IT IS NOT EXCITING, but after nearly $7 billion in campaign spending, dramatic debates, a global pandemic, a summer of protests for racial justice, and a growing phalanx of conspiratorial misinformation clogging social media, the factors that political scientists call “the fundamentals” did a good job predicting the outcome of the 2020 presidential election. A predictive model from Peter Enns and Julius Lagodny, made more than one hundred days before the election, correctly predicted the results in forty-nine of fifty states, missing only Georgia, a state that surprisingly flipped to Joe Biden’s column on the heels of the voter registration and organizing work of voting rights activist and Georgia politician Stacey Abrams.1 While news coverage and pundits focus on things like candidate personality, strategy, and swing voters, much of what predicts the end result of a presidential campaign is beyond candidates’ control.
There are a variety of ways to predict election results with “structural” variables such as presidential approval, economic growth, aggregated polls, and other fundamental aspects of politics and social life. Table 2-1 summarizes many of the major political science forecasts made months before Election Day. Forecasters use data from these fundamental measures of politics and the economy to predict the election, often before each party has even selected its nominee. The average number of electoral votes predicted for President Donald Trump by the political scientists doing the forecasting was 237, only 5 more than he actually received. The average prediction of his two-party vote total was 47.8 percent; he received 47.7 percent.
Table 2-1
Source: Alan I. Abramowitz, “How Did the Political Science Forecasters Do?,” in Sabato’s Crystal Ball, ed. Larry J. Sabato, Kyle Kondik, and J. Miles Coleman (University of Virginia Center for Politics, December 3, 2020), https://centerforpolitics.org/crystalball/articles/how-did-the-political-science-forecasters-do/.
This chapter examines how the economic and political conditions scholars use give us a baseline from which to understand elections in American politics, how the rules that govern elections can influence the outcome, and the basic strategic choices candidates and other interested groups make in an election season. Thus, this chapter serves as a lens through which we can think about the structural conditions that help us understand how the political behavior of the American electorate works. Everything else that we analyze in this book—political partisanship, public opinion, group characteristics and social identity, the media, and the elements of individual vote choice—does not occur in a vacuum, but takes place in a specific context that, when properly understood, can help us to make fairly accurate predictions about election results months before any ballots are cast. As is the case throughout the book, in this chapter we will use examples from the 2020 election and data showing trends over time to illustrate our points.
Learning objectives for Chapter 2 include:
· Understanding how fundamentals like the state of the economy and presidential approval are related to the presidential vote
· Learning how the administrative “rules of the game” influence how elections are conducted
· Exploring how electoral structures influence campaign strategy
· Engaging with how campaigns deal with unexpected factors that have the potential to interact with or even upend “the fundamentals”
· Lay the groundwork to assess how fundamentals interact with other factors to affect election results
THE FUNDAMENTALS
For decades, the state of the American economy has been the most celebrated fundamental factor related to presidential vote choice, and 2020 provided a particularly complicated test of the relationship between the economy and vote choice. The COVID-19 pandemic led the nation’s gross domestic product (GDP) to an astronomical 31.4 percent drop in the second fiscal quarter of 2020 and an equally huge 33.4 percent spike in the third fiscal quarter of 2020.2 Typically, change in the nation’s GDP ranges between −3 percent and +7 percent. How voters would apply their experience of the yo-yoing economy to their evaluations of President Trump was a great unknown heading into the 2020 election.
In nonpandemic times, presidents are keenly aware of how the performance of the economy is tied to their electoral fortunes. About one month into his first term as commander in chief, Barack Obama said on the Today show, “If I don’t have this [improving the economy] done in three years, then there’s gonna be a one-term proposition.”3 As far back as 2014, political scientists argued that the slowly growing economy and President Obama’s relatively low approval rating at the time gave Democrats an “uphill battle” to 270 electoral votes.4 By the beginning of 2016, the economy had picked up slightly and President Obama’s approval was north of 50 percent, but neither measure had improved enough to make most political scientists think the 2016 election was much more than a coin flip. Political scientist Larry Bartels wrote that the 2012 race was a coin flip election that the Democrats won, while the 2016 race was a coin flip race the Republicans won.5
In a world of campaign war rooms, instantaneous media coverage, billions of dollars of television advertising, dramatic debates, raucous campaign rallies, gaffes, special interest spending, and messaging, can a few elements of data correctly predict the winner of almost every election of the past sixty years, months before any ballots are counted?
Historically, the American people have responded to a growing economy by awarding the incumbent or the incumbent’s political party another term in the White House. A clear relationship exists between growth in the nation’s GDP (and/or real disposable income per capita, or RDI) and the percentage of the two-party vote earned by the incumbent’s party in presidential elections. People generally reward the occupant of 1600 Pennsylvania Avenue at the ballot box when the economy is on the rise. Figure 2-1 shows a linear relationship with a growing GDP from January to June of the election year and an increase in the incumbent party’s share of the two-party presidential vote. If we look at just 1956, 1964, 1972, 1980, 1984, 1992, 1996, 2004, and 2012 (the years the incumbent president himself was seeking reelection), the relationship is even stronger than it is for years when a presidential hopeful from the same party is running for president.
There are exceptions to this rule. Figure 2-1 shows that the economy was growing in 1976 when Republican president Gerald Ford lost to Democrat Jimmy Carter. In 1992, the economy was actually rebounding out of the recession that doomed President George H. W. Bush’s reelection bid when third-party candidate Ross Perot achieved a strong showing and Bill Clinton rode his “it’s the economy, stupid” campaign strategy into the Oval Office. But in general, a growing economy is good for the president seeking reelection. Indeed, of the sixteen presidential elections examined in Figure 2-1, only five saw the incumbent or incumbent’s party earn less than 50 percent of the popular vote if the economy was growing at all.
Using second-quarter GDP change in 2020 (−31.4 percent), President Trump, even though he lost, dramatically overperformed what we would expect of him. But using third-quarter GDP growth (+33.4 percent), the president dramatically underperformed what we would expect, given the previous seventy years of presidential elections. Given the incredibly large GDP shifts in 2020, we do not include 2020 data in Figure 2-1 as it would skew how we interpret the previous seven decades of data.
