Voter Landscape Analysis for Final Five Summary 1

Key Insights 1

National Landscape 1

State Picture 2

Most Supportive Voters 2

Least Supportive or Unsure Voters 2

Further Questions to Explore 2

Research References 3

State-by-State Support in Detail 5

Best Positioned States for 6

Opportunities to Expand Support 6

Winning Coalitions 7

Alaska 8

Massachusetts 8

Florida 9

New York City 9

Maine 10

Further Questions to Explore 11

In which States and Precincts is Reform most Likely to be Successful? 11

Do Voters Understand the Problem (and Solution)? 11

Idiosyncrasies of Previous Victories 12 1

Conclusion 12

Summary Over the last 15 months, as part of Citizen’s broader research on democracy reform, we have conducted survey research measuring public support for electoral reforms--both nationally and in key states. Specifically, we measured support for non-partisan primaries, ranked choice , a package of reforms proposed (and subsequently adopted) in Alaska similar to the “Final Five” reforms. Our goal has been to identify and understand on a detailed level the base of support for these reforms and to help our partners understand what states have the best pre-conditions for successful campaigns.

In this memo, we examine the electorate’s views on democracy reform proponents as well as its broader beliefs and opinions, as one important factor of many in the due diligence process. In addition to survey responses, we conduct a precinct-level electoral analysis to identify possible key constituencies who supported and opposed the measures in states such as Alaska and Florida. Our intent with this memo and subsequent research on these policies is to provide the electoral reform community with a shared understanding of current baseline support for reforms, so that our community of partners can focus energy on making progress rather than debating known facts.

Key Insights National Landscape ● At the national level, as recently as January 13-15, 2021, we see that while most respondents favor some kind of electoral reform, they are not yet broadly supportive of reforms with the exception of those that limit the role of money in politics; redistricting reform follows as a close second. ● While non-partisan primaries consistently receive support from a majority of voters, ranked choice voting is less popular, particularly as a standalone policy, and with strong Republican partisans and very conservative voters. ● Opinion may still be unconsolidated for RCV with one-third of voters across multiple surveys saying they’re neutral or need to learn more; however, partners looking to pursue RCV as a standalone or even to use the “RCV” terminology should be aware that very conservative and solid Republican voters are consistently more opposed than supportive.

State-Level Picture ● Support is very high for s ome kind o f reform to make representatives more accountable to their constituencies. The top ten jurisdictions in which voters have expressed strong interest in reform are Minnesota, DC, Rhode Island, Alaska, Washington, Hawaii, Oregon, Vermont, Massachusetts, and Wisconsin.

2

● Support for open primaries was highest in Alaska (+38%), followed by Hawaii (+37%), Oregon (+34%), Washington (+32%), and Minnesota (+31%). These states were also among the highest in general support for electoral reform as well. ● Based on aggregate survey data in states that allow ballot initiatives, states that present the best opportunities for reform at present include Washington, Oregon, Utah, Montana, Ohio, and Arizona.

Most Supportive Voters

● Generalizing support measures across several surveys, the following types of voters are the friendliest to electoral reforms, and specifically FFV’s reforms as a package: ○ Democrats, and self-described non-partisans; ○ Young voters, particularly those under 34; ○ Hispanic voters; and ○ Individuals at either end of the income spectrum: (those making less than $25,000 or over $150,000 per year).

Least Supportive or Unsure Voters

● In general, the demographic groups that are relatively more likely to indicate opposition to electoral reforms include: ○ Older voters, particularly those over the age of 65; ○ Less educated voters, particularly those with only a high school education or less; and ○ Self-identified Republicans. ● Those who are unsure are very evenly spread across demographic categories and generally represent about 30% of respondents, regardless of income, race, education, or age.

