Sasha Issenberg, writing in The New Republic; Harry Enten, on 538.com; and a host of other journalists and bloggers have recently waded into a long running scholarly debate. The argument harks back decades (Campbell 1960; Campbell et al. 1960; Key 1966) and more (Lazarsfeld, Berelson and Gaudet 1948) when researchers asked the core questions: Do political campaigns matter, and if they do, do they matter by persuasion or by mobilization? Do campaigns win by persuading fence-sitters – swing voters, in our current discourse – to vote for them, or do they win by turning out their core partisans in high enough numbers to bring them victory? Or, are elections all about the underlying economic and partisan fundamentals and almost completely outside the control of the candidates themselves?
People in the media and political world are suddenly paying attention to a debate that is rooted in decades of political science literature. This debate has not only defined how political scientists have explained election outcomes, but it also is now shaping how political consultants and candidates spend their resources and run their campaigns. Yet there is a fundamental problem with the way the debate – scholarly and popular – has unfolded through the years and now surfaced in the popular press in our own times. If our political discourse lacks nuance, then it is no small irony that in this debate, prominent political scientists and political journalists often divide themselves into rigid camps that are likewise almost completely lacking in nuance. The dogmatism of our politics is mirrored in the dogmatism of our scholarly and journalistic debates.
The pure arguments put forward by each side tend to obscure the complicated ways in which the conduct of campaigns – at the margins – can mean the difference between victory and defeat. By single-mindedly focusing on persuasion, mobilization, or fundamentals as the explanatory factor in elections, to the exclusion of almost everything else, some political scientists and journalists are guilty of creating a narrow analytic framework that ignores mounting evidence that all three schools of thought have something important to say about why campaigns matter and how certain candidates win elections. While all three schools of thought uncover truth about why candidates win and lose elections, none of these schools in and of themselves provides a sufficient theoretical model for explaining election outcomes in contemporary American politics.
A Popular Argument with Scholarly Roots
For Issenberg (2014), drawing on political science, it’s all about turnout and the composition of the electorate, which he describes by using state-level voter-file data to calculate party predispositions. Political scientist Lynn Vavreck (2014) actually makes a similar argument in a post on the New York Times’ Upshot, where she uses a national survey to argue that there are almost no swing voters. Convinced about the importance of the composition of the electorate, and fervently believing in the ability of mobilization efforts to influence turnout (Gerber and Green 2000; Green and Gerber 2008), both recommend – though Issenberg really only makes recommendations to Democrats – that 2014 candidates focus on their respective bases.
For Enten (2014a), who himself has picked up on themes in the literature, it is all about performance. In other words, it is not so much what the composition of the electorate looks like, but the “figures within the cells”: how candidates perform among whites, blacks, young, old, etc., and the ability of candidates to reach these distinct blocs with their ideas and messages – and to sway the small percentage of swing voters who will put them in office. Enten uses the Current Population Survey to describe the demographic composition of the electorate and exit poll results to calculate how various demographic groups voted.
Of course, in addition to these two popular and scholarly camps, there is a third academic camp, one that is by now simply exhausted by the whole discussion of whether political campaigns matter. For this camp, every campaign pretty much comes down to a question of economic fundamentals and predispositions among other basic factors outside of the control of the candidates and their respective campaign strategists. Writing in these pages, Fiorina (2013) just gives up on the debate altogether: “I shall say no more about this, because, given the long history of the disjunction, it is doubtful that academics could change journalists’ minds about this subject if they had a whole semester, not just a seminar. Who are they going to believe: academic researchers, or their own eyes and ears?”
Likewise annoyed with conversations she must endure with journalists and other non- political scientists who claim that media campaigns shape voter preferences and affect election outcomes, Mutz (2012) posits that the multi-million-dollar media strategies employed by candidates have “small to null effects” on actual voters. “Those most likely to change their vote choice are the least likely to be heavily exposed to political media,” she argues. “Although advertising is just one form of election media, conclusions about the impact of the news are similarly underwhelming, unless one looks for effects other than a change in vote preference, or if one looks at low-profile, local races. The scholarly consensus, specifically on direct persuasive effects of media on vote choice – the type of effect that most fascinates the public and the media – is still that media’s impact is marginal at most.” She goes on to argue that “there is little evidence of direct persuasive effects” on national campaigns from “the slick, highly professional advertising that most Americans think of as powerful.”
Yet there is another, bigger point here: Each of these schools employs the other school as a foil. One does not have to read far to meet up with dogmatic lines such as, “the myth of the swing voter (Vavreck 2014),” “it’s all about turnout, except when it isn’t (Enten 2014a),” and “attempting to study media impact in settings with very stable prior opinions is a social scientist’s equivalent of attempting to count galaxies through the wrong end of the telescope (Bartels 1993).” Still, each school is correct – to a degree.