Not all economic indicators are created equal, however. GDP and RDI are generally good estimators that help predict election results. Some popular data providing evidence about the economy are less well suited to the task of helping students of American politics understand elections. For example, hundreds of stories were written about how the nation’s high unemployment in 2011 and 2012 spelled disaster for Obama. However, unemployment has what former New York Times election blogger Nate Silver called a “maddening” relationship with election results. For example, unemployment was relatively high (over 7 percent) for Ronald Reagan’s popular and electoral vote thumping of Walter Mondale during Reagan’s reelection victory in 1984. Unemployment rose from the beginning of the terms to Election Day for Dwight Eisenhower’s 1956 reelection victory, Richard Nixon’s reelection in 1972, and George W. Bush’s reelection in 2004.6 Unemployment actually dropped during Jimmy Carter’s term, but he was not able to defend the White House from Reagan’s advances.
Description
Figure 2-1 GDP Growth and the Popular Vote for Presidential Candidates of the Incumbent’s Party, 1952–2020
Sources: GDP data from the U.S. Department of Commerce Bureau of Economic Analysis, National Economic Accounts, Gross Domestic Product, at http://www.bea.gov/national/; presidential voting data from John Woolley and Gerhard Peters, The American Presidency Project, at https://www.presidency.ucsb.edu/statistics/elections.
Unemployment rates are certainly an important indicator about the health of the economy, but they are not highly correlated with presidential election results. Trump actually performed better in 2020 in locales with high unemployment.7 In 2012, media coverage regularly pointed out that one “reason” that Obama was in trouble was that no president in the post–World War II era had ever won reelection with an unemployment rate above 7.4 percent. Of course, and much to our dismay, no one in the history of humankind has ever won any election after the Minnesota Vikings won the Super Bowl.8 In other words, some claims are factually accurate without being substantively meaningful.
Gas prices are also a popular indicator of the state of the economy. Republican presidential primary candidate Newt Gingrich went as far as to promise that gas prices would drop to $2.50 per gallon if he were elected president, something that even HBO’s The Newsroom mocked in its dramatic, fictional treatment of the 2012 campaign in which the anchorman character Will McAvoy pointed out the president does not control gas prices. On a more serious note, political scientist John Sides has argued that while rising gas prices can affect presidential approval, they are not a major player in predicting presidential elections.9
The volatile changes in the economy, caused by the national response to the COVID-19 pandemic, meant that any Republican presidential candidate would have a difficult path to victory in 2020. But the state of the economy is not the only fundamental factor that helps us predict election results. In July 2020, political scientists Olivia Quinn, Amanda Brush, and Eric R. A. N. Smith noted, “Without a significant change in Trump’s approval rating and a swift recovery of the pandemic-depressed economy, Republicans seem unlikely to prevail in November.”10 In other words, presidential elections are not just a function of economic performance; they are also structured by the nation’s approval of the job the president is doing.
Presidential approval plays a key role in forecasting elections. It should not be terribly surprising that public approval of the job the president is doing has something to do with whether the president gets a chance at a second term. Figure 2-2 shows the relationship between presidential approval in an election year and the incumbent party’s share of the two-party vote. If you recall from Figure 2-1 that there were five elections in which a growing economy was not enough for the incumbent party to earn 50 percent of the popular vote, a quick look at Figure 2-2 highlights the fundamental importance of presidential approval as an independent indicator of presidential elections. In 1952, Democratic president Harry Truman left office with a 32 percent approval rating, which helped drag down the effect of a growing economy and allowed Republican Dwight Eisenhower to win the election. It is true that in 1992, President George H. W. Bush lost some support to the surprisingly strong third-party candidacy of Ross Perot, but he was also hurt by a dismal approval rating. Gerald Ford, who was appointed (not elected) to be vice president and ascended into the Oval Office after Richard Nixon resigned in disgrace, presided over a booming growth in GDP, but his pardoning of Nixon and problems with high inflation sank his approval rating and his bid for a presidential victory in his own right. Hubert Humphrey has been blamed for underperforming as a candidate, but his commander in chief, Lyndon Johnson, had the approval of only 41 percent of the country in 1968.
Description
Figure 2-2 Presidential Approval and the Two-Party Presidential Vote for the Incumbent Party, 1952–2020
Source: Presidential approval and vote share data are from John Woolley and Gerhard Peters, The American Presidency Project, Presidential Job Approval, http://www.presidency.ucsb.edu/data/popularity.php.
Percentage of two-party vote for incumbent presidential party data from John Woolley and Gerhard Peters, The American Presidency Project, https://www.presidency.ucsb.edu/statistics/elections.
With respect to presidential approval in 2020, President Trump peaked at 49 percent—with only 43 percent approving of his job performance on Election Day. While presidents typically see their approval rating rise on the way to their reelection bid, President Trump’s dropped from a high point in May to rest between 39 and 43 percent from June to Election Day.11 In 2016, President Obama’s approval rating rose steadily during his last year in office. While an unpopular George W. Bush was largely kept off the campaign trail in 2008, the popular Obama was a key surrogate for Hillary Clinton, visiting several swing states in the waning days of the election season. Clinton regularly referenced her time in the Obama administration as secretary of state, trying to tie her candidacy to Obama’s service and legacy. Political scientist Drew Linzer suggested that the 2016 election forecasting showed the election was a normal one as Clinton netted almost exactly the percentage of the two-party vote as would be expected given President Obama’s approval rating.12 Figure 2-2 shows the vote share of all voters that presidential candidates of the president’s party earned as compared to the approval rating of the president. Looking at the two-party vote as a function of presidential approval suggests Trump slightly underperformed in 2020.
In addition to economic fundamentals and those related to approval of the president, political fundamentals can influence election results. This is especially the case in congressional elections. Putting election results into the proper context requires understanding the political structure heading into the election and the complexion of the electorate within electoral boundaries, such as a congressional district or a state.
For an example of the political structure’s influence on elections, we can turn to the 2010 congressional contests. The Democratic Party went into the 2010 midterms with a solid majority, holding 255 seats in the 435-member House of Representatives. Democrats had taken back both houses of Congress in 2006 and brought their party back into the White House in 2008. However, one political fundamental of congressional elections is that both success in previous elections and the size of a party’s majority in a house of Congress tend to beget more vulnerable seats in the next election. For example, forty-eight Democratically represented House districts voted for John McCain in the 2008 presidential election. Districts in which the congressional representative is of a different party from the district’s preferred presidential candidate are generally the hardest to hold in midterm contests.
