Driven to Extremes: Donald Trump's Extraordinary Impact on the 2020 Elections PDF

ARTICLE Driven to Extremes: Donald Trump’s Extraordinary Impact on the 2020 Elections GARY C. JACOBSON Reactions to the divisive person and presidency of Donald Trump dominated voting decisions in the 2020 elections, which consequently set new records for electoral continuity, party loyalty, nationalization, polarization, and presidential influence on the down-ballot vote choices, to the point where local factors such as incumbency, candidate quality, and campaign spending barely registered in the congressional election results. Evaluations of Trump’s leadership in the national crises besetting the United States in 2020 were assimilated almost entirely into existing attitudes toward the president, limiting the impact of these events. The Democrats achieved unified control of the government, but by the narrowest of margins, and the political configuration that had emerged from Trump’s election in 2016 remained largely intact. Republicans’ responses to Trump’s seditious efforts to nullify Biden’s victory promise to extend his influence through the 2022 elections and beyond. The 2020 elections extended several long-term trends in American electoral politics that were driven to new extremes by the singularly divisive person and presidency of Donald J. Trump. The election set new records for electoral continuity, party loyalty, national- ization, polarization, and presidential influence on the down-ballot vote choices, to the point where local factors such as incumbency, candidate quality, and campaign spending barely registered in the congressional election results. Never before have national election results been shaped so comprehensively by a single political figure. Although the election occurred amidst severe national crises brought on by the COVID-19 pandemic and police killings of Black citizens, responses to the questions they raised about Trump’s leader- ship were assimilated almost entirely into existing attitudes toward the president and, by extension, the parties, and therefore did little to alter the political configuration that had emerged from Trump’s election in 2016. My purpose in this article is to document these points with a variety of aggregate and survey data that, as will become evident, Gary C. Jacobson is a Distinguished Professor of Political Science Emeritus at the University of California, San Diego. He specializes in the study of U.S. elections, parties, campaign finance, public opinion, and Congress. His most recent book is Presidents and Parties in the Public Mind. 1 Presidential Studies Quarterly DOI: 10.1111/psq.12724 Vol. 0, No. 0, May 2021, 1–30 © 2021 Center for the Study of the Presidency and Congress 2 | JACOBSON TABLE 1 Results of the 2020 Federal Elections Democratic Candidate Republican Candidate President 2016 Popular vote 65,852,514 62,984,828 Popular vote percentage 48.2 46.1 Electoral vote 232 306 2020 Popular vote 81,268,924 74,216,154 Popular vote percentage 51.3 46.9 Electoral vote 306 232 Republicans Democrats Independents House of Representatives Prior to the 2020 election 201 233 1 Elected in 2020 213 222 Incumbents reelected 166 205 Incumbents defeated 0 13 Open seats retained 33 13 Open seats lost 3 1 Senate Prior to the 2020 election 53 45 2* After the 2020 election 50 48 2* Incumbents reelected 14 10 Incumbents defeated 4 1 Open seats retained 3 1 Open seats lost 0 0 *The Senate independents caucus with the Democrats. Source: Compiled by author. point uniformly to Trump’s pervasive influence on the 2020 elections. I also consider how Republicans’ responses to Trump’s seditious efforts to nullify Joe Biden’s victory might extend his influence through the 2022 elections and beyond. Table 1 summarizes the election results. Joe Biden outpolled Trump by 7.05 mil- lion votes, giving him a margin of 4.4 percentage points in the popular vote (compared to Hillary Clinton’s 2.86 million vote and 2.1-point margins in 2016), flipping five states and thereby inverting the 2016 Electoral College count for a 306– 232 victory.1 Republicans nonetheless gained seats in the House of Representatives, reducing the Democrats’ majority to single digits. This was not, however, a consequences of wide- spread ticket splitting; the reality, documented below, was very much the opposite. Democrats also assumed control of the Senate when they won the two Georgia runoff elections held in January 2021 to tie the Senate at 50-50, with Vice President Kamala Harris serving as tiebreaker. The 2020 election thus delivered a unified Democratic gov- ernment, but barely. If Trump had won three states where Biden’s margin was less than a 1. The five states were Arizona, Georgia, Michigan, Pennsylvania, and Wisconsin. Driven to Extremes | 3 percentage point, he would still be president;2 without Trump’s post-election assault on the integrity of the vote, the tight Georgia runoffs might easily have gone the other way (Blake 2021); and had Republicans won