It’s Election Day and millions of Americans are heading to the polls. Millions have already voted, too, in what’s shaping up to be another high-turnout general election. To help shed light on the electorate, Catalist is publicly sharing data on early voting trends as well as results from polling the organization has contributed to over the past several months.
Catalist is a progressive data utility that maintains the longest-running voter file outside the major political parties. Each election cycle, Catalist analyzes publicly available data from election officials and matches it to its voter file, along with demographic information, and modeled estimates for vote choice, vote propensity and other variables.
Voter-file based data is one of the most robust tools we have for understanding the electorate, but takes time to process and analyze. Catalist has created an Election Data FAQ to help shed light on what we will know about the electorate and when over the coming months.
In a typical midterm election year, we would expect Democrats to perform poorly given backlash against the party in power — just as Republicans underperformed in 2018 — but Democratic voters became increasingly engaged over the summer after Republican-appointed Supreme Court justices overturned Roe v. Wade. Instead, we see evidence for tightly contested races across battleground states.
Early Vote Data
Today’s report is a snapshot of the 2022 absentee and early vote based on publicly available information that has been matched to Catalist’s voter file as of this morning, Tuesday, November 8, 2022. Catalist’s data will update over time as election administrators count and report more ballots, but this web page will remain static. Catalist will publish more information about the electorate in the coming months as part of its What Happened series.
Total returns and turnout
At least 39.5 million ballots have been cast in the 2022 election. At this same point in the 2018 cycle, 35.4 million ballots had been cast. Because Catalist processes early vote data as it’s made available from election administrators, this snapshot lags the real-time early vote total. Nationally, early vote turnout has been remarkably high, both as a result of higher engagement among voters as well as increased adoption of mail and early voting throughout the 2020 cycle. Whether turnout this year will break 2018 records will come down to how engaged remaining potential voters are on Election Day.
Figure: Total Ballots Cast (2018, 2020, 2022)
Catalist’s snapshot report focuses on battleground states with competitive state-wide elections, including for U.S. Senate and governor.
Table: Ballots Cast in Battleground States (2018, 2020, 2022)
|State||2022 Early Ballots Cast||2020 Early Ballots Cast||2018 Early Ballots Cast|
Democratic Share and Vote Propensity
Early and mail voting has leaned Democratic, continuing trends we saw in 2020. In the 2021 general elections and 2022 primaries, we also saw many Democrats return to Election Day voting, making it harder to estimate how much early voting will contribute to final election results.
Democratic support among early voters remains generally higher than in 2018 but lower in 2020.
- In AZ and WI, Democratic support is higher among early voters than 2020 and 2018. NM early voters are supporting Democrats more than in 2020 but equal to 2018.
- In FL, NC, and OR, Democratic support is lower among early voters than 2018.
Catalist data presented below are based on modeled scores for vote choice and likelihood of voting in this election.
Vote Choice Index (VCI) — Vote Choice Index is a ranked, modeled score running from 0 to 100 estimating how likely a voter is to cast a ballot for a Democratic candidate at top-of-the-ballot races. In states with partisan voter registration, Republican or Democratic or independent status is a matter of public record. These data inform Vote Choice Index scores when they are available.
Vote Propensity — Vote Propensity is a ranked, modeled score running from 0 to 100 estimating how likely a voter is to cast a ballot in a given election.
Table: Democratic Support in Early Vote (2018, 2020, 2022)
|State||2022 Early Ballots Cast Democratic Support||2020 Early Ballots Cast Democratic Support||2018 Early Ballots Cast Democratic Support||Difference in Early Ballot Democratic Support, 2020 to 2022||Difference in Early Ballot Democratic Support, 2018 to 2022|
In the battleground overall, the Democratic lead in the early vote is largely thanks to more turnout among Democratic middle propensity voters. Pennsylvania is the exception, with many high-propensity Democrats having already cast ballots.
Figure: Battleground States Ballots Cast by Modeled Vote Propensity and Vote Choice
Figure: Arizona Ballots Cast by Modeled Vote Propensity and Vote Choice
Figure: Georgia Ballots Cast by Modeled Vote Propensity and Vote Choice
Figure: Nevada Ballots Cast by Modeled Vote Propensity and Vote Choice
Figure: Pennsylvania Ballots Cast by Modeled Vote Propensity and Vote Choice
Midterms see lower turnout than presidential elections. Democrats generally have an advantage among all registered voters, but many of those voters are less likely to participate in midterm elections.