These “swing districts” are an example of how the partisan complexion of a district can affect election results. Since they often have an incumbent who represents a district in which the majority of voters identify with the opposite political party, more qualified challengers are often inspired to enter the race. In competitive districts in 2010, the Republican Party fielded more quality challengers (those who have previously held elective office) than ever before. Over 39 percent of Democratic representatives holding competitive seats faced a quality Republican challenger in 2010, compared to 28 percent in 2008. Regardless of whether an incumbent faces a quality challenger, the fundamental truth is that a district’s general ideology holds great predictive power in congressional elections. As we noted in our supplement to this volume, Political Behavior in Midterm Elections, 2015 edition,13 simply knowing whether a congressional district voted for Obama or McCain, and nothing else, correctly predicted the results of 85 percent of the races for the House of Representatives in 2010.
In 2020, changes in real gross national product in the first and second quarters, the president’s approval rating in July, and the fact that it was a presidential year predicted modest gains in the House and Senate for the Democrats, albeit with quite a bit of variance.14 In the end, Democrats captured the Senate majority, gaining 3 seats, and held onto the House majority despite losing 12 seats.
THE RULES OF THE GAME
One of us authors likes to play pickup basketball with fellow faculty members, graduate students, and understanding undergraduates who don’t mind a slower game consisting of older players with rapidly diminishing skills. He is most valuable to his team when the game rewards three-point baskets (rather than counting all baskets by ones, which is a conventional way of scorekeeping in pickup games). This is because he is slow-footed, middle-aged, short, and a loafer on defense. When the rules of the game reward long-distance shooting, the one element of the game in which he has some skill, his teams are more likely to win. The rules of the game are important in the confines of pickup basketball at public universities, but they are even more important (and more consequential!) in the world of elections.
Election rules affect access to the ballot, the ability to participate in debates, the schedule of primary elections, the date of the general election, who can run for office, and what type of vote is most important (the popular vote or that of the Electoral College). Rules are especially important in primary elections. The Democratic and Republican Parties award their delegates differently, which leads to different campaign strategies in primary elections for candidates of both parties. The Democrats award their delegates proportionately (so long as the candidate wins above 15 percent of the popular vote in the primary) so that a candidate who wins 35 percent of the vote will win 35 percent of the delegates. Thus, it is in Democratic candidates’ interest to campaign hard in every state in which they might win enough of the vote to earn some delegates. The Texas primary/caucus is so unique that even though Hillary Clinton won more votes than Barack Obama in the 2008 Texas primary, his strong performance in the caucus there resulted in Obama winning more delegates. Some analysts attributed Obama’s defeat of Clinton to his campaign’s more strategic understanding of the party’s rules regarding the awarding of delegates, especially in states that held caucuses rather than primaries. Democrats also can win “superdelegates”—typically, lawmakers and high-ranking officials in the party. These superdelegates are not bound to the popular vote in their state’s primary when deciding which candidate they will vote for at the Democratic National Convention. Clinton dominated her chief competitor, Bernie Sanders, in superdelegates by a margin of 570½ to 44½.
Some Republican primaries are winner take all. That is, whoever wins the most votes gets all the delegates. Other Republican primaries elect to award delegates on a proportional basis. The different rules for these primaries can affect the number of visits a candidate makes to a state, the size of an advertising buy in that state, and the media momentum a candidate might enjoy (or suffer) after all the votes are counted.
Regardless of how delegates are awarded, Marty Cohen, David Karol, Hans Noel, and John Zaller have persuasively argued that party leaders typically unite behind a candidate, and when they do, that candidate becomes the nominee.15 To be sure, there are times when party leaders stay on the sidelines, perhaps wary of the risk involved in supporting a candidate who falters, but in general, even in the era of primary elections and caucuses, “the party decides” who their nominee will be for president. Even though the rules of party nominations have changed a great deal, the somewhat hidden “fundamental” of endorsements from party insiders plays a crucial role in selecting nominees for president.
Of course, authors Cohen et al. of The Party Decides took a bit of a beating in 2016 because the Republican Party establishment did not want Donald Trump to be their party’s nominee. This simple fact led many to declare the “party decides” thesis to be dead. The question we should ask as political scientists is whether 2016’s Republican primary results suggest that the “party decides” theory is incorrect or whether there were other factors at play that uniquely contributed to the results. Presidential scholar Julia Azari argued that the Republican Party simply decided not to decide.16 The Grand Old Party (GOP) fielded a large number of qualified candidates for president, and no single candidate ran away with the endorsement primary. The party never coalesced around Jeb Bush, Scott Walker, John Kasich, or Ted Cruz. Political parties scholar Seth Masket reminded election watchers that the “party decides” thesis was alive and well on one side of the aisle.17 Democratic Party elites clearly settled on Hillary Clinton—and she won her closely contested race with Bernie Sanders. They did the same thing in 2020, preferring Joe Biden in the crowded field.
Perhaps no other rule shapes presidential campaigns as much as the awarding of Electoral College votes. Forty-eight states award electoral votes on a winner-take-all basis. Nebraska and Maine give a candidate two electoral votes for winning the state’s popular vote and one electoral vote for each congressional district the candidate wins. In 2008, Barack Obama won the Second Congressional District in Nebraska, but lost the Cornhusker State’s popular vote and the contests in the First and Third Congressional Districts, giving him one electoral vote from Nebraska to John McCain’s four. Obama’s 2008 campaign manager David Plouffe called the history-making result in Nebraska’s Second Congressional District his favorite electoral vote. In 2016, Maine awarded one electoral vote to Donald Trump, who won the Second Congressional District, and Hillary Clinton won the statewide contest and the First Congressional District, netting three electoral votes. In 2020, once again Trump won an electoral vote in Maine while Biden collected Nebraska’s second district for the Democrats.
Given the Electoral College’s structure, candidates spend the most amount of time in swing states that possess a large number of electoral votes. Even the most casual observer of American politics can name a few of the states that regularly enjoy a high volume of candidate visits and an even higher volume of campaign advertising—states such as Ohio, Florida, Wisconsin, Pennsylvania, Michigan, and Virginia. A candidate must win 270 electoral votes to become the president. However unlikely it may be, given that there are 538 total electoral votes up for grabs, a 269 to 269 stalemate is possible.