any five of the six House seats where the winning Democrat’s vote share fell below 51.2%, they would hold a majority in that chamber. With only small changes in the vote totals, the Republicans’ structural advantages in states, districts, and the Electoral College could have delivered a unified Republican government.3 The Trump Effect The 2020 elections were dominated at all levels by popular reactions to Donald Trump, which have been deeply divided along party lines from the onset of his 2016 can- didacy through his time in office and afterward. Why Democrats and Republicans devel- oped such antithetical and intensely held opinions of Trump is an important topic in its own right, and I have reviewed some explanations elsewhere (Jacobson 2020a, 2020b). I will not repeat that exercise in this article; the focus is rather on showing how profoundly these divergent opinions of Trump, however derived, shaped the 2020 elections. Partisan differences in assessments of presidents have been growing for four decades, to be sure, but Trump drove them to unprecedented extremes. Opinions of a president’s job performance tend to be most polarized during the year the president seeks reelection;4 2020 was no exception, and Trump outdid all of his predecessors in this regard, inspiring an ap- proval gap averaging 85 points in Gallup Polls taken in 2020 (approval by 91% of Republicans, 6% of Democrats) and 10 points wider than the previous election-year record set by George W. Bush and matched by Barack Obama (Figure 1). The gap approached its arithmetic limit to reach 92 points in the October 16–27 Gallup Poll, with 95% of Republicans but only 3% of Democrats approving of Trump’s performance.5 Surveys prob- ing the strength of approval or disapproval also found very large Republican majorities approving “strongly” and even larger Democratic majorities disapproving “strongly.”6 Extreme polarization coincided with and contributed to an unusually stable distri- bution of opinion on Trump over the course of his presidency; neither good news (e.g., very low unemployment and a record-setting stock market) nor bad news (e.g., the at- tempt to extort Ukraine that led to his first impeachment, revelations of payoffs to silence paramours, or the lethally deficient response to the COVID-19 pandemic) moved the 2. Had Biden not won Arizona, Georgia, and Wisconsin, the electoral vote would have been tied 269–269, and Trump would have prevailed in the House of Representatives, where Republicans controlled a majority of state delegations. 3. The advantage derives from their more efficient distribution of voters across states and districts, augmented by partisan gerrymandering (Jacobson and Carson 2020) and the overrepresentation of low- population rural states in the Senate and thus the Electoral College. 4. This is true for every president since Gerald Ford. 5. See https://news.gallup.com/poll/323234/national-satisfaction-rises-election-nears.aspx. 6. In the October 2020 Quinnipiac Poll, for example, 81% of Republicans approved strongly, whereas 93% of Democrats disapproved strongly; see https://poll.qu.edu/natio nal/relea se- detai l?Relea seID=3681. 4 | JACOBSON 100 90 80 70 Partisan Difference 60 50 40 30 20 10 0 Ford 1976 Eisenhower 1956 Johnson 1964 GHW Bush 1996 Trump 2020 GW Bush 2004 Obama 2012 Nixon 1972 Carter 1980 Reagan 1984 Clinton 1996 FIGURE 1. Average Partisan Difference in Presidential Approval, Reelection Years, 1956–2020 (Gallup Polls). Source: Compiled by author. needle much on Trump; his overall approval ratings varied by only a few points around the 40% mark over his entire time in office, the least variation for any president on record (FiveThirtyEight 2021; Jacobson 2021a). Polarized views of Trump bred polarized views of the parties, candidates, and parti- sans. Opinions of Trump shaped opinions of his party and its rival more strongly than had opinions of any of his recent predecessors (Jacobson 2019a, 2020b), and consequently, partisan differences in evaluations of the parties widened to record levels. For example, the partisan gap in favorable opinions of the congressional parties had averaged 48, 45, and 47 percentage points, respectively, during the Clinton, G. W. Bush, and Obama presidencies; under Trump, the gap grew steadily, from 52 points in 2017 to 72 points in 2020.7 Negative partisanship—hostility to the rival party as measured by the American National Election Study (ANES) 100-point feeling thermometers (Abramowitz and Webster 2018)—also reached a new maximum by the end of Trump’s presidency; in the pre-election ANES survey for 2020, partisans rated the rival party on average an icy 20 degrees, about 10 degrees below the previous low reported in 2016.8 Thermometer rat- ings of the rival party’s House candidates were also the lowest, party differences in 7. Calculated from 646 surveys by ABC News/Washington Post, CBS News/New York Times, Fox News, Gallup, NPR/Marist, Morning Consult, NBC News/Wall Street Journal, Pew, Quinnipiac, and Economist/YouGov polls. 