Further Questions to Explore

In the future, we recommend more 1) state specific polling, including message and language testing, to understand how best to position measures for success in states with unique voter bases, 2) r esearch to probe how much voters agree with reformers’ diagnoses of the problems ailing our (let alone the proposed solutions), and 3 ) building national or state-specific model(s) based on what we know about the winning voter base for these reforms to understand early which voters warrant the most attention. Understanding the base motivations and bel iefs of undecideds and Neutral voters should be a high priority since those are the groups that will be decisive for campaign victories. In general, knowing which voters or “audiences” will be most important for victory early on

3 should drive campaign strategy including decisions about staffing, surrogates, messaging and budget e.g. which Republicans are persuadable.

Research References Linked here please find the survey toplines for the following: ● January 2021 “State of Play” Survey ● November 2020 Alaska After Action Survey ● September 2020 National Vote by Mail Survey: Reform Results

Citizen Survey Results

On two separate occasions in the last six months, Citizen Data has fielded large nationwide surveys that collected public opinion on the need for non-partisan, open primaries, for ranked choice voting elections, as well as the need for some kind of electoral reform generally.

The first survey, fielded between September 3 - 8, 2020, collected 30,000 responses from a representative sample of likely voters from all 50 states and the District of Columbia. The second survey was fielded between January 13 - 15, 2021, and collected 3,000 responses from a nationally representative sample of 2020 voters.

In general, the surveys found that while the vast majority of respondents are strongly in favor of some kind of reform, they are less supportive at this stage of the specific reforms proposed in the surveys. Non-partisan primaries received majority support in both surveys, while ranked choice voting was substantially less popular among those who expressed an opinion. However, in the case of both reforms, about a third of respondents were undecided or said they may need to learn more.

Support for non-partisan primaries and ranked choice voting broke down similarly across both surveys and between proposed types of reform. In general, Biden supporters and self-identified Democrats were more likely to support these reforms than Republicans and Trump supporters. Levels of support were largely stable across racial and education strata, though the most educated were more likely to support an increase of accountability among their representatives and least likely to support ranked choice voting in particular. One notable trend we observed is that younger voters are more likely to support reforms (including ranked choice voting and open primaries specifically), while older voters are comparatively unlikely to do so.

Additionally, Citizen broke down support for each measure by whether the respondents voted in 2018 (a midterm year) or 2016 (a presidential year). In general, the support among voters who voted in the most recent presidential election for which we have complete data is indistinguishable from those who voted in the most recent midterm

4 year. Though we would need to conduct further research to offer better support for the claim, it appears that a campaign for these measures would probably fare about the same regardless of whether they were offered in 2022 or 2024. The results of those breakdowns are presented in the table below.

2016 2018 2016 2018 2016 2018

Electorate Electorate Electorate Electorate Electorate Electorate

Need for Reform* Open Primary Reform* RCV Reform* Support 65% 65% 43% 43% 18% 19% Not Sure 27% 27% 29% 29% 29% 29% Oppose 8% 8% 28% 28% 52% 52%

*differences between each category a re statistically indistinguishable

As seen in these tables, support does not vary between those voters who participated in the 2016 election compared to those who participated in the 2018 election. The two types of voters both report an indistinguishably high need for reform, a moderate openness to open primary reform and demonstrate an equivalently low baseline support for ranked choice voting. The data summarized in the tables above offer some preliminary evidence that a ballot in a midterm year would have about the same chance of success compared to one in a presidential election year.

State-by-State Support in Detail The survey results visualized state by state in the series of tables below demonstrate the support and opposition percentage breakdowns from the national 30,000 response survey from September of 2020.

The following analysis is an initial surface-level look into state-by-state support for each of these measures. Bear in mind that the data here was part, but not the main focus of, a longer survey focused on early in-person and vote by mail turnout in the 2020 election. Further, this survey did not test messaging or seek to educate survey respondents on the details of any of the measures. In combination, these three factors indicate that the results presented below should be viewed as preliminary findings that warrant further research and investigation.

Further, as we’ll discuss in the section below, there remains a significant portion of the population that remains “unsure” or expresses a desire to learn more about the measures

5 before taking a stance for or against them. Nonetheless, some interesting trends are revealed when looking at baseline support across the top states.1

The first question asked whether respondents would support some kind of change to election systems to hold members of Congress more accountable to all voters in their districts, regardless of . The table of state-by state responses labeled “Need Reform” shows that support is very high for s ome kind of electoral reform.

