Electability Reflects What Americans Believe is Possible

How beliefs about gender in America drive perceived electability

Between May 30-June 3, we conducted a listening survey to 1,871 registered voters to understand how the concept of electability is influencing the beliefs and behavior of the American electorate. While traditional research has struggled to define and measure electability, our unique methodology provides unique insight into this complex concept.

Likely Democratic voters feel a mix of extreme urgency and uncertainty about the chances of winning in 2020. This mindset helps to explain why perceptions of electability are such a strong force this cycle. The impact of perceived electability can be seen in the gap between the primary candidate voters choose in a typical horse race question compared to the candidate that they most wish could be the next president.

The electability gap is driven in large part by beliefs about how gender is perceived in America today. Those who see gender as a barrier to electability consistently express beliefs that other voters will not vote for a woman. These findings suggest that many voters may not need to be convinced that female a candidate is more capable — they must be convinced that Americans are capable of electing them.

When it comes to perceived electability, the greatest messaging challenge for the campaigns of female candidates may be less about the story of their candidate and more about the story of America today.


TOPLINES

Likely Democratic voters are currently experiencing a mix of extreme urgency, relatively low confidence, and unsettled emotions heading into 2020.

  • 97% of likely Democratic voters believe beating Trump is very or extremely important, yet only 28% are certain that Democrats will win.

  • When it comes to 2020, likely Democratic voters are currently feeling emotions of determination and hope, but also frustration, overwhelm, and doubt, whereas likely Republican voters are feeling relatively more excitement, pride, and happiness.

As a result, perceived electability is particularly salient for likely Democratic voters heading into 2020.

  • This study sheds light on the impact of perceived electability by observing which candidates voters favor when they are taking electability into consideration, compared to the candidates they favor when electability is taken off the table.

When perceived electability is removed from consideration, the Democratic primary race tightens significantly.

  • When choosing a preferred president absent considerations of electability, 21% chose Warren, 19% chose Biden, 19% chose Sanders, 16% chose Buttigieg, 12% chose Harris, and 4% chose O’Rourke.

Gender appears to have a greater effect on perceived electability than age, race, ideology, or sexual orientation.

  • 19% of those who would make a female candidate president believe that gender is a primary barrier to increasing electability.

  • Among those who raise gender as an issue, 62% express beliefs that American voters will not elect a woman. Their concerns are not about the capability of female candidates, so much as they are about the willingness of Americans to elect a woman.

  • This data suggests that at the time the survey was conducted, the preferred candidate by a small margin when electability is removed from the equation, is a woman.  


SUMMARY OF FINDINGS

The Democratic electorate is experiencing a mix of extreme urgency, relatively low confidence and a mix of unsettled emotions

  • 97% of likely Democratic voters believe it is very or extremely important to beat Trump

  • Only 28% of likely Democrat voters believe that a Democratic nominee will certainly win in 2020, compared to 61% of likely Republican voters.

Importance + Confidence 2020

Republicans are more positive and excited about the 2020 election, whereas Democrats are more likely to feel a mix of hope and unsettled, negative feelings:

  • 27% of respondents from both parties feel Determination

  • 21% of likely Democratic voters feel Hope, 42% feel negative emotions including Frustration, Anger, Doubt, Hopelessness, Overwhelm, or Sadness

  • Likely Republican voters feel relatively more Excitement (19%), Pride (8%), Happiness, (4%) and No Strong Feelings (8%)

Emotion associated with the 2020 election by likely party vote.

Emotion associated with the 2020 election by likely party vote.

 

Many likely Democratic voters prioritize winning above issue positions.

  • 38% of likely Democratic voters say they would prefer a Democratic candidate that is likely to win, even if they do not agree with them on most issues, compared to just 7% of Republicans.

  • This climbs to 63% for likely Democratic primary voters when the question is focused on beating Trump specifically (not shown).

 
Whether respondents would prefer a candidate that agrees with them on most issues but is less likely to win, or a candidate that they disagree with on most issues but is most likely to win.

Whether respondents would prefer a candidate that agrees with them on most issues but is less likely to win, or a candidate that they disagree with on most issues but is most likely to win.