8. Based on data in the ANES Cumulative Data File and the ANES 2020 time-series pre-election wave; the data series begins in 1978; data are available at https://electionstudies.org/data-center/. Driven to Extremes | 5 TABLE 2 Correlation of Vote with the Previous Presidential Election, 1972–2020 Year States House Districts Richard Nixon 1972 0.414 1976 0.063 0.612 Jimmy Carter 1980 0.842 0.906 Ronald Reagan 1984 0.792 1988 0.862 0.963 George H. W. Bush 1992 0.783 Bill Clinton 1996 0.899 0.952 2000 0.928 0.970 George W. Bush 2004 0.967 2008 0.934 0.971 Barack Obama 2012 0.977 2016 0.939 0.963 Donald Trump 2020 0.993 0.989 Note: House district correlations cannot be calculated across redistricting years. Source: Compiled by author. thermometer ratings of the presidential candidates were the widest,9 and correlations between the thermometer ratings of the presidential candidates and their parties were the highest for any of the 11 elections covered by ANES data. Acute polarization had been essential to Trump’s victory in 2016. Although many Republicans were less than enthusiastic about their nominee (Trump received colder ther- mometer ratings from his own partisans than any previous Republican candidate), their much greater dislike of Hillary Clinton and her party minimized defections (Jacobson 2017). Once in office, Trump’s conduct and policies drove partisans even further apart, hardening the mutual antipathy between people who embraced his vision of what con- stituted American greatness and people who were repelled by it (Deane and Gramlich 2020; Pew Research Center 2019). Trump governed as divisively as he had campaigned, resulting in extraordinary continuity between the first and second elections with him on the ballot. Continuity with 2016 Despite Trump’s tumultuous presidency and the massive influx of new voters it inspired (Table 1), the rival electoral coalitions backing him and Hillary Clinton in 2016 reappeared, albeit with a few of their salient elements strengthened, in 2020. Both aggre- gate and survey data reveal unusually high levels of continuity between the two elections. Table 2 displays the correlations between the vote shares cast in the current and previous 9. Republicans rated Biden at 23 degrees, higher than Hillary Clinton’s 16 degrees in 2016; Democrats also rated Biden higher, 80 degrees compared to 68 degrees. Democrats rated Trump at 7 degrees, 10 degrees colder than in 2016; Trump’s ratings among Republicans was 75 degrees, up 16 degrees from 2016. Thus, the partisan gap was 57 degrees in 2020, up from the previous record of 46 degrees set in 2016. 6 | JACOBSON ϭϬϬ ϵϬ ϴϬ ϳϬ ƌс͘ϵϴϵ ϲϬ WĞƌĐĞŶƚϮϬϮϬ ϱϬ ϰϬ ϯϬ ϮϬ KƚŚĞƌ^ƚĂƚĞƐ ŝƐƉƵƚĞĚ^ƚĂƚĞƐ ϭϬ Ϭ Ϭ ϭϬ ϮϬ ϯϬ ϰϬ ϱϬ ϲϬ ϳϬ ϴϬ ϵϬ ϭϬϬ WĞƌĐĞŶƚϮϬϭϲ FIGURE 2. District-Level Presidential Vote, 2016 and 2020. Source: Compiled by author. presidential election across states and, for years when redistricting does not preclude com- parisons, congressional districts in elections since 1972. The presidents seeking reelection are listed by name. These correlations have generally risen over the past half century but reached remarkable new highs in 2020 of 0.993 and 0.989 in 2020. A plot of the district-level major party presidential vote in 2020 against the 2016 vote (Figure 2) illustrates this continuity but also reveals a handful of informative out- liers. (Districts in states where Trump sought to nullify the results are highlighted in red for later discussion.) Six districts where Biden’s vote was well below Clinton’s vote are highlighted with green markers. These districts, three in South Florida and three in South Texas, were heavily Hispanic (between 70 and 83% of district populations), and Trump’s gains among Hispanics in these states was one notable departure from 2016. Above average shifts away from Trump were most common in upscale suburban districts. Thus, despite extraordinary geographic continuity, the demographic composition of the presidential coalitions was not entirely static across the two elections. Table 3 displays the share of votes won by Trump in 2016 and 2020 among various demographic subgroups according to the CCES (2016) and CES (2020) surveys (Ansolabehere and Schaffner 2017; Schaffner, Ansolabehere, and Luks 2021). The breakdowns are very similar for the two elections, but with a few notable differences (those exceeding +/− 2.0 percentage points are in bold). The comparisons suggest that 4 years of Trump largely reinforced patterns set in 2016. His support among well-educated white women, already low, fell further, while his support among voters with no more than a high school education grew further. As Driven to Extremes | 7 TABLE 3 Demographics of the Vote for Donald Trump, 2016 and 2020 2016 2020 Difference 2016 2020 Difference All 48.4 47.5 −0.9 White Voters Men 54.2 52.0 −0.4 High School or Less 67.5 69.4 1.8 Women 44.5 43.4 −1.1 Men 69.6 70.7 1.1 White 57.3 55.3 −2.0 Women 65.1 68.0 2.9 Men 60.5 59.3 −1.2 Some College 58.9 59.9 1.0 Women 