Table 2-2 shows that while there is a correlation between electoral votes and popular votes, these two measures of campaign success are not the same. For example, John F. Kennedy earned more electoral votes in 1960 than Jimmy Carter did in 1976 and George W. Bush did in 2004 even though Kennedy won less of the popular vote. Bill Clinton won more electoral votes in 1992 than presidents in seven other elections—all seven of whom got a higher percentage of the popular vote than he did. Of course, you will recall that Clinton’s share of the popular vote was diminished as he beat President George H. W. Bush and strong third-party candidate Ross Perot in that election. One reason political scientists are far more skeptical about declaring that a winning presidential candidate has a “mandate” is that Electoral College blowouts, like the ones in 2012, 1996, 1992, and 1980, mask much narrower victories in the popular vote. In 2016, Donald Trump’s big Electoral College win came while losing the popular vote, just as George W. Bush’s victory in 2000 did.
Table 2-2
Source: John Woolley and Gerhard Peters, The American Presidency Project, available at https://www.presidency.ucsb.edu/statistics/elections.
Campaign finance rules are also important to understand. Individuals are limited in the amount of money they can give to a campaign, but candidates face no such limit, in terms of how much they spend or how much they can give themselves. Some, notably John Kerry in 2004 and Mitt Romney in 2008, have spent substantial amounts of their own money in (unsuccessful) bids for the White House. Donald Trump spent $66.1 million of his own money in his successful White House bid in 2016. Self-financed candidates have been more successful in congressional races; Jon Corzine of New Jersey spent nearly the same amount, about $62 million, of his own money to get to the Senate in 2000. Despite Corzine’s victory, self-funded candidates are by no means guaranteed victory. Wrestling mogul Linda McMahon has spent nearly $100 million in unsuccessful bids for U.S. Senate seats in Connecticut.
In fact, enormous campaign spending by incumbent members of Congress, whether they are self-financed or financed conventionally by individuals and political action committees (PACs), is negatively correlated with vote totals. This is because the more members of Congress have to spend to get reelected, the more likely it is that they represent a swing district, are facing a stiff challenge, or both. In other words, the more money one has to spend to hold onto a congressional seat, the more danger that seat tends to be in during an election. On the other hand, plenty of incumbents who spend a ton of money do win reelection, and spending too much money is far more preferable than not having enough to compete.
Though the rules of the game are difficult to change, the times that they are altered introduce greater uncertainty into our ability to understand how elections will work. One example of a recent major change comes from a controversy commonly referred to as “Citizens United.” The Federal Election Commission (FEC) had objected to the airing of Hillary: The Movie, a movie critical of Hillary Clinton, within thirty days of the 2008 presidential primaries. The FEC claimed that the movie, produced by a nonprofit organization called Citizens United, was electioneering and thus prohibited by law to air the movie that close to an election. The Supreme Court’s decision in Citizens United v. Federal Election Commission determined that while corporations and unions are not literal humans, they do have the constitutional right to free speech afforded to “associations of persons.” Thus, banning corporations and unions from spending money to engage in electioneering at any point in a campaign was unconstitutional.
Citizens United, and other subsequent decisions from the court, opened the door to major spending from PACs and “super PACs,” such as Restore Our Future (which supported Mitt Romney) and Priorities USA Action (which supported Barack Obama), to the tune of $142 million and $65 million, respectively, in 2012.18 PACs are outside groups not officially connected with a campaign. Spending was up in congressional races too, though it should be noted that independent expenditures began their rapid rise after the Bipartisan Campaign Reform Act (BCRA) in 2002 and not after Citizens United. Regardless, it is worth asking whether all the spending was consequential. The early returns on that question suggest that spending did not do much to affect the result of the 2020 presidential election.
Figure 2-3 shows the total presidential and congressional expenditures in millions of dollars from 1976 to 2020. Before the BCRA was passed, the average increase in spending from one congressional election to the next was $34 million. The average increase after the BCRA was $83.75 million. As Figure 2-3 shows, the difference on the presidential side is even greater.
Description
Figure 2-3 Total Presidential and Congressional Campaign Expenditures (in Millions), 1976–2020
Sources: Presidential campaign expenditures data from OpenSecrets, at http://www.opensecrets.org/pres16/totals.php, and for 2016 “2016 Presidential Race,” OpenSecrets, at http://www.opensecrets.org/pres16/index.php; congressional campaign expenditures data from “House Campaign Expenditures, 1974–2016,” Campaign Finance Institute, available at http://www.cfinst.org/data.aspx, “Elections Overview” at https://www.opensecrets.org/elections-overview?cycle=2018, and “2020 Presidential Race” at https://www.opensecrets.org/2020-presidential-race/.
STRATEGY
Even if candidates can accurately assess whether the economic, partisan, and structural fundamentals favor them or their opponents, candidates and their advisers need to develop and execute strategies designed to help them win elections. Increasingly, these strategies—at least at the presidential level—appear to be rooted in empirical evidence regarding what drives voter turnout (see Chapter 3), who is most likely to be persuaded to change their vote, and where the most efficient use of television advertising might be located.
Given the crucial role that economic conditions play in affecting presidential and congressional elections, successful candidates are more likely to rely on economic messages when the economic fundamentals favor them and focus on meeting other criteria when the prevailing economic headwinds are bad news for their campaigns. When candidates ignore a growing economy (recall Figure 2-1), as Vice President Richard Nixon did by focusing on foreign policy in his 1960 campaign against John F. Kennedy, opponents can use their media megaphone, as Kennedy did, to turn a potential strength into a problem area. Kennedy’s focus on the decline in school quality, housing issues, and welfare helped propel him to victory, according to Lynn Vavreck.19 Candidates are often adept at ignoring issues they would rather not discuss in favor of issues that the campaign believes are a greater area of strength. As we note in Chapter 7, candidates even do this in presidential debates, doing their best to reverse course on the topic of a question about a strategically disadvantageous issue to discuss an issue that fits the campaign’s theme or area of strength.