The next question asked whether respondents would support replacing the partisan primary system with a non-partisan primary system wherein the top five vote-getters would proceed to the . This question gained net positive support in almost all states. The table labeled “Open Primary Reform” details that support was highest in Alaska, where positive messaging and education was in full swing at the time of the survey. Closely following Alaska, however, were Hawaii, Oregon, Washington, and Minnesota. These states were among the highest in support for general reform as well.

Finally, Citizen asked whether respondents would support the ranked choice voting reform specifically. The table labeled “RCV Reform” shows that while RCV failed to gain net support, this question also reveals the highest levels of undecideds. This suggests that

1 Rhode Island’s, Alaska’s, Wyoming’s, DC’s, and Delaware's results are based on fewer than 100 responses

6 the topic may be too unfamiliar or complex for respondents to commit to support or opposition before learning more. In any case, the same states that are among the highest levels of net support for general reform and open primaries are among the lowest net opposition for ranked choice voting. Minnesota, Washington, and Oregon all feature highly on this list, validating their strength as states to pass popular reform.

Opportunities to Expand Support Between Citizen’s two national surveys, fielded about three months apart, we posed five electoral reform questions to about 33,000 total respondents. From those responses, we have assessed the rough patterns of support, opposition, and need for further education.

The demographic groups detailed below are those that are mostly likely to align themselves in favor of the reforms they were asked about. Though further research would be required to confirm, it initially appears that the following groups would be more likely to support these combinations of reforms relative to their counterparts:

● Democrats and self-described non-partisans; ● Young voters, particularly those under 34; ● Hispanic voters ● Individuals at either end of the income spectrum: (those making less than $25,000 or over $150,000 per year).

In addition to the respondents who reported supporting reform, there remain a significant percentage who wish to learn more before taking a stance. About 30% of those responding to the question on Open Primaries and 28% of those responding to a question on Ranked Choice Voting are “Not Sure” whether they support or oppose the measure. Voters not sure yet are evenly distributed across demographic categories and therefore are reachable uniformly throughout the population.

The groups detailed below are those who, in general would fall on the side of opposition in the absence of education or positive messaging. They are the groups that would most likely need to be flipped in a successful campaign:

● Older voters, particularly those over the age of 65; ● Less educated voters, particularly those with only a high school education or less; and ● Self-identified Republicans.

7

Potential Winning Coalitions Five jurisdictions have offered ballot initiatives that have attempted to pass some or all of the components of Final Five Voting reform. Those five jurisdictions are Alaska, Maine, Massachusetts, Florida, and New York City. Citizen Data has collected the precinct level returns for those ballot initiatives from each of the five jurisdictions and combined those precinct level returns with the demographic characteristics of each precinct using Citizen’s in-house voter data file.

The analyses below provide insight into the most important demographic characteristics that influenced each of those races. Citizen implemented linear regression models to estimate the relationship between demographic variables (e .g., r ace, income, partisanship) with support for electoral reform ballot measures at the precinct level. Because we cannot directly observe the vote choices of individual voters, particularly in states that Citizen didn’t specifically poll, leveraging precinct-level data allows us to employ the most granular level of analysis possible in understanding potential contributing factors in support for (or opposition to) electoral reform measures. These analyses are summarized in the table below (with more detailed discussion of state-by-state results in the following sections).

