek clearer explanations of the counting process and reassurance that their votes will be counted correctly. Page 11 of 12 Single Ranking Misconception Roughly 9% of voters expressed concern that ranking more candidates would hurt their first choice – a misconception that persists two years into Arlington's use of RCV with the general electorate. Though the share of voters who hold this view is relatively small, the County could continue to explore new approaches for addressing this point of confusion, including revised ballot instructions and visuals, oral instructions by pollworkers, and focused outreach with trusted community groups to dispel this concern. Nonpartisan Candidate Information Because RCV encourages voters to consider more candidates, the process can also increase demand for reliable information about the people running for office. In the interest of keeping the survey short, we did not ask voters about the information sources they used to learn about the election, but our work offers some signs that voters are eager for more opportunities to compare the candidates: ー Roughly 10% of all voters ranked only one candidate because they didn't know enough about the others. ー More than 3,700 unique devices visited our Arlington web page during the election period, which is substantial traffic for a local County Board race. Several jurisdictions across the US provide official voter guides as a routine part of election administration. California law requires all counties to produce guides that include candidate statements for local nonpartisan offices, and Washington tasks county auditors with publishing similar content. Securing legal and financial support for a state voter guide program in Virginia would no doubt be a multi-year endeavor. Still, the County could explore more modest initiatives with local organizations to support voters in accessing neutral, credible candidate information ahead of elections. Page 12 of 12

Source: Arlington County website. First collected Oct 18, 2026.