Deciding on the issue content of campaigns is important at the congressional level as well. While constant press attention, millions of dollars in attack ads, and the like make it difficult for presidential candidates to completely ignore a major issue on the campaign trail or in a debate, candidates in lower-visibility races have a much easier time focusing on a small set of issues that are good for their campaign. Jamie Druckman and his colleagues have shown, via an analysis of campaign websites, a stunning lack of engagement between congressional campaigns on issues.20 Simply put, candidates do not spend much time engaging in a back-and-forth debate over the same issues. Rather, they focus on an almost completely different set of issues, making it more difficult for the public to sort through which candidate is the best fit. Candidates tend to avoid engaging their opponent unless the issue is so salient to the voters that avoiding the issue would be campaign suicide.
How do candidates know upon which issues to base their campaigns? In Chapter 7, we discuss the issue ownership and trait ownership hypotheses, which demonstrate that the public believes that one party handles particular issues better or exudes particular traits more than the other party. If candidates can get people to focus on the issues their party owns, they can improve their electoral outcomes. While it is difficult to steal issues from the opposing party, it is far easier to steal traits to improve their electoral fortunes. For example, though Democrats are viewed as the more empathetic party, George W. Bush ran for office in 2000 as a “compassionate conservative.” While Republicans get higher marks from the public on the trait of strength, Barack Obama was fond of saying on the campaign trail, “We never throw the first punch, but we’ll throw the last.”21 In addition to searching for an issue focus, candidates work hard to get out the vote (GOTV). The Obama campaign’s GOTV efforts in 2012 relied on the analysis of “big data,” massive voter files that were merged with other bits of data exploring potential voters’ previous voting behavior, party identification, spending habits, and, if known, favored candidate in 2012. As discussed in Chapter 3, after the campaign made predictions about which supporters were least likely to vote, volunteers were sent to knock on those people’s doors, call them to remind them to vote, and send them mail that was tailored to their particular profile. The campaign even ran experiments to see whether volunteers were more successful at persuading undecided voters if the volunteer callers read a script about Obama’s economic performance or asked the undecided voters which issues they wanted to discuss (the economic script performed better). It has been harder to discern the role of data analytics in the 2016 elections. Donald Trump employed a company previously used by failed GOP presidential candidate Ted Cruz to help the campaign identify potential voters via their personality characteristics and discourage potential opponents from turning out to vote. Hillary Clinton’s campaign relied on an algorithm called “Ada” that ran hundreds of thousands of simulations of the battle between Clinton and Trump, helping to guide the campaign’s behavior. The performance of the algorithm was spotty, as it correctly identified Pennsylvania as an important state to visit (Clinton lost the state Democrats had won from 1992 to 2012), but it did not send Clinton to Wisconsin—a state Trump captured for Republicans for the first time since 1984.
Candidates also must consider where to buy television advertisements and what to say in them. Contrary to advertisers who want to get their brand in front of as many eyes as possible, presidential campaigns are more interested in where their audience lives than in reaching the highest number of people. As you might have guessed, this means that candidates almost exclusively focus their advertising spending in a small number of states. In the last month of the 2012 presidential campaign, Barack Obama aired television ads in just over one-quarter of the country’s media markets, ignoring the largest ones in New York City and Los Angeles. Mitt Romney aired ads in about one-third of the nation’s markets over the course of the campaign, but his choices in the last month of the 2012 contest suggested that he spent money in places he should not have. In the summer and fall of 2012, Romney put ads up in seventy-three media markets, but was down to fifty in the final weeks of the campaign. Obama aired ads in sixty-two markets and was still airing them in fifty-seven during the last month of the campaign. In 2016, Hillary Clinton dominated the airwaves, but her advertising strategies may not have served her well in the end. In 2020, Joe Biden aired nearly double the television ads Donald Trump did during the critical period from the end of September to mid-October.22
Most congressional candidates who can afford television advertisements do not have to worry as much about where to air their ads. Many districts overlap into more than one media market, but typically, most of a district lies within a single market. Of course, the ranking of the market determines, in part, how expensive the advertisements will be. For example, the First Congressional District in Nebraska does not have its own NBC station. One NBC station is located in Hastings, a small city of twenty-five thousand people in the Third Congressional District. The other NBC station is in the Omaha media market, which is in the Second Congressional District. Omaha is the 76th-ranked media market in the country, while Lincoln, located in the First District, is part of the 106th market, allowing Omaha stations to charge more for their ads. Thus, ads aired in First District races during the local and national news on the NBC affiliate reach far more voters who live outside the district than voters who live within the First District boundary, while having the added problem of costing more precious campaign dollars.
The tone of the ads the candidates air is largely determined by the closeness of the race. A great deal of evidence shows that incumbents who are running against opponents who are not well funded, have not held elective office, and so forth tend to air positive advertisements. Congressional incumbents typically air negative ads only in close races or after they have been attacked. However, well-financed candidates fearing future attacks may try to gain the upper hand by going negative first. In Wisconsin’s 2012 Senate race, Democrat Tammy Baldwin was on the air first with a series of ads blasting her opponent, Republican and former governor Tommy Thompson, as someone who is “not for you anymore.” By the time Thompson got on the air to attack Baldwin, she had climbed ahead in the polls, eventually upsetting the popular Thompson in November. Showing how much things can change from one election to the next, Baldwin’s reelection bid against state lawmaker Leah Vukmir was not competitive, leaving Baldwin to air mostly positive ads in 2018.
In presidential races, the question is not whether to go negative, but how much. Kathleen Hall Jamieson makes a distinction between advertisements that purely attack the opponent and those that contrast the opponent’s record or position on an issue with the candidate’s own record or views.23 As shown in Table 2-3, contrast ads made up about a quarter of all presidential campaign ads in 2016. The biggest change in ad tone from 2012—the most negative election of the last five—was the 11-percentage-point increase in positive ads and concomitant 12-percentage-point drop in negative ads. What we cannot see in Table 2-3 are the incredible differences in advertising strategy deployed by Hillary Clinton and Donald Trump in 2016 and Joe Biden and Trump in 2020. While Mitt Romney aired more ads on cable television than Barack Obama (260,210 to 216,363) in 2012, Trump essentially did not air ads on cable television in 2016. Meanwhile, Clinton aired 332,817 cable ads. In 2020, Trump and Biden spent similar amounts on both national and local cable ads.
