Voters nationally give Vice President and Democratic nominee Kamala Harris an edge over former President Donald Trump in November’s election by a seven-point margin (50 to 43), but race and gender remains central to the vote. When voters are made to think about the race or gender of the candidates, Harris’ lead grows substantially; when they’re not, support is essentially tied. Harris is also helped by strong support among the slightly less than half of men who reject traditionally masculine identities. Trump’s strongest support is among men who hold traditionally masculine identities, while women and other men strongly favor Harris.

“Trump has built his political career around a very specific performance of whiteness and masculinity,” said Dan Cassino, a professor of Government and Politics at Fairleigh Dickinson, and the Executive Director of the poll. “In the past, that’s been seen as a strength, but it’s no longer clear that it’s working.”

In the overall sample, Harris leads Trump 50 to 43 among likely voters, with 7 percent saying that they will support someone else in November’s election (with undecided voters included, that lead is 47 to 40). Both Trump and Harris have the support of 95 percent of their partisans, and Harris leads Trump 38 to 33 among independents who do not lean towards either party. Similarly, Harris has a strong lead among self-identified liberals (87 to 10), progressives (93 to 5) and moderates (62 to 30). Trump leads among conservatives (76 to 19), and MAGA voters (95 to 4).

To examine the effect of race and gender on vote choice, an experiment was embedded in the survey.

In the survey, before being asked who they were supporting in the Presidential race, respondents were given a list of five issues, and asked which ones were important to their vote. The issues included Tax Policy, Immigration Policy, Climate Change, Abortion and Foreign Policy. But not all respondents got the same list. In addition to randomizing the order of the issues, one-third of respondents were given “The Race or Ethnicity of the Candidate” as the last issue before the vote choice question, and one-third were given “Whether the Candidate is a Man or a Woman” as their last issue. The remaining one-third got all five of the issues, in no particular order.

This survey experiment means that we can compare the voters who were primed to think about race or sex with those who were not, and because the assignment to these conditions is done at random, we can be confident that any differences between the groups are a result of the priming, and not other factors. The effects are enormous.

Among voters who were not primed to think about the race or sex of the candidates, Harris and Trump are tied (47 to 48). When the list of issues mentions the sex of the candidates, Harris pulls ahead, 52 to 42. And when the race of the candidates is mentioned, Harris holds a 14-point lead, 53 to 39, a 15-point shift from the baseline condition.

“When voters are thinking about race or sex, Trump’s support just plummets,” said Cassino. “All the time, we hear strategists and pundits saying that Democratic candidates shouldn’t talk about identity, but these results show that making race and sex salient to voters is bad for Trump and boosts Harris.”

The movement in the race priming condition is largely due to changes in support among non-white voters. Fifty-five percent of non-white voters in the unprimed condition say that they’ll support Harris, with 39 percent supporting Trump. But in the race primed condition, Trump’s support among non-white voters drops by 10 points to 29 percent, while Harris’s support rises by 10, from 55 to 65 percent. All told, mentioning the race of the candidates moves Harris’s lead among non-white voters from 16 points (55 to 39) to 36 points (65 to 29).

This size of the effect is supplemented by a shift away from Trump among white voters in the race primed condition. In the unprimed condition, Trump leads Harris among white voters by 11 points, 53 to 42. In the race primed condition, the two are tied, with Harris marginally ahead among white voters, 47 to 44.

The largest shifts due to this priming condition appear to be among Black voters, but analysis of these effects is limited by the sample size of the survey.

“Race matters in elections, but it’s not inevitable that voters are thinking about it,” said Cassino. “Trump does reasonably well among non-white voters so long as they’re not thinking about race: once they are, we see a huge shift to Harris.”

Mentioning the sex of the candidates also impacts vote choice, with the gender prime moving women’s support away from Trump. Mentioning the sex of the candidates has no real effect on men’s votes: it increases support for Harris by 5 points and reduces support for Trump by 2. But among women, mentioning the sex of the candidates drops Trump’s support by 7, from 40 percent in the unprimed condition to 33 percent in the gender primed condition. The net effect is a change from women favoring Harris by 16 (56 to 40) in the unprimed condition, to favoring her by 26 (59 to 33) in the gender primed condition.

The gender dynamics of the race are also clear from a question asking respondents about how masculine or feminine they consider themselves to be. On this question, a little more than half of men say that they’re “completely masculine,” and a little less than half describe themselves some other way (as “mostly masculine,” “slightly masculine” or in one of the feminine categories). The men who put themselves in the “completely masculine” category favor Trump over Harris by a wide margin, 64 to 30. All other men favor Harris by a 20-point margin, 55 to 35. Among women, “completely feminine” women are little different than other women: both favor Harris by about a 20-point margin. While there is a big difference between men and women overall, that difference is driven entirely by the men who say that they are “completely masculine.” Those men favor Trump; all other sex and gender groups favor Harris by about the same 20-point margin.