54.6 51.6 −3.0 Men 61.2 63.4 2.2 Black 8.6 8.9 0.3 Women 57.0 56.8 −0.2 Men 13.6 13.1 −0.5 College Graduate 45.1 41.9 −3.2 Women 4.4 5.6 1.2 Men 49.3 47.8 −1.5 Hispanic 29.4 32.5 3.1 Women 40.8 36.5 −4.3 Men 31.9 36.9 5.0 Women 26.9 28.4 1.5 Born-again Christians 80.7 82.4 1.7 Other 41.7 40.3 −1.4 Other 47.0 41.4 −5.6 Men 50.1 44.7 −5.4 Women 32.3 36.4 4.1 Republican 93.7 95.4 1.8 Liberal 5.5 3.7 −1.8 Independent 64.4 53.0 −11.1 Moderate 38.0 36.0 −2.0 Democrat 7.9 3.3 −5.6 Conservative 85.2 91.1 5.9 Source: 2016 Cooperative Congressional Election Study, 2020 Cooperative Election Study. suggested by the aggregate data, Trump gained support among Hispanic voters, espe- cially men. However, in contrast to some reports based on exit polls, Trump’s support among Blacks did not, according to these data, grow at all, although he once again did considerably better among Black men than among Black women. His support was great- est among Black males under 45—17%—but this was only a point higher than in 2016.10 On the whole, the data point to basic stability in presidential coalitions across the two elections. Women were responsible for the Democrats’ majority vote in both, with a gender gap only slightly smaller in 2020 than in 2016. Whites favored Trump, nonwhites Biden. The education gradient among whites, already steep, became a little steeper. The age gradient (not shown) was about the same in both elections, with Trump’s support 23 points higher among voters over 65 than among voters under 30; in 2016, the difference was 22 points. Trump’s support among white born-again Christians, his strongest demographic, rose 1.7 points to reach 82%; among other white voters it was down 5.6 points. These and other survey data (Jacobson 2021a; Pew Research Center 2020a) confirm the image of Trump’s Republicans as the party of older, white, religiously observant, less educated, and male voters residing outside metropolitan areas, and Biden’s Democrats as the party of younger, nonwhite, secular, better educated, and female voters living in urban areas and metropolitan suburbs. The rival coalitions became a bit more 10. Trump’s support among voters in the “other” ethnic category also declined slightly, but with notable differences across gender categories, which show an anomalous shift among men to Biden and among women to Trump. This is probably an artifact of the improbably large gender gap reported for this category in 2016 (nearly 18 points) regressing to a more plausible size. 8 | JACOBSON 100 90 80 70 60 Percent 50 40 30 20 10 0 1952 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 ANES Face-to-Face CES Economist/YouGov Exit Polls ANES Social Media Study FIGURE 3. Party Loyalty in Presidential Elections, 1952–2020. differentiated along these lines over the course of Trump’s presidency, contributing to the shifts in the ideological composition of the presidential coalitions shown in the last section of the table. Compared to 2016, a smaller proportion of self-identified liberals and moderates and a larger proportion of conservatives voted for Trump in 2020, expanding an already yawning ideological gap. Party loyalty increased between the two elections, especially among Democrats, reaching its highest level ever measured in 2020. Pure independents (those not leaning toward either party) report the largest drop in support for Trump (−11 points), making an important if limited contribution to the swing to Biden, as they comprised only about 7% of the 2020 voters in this survey.11 Trump’s ratings among independents in major polls taken in September and October averaged only 37%, accounting for his loss of sup- port from this group. The growth of party loyalty to unprecedented levels in 2020 is confirmed by mul- tiple surveys. Figure 3 displays the proportion of major party voters supporting their own 11. For comparison, the Economist/YouGov post-election poll (November 8–10, 2020 (https://docs. cdn.yougov.com/9j7sr0my95/econTabReport.pdf), shows a larger drop (−15 points), and the National Exit Poll (https://www.cnn.com/election/2020/exit-polls/president/national-results), which treats leaners as inde- pendents, shows a smaller drop (−7 points). I am obliged to Douglas Rivers for making the Economist/YouGov surveys available for secondary analysis. Driven to Extremes | 9 TABLE 4 Consistency of Presidential Vote with Previous Vote and Presidential Approval (Percent) Year Vote Same as Recalled Vote Vote Congruent with Approval 1972 80.8 84.0 1976 78.2 80.0 1980 79.1 81.5 1984 86.1 88.9 1992 89.1 85.6 1996 90.5 89.4 2004 90.2 92.2 2012 91.1 93.8 2020 94.7 94.9 Source: 1972–2012: ANES Time Series; 2020: ANES Social Media Study. party’s candidate rather than the rival party’s candidate in presidential elections since 1952. The ANES series covers 1952 through 2016 (the last year with face-to-face inter- views), the exit poll series covers 1980 onward, and the CCES series 2008 onward. Results from the Economist/YouGov post-election survey are added for 2016 and 