Precinct-Level Demographics AK BM2 FL A3 MA BM2 ME BQ5 NYC BQ1 and Electoral Reform Support 2020 2020 2020 2016 2019 -0.02 0.04* * -0.16* ** 0.14* -0.18* * % Voters of Color (0.05) (0.02) (0.05) (0.08) (0.05) -0.04* * 0.02 0.06* ** 0.15* ** 0.01 % Income > $100K (0.02) (0.01) (0.02) (0.05) (0.05) 0.13* * -0.28* ** -0.08 0.06 -0.02 % College Graduates (0.05) (0.08) (0.06) (0.06) (0.11) -0.50* ** -0.14* * -0.62* ** -0.66* ** -1.66* ** % Registered Republicans (0.07) (0.06) (0.14) (0.07) (0.20) 0.28* ** -0.02 0.77* ** Biden Share of 2-Party Vote (0.02) (0.04) (0.06) 0.43* ** 0.68* ** 0.01 0.61* ** 1.01* ** Constant (0.02) (0.04) (0.04) (0.03) (0.05) Observations 440 5,924 2,175 483 3,466 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01

8

These regression analyses allow us (based on a series of important assumptions) to estimate relationship of each demographic variable with the precinct-level support for electoral reform ballot measures. Results from these analyses are summarized in the figures below. Estimates of the individual predicted effect of each variable on ballot measure support at the precinct level are plotted as points (surrounded by confidence intervals at the 0.95 and 0.90 levels). The points—or regression coefficients—plotted in the figures below should be interpreted as follows: each point represents the predicted effect of a one percentage point increase in that demographic characteristic on the level of support for the relevant electoral reform ballot measure in an average precinct.2

Alaska In Alaska in 2020, the average precinct-level support for Ballot Measure 2 was about 45% (BM2 was passed with 50.6% percent of the vote, meaning that larger precincts had disproportionately higher levels of BM2 support). On average, we find that higher proportions of college graduates and Biden voters are associated with significantly higher precinct-level support for BM2. Conversely, precincts with a higher proportion of registered Republican voters in particular, are associated with significantly (and very substantially) lower precinct-level support for BM2.

2 A +1.0 regression coefficient for “percent income > $100K” indicates that a one percentage point increase in the number of voters from households with >$100,000 in annual income is associated on average with a one percentage point increase in precinct-level ballot measure support.

9

Massachusetts In Massachusetts in 2020, the average precinct-level support for Ballot Measure 2 was about 44% (BM2 was defeated, garnering 45.2% of the vote statewide, meaning that larger precincts had marginally higher levels of support for BM2 on average). Support for MA BM2 at the precinct level was significantly and substantially correlated with higher proportions of Biden voters, while opposition was similarly highly correlated with higher shares of registered Republican voters. Precincts with higher shares of voters of color also had significantly lower levels of support for BM2 on average.

Florida In Florida, Amendment 3 received 57% of the final vote (meaning that precinct size was not associated on average with higher or lower levels of support for the measure). However, in Florida, a ballot measure requires at least 60% of the vote in order to pass. So, while the coalition here would have been winning in most other states, it was not in Florida. In part because the overall level of support for Amendment 3 was so high, the tested precinct-level demographic characteristics do not predict substantially higher levels of ballot measure support. That said, precincts with higher proportions of voters of color had significantly higher Amendment 3 support on average. Conversely, higher precinct-level shares of college graduates and registered Republican voters predict significantly lower levels of Amendment 3 support.

10

New York City New York City Ballot Question 1 received about 73% support at the precinct level on average in 2019, marginally lower than the 73.6% support it passed with citywide. Likely due to the very high overall rate of support, none of the tested precinct-level demographic variables were predictive of higher BQ1 support. However, the proportion of registered Republican voters was extremely highly correlated with lower BQ1 support, such that on average, a one percentage point increase in the precinct-level share of registered Republicans was associated with a nearly 1.7 percentage point decrease in BQ1 support.

Maine Maine Ballot Question 5 passed with 52.1% of votes statewide in 2016. In comparison, a precinct-level analysis shows that the average precinct supported BQ5 at about 45%, suggesting that supporters were disproportionately concentrated in larger voting precincts. In addition, we find that precincts with higher proportions of higher-income voters and voters of color were significantly associated with higher levels of support for BQ5. In contrast, higher proportions of registered Republican voters in a precinct is both substantially and significantly predictive of lower support for BQ5.