The other major difference in the 2016 presidential campaign ads as compared to years past went beyond issues of volume—where Clinton enjoyed a 3:1 advantage—to placement. The Wesleyan Media Project reported that Clinton did not air many ads in three key states that she lost—Michigan, Wisconsin, and Pennsylvania—until the last week of the election. Moreover, the content of Clinton’s ads was far different from her opponent’s and that of presidential candidates in previous years. Typically, about 60 percent of candidates’ ads focus on some kind of policy discussion while other ads focus on personal features of the candidates or both the policy and the personal. Erika Franklin Fowler and her colleagues reported that only about 25 percent of Clinton’s ads had policy content. Most of her ads focused on her campaign’s belief that Trump was personally unfit for office.24
In 2020, analysis from the Wesleyan Media Project, the premier U.S. political advertising tracking project in political science, showed that Trump spent more resources on digital advertising on outlets like Facebook and Google than Biden did while Biden continued the Democratic Party’s dominance of the television airwaves.25 Most of Biden’s ads were about taxes, COVID-19, and social security. Most of Trump’s ads were about jobs, COVID-19, and China.26
Table 2-3
Source: Based on initial Wesleyan Media Project coding of Kantar/CMAG data. See “Political Ads in 2020: Fast and Furious,” Wesleyan Media Project, March 23, 2021, https://mediaproject.wesleyan.edu/2020-summary-032321/.
Note: Totals include all sponsors, including outside groups.
Beyond the tone, advertising can prime voters to consider specific factors when preparing to cast their ballots. For example, an analysis by Mirya Holman, Monica Schneider, and Kristin Pondel showed that women candidates could prime women voters’ gender identity whereas men running for office could not.27
Candidates also want to find new ways to reach voters, especially potential voters. The Obama campaign tried to reach out to young voters in a variety of ways in 2008 and 2012. For example, the campaign broke the news of Obama’s choice of Joe Biden as his running mate in 2008 to supporters via text message. In 2016, Donald Trump used Twitter to communicate to voters. One analysis of the 2016 primaries revealed that when Trump’s news attention started to dip, he could generate increased media coverage by going on “tweet storms” where his provocative use of social media, and the thousands of retweets he generated from his followers, inevitably resulted in more media attention—attention his opponents in the primary were not getting.
Campaigns spend an enormous amount of resources on advertising, but they also pay for pollsters, consultants, advance staff, administrative staff, yard signs, stickers, buttons, travel, and more. Red “Make America Great Again” hats were a best seller for Trump in 2016. When one of us authors asked a top Obama campaign staffer about why the campaign was trying so many different strategies to reach voters, he replied, “We know that some of these things are working, but we aren’t 100% sure about which ones they are. So we’ll keep doing them all.”
THE UNEXPECTED
Campaigns cannot plan for everything. Much of the day-to-day life of the campaign falls outside considerations of how closely the strategic messaging hews to the latest relevant economic data. Rather, campaign staff spend a great deal of time responding to press inquiries, reacting to moves made by their opponent, and dealing with problems and opportunities they had not anticipated. While 2020 had fewer dramatic moments than 2016 did, the surprises in 2020 were longer lasting. In 2020, the COVID-19 pandemic cast a shadow over the entire general election campaign. President Trump gave wildly inconsistent messages about the seriousness of the pandemic and what the government’s response would be, and he even made dozens of false statements about COVID-19, how it spreads, how it can be treated and prevented, and how dangerous it is. While at times the president took the coronavirus seriously, he spent just as much time downplaying the severity of the crisis, leading Republicans to follow their party’s leader—compared to Democrats, Republicans were less likely to wear masks, less likely to socially distance, and less likely to want to get a COVID-19 vaccine.
Figure 2-4 shows the strong, positive relationship between the percentage of people in each state fully vaccinated as of June 1, 2021, and the percentage of people who voted for Biden in the 2020 election. A higher Biden vote is clearly associated with a higher vaccination rate. Biden’s emphasis on vaccination during the campaign and in the early days of his presidency integrated the structural challenges of the COVID-19 pandemic and individual voters’ assessments of who would handle the pandemic better.
Description
Figure 2-4 2020 Biden Vote and Percentage of COVID-19 Vaccinated Citizens
Sources: David Wasserman, Sophie Andrews, Leo Saenger, Lev Cohen, Ally Flinn, and Griff Tatarsky, “2020 National Popular Vote Tracker,” Cook Political Report, accessed June 28, 2021, https://cookpolitical.com/2020-national-popular-vote-tracker.
Another unexpected sustained moment in 2020 involved the racial justice protests in support of the Black Lives Matter movement that began after a Minneapolis police officer named Derek Chauvin murdered a Black man named George Floyd. Protests broke out over the whole country, with 55 percent of American adults supporting the movement in September—after a summer of high-profile protests (discussed in greater depth in Chapter 4).28
Sometimes, the unexpected events fit well with the theme a campaign is trying to portray. Regarding COVID-19, Biden portrayed himself as a president who would follow the science, saying in his convention speech that, “as president, the first step I will take will be to get control of the virus that’s ruined so many lives.”29 On racial justice issues, Biden promised to root out systemic racism.
As noted, though, 2016 had more than its share of unexpected moments. Trump’s use of Twitter was so unconventional that the New York Times actively kept a list of the “people, places, and things” that Trump had insulted on Twitter.30 The leak of the Access Hollywood video in which Trump told TV host Billy Bush how he could sexually assault women because he is a celebrity was one of many moments that caused pundits to declare that what had just occurred was something from which the Trump campaign could not recover. But recover the campaign did. Whether led by spirited defenses from his chief spokesperson Kellyanne Conway or Trump himself, the campaign was always able to explain away the latest controversy, blame the media for it, blame his opponents for it, blame the electoral system for it, or change the subject. Hillary Clinton’s campaign was not immune from unexpected moments that were damaging as well. Clinton’s husband, former president Bill Clinton, generated controversy on multiple occasions, most notably when he met privately with Attorney General Loretta Lynch—which was seen as an attempt by the former commander in chief to influence the Justice Department’s investigation of Hillary Clinton’s use of a private email server while serving as secretary of state. FBI director James Comey said that the meeting spurred his decision to inform the public of his investigation of Clinton in the waning days of the campaign—a dramatic development Clinton partially blamed for her loss on Election Day. The release of the Comey letter in the fall of 2016 fit Trump’s narrative that Clinton was a corrupt insider who was part of a rigged system. Trump’s Access Hollywood video fit Clinton’s narrative that Trump was a misogynist who was unfit for office.