A significant segment of likely Democratic voters are choosing their preferred candidates based on their perceptions and beliefs about who can win - about electability. While electability has proven difficult to define and measure in the abstract, this study provides evidence to understand the impact of electability considerations by observing which candidates voters favor when they are taking electability into consideration, compared to the candidates they favor when electability is taken off the table.

First we asked a typical “horse race” question about who the respondent would support if the primary election were today. Later in the survey, we asked respondents to imagine that they have a magic wand and can make any of the candidates president --  they don’t have to beat anyone or win the election -- and then asked who the respondent would choose.

Comparing who respondents would vote for today with who they would choose to “magically” make president reveals the impact of electability concerns for each candidate:

Who likely democrats would vote for if the election were today (grey) compared to who they would make president with a “magic wand” (blue).

Who likely democrats would vote for if the election were today (grey) compared to who they would make president with a “magic wand” (blue).

To understand the attributes contributing to perceived electability, we asked respondents to tell us what their “magic wand” candidate could change to be more likely to win. What supporters believe should be changed to increase electability suggests what supporters see as the primary barriers to greater electability.

We looked specifically for mentions of many qualities frequently discussed as potentially driving electability including ideology, age, race, sexual orientation, and gender.

We compare the frequency of mentions of each attribute by supporters of candidates with different metrics on that attribute. For example, comparing mentions of age by supporters of older candidates, to mentions of age by supporters of younger candidates, sheds light on whether supporters of older candidates see age as a relatively greater driver of electability concerns.

  • Age is mentioned by only 5% of supporters of candidates over 65 years of age suggesting it is not explicitly a primary concern.

  • Ideology is mentioned by 8% of supporters of more progressive candidates and 4% of supporters of more moderate candidates, suggesting ideology is not actually perceived as a major driver of perceived electability.

  • Sexual orientation is mentioned by 9% of supporters of Buttigieg.

  • Race is mentioned by 7% of supporters of both white candidates and candidates of color, showing that voters perceive challenges on both sides of this issue.

  • Gender is mentioned the most of any factor, and by 19% of supporters of female candidates.

  • Gender is mentioned by 24% of respondents who chose a male candidate in the horse race, but a female “magic wand” president, suggesting a heightened concern with gender among these vote switchers.  

This comparison reveals both the overall degree to which any of these attributes are driving electability, and how it impacts different groups of candidates differently.

Frequency of mentions of age, ideology, sexual orientation, race, or gender when “magic wand” supporters of different candidates are asked what that candidate could change to become more likely to be elected.

Frequency of mentions of age, ideology, sexual orientation, race, or gender when “magic wand” supporters of different candidates are asked what that candidate could change to become more likely to be elected.

When it comes to electability, gender plays a unique role:

  • Gender is the most cited attribute

  • Among the respondents who mention gender, 69% are female

  • 19% of the respondents who choose to make a female candidate president given a “magic wand” explicitly reference gender as a perceived electability issue

  • Another 27% of respondents who choose one of the female candidates to be president given a “magic wand,” use highly gendered language or reference stereotypes to describe other attributes like strength and likeability

  • Among the 91% of respondents who select one of the six leading candidates, 63% of respondents choose a male candidate to be president in a typical horse race question

  • When given a magic wand, only 58% of respondents choose a male candidate to be president and Elizabeth Warren emerges as the preference by a 2% margin

Voters doubt the ability of American voters to elect a woman, not the capabilities of the candidate. 62% of those who mention gender explicitly assert a belief that American voters are not capable of electing a woman. Here’s what those statements sound like:

 

“I'm not sure a woman can beat Trump. Honestly the misogyny and racism in this country right now is heartbreaking. I hope one day she’s president though. She is amazing.” 39 year old female from Kentucky

“I don’t know, I don’t think men will vote for her or any woman.” 75 year old female from North Carolina

“She needs to change that she's a woman, because America can't accept truth from a woman when the lies of men are so much more soothing.” 49 year old male from Washington

“She's a woman, so the media will never give her equal time or focus on her plans. They will distort her life, her work, her looks, her ambition, her voice, her passion.” 39, Female, Texas

“Rally liberals to the polls, and eliminate fear that a woman can win the election.” 44, year old male from Florida

 

While the data shows a strong connection between perceived electability and gender, it  also shows that, at the time this survey was conducted, voters were most likely to select a woman for president given the opportunity.