11

Best Positioned States for Ballot Initiatives

By combining its analysis of survey results with a review of jurisdictions that allow ballot initiative referenda, Citizen has formed a recommendation of jurisdictions that would be most optimal to target for reform in 2022. Based on these states’ aggregated survey differentials and ability to pass ballot initiative reform, Citizen recommends that election reformers pursue additional research in:

● Washington, DC ● Washington ● Oregon ● California ● Utah ● Montana ● Ohio ● Arizona

To arrive at these recommendations, Citizen summed up the total differential between support and opposition within each state across all of the electoral reform questions asked on our survey and created a rank order of states from that list. After filtering to those states that allow ballot initiatives to be brought in general elections, these seven states were among the top that hadn’t already implemented some kind of reform through ballot initiative.

Further Questions to Explore This memo is intended to serve as an initial analysis of the future directions for FFV support campaigns. However, there remain many lingering questions and analyses that would need to be undertaken in order to have a more robust and thorough understanding of viability. This section endeavors to explore those known unknowns and offer recommendations on the analytical approaches that would resolve their uncertainty. Some key questions that could warrant further research:

A. What messaging best connects electoral reforms as solutions to various types of voter dissatisfaction? B. How does a voter’s views on the legitimacy of recent elections impact reform support? C. How does a Biden administration and a Trump loss shift voter attitudes toward reform? D. How much is Final Five’s viability dependent on even more popular reforms e.g., by combining a FFV reform with limiting the role of money in politics?

12

In which States and Precincts Is Reform Most Likely to Be Successful? Citizen has compiled the precinct-level election returns of each state or municipality that has recently attempted the passage of some or all of the components of FFV reform. These areas are described above, but include Florida, Massachusetts, Alaska, New York City, and Maine. Using Citizen’s in-house voter file, we have matched the precinct level results to the demographic characteristics of each precinct.

Leveraging its national voter file and these matched results, Citizen proposes profiling the success of FFV reform in every precinct across the country. Organizations that would want to leverage this model at any subsequently larger geographic level would be able to pull from this predictive model to understand the baseline success probability metrics to better focus their efforts.

Do Voters Understand the Problem (and Solution)? The champions of Final Five Voting understand the core problems that it aims to solve. However, the rest of the voting public might not understand those problems. Even if they do, those voters might not connect the proposed solution that FFV provides with those problems.

Citizen notes the need to conduct additional research on whether voters understand the problem in the first place, how to get them to understand FFV voting intuitively, and that FFV resolves the problem.

Idiosyncrasies of Previous Victories Alaska, Maine, and New York City have most recently passed some or all of the components of Final Five Voting. However, we would like to understand the extent to which these districts are distinctive from the rest of the country. Each one of these areas have unique political landscapes and cultures that may have contributed to the success of each campaign.

Further research is needed to determine whether the campaigns that were successful in these areas are repeatable. As a start, Citizen could attempt to test successful messaging from each of these areas in other prospect states to see if voters in prospect states are receptive to the same messages that won over the constituencies of New York City, Alaska, and Maine.

Conclusion Recent Citizen Data research and the results of RCV and open primary ballot initiatives in states since 2018 provide a picture of strengths, weaknesses, and opportunities for the movement to implement electoral reform across the country. Voters are receptive to the need for reforms, and now there’s much work to be done to persuade voters “Final Five” is

13 the solution to their concerns. Non-partisan primaries are popular whereas ranked choice voting is unpopular--particularly among Republicans--although a large number of voters are neutral or need to learn more before forming an opinion. Taking on “dark money” in politics and redistricting reform are more well known and more popular at this stage, and their popularity may help “sweeten” Final Five reforms if voters perceive them to solve the same problem(s). On average, younger and more progressive voters are most supportive of Final Five reforms whereas very conservative older voters are least supportive. Areas that warrant the most additional research are message and language testing with groups that are not yet supportive but critical to success in the states e .g., Independents and Republican leaners, as well as voter file analysis that predicts which voters will be supportive, opposed and persuadable in order to inform early campaign strategy from coalition building and branding to media and targeting.