The unexpected developments of the 2016 race even led to a renaissance for the comedy program Saturday Night Live. With cast member Kate McKinnon playing Hillary Clinton and movie star Alec Baldwin impersonating Donald Trump, the presidential election was played out each week on the news, and the satirical version of the stunning developments of the week found a large audience on Saturday night. Trump even hosted Saturday Night Live a year before the election while Clinton guest starred in one sketch in the fall of 2015.
Perhaps the most well-known media event from the 2012 campaign was the secret video taken of Mitt Romney referring to the “47 percent” of voters who would not be voting for the GOP in 2012. The video and subsequent coverage fit into the Obama campaign’s argument that Romney was a wealthy, heartless businessman who would leave middle America behind. Although Romney later offered an apology for the comment, it largely fell on deaf ears.
In 2008, both Hillary Clinton in the Democratic primary campaign and John McCain and Sarah Palin in the general election painted a picture of then senator Barack Obama as an elitist who was out of touch with “real Americans.” When audio surfaced late in the primaries of Obama saying that some people would “cling to guns or religion or antipathy to people who aren’t like them” in difficult times, his opponents were quick to jump on the gaffe as an example of the “real Obama.” However, the kerfuffle did not affect Obama’s campaign for long. Of course, the cratering of the nation’s economy was the major event of the 2008 campaign. McCain filled in his own narrative that he was a “maverick” by suspending his campaign and calling for a postponement of a debate with Obama. Obama countered that there are no time-outs for the president and that chief executives have to deal with many problems at once. McCain relented and participated in the debate, which helped amplify Obama’s efforts to steal the “strength” trait owned by Republicans.
A few days before the 2004 election, Osama bin Laden released a video taking responsibility for the September 11 attacks against the United States. The video might have reminded voters that they gave high marks to President George W. Bush on his handling of the war on terror even though an increasing number of voters were critical of his leadership of the war in Iraq. Meanwhile, trying to drum up votes in Ohio, Bush’s opponent John Kerry went hunting and was widely ridiculed for pandering to rural voters as the photos from that event showed Kerry, who had an F rating from the National Rifle Association, decked out in full camouflage, which was consistent with the Bush team’s argument that Kerry was a flip-flopper who would say anything to get elected.
UNDERSTANDING FUNDAMENTALS AND WINNING ELECTIONS
After billions of dollars of spending and nearly two years of campaigning, Joe Biden won back typically Democratic Party strongholds Pennsylvania, Michigan, and Wisconsin from Donald Trump. He also won the swing states of Georgia, Arizona, and Nevada. Republicans lost the Senate, after both Senate elections in Georgia went to Democratic Party candidates, giving Vice President Kamala Harris the tiebreaking vote in a 50-50 Senate. Republicans gained seats in the House of Representatives, but not enough to get the 218 they needed to take back a majority. Overall, the fundamentals predicted a close presidential election in which the Democrat would narrowly win the popular vote. That is just what happened. The fundamentals also predicted that the vast majority of congressional incumbents would win, which they did. This is not to say that the campaigns did not matter. Each of the campaigns turned out an enormous number of voters, had evidence that they had persuaded fence-sitters to their side, and thought that they were going to win on Election Day. In the end, whether by design or by luck, the Biden team was successful. In the remainder of the book, we take detailed looks at the context of these conclusions. We also engage in examinations, over time, of American political behavior and public attitudes. While the rest of the book shows analyses of American partisanship, public opinion, group identities and social characteristics, media use, and determinants of vote choice, keep in mind that each of those factors occurs in the context of the framework described in this chapter.
Study Questions
1. What are the major structural, or fundamental, factors that influence presidential election results?
2. In what ways was the 2020 election a typical election? In what ways did it differ from the norm?
3. How do the ways in which candidates react to the unexpected affect how U.S. presidential elections turn out?
Suggested Readings
Cohen, Marty, David Karol, Hans Noel, and John Zaller. The Party Decides. Chicago: University of Chicago Press, 2008. A new classic in election research.
Issenberg, Sasha. The Victory Lab: The Secret Science of Winning Campaigns. New York: Crown, 2012. An accessible account of the “analytical revolution” in campaign management.
Shaw, Daron R. The Race to 270: The Electoral College and the Campaign Strategies of 2000 and 2004. Chicago: University of Chicago Press, 2006. A careful treatment of how and why Republicans and Democrats targeted specific media markets in presidential campaigns.
Sides, John, Michael Tesler, and Lynn Vavreck. Identity Crisis: The 2016 Presidential Campaign and the Battle for the Meaning of America. Princeton, NJ: Princeton University Press, 2019. A book that shows how the stories claiming the 2016 election violated fundamentals of American politics were wrong.
Theiss-Morse, Elizabeth, and Michael W. Wagner. 2018 Congressional Elections. Washington, DC: CQ Press, 2018. A short, and readable, take on the 2018 midterms.
NOTES
1. Peter K. Enns and Julius Lagodny, “We Predicted the States Biden Would Win 100 Days Before the Election,” Monkey Cage (newsletter), Washington Post, November 12, 2020, https://www.washingtonpost.com/politics/2020/11/12/we-predicted-states-biden-would-win-100-days-before-election/.
2. “Gross Domestic Product, First Quarter 2021 (Advance Estimate),” Bureau of Economic Analysis, April 29, 2021, press release, https://www.bea.gov/sites/default/files/2021-04/gdp1q21_adv.pdf.
3. Barack Obama, interview with Matt Lauer, Today, NBC, February 2, 2009.
4. John Sides, “The Democratic Party’s Uphill Battle to 270 Electoral Votes,” Monkey Cage (newsletter), Washington Post, January 18, 2014, https://www.washingtonpost.com/news/monkey-cage/wp/2014/01/18/the-democratic-partys-uphill-battle-to-270-electoral-votes-in-2016/?tid=a_inl&utm_term=.fa6fe3c8bc7a.
5. Larry Bartels, “2016 Was an Ordinary Election, Not a Realignment,” Monkey Cage (newsletter), Washington Post, November 10, 2016, https://www.washingtonpost.com/news/monkey-cage/wp/2016/11/10/2016-was-an-ordinary-election-not-a-realignment/?utm_term=.51e27306ed05.