Among those who see gender as a barrier to electability, the pervasive belief isn’t that female candidates are doing anything wrong - but that even if they do everything right, Americans won’t elect a woman. These findings suggest that many voters may not need to be convinced that female a candidate is more capable — they must be convinced that Americans are capable of electing them.

When it comes to perceived electability, the greatest messaging challenge for the campaigns of female candidates may be less about the story of their candidate and more about the story of America today.


 

METHODOLOGY

Avalanche provides deep and accurate insight into the values, emotions, and beliefs that drive behavior on key issues. Our technology leverages both human expertise and natural language processing to quickly and accurately interpret and analyze large volumes of qualitative data. We deliver the depth of a focus group, with the scale of polling.

Since launching in 2017, Avalanche gained the backing of Higher Ground Labs and supported strategy and messaging for leading organizations in the progressive ecosystem including AFL-CIO, NARAL Pro-Choice America, Run for Something, and campaign committees.

An Avalanche Listening Survey combines open and closed-ended questions to deeply understand how people relate to issues. Close-ended questions provide additional clarity and comparability with existing research, while open-ended questions provide rich and wholly novel data sets to understand the beliefs, values, and emotions through which segments of the public are currently relating to issues.  Avalanche works with best-in-class data collection partners to gather targeted representative samples of respondents.

The Civiqs Panel: The data in this report comes from a listening survey of 1,871 registered voters conducted using the Civiqs panel. Avalanche designed a Listening Survey, which Civiqs fielded to their online panel to collect responses. Raw data and weights were then provided from Civiqs to Avalanche for analysis. The Civiqs panel is a nationally representative panel of randomly selected U.S. adults. Civiqs employs a quota sampling procedure to ensure that the sample is demographically representative of the target population. Panelists are randomly recruited and not financially incentivized to complete surveys.

Data Collection Methodology: Data for this report was collected from the panel between May 31- June 3, 2019. To ensure representative samples from many specific subsets in addition to the national sample, the survey was emailed to 11,997 participants reflecting a 15% completion rate. More details on the Civiqs Panel and Targeted Survey methodology can be found here.

Processing Methodology: Avalanche uses a proprietary natural language processing system to analyze open-ended responses. First, important themes are identified through a combination of behavioral theory, computational techniques, and human expertise. These themes may include stories, tone, metaphor, or other elements of language. The chosen themes are then identified at scale with algorithms that ensure accuracy and rigor.

While particular elements may change from project to project, our technology consistently identifies the core aspects of how a person relates to an issue in terms of: priorities (the most salient aspects of the issue), values (the moral reasoning driving position), and emotion (the way a person feels about the issue). We customize the analysis for every project based on the insights of expert human reviewers to dynamically identify emergent themes. This allows humans to define the elements of speech that matter most, and the technology to identify those parts accurately across large datasets.

Survey Sample Detail

The total sample for this survey project is 1,871 registered voters. Civiqs constructed this sample with several key subsets to allow us to compare both likely Democratic supporters to likely Republican supporters, and to examine Democratic primary candidate support within likely Democrats. To this end, the sample includes:

  • 1109 likely Democratic voters (DEM)

  • 628 likely Republican voters (GOP)

  • 143 Non-voters (Not included in analysis)

We define likely party voters based on whether the respondent would vote for a Democrat or Republican if the election were held today. Non-Voters are not included in the analysis and are defined as respondents who say that they would not vote if the election were held today.

While it is not possible to accurately forecast a 2020 electorate with confidence at this stage, we can compare the demographic distribution of this sample to the demographic distribution of the 2016 Electorate according to Pew Research. The sample Civiqs provided for this study closely matches those distributions, and represents a highly plausible electorate.

We break the sample down into Democrats and Republicans, and show the percentage of each category in our sample compared with the percentages reported by Pew. Given the greater sample of Democrats, due to our focus on them in this work, the distribution of Democrats from Civiqs more closely matches the reported distribution.

 
SampleDetail.png
 

Overall, we believe this sample is generally consistent with a reasonable expectation of likely supporters of either party at this early stage in the election. The most significant difference is that this sample under estimates the proportion of women in a likely 2020 electorate, suggesting that findings related to gender may be conservative. As with every poll, this instrument can and should be re-fielded in the future with both expanded and longitudinal samples, as estimates of likely voter demographics solidify leading up to the election.