proposal

Analysis of the Trump Vote in the 2024 U.S. Presidential Election

Takeshi Iida

1. Introduction

In the 2024 U.S. presidential election, Republican candidate Donald Trump outperformed Democratic candidate Kamala Harris not only in the number of electors he won but also in the popular voting ratio, realizing his return to the White House. It was the first time since 2004, amid the Iraq war, and only the second time since the 1990s, that a Republican candidate outperformed a Democratic candidate in the popular vote in any U.S. presidential election. Trump also won in all the battleground states of Michigan, Wisconsin, Pennsylvania, North Carolina, Georgia, Nevada, and Arizona. That was astonishing given the growing dominance of liberal ideas about major social issues at stake in American society, such as abortion and same-sex marriage,1 and the continuing rise in the percentage of racial minorities in the population, considered to constitute the Democratic Party's long-standing support base. How was Trump able to win so many votes and get reelected?

According to prior studies (e.g., Sides et al. 2019), Trump's victory in the 2016 presidential election was credited to the shift of white working-class votes from the Democrats to the Republicans in the Midwest battleground states, allowing him to win these states by a narrow margin.2 The initial argument was that economic factors such as widening income disparity may have been the reason why the white working class voted for Trump. Still, subsequent research revealed that racial factors had a greater impact than economic factors. In other words, the main factors inducing white working-class voters to support Trump were antipathy toward racial minorities, antipathy toward immigrants, and the perception that whites were treated unfairly. In addition, given that Hillary Clinton was the Democratic candidate in the 2016 presidential election, misogyny is considered to have contributed to increasing the Trump vote.

Taking into account the findings of previous studies, this paper analyzes the determinants of the Trump vote in the 2024 U.S. presidential election. It also discusses differences in policy opinions between Trump voters and Harris voters, particularly regarding security policy. This paper is a preliminary analysis made less than three weeks after the presidential election; a more detailed analysis will be undertaken by researchers and media outlets in, first of all, the United States.

2. Substantial Increase in the Trump Vote

First, we begin with an overview of the increase in the Trump vote using state-level data. Figure 1 shows the rate of change of the Trump vote by state: Trump has increased votes in his favor in more than two-thirds of the states, winning back six states he lost in the 2020 presidential election. Turnout in the recent presidential election is the second-highest in the postwar period, but it is about two points lower than in the previous election; this suggests that the rise in Trump's votes was due to increased support for him.

Fig. 1: The Rates of Change of Trump's Votes Comparing 2020 and 2024

Note: California, Colorado, Mississippi, New York, Oregon, Rhode Island, and Utah are excluded because the vote count in these states has not been finalized as of November 21, 2024. Vote change rate (%) = (2024 votes / 2020 votes) × 100.

Then, in which states did the Trump vote increase? Figure 2 plots the percentage of the Hispanic population versus the rate of change of Trump's votes in each state. The figure suggests that states with a higher percentage of Hispanic population witnessed a significant increase in Trump votes. This is just an indication of correlation, not causation. However, a similar relationship was not found regarding other racial factors, such as the percentage of the white population or that of the black population.

Furthermore, Figure 3 plots the inflation rates for the immediate one-year period from October 2023 to October 2024 versus the rate of change of Trump's votes; almost no meaningful relationship is identified between them. Figure 4 plots the inflation rates between January 2021, when the Biden administration took office, and October 2024, versus the rate of change of Trump's votes; there is no relationship between them. In other words, as far as these figures suggest, it does not appear that Trump has increased votes in his favor in the states with high inflation rates.

That said, making inferences about individual voting behavior by an analysis of state-level aggregate data is fraught with problems. Above all, the states are too large a unit of analysis. Consequently, analysis in the next section will deal with individual-level data.

Fig. 2: The Percentages of Hispanic Population versus the Rate of Change of Trump's Votes

Note: Hispanic population percentages were calculated using the "Hispanic or Latino" population data included in the 2022 data from the U.S. Census Bureau's American Community Survey.

Fig. 3: Inflation Rates in the Immediate One-year Period versus the Rate of Change of Trump's

Note: The inflation rates in the most recent one-year period were calculated using inflation rate data for October 2023-October 2024 included in the "State Inflation Tracker" run by the Joint Economic Committee of the U.S. Congress.