2020, and ANES Social Media Study (American National Election Studies 2021) results are included for 2020. The data tell a very consistent story, with party loyalty ranging from 94.8 to 96.1 and averaging 95.4% in 2020; the previous highest average, in 2016, was 93.6%.12 Two other data series provide further evidence of unusual electoral stability and coherence in 2020. The first column in Table 4 lists the proportion of voters whose cur- rent vote matches the vote they remember casting 4 years earlier in elections involving presidents seeking election since 1972. Obviously, voters’ memories may be imperfect, and a need for cognitive consistency will lead some people to “remember” voting for their current favorite, but assuming that this would apply in every election, comparisons over time can be informative. Fewer voters reported changing their vote in 2020 than in any previous election in the series. The second column reports the congruence between pres- idential approval and the presidential vote (for the president if approving, for the chal- lenger if disapproving). A great deal of congruence is of course expected, but in earlier decades—specifically, 1972–1992 in these data—averages of 4.6% of voters disapproved of a president they nonetheless voted for, and 11.2% approved of a president they none- theless voted against. There were very few such conflicted voters in 2020 (2.5 and 2.7%, 12. This result was anticipated by pre-election horserace polls, with party loyalty averaging 96.5% from July onward (N = 44). I use the ANES Social Media Study rather than the ANES Time Series Study for data on voting in this article because some serious anomalies in the study’s preliminary release raise doubts about its accuracy. For example, 10.1% of respondents report changing their House vote preference between the pre-and post-election waves, a far larger percentage of supposed switchers than in any other comparable study; the average for respondents in the 2012 and 2016 ANES face-to-face respondents, the 2008–2020 CCES, and the ANES Social Media Study is 3.9%, with a maximum of 6.3%. The study also reports far higher levels of partisan defection in the House elections than any other poll, 12.5%. No other post-election poll has House defection rates exceeding 5.0%; in 137 pre-election polls, the average was 3.4%, with a max- imum in any survey of 6.8%. The levels of instability and partisan defection in the ANES Time Series Study are so thoroughly at odds with both the electoral climate and all other data for 2020 that I opted not to use its vote measures in this article. 10 | JACOBSON respectively).13 Trump’s sub-50% approval rating among 2020 voters was thus instru- mental to Biden’s national majority.14 The COVID-19 Effect The degree of continuity and partisan consistency in 2020 are all the more remark- able for an election that took place during the massively disruptive crisis generated by the COVID-19 pandemic that arrived earlier in the year. By November, the disease had killed 230,000 Americans and upended social and economic lives everywhere. Trump addressed the pandemic with characteristic deceit, callousness, and denial of responsibility (Jacobson 2020a), and his administration’s response to the crisis was deficient on multiple fronts (Abutaleb et al. 2020). A substantial majority of the public evaluated Trump’s handling of the pandemic negatively,15 and the Biden campaign made his failings a major target. Yet, the issue’s potency was sharply limited by partisan priors. The great majority of Republicans maintained their habitual resistance to information that might call their support for Trump into question and rated his handling of the crisis positively, with av- erage approval of it rising from 77% in July to 84% in October, only a few points below his very high overall approval ratings at the time. Democrats were overwhelmingly crit- ical, so reactions to Trump’s pandemic performance were largely assimilated into the ex- isting polarized configuration of opinions of Trump’s presidency.16 That said, and despite the prevalence of party-line voting, the pandemic left some tangible traces in survey data on presidential voting in 2020. Table 5 displays the margin- als and cross-tabulations for opinions of Trump’s handling of the pandemic and partisan voting in the ANES Social Media Study and the post-election Economist/YouGov survey. Partisan loyalty was very high in both surveys, especially among Democrats. Evaluations of Trump’s pandemic performance were also decisively partisan, although a larger pro- portion of Republicans than Democrats dissented from their party’s consensus. Partisan defections were heavily concentrated among these dissenters. Those sharing their party’s consensus almost never defected (0.3–1.3%), while substantial portions of the dissenters did so (51.6–30.8% among Republicans, 47.5–69.8% among Democrats). As a conse- quence, dissenters comprised large majorities of defectors (65.8–96.9%, depending on the survey). In short, although partisan disloyalty was very rare in 2020, its occurrence 13. Vote congruence with the remembered 2016 vote and with approval was even higher in the CCES, at 95.7% and 96.9%, respectively; in the post-election Economist/YouGov poll, the figures were 94.6% and 96.3%. 