When we look at registered voters in the battleground who have yet to cast a ballot in 2022, we see that Democrats slightly outnumber Republicans. But Republicans have more high-propensity registered voters who have not yet cast ballots in 2022 while Democrats have slightly more moderate-propensity registered voters who have yet to vote.
Figure: Modeled Vote Propensity and Vote Choice for Registered Voters Who Have Yet to Vote in 2022 Battleground States
State practices vary for reporting vote method. Of the battleground states, the following provide us enough information to distinguish between a mail ballot and an early in person ballot in both 2022 and 2020: FL, GA, ME, NC, NM, NV, OH, and WI.
In 2020, voters who cast ballots by mail were more likely to support Democrats than those who voted early in person or on Election Day.
In GA, early in person voting has grown compared to 2020. The switch from voting by mail to voting early in person has been most prominent among Democrats, who went from 64% early in person voting in 2020 to 90% in 2022. Republicans shifted from 73% voting early in person in 2020 to 93% in 2022.
In NV, we see shifts towards voting by mail. Democrats went from 56% voting by mail in 2020 to 69% using mail in 2022. Republicans increased mail voting from 43% in 2020 to 55% in 2022.
Figure: Percent of Early Ballots Cast in Person (2020, 2022)
Catalist calculates age based on publicly available information about date of birth on voter files, where available. Examining data by birth year also lets us group voters by generational rather than age cohorts. This is especially important as Baby Boomers, a relatively large generational cohort, start to move into some of the older age buckets, such as 75+, which are commonly used in election analysis. Generational birth year ranges are based on the Pew Research Center’s definitions.
The youngest voters make up a smaller portion of all early ballots cast than in 2020 and 2018. Voters born between 1945 and 1960 have increased their share of the early vote the most since 2018.
Table: Battleground Early Vote Share by Birth Year
Young voters support Democrats at higher levels than do older voters. We estimate that in the battleground, 62% of Gen Z voters who have already cast ballots and 64% of Millennials who have voted early supported Democratic candidates.
Table: Modeled Democratic Share Among Early Voters by Generation
|Gen Z (1997 and younger)||62.4%|
|Gen X (1965-80)||53.9%|
|Greatest (before 1945)||49.8%|
In a few states, voters self-report race on their voter registration forms. In most states, race is modeled based on the Census, polling, and other data. The “Other” category includes multi-racial, Native American, and other groups. In the battleground overall, we see white voters increasing their vote share from 2022 by almost 3pp, with Latino voters decreasing their share the most, by 1.4pp since 2020. Compared to 2018, white voters are 1.2pp more of the early vote. Voters of color continue to lean heavily toward Democrats and remain a vital part of their multi-racial coalition.
Table: Battleground Early Vote Share by Race and Modeled Vote Choice
|Race / Ethnicity||Share of Early Vote, 2018||Share of Early Vote, 2020||Share of Early Vote, 2022||Difference in Early Vote Share, 2022-2020||Difference in Early Vote Share, 2022-2018|
When we disaggregate by state, we can more clearly see that voters of color comprise a smaller percent of the early vote than the cycle in nearly every state. Compared to 2018, some states have a higher share of voters of color, while other states have a lower share.
- GA is the only state where voters of color are approximately at the same level of early vote share compared to 2020.
- The biggest drop off in early vote share for voters of color is FL, where voters of color are 7.2pp less of the early vote in 2020 and 3.6pp less than in 2018.
- NV voters of color show the biggest positive shift from 2018, increasing their vote share by 2.6pp compared to 2018. In 2018, however, Nevada had very little mail voting. This year, every voter has been mailed a ballot.
Table: Battleground Early Vote Share by Race
|State||Voters of Color Share of the Early Vote, 2022 - 2020||Voters of Color Share of the Early Vote, 2022 - 2018|
Figure: Battleground Early Vote Share by Race and Modeled Vote Choice
In many states, gender data is available from voter registration forms. A handful of states are also starting to let voters pick non-binary gender designations, but this is not yet reflected in election administration data.
Over the summer, Catalist and other analysts saw a spike in women registering to vote in response to the Dobbs decision, particularly in states with competitive elections and in Kansas, where voters defeated an anti-abortion ballot initiative. Regardless of the registration spikes we saw this summer, currently, in the majority of battleground states, women comprise a smaller share of early ballots cast than in the 2020 and 2018 elections.
We see the largest declines in NV from 2018, when the state had very little mail voting. This year, every NV voter has been mailed a ballot.