6. Nate Silver, “On the Maddeningly Inexact Relationship Between Unemployment and Re-election,” FiveThirtyEight (blog), New York Times, June 2, 2011, http://fivethirtyeight.blogs.nytimes.com/2011/06/02/on-the-maddeningly-inexact-relationship-between-unemployment-and-re-election/?_r=0.
7. Elliott Ramos, “Where Did Trump Make Election Gains? Unemployment Data Tell a Surprising Story,” NBC News, November 16, 2020, https://www.nbcnews.com/news/us-news/where-did-trump-make-election-gains-unemployment-data-tells-surprising-n1247935.
8. Something the Vikings have never done.
9. John Sides, “The Political Consequences of Gas Prices,” Monkey Cage (blog), March 12, 2012, http://themonkeycage.org/2012/03/12/the-political-consequences-of-gas-prices/.
10. Olivia Quinn, Amanda Brush, and Eric R. A. N. Smith, “A Simple Forecast Suggests a Democratic Sweep in 2020,” Monkey Cage (newsletter), Washington Post, July 7, 2020, https://www.washingtonpost.com/politics/2020/07/07/simple-forecast-suggests-democratic-sweep-2020/.
11. “Presidential Job Approval Center,” Gallup, accessed June 28, 2021, https://news.gallup.com/interactives/185273/presidential-job-approval-center.aspx.
12. Drew Linzer, “The Forecasts Were Wrong. Trump Won. What Happened?,” Votamatic, November 16, 2016, https://votamatic.org/2016/11/.
13. Elizabeth Theiss-Morse, Michael W. Wagner, William H. Flanigan, and Nancy H. Zingale, Political Behavior in Midterm Elections, 2015 ed. (Washington, DC: CQ Press, 2015).
14. Quinn et al., “Simple Forecast Suggests a Democratic Sweep in 2020.”
15. Marty Cohen, David Karol, Hans Noel, and John Zaller, The Party Decides (Chicago: University of Chicago Press, 2008).
16. Julia Azari, “What If the Party Decided Not to Decide?,” Vox, January 26, 2016, https://www.vox.com/mischiefs-of-faction/2016/1/26/10834512/party-decides-establishment.
17. Seth Masket, “The 2020 Invisible Primary in Light of 2016,” Vox, January 7, 2019, https://www.vox.com/mischiefs-of-faction/2019/1/7/18170894/2020-invisible-primary-2016-democrats.
18. “Super PACs,” OpenSecrets, accessed June 28, 2021, http://www.opensecrets.org/pacs/superpacs.php.
19. Lynn Vavreck, The Message Matters: The Economy and Presidential Campaigns (Princeton, NJ: Princeton University Press, 2009).
20. James N. Druckman, Cari Lynn Hennessy, Martin J. Kifer, and Michael Parkin, “Issue Engagement on Congressional Candidate Web Sites,” Social Science Computer Review 28, no. 1 (2010): 3–23.
21. ABC News, “Obama: ‘We Don’t Throw the First Punch, but We’ll Throw the Last,’” HuffPost, accessed June 29, 2021, https://www.huffpost.com/entry/post_169_n_132323.
22. “Biden Continues to Dominate Advertising” (Table 1: Top Spenders in Presidential Race), Wesleyan Media Project, October 15, 2020, https://mediaproject.wesleyan.edu/releases-101520/#table1.
23. Kathleen Hall Jamieson, conversation with Bill Moyers, Bill Moyers Journal, PBS, February 29, 2008, http://www.pbs.org/moyers/journal/02292008/profile.html.
24. Erika Franklin Fowler, Travis N. Ridout, and Michael M. Franz, “Political Advertising in 2016: The Presidential Election as Outlier?,” De Gruyter, February 22, 2017, https://www.degruyter.com/document/doi/10.1515/for-2016-0040/html, as cited in “2016 Election Study Published,” Wesleyan Media Project, March 6, 2017, http://mediaproject.wesleyan.edu/blog/2016-election-study-published/.
25. “Biden Continues to Dominate Advertising” (Table 2: Digital Ad Spending by Presidential Candidates and Single-Candidate Super PACs), Wesleyan Media Project, October 15, 2020, https://mediaproject.wesleyan.edu/releases-101520/#table2.
26. “Biden Continues to Dominate Advertising” (Figure 1: Advertising Advantage in the Presidential Race on Television), Wesleyan Media Project, October 15, 2020, https://mediaproject.wesleyan.edu/releases-101520/#fig1.
27. Mirya R. Holman, Monica C. Schneider, and Kristin Pondel, “Gender Targeting in Political Advertisements,” Political Research Quarterly 68, no. 4 (2015): 816–829.
28. Deja Thomas and Juliana Menasce Horowitz, “Support for Black Lives Matter Has Decreased Since June but Remains Strong Among Black Americans,” Pew Research Center, September 16, 2020, https://www.pewresearch.org/fact-tank/2020/09/16/support-for-black-lives-matter-has-decreased-since-june-but-remains-strong-among-black-americans/.
29. Jacob Pramuk, “Read Joe Biden’s Full 2020 Democratic National Convention Speech,” CNBC, August 21, 2020, https://www.cnbc.com/2020/08/21/joe-biden-dnc-speech-transcript.html.
30. Jasmine C. Lee and Kevin Quealy, “The 359 People, Places and Things Donald Trump Has Insulted on Twitter,” Upshot (newsletter), New York Times, January 28, 2016, updated August 15, 2017, https://www.nytimes.com/interactive/2016/01/28/upshot/donald-trump-twitter-insults.html?_r=0.
Descriptions of Images and Figures
Back to Figure
The horizontal axis is labeled G D P growth between January and September, and ranges from negative 3.0 to 7.5 in increments of 0.5. The vertical axis is labeled percent vote for incumbent party and ranges from 44 to 62. The approximate data from the graph are tabulated below.
Back to Figure
The horizontal axis is labeled presidential approval and ranges from 25 to 75 in increments of 5. The vertical axis is labeled percent vote for incumbent party and ranges from 44 to 62. The approximate data from the graph are tabulated below.
Back to Figure
The horizontal axis ranges from 1974 to 2020 in increments of 2. The vertical axis shows millions of dollars and ranges from 0 to 4500 in increments of 500. The approximate data from the graph are tabulated below.
Back to Figure
The horizontal axis shows states. The vertical axis ranges from 0 to 100 in increments of 10. The approximate data from the graph are tabulated below.