14. His approval rating was 42% among voters in the ANES Social Media Study and 47% in the CCES and Economist/YouGov surveys. 15. On average, in 140 polls that asked the question between July and election day, 39.0% approved and 56.8% disapproved of Trump’s handling of the pandemic; see Jacobson (2021a, Figure 4). Data are from ABC News/Washington Post, CBS News, CNBC, CNN, Fox News, Ipsos, Kaiser, NPR/Marist, Morning Consult, NBC News/Wall Street Journal, Quinnipiac, Suffolk, and YouGov/Yahoo polls. 16. In 19 polls with partisan approval data on both handling of the pandemic and overall approval taken during October and early November, the respective Republicans’ approval ratings were 84 and 89%. The comparable averages among Democrats were 5 and 6%, and among independents, 35 and 39%. Driven to Extremes | 11 TABLE 5 Opinions of Trump’s Handling of the COVID-19 Pandemic and the Partisan Vote in 2020 (Percent) Republicans Democrats Approve Disapprove Approve Disapprove ANES Social Media Study 85.3 14.7 2.8 97.2 Loyal Partisans 92.2 85.0 7.1 96.8 1.9 96.4 Defectors 7.8 0.3 7.6 3.2 1.0 0.8 % Defecting 0.3 51.6 69.8 0.8 % of Defectors 3.1 96.9 65.8 34.2 Economist/YouGov Poll 88.0 12.0 5.3 94.7 Loyal Partisans 95.0 86.7 8.3 97.0 2.8 94.2 Defectors 5.0 1.3 3.7 3.0 2.5 0.5 % Defecting 1.5 30.8 47.5 0.5 % of Defectors 25.6 74.4 84.5 15.5 Source: ANES Social Media Study; Economist/YouGov poll of November 8–10, 2020. was strongly related to views of Trump’s handling of the pandemic, with the net effect benefiting Biden. Trump’s inept handling of the pandemic harmed him less than it might have be- cause Republicans who viewed it negatively defected only if this also led to negative views of his overall performance. In the Economist/YouGov survey, about half the Republicans disapproving of his pandemic performance nonetheless approved overall; all of them reported voting for Trump, whereas two-thirds of the rest voted for Biden. In the ANES Social Media Study, about 35% of Republicans disapproving on the pandemic nonetheless approved of Trump’s overall performance; 91% of them voted for Trump, compared to 25% of the rest. Trump’s ability to retain the good opinion of a substantial portion of Republicans who were critical of his handling of the COVID-19 crisis thus limited its damage to his cause.17 For similar reasons, Trump’s comparatively high approval rating on managing the economy proved to be of limited benefit to his cause. During his time in office, Trump’s ratings on the economy typically ran 10–15 points higher than his overall ratings among all voters.18 Among Democrats, his approval rating on “jobs and the economy” in the post-election Economist/YouGov survey, at 14% (matching the average from 20 July– November election-year polls), was twice as high as his overall approval rating, 7%. Of those who disapproved, 0.3% reported a vote for Trump; of those who approved, 18% 17. Among the pure independents in these two surveys, very large majorities voted consistently with their views of Trump’s handling of the pandemic: 88.5% in the Economist/YouGov survey and 93.8% in the ANES Social Media Study. The majority of independents gave his performance negative reviews, contribut- ing to their net swing toward Biden; on average, in October through early November surveys, 35.3% of in- dependents approved and 58.5% disapproved of Trump’s handling of the pandemic. 18. See the data at https://pollingreport.com/trump_ad.htm. 12 | JACOBSON voted for Trump, comprising 90% of disloyal Democrats in this survey. But approval on the economy helped Trump only if it translated into broader approval; among the Democratic “jobs and economy” approvers, 31% also approved generally, and 54% of them voted for Trump; of the remaining 69% who disapproved generally, 2% voted for Trump. Thus, the economy, which had been doing well before the pandemic and was still a plus for Trump despite the hit it absorbed, generated few crossover votes from Democrats because it was overshadowed by their negative overall views of the president. It did, how- ever, discourage Republican defections; nearly 96% of Republicans approved of Trump’s economic performance, and 98% voted for him. This closely paralleled the reinforcing effect of opinions of Trump’s handling of the COVID-19 crisis among Democrats; nearly 95% disapproved, and virtually all of them voted for Biden. The more general point of this brief examination of the electoral effects of two sa- lient national concerns is not that Trump’s handling of the pandemic and the economy were not major considerations for voters in 2020, but that voters’ evaluations of his per- formance were so thoroughly dominated by prior opinions of Trump that they were, with very