“We talk about the gender gap in voting as being between men and women,” said Cassino. “But it’s not. The real gender gap is between men who are holding to traditionally masculine identities, and everybody else. Identity isn’t just about race and sex: Trump’s appeal to a traditional form of masculine identity is the only thing keeping this race close.”

Methodology

The survey was conducted between August 17 and August 20, 2024, using a voter list of registered voters nationwide carried out by Braun Research of Princeton, New Jersey. Respondents were contacted via either live caller telephone interviews, or text-to-web surveys sent to cellular phones, resulting in an overall sample of 801 registered voters nationally. 520 of the surveys were carried out via live caller telephone interviews to landlines (30%) and cellphones (70%) and the remainder (281) were done on a web platform via weblinks sent via SMS to cell phones. Surveys were conducted only in English.

The data were weighted to be representative of the population of voters in the 2020 US Presidential Election, according to data from AP VoteCast. The weights used, like all weights, balance the demographic characteristics of the sample to match known population parameters. The weighted results used here are balanced to match parameters for sex, age, education and race/ethnicity. Individuals were considered likely voters in the upcoming Presidential Election if they (a) were registered voters, who (b) said that they planned to vote in the upcoming election, and (c) had a candidate preference in that election.

SPSSINC RAKE, an SPSS extension module that simultaneously balances the distributions of all variables using the GENLOG procedure, was used to produce final weights. Weights were trimmed to prevent individual interviews from having too much influence on the final results. The use of these weights in statistical analysis helps to ensure that the demographic characteristics of the sample approximate the demographic characteristics of the target population. The size of these weights is used to construct the measure of design effects, which indicate the extent to which the reported results are being driven by the weights applied to the data, rather than found in the data itself. Simply put, these design effects tell us how many additional respondents would have been needed to get the weighted number of respondents across weighted categories: larger design effects indicate greater levels of under-representation in the data. In this case, calculated design effects are approximately 1.3, largely driven by the weights used on the race/ethnicity variable.

All surveys are subject to sampling error, which is the expected probable difference between interviewing everyone in a population versus a scientific sampling drawn from that population. Sampling error should be adjusted to recognize the effect of weighting the data to better match the population. In this poll, the simple sampling error for 801 registered voters is +/-3.5 percentage points, at a 95 percent confidence interval. Including the design effects, the margin of error would be +/-4.6 percentage points, though the figure not including them is much more commonly reported.

This error calculation does not take into account other sources of variation inherent in public opinion studies, such as non-response, question wording, differences in translated forms, or context effects. While such errors are known to exist, they are often unquantifiable within a particular survey, and all efforts, such as randomization and extensive pre-testing of items, have been used to minimize them.

The FDU Poll is a member of the AAPOR Transparency Initiative and is devoted to ensuring that our results are presented in such a way that anyone can quickly and easily get all of the information that they may need to evaluate the validity of our surveys. We believe that transparency is the key to building trust in the work of high-quality public opinion research, and necessary to push our industry forward.

Weighted Telephone Sample Characteristics

801 Registered Voters Nationally

Figures do not include individuals who declined to answer demographic items.

Man                                 48%                 N = 380

Woman                             52%                 N = 412

Some Other Way           >1%                N = 7

18-30                           17%                N = 136

31-44                           23%                 N = 184

45-64                           38%                 N = 304

65+                               22%                 N = 176

White                                            67%                N = 526

Black                                               13%                N = 102

Hispanic/Latino/a                                      13%                N = 102

Asian                                        4%                  N = 31

Other/Multi-racial                                     3%                  N = 24

No college degree                       59%                N = 468

College degree or more              41%                N = 326

Question Wording and Order

First off, we’d like to ask you a few questions about the upcoming Presidential Election

P1. Which of the following would you say are important to your vote in November? For each one, just say if you think it’s important or not.

[All respondents get five of these. One third of respondents get A-E, shuffled. Another third get 4 of A-E, in shuffled order, with F as the last one. The Last third get 4 of A-E, in shuffled order, with G as the last one]

1. Foreign Policy
2. Immigration Policy
3. Abortion
4. Climate Change
5. Tax Policy
6. [One third of respondents get this option] The Race or Ethnicity of the Candidate
7. [One third of respondents get this option] Whether the Candidate is a Man or a Woman

P2. [Shuffle order of top candidates] In November’s Presidential election, do you intend to vote for Kamala [COMMA-lah] Harris, the Democrat, Donald Trump, the Republican, for someone else, or do you not plan on voting?