Fig. 4: Inflation Rates under the Biden Administration versus the Rate of Change of Trump's

Note: The inflation rates in the most recent year were calculated using inflation rate data for January 2021-October 2024 included in the "State Inflation Tracker" run by the Joint Economic Committee of the U.S. Congress.

3. White People's Discrimination Awareness, Misogyny, Approval/Disapproval of Deportation of Illegal Immigrants, and the Trump Vote

The research team led by the author conducted surveys before and after the recent presidential election to obtain individual-level data on American citizens' voting behavior.3 The pre-election survey was conducted on the internet from October 31, 2024, to November 3, 2024, two days before the election, on a sample of 1,500 U.S. citizens aged 18 and older, selected from Dynata's registered monitors in a way to represent a microcosm of the American electorate in terms of gender, age, and area of residence. The post-election survey was conducted with the 1,500 people, who had responded to the pre-election one, from November 6, hours after Harris's concession speech, through November 11. A total of 801 of them responded.

Using these data, the research team conducted a multivariate analysis employing a logistic regression model with vote choice (Trump votes or Harris votes) as the dependent variable.4 The independent variables include race (white or non-white), gender (male or non-male), age, education (college graduate or non-college degree), income (above/below median household income), Democratic identification (yes or others), Republican identification (yes or others), household economic status anxiety (ranging from 1: "Not at all worried" to 5: "Extremely worried"), awareness of the existence of discrimination against blacks (ranging from 1: "None at all" to 5: "A great deal"), awareness of the existence of discrimination against Hispanics (ditto), awareness of the existence of discrimination against Asians (ditto), awareness of the existence of discrimination against whites (ditto), the temperature of feelings toward feminists (1 to 100), the temperature of feelings toward sexual minorities (ditto), and agreeing or disagreeing with the deportation of illegal immigrants (ranging from 1: "Strongly disagree" to 5: "Strongly agree"). Their individual effects were estimated while taking each other's effects into account. The results indicate that while the influence of anxiety about one's household economic situation was not significant, awareness of the existence of discrimination against whites, the temperature of feelings toward feminists, and agreeing or disagreeing with the deportation of illegal immigrants were particular factors that significantly split votes between Trump and Harris.

Figure 5 shows how the predicted probability of voting for Trump changes as voters become aware of discrimination against whites. Regarding the question asking how much discrimination exists against whites in the United States today, respondents who answer "A great deal" tend to vote for Trump. For example, as for respondents who answer that discrimination against whites does not exist at all, their probability of voting for Trump is about 25%; as for respondents who answer that a great deal of such discrimination exists, their probability of voting for Trump rises to over 60%. Figure 6 shows the relationship between awareness of the existence of discrimination against whites and the Trump vote, dividing respondents into white voters and other voters: white voters are more likely than other voters to vote for Trump when they become aware of the existence of discrimination against whites; white voters who answer that there is a great deal of discrimination against whites vote for Trump with a probability of as high as 70%. Since a similar relationship is not observable for voters other than white working-class voters, the awareness of the existence of discrimination against whites encourages not only white working-class voters but also white voters as a whole to vote for Trump.

Fig. 5: Voters' Awareness of the Existence of Discrimination against Whites versus the

Note: Created using data from the research team led by the author. The predicted probability was calculated by fixing other independent variables at their mean values. The grey area represents the 95% confidence interval for the predicted probability estimates.

Figure 7 illustrates how antipathy toward feminists affects the predicted probability of voting for Trump. Respondents were asked about the temperature of their feelings toward feminists by describing it on the feeling thermometer, where 0 represents the coldest feelings toward feminists, 100 the warmest feelings, and 50 neither sympathy nor antipathy. As the temperature of respondents' feelings toward feminists is nearer to 0 or they harbor more intensive antipathy against feminists, the probability of their voting for Trump rises. For example, if a respondent harbors strong antipathy against feminists or the temperature of his/her feelings toward feminists is 0, the probability of voting for Trump exceeds 60%; the probability of voting for Trump drops to just over 20% with those who cherish strong sympathy toward feminists or the temperature of their feelings toward feminists is 100. Figure 8 examines the relationship between the temperature of feelings toward feminists and the Trump vote, dividing respondents into white working-class and others. White working-class respondents are more likely than other respondents to vote for Trump because of their antipathy toward feminists; a portion of them, the temperature of whose feelings toward feminists is 0, has over 70% probability of voting for Trump. From the above, it is apparent that misogyny is a critical factor in splitting votes between Trump and Harris.