limited exceptions, simply assimilated into the existing configuration of partisan attitudes and therefore electoral preferences. The same holds for opinions of Trump’s han- dling of “civil rights and civil liberties” after the racial justice protests sparked by police killings of Blacks in the spring of 2020. The signal events of 2020 thus reinforced far more than they undermined existing partisan loyalties. Letting Trump Be Trump The Biden campaign’s exploitation of Trump’s lethally deficient handling of the pandemic (Lancet 2021) was part of a broader strategy aimed at keeping Trump front and center. Biden won the nomination by convincing Democrats that, whatever else they thought about him, he would be the strongest candidate against Trump, an ar- gument fully backed by the data (Jacobson 2021a). Once Biden won the nomination, shared loathing for Trump united the Democrats behind his candidacy, with none of the discord that had plagued Clinton in 2016. Biden’s campaign presented him as the anti- Trump, emphasizing the contrasts in character, temperament, competence, and respect for institutions and science along with fundamental differences on national policy. Biden patiently let the focus remain on Trump, who obliged him by conducting an ugly, men- dacious, and polarizing campaign that turned out his admirers in great numbers but his revilers in even greater numbers (Jacobson 2021b). In Pew’s post-election survey, 54% of Biden voters said their vote was mainly against Trump rather than for Biden, a finding that should neither surprise nor upset Biden’s campaign strategists. The more the elec- tion was a referendum on Trump, the better for Biden. Starkly polarized responses to Trump dominated down-ballot as well as presi- dential voting, producing record levels of party loyalty and straight-ticket voting and an extremely close connection between presidential and congressional voting at both aggregate and individual levels. Both the House and Senate election results were, with only a few exceptions, tightly bound to Trump’s popular standing and electoral Driven to Extremes | 13 1 0.9.987.95.95 0.8.86.83.85.83.84.80 0.7.76 ͘ϲϴ.71.70 0.6.67.67 Correlation.65.62 0.5.54 0.4 0.3 0.2 0.1 0 1952 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 FIGURE 4. Correlations between Congressional and Presidential Voting at the District Level, 1952– 2020. Source: Compiled by author. performance in the districts and states. Despite the staggering sums spent pursuing congressional victories, the results echoed the vote for the top of the ticket far more than any local factors. The House Elections At the aggregate level, the impact of the presidential contest on House election results in 2020 can scarcely be overstated. To begin, the correlation between the district- level House and presidential vote, which had been growing for two decades, reached its highest point ever in 2020, an astonishing 0.987 (Figure 4).19 Figure 5 plots the relation- ship between the variables for 1980, 2000, 2012, and 2020 to illustrate just how much tighter the link was in 2020 compared to earlier elections, including the previous record holder, 2012. Second, as Figure 6 shows, the proportion of districts delivering split outcomes—majorities for the presidential candidate of one party and for the House can- didate of the other party—reached a postwar low in 2020. Only 16 districts produced split outcomes— seven won by House Democrats, nine by House Republicans— amounting to a mere 4% of the total. Third, the 2020 House elections were the most nationalized in history. A simple measure of nationalization is provided by the standard 19. The correlation is between the two-party presidential vote and the major party House vote in districts with elections contested by both major parties. 14 | JACOBSON ŝƐƚƌŝĐƚ>ĞǀĞůWƌĞƐŝĚĞŶƟĂůŶĚ,ŽƵƐĞsŽƚĞ͕ϭϵϴϬ ŝƐƚƌŝĐƚ>ĞǀĞůWƌĞƐŝĚĞŶƟĂůŶĚ,ŽƵƐĞsŽƚĞ͕ϮϬϬϬ ϭϬϬ ϭϬϬ ϵϬ ϵϬ ,ŽƵƐĞǀŽƚĞ;йĞŵŽĐƌĂƟĐͿ ,ŽƵƐĞǀŽƚĞ;йĞŵŽĐƌĂƟĐͿ ϴϬ ƌс͘ϲϳ ϴϬ ƌс͘ϴϬ ϳϬ ϳϬ ϲϬ ϲϬ ϱϬ ϱϬ ϰϬ ϰϬ ϯϬ ϯϬ ϮϬ ϮϬ ϭϬ ϭϬ Ϭ Ϭ Ϭ ϭϬ ϮϬ ϯϬ ϰϬ ϱϬ ϲϬ ϳϬ ϴϬ ϵϬ ϭϬϬ Ϭ ϭϬ ϮϬ ϯϬ ϰϬ ϱϬ ϲϬ ϳϬ ϴϬ ϵϬ ϭϬϬ WƌĞƐŝĚĞŶƟĂůsŽƚĞ;йĞŵŽĐƌĂƟĐͿ WƌĞƐŝĚĞŶƟĂůsŽƚĞ;йĞŵŽĐƌĂƟĐͿ ŝƐƚƌŝĐƚ>ĞǀĞůWƌĞƐŝĚĞŶƟĂůŶĚ,ŽƵƐĞsŽƚĞ͕ϮϬϭϮ ŝƐƚƌŝĐƚ>ĞǀĞůWƌĞƐŝĚĞŶƟĂůŶĚ,ŽƵƐĞsŽƚĞ͕ϮϬϮϬ ϭϬϬ ϭϬϬ ϵϬ ϵϬ ,ŽƵƐĞǀŽƚĞ;йĞŵŽĐƌĂƟĐͿ ,ŽƵƐĞǀŽƚĞ;йĞŵŽĐƌĂƟĐͿ ϴϬ ƌс͘ϵϱ ϴϬ ƌс͘ϵϴϳ ϳϬ ϳϬ ϲϬ ϲϬ ϱϬ ϱϬ ϰϬ ϰϬ ϯϬ ϯϬ ϮϬ ϮϬ ϭϬ ϭϬ Ϭ Ϭ Ϭ ϭϬ ϮϬ ϯϬ ϰϬ ϱϬ ϲϬ ϳϬ ϴϬ ϵϬ ϭϬϬ Ϭ ϭϬ ϮϬ ϯϬ ϰϬ ϱϬ ϲϬ ϳϬ ϴϬ ϵϬ ϭϬϬ WƌĞƐŝĚĞŶƟĂůsŽƚĞ;йĞŵŽĐƌĂƟĐͿ WƌĞƐŝĚĞŶƟĂůsŽƚĞ;йĞŵŽĐƌĂƟĐͿ FIGURE 5. District-Level Presidential and House Vote, 1980. District-Level Presidential and House Vote, 2000. District-Level Presidential and House Vote, 2012. District-Level Presidential and House Vote, 2020. deviation of the major party vote swing from the previous election across stable, contested districts: The smaller the standard deviation, the more uniform the swing across districts, and thus the more nationalized the election. As Figure 7 shows, the standard deviation of the swing between 2018 and 2020 is the smallest yet observed: 3.0.20 As with the