1. Kamala [COMMA-lah] Harris, the Democrat
2. Donald Trump, the Republican
3. Someone else
4. Not going to vote
5. [Vol] Don’t know/ Refused

P3. In the US today, do you feel that you can openly express your social and political opinions, that you can express your opinions sometimes, or do you have to be careful not to express your opinions?

1. I can only express my opinions
2. I can only sometimes express my opinions
3. I have to be careful not to express my opinions
4. [Vol] Don’t Know/Refused

P4. How commonly do you see your views about social and political issues represented in the media?

1. Often
2. Sometimes
3. Rarely
4. Never
5. [Vol] Don’t Know/Refused

P5. Regardless of where you get your news about politics and current events, which of these sources of information do you consider to be trustworthy?

[You can pick as many as you like]

1. Local newspapers
2. National Newspapers like the New York Times or Washington Post
3. Local TV News
4. National TV News
5. CNN
6. MSNBC
7. Fox News
8. Other Cable News Channels
9. Social Media

P6. Gender roles in the US have been changing rapidly, and we’d like to hear what you think about how men should act today. For each of the following statements, tell me whether you strongly agree or disagree, agree or disagree but not strongly, or have no opinion.

1. Strongly Agree
2. Agree
3. No opinion
4. Disagree
5. Strongly Disagree
6. [Vol] Don’t know/ Refused

1. Men should watch football games instead of soap operas
2. Boys should prefer to play with trucks rather than dolls.
3. A man should always be the boss

[Please pick Disagree in this row]

1. I think a young man should try to be physically tough, even if he’s not big.
2. Men should not be too quick to tell others that they care about them

Intervening questions held for later release

Demos

Just a few more questions, for statistical purposes

D1. In politics today, do you consider yourself a Democrat, Republican, Independent, or something else?

1. Democrat
2. Republican
3. Independent  [ASK D1A]
4. Something Else/Other
5. DK/Ref [vol]

D1A. [Ask only if D1 is 3] Which way do you lean?

1. Democrat

1. Republican
2. Independent
3. Something Else/Other
4. DK/Ref [vol]

D1B. In addition, which of the following terms would you use to describe your political views? You can choose as many as you like. [Shuffle Order]

1. Liberal
2. Moderate
3. Conservative
4. Socialist
5. Progressive
6. Libertarian
7. Make America Great Again or MAGA
8. Nationalist

D2A. To ensure we are reaching people of all ages, would you please tell me your age?

____    (ENTER AGE: 98=98+, 99 = REFUSED)

[IF Don’t Know/REFUSED IN QD1, ASK:]

D2B.  Would you be willing to tell us whether it’s between…?

1. Under 30

1. 31 to 44
2. 45 to 64
3. 65 or over
4. [Refused]

D3. What was the last grade in school you completed? [CODE TO LIST]

1. Did not complete High School
2. High School Diploma or equivalent
3. Vocational or Trade School
4. Some college, but no degree
5. Associates, or other 2 year degree
6. Bachelor’s Degree
7. Graduate work, such as Law, MBA, Medical School, or similar
8. Refused (VOL)

D4. How would you describe your sex? Do you describe yourself as …

1. A Man

1. A Woman
2. Some other way
3. [DK/REF]

D5. How would you describe your racial and ethnic background? You can pick as many as you’d like.

1. White

1. Black
2. Asian
3. Hispanic/Latino/a/Spanish
4. Other or Multi-Racial
5. [Dk/Ref]

Have you ever used financial trading programs or apps, like RobinHood or ETrade to buy and sell stocks, options, forex or cryptocurrency on your own, without using a broker or financial advisor?

1. Yes
2. No
3. [Vol] Don’t Know/Refused

D6. Have you ever owned any cryptocurrency, NFTs, or other similar digital products? It’s fine if you don’t know what those are.

1. Yes
2. No
3. Don’t know what those are
4. [Refused]

D7. The traits that we see as being masculine or feminine are largely determined by society, and have changed dramatically over time. As a result, everyone has some combination of masculine and feminine traits, which may or may not correspond with whether they’re male or female. How do you see yourself? Would you say that you see yourself as…

1. Completely Masculine
2. Mostly Masculine
3. Slightly Masculine
4. Slightly Feminine
5. Mostly Feminine
6. Completely Feminine
7. [Dk/Ref]

Thanks so much for your participation – you’ll see the results in the news in the next few weeks

Release Tables

Note: respondents who did not have a preference or said that they would not vote in the upcoming election were not considered likely voters, and are excluded from these tables.

For purposes of comparison, this set of tables includes respondents not considered likely voters because they said that they would not vote, or did not have a candidate preference