We see gains in women’s share of the early vote in MI and PA compared to 2020. In ME, we see an increase in the women’s share of the early vote of 0.2pp from 2018.
Overall, in the battleground states, we expect 57% of women who have voted early to support Democratic candidates, compared to 49% of men who have cast ballots early.
Table: Women's Share of Early Ballots Cast
|State||Women's Share of Ballots Cast 2018||Women's Share of Ballots Cast 2020||Women's Share of Ballots Cast 2022||Women's Share of Ballots Cast, 2022 - 2020||Women's Share of Ballots Cast, 2022 - 2018|
Across the battlegrounds, the dropoff in womens’ vote share also shows disproportionate changes by race, with women of color dropping off from their 2020 share more than white women in nearly every state.
GA is an exception, where women of color comprise 0.3pp more of the early vote than at this point in 2020 and 2018. White women, meanwhile, have slightly declined in vote share since 2020 by 0.8pp and 0.3pp compared to 2018. In OR, both women of color and white women have seen a decrease in their share of the early vote compared to 2020.
We estimate that 78% of women of color voters in the battleground have cast ballots for Democratic candidates; meanwhile, we expect around 51% of white women voters to have voted for Democrats this cycle.
Table: Women's Share of Early Ballots Cast by Race
|State||Women of Color Share of Early Ballots Cast, 2022 - 2020||White Women Share of Early Ballots Cast, 2022 - 2020||Women of Color Share of Early Ballots Cast, 2022 - 2018||White Women Share of Early Ballots Cast, 2022 - 2018|
Education is modeled based on the Census, polling and other data. While education levels are correlated with class, income and wealth, they are not equivalent.
We see the largest increases in the early vote share from 2020 among white non-college voters. The exceptions are GA and KS, where the white non-college vote share decreased 0.6pp and 0.2pp, respectively, from 2020. We see the largest increase in FL, where white non-college voters are 4.3pp more of the early vote compared to 2020 and 2.2pp more compared to 2018.
We estimate that 45% of white non-college voters in the battleground who have already cast ballots supported Democratic candidates, compared to 51% of white college voters, 74% of college-educated voters of color, and 76% of non-college educated voters of color.
Table: Early Vote Share by Race and Education
|State||VOC College Share of Early Ballots Cast, 2022 - 2020||VOC College Share of Early Ballots Cast, 2022 - 2018||VOC Non-College Share of Early Ballots Cast, 2022 - 2020||VOC Non-College Share of Early Ballots Cast, 2022 - 2018||White College Share of Early Ballots Cast, 2022 - 2020||White College Share of Early Ballots Cast, 2022 - 2018||White Non-College Share of Early Ballots Cast, 2022 - 2020||White Non-College Share of Early Ballots Cast, 2022 - 2018|
State by State Data
Catalist is analyzing early vote data from 40 states this cycle. State-based estimates are available below.
Table: Early Vote Ballots Cast by State
|State||2022 Ballots Cast||2022 Democratic Share||2020 Ballots Cast||2020 Democratic Share||2018 Ballots Cast||2018 Democratic Share||Difference in Democratic share, 2022-2020||Difference in Democratic share, 2022-2018|
Catalist has worked with pollsters and organizations sponsoring polls for many years. In recent election cycles, we’ve seen a dramatic drop-off in response rates to polls. Voter files can help improve the representativeness of polls by accurately identifying registered and likely voters and weighting analyses by demographics and self-reported engagement in elections.
For campaigns, polling information is often private because assumptions about the electorate are important competitive information, but there’s also a strong public interest in understanding how different methods yield different results for campaigns and other practitioners since polls are such an important tool for understanding public opinion in a democracy. To that end, we are sharing these topline results from two of our major polling projects this cycle before polls close.
The Omnibus Project (TOP)
The Omnibus Project (TOP) is sponsored by major unions and progressive organizations. Catalist administers the poll and works with GQR, a major Democratic-aligned polling firm, to conduct these polls. Throughout the cycle, the TOP largely mirrored national trends we have seen in other major polls, including a statistical dead heat in the generic Congressional ballot. Each poll had about 1,700 respondents with oversamples of Black, Latino, and Asian American and Pacific Islander voters. A smaller portion of respondents, usually around 1,200 could be reliably matched to the voter file in a given cycle. The follow results are from the final pre-election TOP poll, which was in the field from October 7th to 25th:
- Motivation to vote remains extremely high. Democrats and Republicans report high motivation to vote, as do independents, though to a lesser extent. 79% of Democrats are highly motivated compared to 75% of Republicans, a slight increase for Democrats from September when the parties were more evenly matched. The Dobbs decision continues to motivate Democrats with 77% reporting an increased likelihood of voting as a result of the ruling. The generic Congressional ballot remains virtually tied, as it has for months.