Fig. 6: White Voters' Awareness of the Existence of Discrimination against Whites versus the

Note: Created using data processed by the research team led by the author. The predicted probability was calculated by fixing other independent variables fixed at their mean values.

Figure 9 illustrates how the predicted probability of voting for Trump varies following the strength of one's approval of deporting illegal immigrants. When a person endorses the deportation of illegal immigrants, the probability of his/her voting for Trump rises: the likelihood of voting for Trump is over 70% with a person who strongly agrees with deporting undocumented immigrants. Figure 10, which examines this relationship by dividing respondents into white working-class and others, also reveals that among white workers, agreeing or disagreeing with deporting illegal immigrants, in particular, is closely related to whether or not to vote for Trump. In other words, among white workers, the deportation of illegal immigrants is a factor that helps to split votes between Trump and Harris.

Fig. 7: Temperature of Feelings toward Feminists versus the Predicted Probability of Voting for

Note: Created using data processed by the research team led by the author. The predicted probability was calculated by fixing other independent variables at their mean values. The grey area represents the 95% confidence interval for the predicted probability estimates.

Fig. 8: Temperature of Feelings of White Working-Class Voters toward Feminists versus the

Note: Created using data processed by the research team led by the author. The predicted probability was calculated by fixing other independent variables at their mean values. The white working class is defined as white people with no college degree and household income below the median.

Fig. 9: Agreeing or Disagreeing with Deporting Illegal Immigrants versus the Predicted

Note: Created using data processed by the research team led by the author. The predicted probability was calculated by fixing other independent variables at their mean values. The grey area represents the 95% confidence interval for the predicted probability estimates.

Fig. 10: White Working-Class Voters' Agreeing or Disagreeing with Deporting Illegal

Note: Created using data processed by the research led by the author. The predicted probability was calculated by fixing other independent variables at their mean values. The white working class is defined as white people with no college degree and household income below the median.

4. Differences of Opinion over Policy between Trump and Harris Voters

Lastly, we briefly review the differences in opinion over policy between Trump and Harris voters. Figure 11 above shows the distribution of responses from Trump and Harris voters, respectively, to the question, "Should the United States promote economic policies that prioritize U.S. economic interests, even if such policies hurt U.S. relationships with its military allies?" 62.5% of Trump voters answered "Agree," much higher than 40.3% of Harris voters who answered the same. Figure 11 below shows the distribution of responses to the earlier question about agreeing or disagreeing with deporting unauthorized immigrants, presented by voters in favor of each presidential candidate: the percentages of voters supporting each candidate who answered in the affirmative widely vary. "Strongly agree" accounts for the highest ratio of Trump voters' responses; on the other hand, "Strongly disagree" accounts for the highest percentage of Harris voters' responses, thus indicating a complete polarization of opinion between Trump and Harris voters over the deportation of illegal immigrants.

Fig. 11: Differences of Opinion over Policy between Trump and Harris Voters (Distribution of Responses)

Note: Created using data processed by the research team led by the author.

Our research quantified responses about eight policy opinions (including the two above) into five values from 1: "Strongly disagree" to 5: "Strongly agree." Figure 12 shows differences in the mean values of quantified responses between Trump and Harris voters. The horizontal bar represents a 95% confidence interval for the estimated difference in mean values. When it straddles the vertical line representing "0," the difference in the mean values of quantified responses between Trump and Harris voters is statistically significant at the 5% level, at least. Since our research subtracts the mean values of quantified responses of Harris voters from the mean values of quantified responses of Trump voters, a positive difference means that Trump voters agree with a specific policy opinion more than Harris voters; a negative difference means that Harris voters agree with a specific policy opinion more than Trump voters. When examining the policy opinions widely at odds between the two groups, Trump voters are more likely than Harris voters to: oppose free trade, prioritize U.S. economic interests over those of US allies, reject the idea of respecting international treaty obligations at the expense of U.S. national interests, support strict immigration policies, agree with deporting unauthorized immigrants, and oppose accepting refugees.