district presidential vote (Table 2), the correlation between the current and prior district vote for representative in 2020 was the highest ever recorded, 0.983; the previous high was 0.968 in 2018. As elections have grown increasingly nationalized and the presidential choice has become such a dominant electoral force, the value of incumbency, measured in vote shares, has dropped to a mere 1.5 percentage points, one-sixth what it was at the beginning of 20. Kawato (1987, 1247) lists the variances of inter-election district vote swings across election pairs from 1844 through 1980; taking the square roots of these variances gives us their standard deviations, and none is as small as the figure for 2020. The measure cannot be computed for years ending in “2” because re- districting destroys comparability. Driven to Extremes | 15 250 192 196 200 145 148 150 139 143 130 124 114 110 103 100 86 84 83 59 50 35 26 16 0 1952 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 FIGURE 6. Number of Districts with Split Results in Presidential and House Elections, 1952–2020. Source: Ornstein, Mann, and Malbin (2008, Tables 2–16); data for 2008–2020 compiled by author. the century (Figure 8).21 For the same reason, the effect of candidate quality as measured by previous experience in elected office has also atrophied; its estimated value was only 0.7 percentage points in 2020, the lowest for the postwar period (Figure 9).22 Although the incumbency advantage measured in votes was small, House incum- bents were very successful in 2020, with incumbent Republicans winning all of their general election contests and incumbent Democrats winning 94% of theirs. This was not the result of incumbency, however, but of running in districts that favored their party in presidential voting. Nine of the 13 losing Democratic incumbents represented districts won by Trump in 2020. All three Democratic open-seat takeovers were in districts where Biden outpolled Trump, so that, altogether, 13 of the 17 seats that changed party hands in 2020 went to the party of the district presidential winner. Of the 50 open seats con- tested in 2020, the party of the presidential winner took 49. Individual Voting Survey data confirm the extraordinarily powerful influence of the presidential vote on the House vote in 2020. Pre-election surveys suggested opinions of Trump would have a huge impact on the individual vote, with about 93% of prospective voters reporting 21. The value of incumbency is estimated here by a modified version of the Gelman-King index, which substitutes the district-level presidential vote in the current or, for midterms, most recent presidential election for the lagged vote, allowing years ending in “2” and districts redrawn between apportionment de- cades to be included; for details, see Jacobson (2015). 22. For an explanation of how quality effects are estimated, see Jacobson and Carson (2020, 66). 16 | JACOBSON ϭϮ ϭϬ ϴ ^ƚĂŶĚĂƌĚĞǀŝĂƟŽŶ ϲ ϰ Ϯ Ϭ ϭϵϰϲ ϭϵϰϴ ϭϵϱϬ ϭϵϱϮ ϭϵϱϰ ϭϵϱϲ ϭϵϱϴ ϭϵϲϬ ϭϵϲϮ ϭϵϲϰ ϭϵϲϲ ϭϵϲϴ ϭϵϳϬ ϭϵϳϮ ϭϵϳϰ ϭϵϳϲ ϭϵϳϴ ϭϵϴϬ ϭϵϴϮ ϭϵϴϰ ϭϵϴϲ ϭϵϴϴ ϭϵϵϬ ϭϵϵϮ ϭϵϵϰ ϭϵϵϲ ϭϵϵϴ ϮϬϬϬ ϮϬϬϮ ϮϬϬϰ ϮϬϬϲ ϮϬϬϴ ϮϬϭϬ ϮϬϭϮ ϮϬϭϰ ϮϬϭϲ ϮϬϭϴ ϮϬϮϬ FIGURE 7. Nationalization of Congressional Elections, 1946–2020 (Standard Deviation of the Vote Swing in Contested House Elections). Source: Compiled by author. House preferences congruent with their opinion of the president (for Republicans if ap- proving, for Democrats if disapproving).23 Such a high degree of congruence is unprece- dented; the previous record holder was Obama, at 87% (Jacobson 2019b). Surveys also found very few voters inclined to split their tickets. In Pew’s October survey, only 4% of voters planned to vote for either Biden and a House Republican or Trump and a House Democrat (Pew Research Center 2020b). In the 26 weekly Economist/YouGov surveys taken between June and November that asked the same question, an average of 97.2% of respondents who reported preferences chose the same party’s candidate in both.24 Straight- ticket voting intentions were of course strongly linked to party loyalty. In the pre-election Economist/YouGov polls, prospective defections among Democratic identifiers averaged only 3.7% for president and 2.9% for the House; the respective figures for Republican identifiers were 5.3% and 3.3%. Similarly low defection rates were reported in other pre- election polls. Consistent with these projections, post-election surveys reported record levels of straight-ticket voting and party loyalty in 2020. Figure 10 displays incidence of straight- ticket voting in elections going back to 1952 in the ANES time series (face-to-face 23. The average is from 114 Morning Consult, Economist/YouGov, Emerson College, and George Washington University surveys. 24. The Economist/YouGov results are reported at https://today.yougov.com/topics/economist/surve y-results. Percent DŽĚŝĮĞĚ'ĞůŵĂŶͲ

Driven to Extremes: Donald Trump's Extraordinary Impact on the 2020 Elections PDF

Document Details

Tags

Related

Summary

Full Transcript