- Economic pessimism rebounds. In a troubling sign for Democrats, economic pessimism is trending upwards after falling earlier this year. Overall, 67% of respondents say the economy is getting worse, an increase of 5 points from the previous poll. While the number of people anticipating a recession is holding steady in the high 80’s, individuals are personally feeling more financial stress. When asked about their ability to handle an unexpected $500 dollar expense, 67% reported this would pose a major problem for them, the highest this metric has been since we began polling this year in March.
- Biden’s job approval improves. Despite the worsening economic outlook, Biden’s job approval ratings improved slightly. While still underwater, Biden’s net favorability rose from -10 in September to -6. His ratings for handling specific issues have also generally risen compared to the summer, with some of his biggest improvements coming from approval of his handling of the issue of women’s rights (from 0 net favorable in July to +10) and crime and drugs (from -20 in July to -12).
Catalist has helped conduct polls in battleground states for a different coalition, in collaboration with Impact Research. These polls have been focused on Michigan, Minnesota, Nevada, Pennsylvania and Wisconsin and aggregated results are presented here.
Midterms often see backlash against the president’s party, especially when that party controls both houses of Congress.
But 2022’s midterm has been marked by much stronger enthusiasm from Democrats than we would usually see, with spikes in Democratic enthusiasm following Republican-appointed justices on the Supreme Court overturning long-standing abortion rights in Roe v. Wade. These spikes were evident in the summer when more women and Democrats registered to vote and voters in Kansas defeated an anti-abortion ballot initiatives. By contrast, when Democrats lost an odd-year gubernatorial election in Virginia in 2021, their reported enthusiasm was 11 points lower than Republicans’. (In the latest poll, voters who have already cast ballots were included with high-enthusiasm respondents, which didn’t significantly change the results.)
Figure: Voter Enthusiasm by Party
Previously Undecided Voters are Breaking for Republicans
Earlier in the cycle, Republicans were initially twice as likely to say they were undecided as Democrats. As the election has drawn nearer, undecided Republicans have started “coming home” to their party’s candidates after initial hesitation, leading to battleground races drawing closer.
In the past several weeks, undecided respondents — as well as a smaller number of respondents who had indicated support for third-party candidates — start to move toward Republican candidates in several topline races. Compared to solid Republican voters, these undecided voters are less conservative, less likely to vote, and less supportive of Trump.
Figure: Undecided Voters by Modeled Vote Choice Over Time
These movements can inform how campaigns estimate the ultimate disposition of the electorate. If we allocate remaining undecided voters based on how other undecided voters are breaking, we would expect to see a more Republican-leaning electorate than the latest polling indicates.
The economy, abortion and crime were the top issues for potential voters, with variance based on partisanship. How each voter conceives of an issue is not fully captured in these broad questions: for instance an anti-choice and pro-choice voter might both identify abortion as a top issue.
Table: Issue Importance by Modeled Voice Choice
|Most Important Issue||Total||Solid D||Swing D||True Swing||Swing R||Solid R|
|Size in Electorate||100%||35%||10%||8%||12%||35%|
|ANY ECONOMIC ISSUE*||42%||16%||35%||44%||57%||63%|
|The Environment & Climate Change||6%||14%||7%||4%||2%||0%|
|ANY OTHER ISSUE**||8%||11%||11%||13%||9%||3%|
* Includes inflation, the economy and jobs, taxes, the rising cost of living (asked in one state)
** The following issues scored <5% when asked: roads & infrastructure, racism, housing, drinkable water & renewable energy, the coronavirus
Republicans More Likely to Want to Challenge Election Results
Democrats overwhelmingly say they want their preferred candidates to accept the election results and concede if they lose. Among Republicans, only 43% say they want their preferred candidates to concede defeat if they lose while 22% say they want their candidate to challenge the election results and 36% say they don't know.
Figure: Preference for Preferred Candidate to Concede or Challenge Results After Losing
What Happened™ Catalist © LLC. All Rights Reserved.
Proprietary data and analysis not for reproduction or republication.
Catalist’s actual providing of products and services shall be pursuant to the terms and conditions of a mutually executed Catalist Data License and Services Agreement.