Fig. 12: Differences in Policy Opinions between Trump and Harris Voters (Mean Values)

Note: Created by using data processed by the research team led by the author. Responses about each policy opinion were quantified into five values from "1: Strongly disagree" to "5: Strongly agree." This figure shows differences in the mean values of quantified responses between Trump and Harris voters. The horizontal bars represent the 95% confidence intervals of the estimated differences in the mean values; when they straddle "0," differences in the mean values of quantified responses between Trump and Harris voters are statistically significant at the 5% level, at least.

The following two figures show Trump and Harris voters' agreeing or disagreeing with U.S. armed intervention under different security situations. Figure 13 above illustrates each camp's voters' reaction to U.S. military intervention to deal with "Russia attacking a NATO ally like Latvia, Lithuania, or Estonia." Although a majority of Trump voters agree with U.S. military intervention in that situation, their percentage is lower than that of Harris voters who support U.S. military intervention under that particular situation; the majority response from Trump voters is "Somewhat agree," suggesting a weak level of agreement. Figure 13 below shows voters' agreeing or disagreeing with U.S. military intervention when China attacks Japan. It suggests that Trump voters are more reluctant to military intervention than Harris voters but without so pronounced a difference between them.

Fig. 13: Difference in Opinion about U.S. Military Intervention between Trump Voters and

Note: Created using data processed by the research team led by the author.

We have quantified responses about U.S. military intervention under seven security situations (including the two above) into five values from 1: "Strongly disapprove" to 5: "Strongly approve." Figure 14 shows differences in the mean values of quantified responses between Trump and Harris voters. Trump voters are more reluctant than Harris voters to intervene in each contingency of "if North Korea attacks South Korea," "if Russia attacks a NATO ally," "if North Korea attacks Japan," or "if China attacks Japan." On the other hand, there is no significant disparity between Trump voters and Harris voters in their approval ratings of U.S. military intervention in each contingency of "if China attacks Taiwan," "if China attacks the Philippines," or "if Iran attacks Israel." To be precise, Trump voters were slightly more supportive of military intervention than Harris voters "if Iran attacks Israel." In general, however, Trump voters are inclined to be more reluctant about military intervention.

Fig. 14 Differences in Opinion about Military Intervention between Trump Voters and Harris

Note: Created by using data processed by the research team led by the author. Responses were quantified into five values from "1: Strongly disapprove" to "5: Strongly approve." The figure shows differences in the mean values of quantified responses between Trump and Harris voters. The horizontal bars represent the 95% confidence intervals of the estimated differences; when they straddle "0," the differences in the mean values of quantified responses between Trump and Harris voters are statistically significant at the 5% level, at least.

5. Conclusion

In this paper, we have confirmed the substantial growth of the Trump vote in the recent presidential election by using state-level data, analyzed drivers of the Trump vote based on individual-level data, and examined differences in policy opinions between Trump and Harris voters. The findings include the following: Trump substantially increased votes in his favor in more than two-thirds of the states; increased awareness of discrimination against white individuals in American society, growing antipathy toward feminists, and a rising support for the deportation of illegal immigrants combined to motivate many voters to cast their ballots in favor of Trump; Trump voters seek more than Harris voters to pursue US interests in the international society and are more reluctant than Harris voters to intervene in armed conflict―not that the majority of Trump voters are opposed to intervention. These findings are consistent with those from previous studies.

Why was Trump able to increase votes in his favor in the recent presidential election to outperform Harris in popular votes? One possible explanation is Harris's attributes. The fact that Harris is a person of color and a woman may have helped amplify feelings of discrimination against whites (i.e., the white consciousness that whites are unfairly treated compared to other racial minorities) and the influence of misogyny on voting behavior. The problem of illegal immigration, which worsened under the Biden administration, may also have helped increase voters in favor of deporting illegal immigrants and, as a result, cultivate new Trump votes.

However, the validity of these claims needs to be verified by a comparative study with the 2020 presidential election. In this sense, they are hypothetical. We look forward to a more in-depth comparative analysis by researchers and the media employing causal inference methods.

(Professor, Doshisha University)

References

  • Carnes, Nicholas and Noam Lupu. 2021. "The White Working Class and the 2016 Election." Perspectives on Politics 19(1):55-72.
  • Sides, John, Michael Tesler, and Lynn Vavreck. 2019. Identity Crisis: The 2016 Presidential Campaign and the Battle for the Meaning of America. Princeton University Press.
proposal
current topics
letter