Of course the context and fundamentals matter. We have no quibble whatsoever with the notion that the fundamental factors of party identification and the nature of the times are the main drivers of election outcomes in congressional and presidential races. In fact, underlying circumstances surely explain almost every last percentage point of election outcomes. The fundamentals structure the election as both performance and turnout – swing voters and base voters – affect election outcomes, and in dynamic, interdependent ways. Still, there is no need to be a purist, especially when the evidence does not warrant purity. If we can understand what we are studying and make sure the telescope is pointed in the right direction, perhaps we can have a clearer view of when candidates and campaigns matter, how they matter, and why they matter.
Why Margins Matter
Consistent with the findings of scores of researchers, the fundamentals greatly matter but they don’t explain every last percentage point of election outcomes. At the same time, while a massive effects “fantasy” world may be the reality of cable television news, it is not the reality for most political consultants we have worked with and is not the reality if one carefully reads much scholarly research (Campbell 1987; Popkin 1991; Finkel 1993; Nagel and McNulty 1996; Martinez and Gill 2005; Shaw 2006; Mayer 2007; Hillygus and Shields 2008; Hill 2014 to cite just a few). These scholars realize that the fundamentals leave ample room for marginal effects of turnout or persuasion. And in American political campaigns, the marginal variance – one, two, or three percentage points – can mean the difference between victory and defeat. Minute changes in the loyalty of partisans (performance) and turnout can move election results by a percentage point or two – and produce numbers that would have flipped Presidential outcomes in 2000 and 2004 – Florida in the former, Ohio in the latter – and some key Senate races over the last dozen years (Missouri and Minnesota in 2002; Virginia and Montana in 2006; Colorado and Illinois in 2010; North Dakota and Nevada in 2012, just to cite a handful of cases).
More generally, election outcomes are driven by partisan turnout, partisan loyalty, and the behavior of swing voters. In a basic calculus that virtually every campaign must make, an election can be won by a particular side by one or more of these combined factors:
Getting a large share of the votes from your own identifiers – if you are a Democrat, lock down the Democratic base; shoot to get 90% plus of the Democratic vote. If you are a Republican; lock down the Republican base and 90% plus of the GOP vote.
Getting high turnout from your own identifiers – Of course, Democrats are going to vote for Democrats and Republicans are going to vote for Republicans. But it is crucial for both political parties to generate as much turnout from their side as possible in order to maximize their vote totals.
Getting some of the other candidates’ supporters – It is very difficult to convince members of the opposite party to vote for your candidate. But if you can, it can make a very big difference; not only do these votes count for your candidate, they are subtracted from the total of your opponent.
Lower turnout among the other candidate’s identifiers – If your opponent’s base is angry or unenthusiastic, it can cost them votes. Getting your opponent’s supporters to vote for your candidate is the best possible scenario, but getting them to refrain from voting at all is the next best option. That’s why laws governing elections and absentee ballots and registration rules matter so much to parties and their strategists. When Republicans and Democrats battle over voter ID rules, the Democrats are not simply fighting to protect democracy and Republicans are not protecting integrity; both sides have a direct electoral stake in such rules because the rules affect the composition of the electorate.
Winning swing votes – Even if one believes that there are relatively few swing voters or voters who split their tickets, there are not zero swing voters, and their behavior can be decisive at the margin in a close election.
The main driver of partisan loyalty, voter turnout, and the division among independent or undecided voters (no matter how few there may actually be) is the nature of the times. But having said that, most campaign consultants are trying to get a little more turnout of their supporters or persuade a few undecideds to come their way. Figures 1 and 2 illustrate the basic possible choices for campaigns in whom to target and why.
Each competitive campaign faces a series of hard choices, illustrated by the five factors described above. It’s important to recognize that the structure of a state or district (and thus the predispositions of the electorate) determines which and how many of these choices will be utilized. While there might be a variety of theoretical solutions to the algebra word problem of 50-percent-plus-one, realistically, given the fundamentals, campaigns can either try to get people who are going to vote for them – but may not be inclined to show up – to go to the polls on Election Day; or they can persuade people who are going to vote but who may not vote for them to vote for them. And the choices are interdependent. If you get a very high percentage of the vote from your own identifiers, and if your identifiers show up to the polls in droves, then you may not even need to worry about the remaining three options on the list.
Say your candidate is a Republican running for statewide office in Utah, which is one of the most solidly GOP states. The only options your campaign will need to focus on are numbers one and two from the above list. Make sure Republican voters support your candidate and get them to the polls. There is no need to worry about Democratic turnout or party loyalty. In fact, in this scenario, there is not even any need to worry about independents. The same math works in states like Rhode Island or Massachusetts, only with Democrats as the dominant party.
Now say your candidate is a Democrat running for statewide office in a heavily Republican state. This Democrat must employ a strategy that draws on four or all five of the criteria identified above. For example, having lost these four states in the 2012 presidential race, Democrats cannot hang onto their Senate seats in Alaska, Arkansas, Louisiana, or North Carolina in 2014 simply by re-creating the composition of the 2012 electorate. If they focus solely on the base, they will somehow need to make these four electorates more Democratic than they were two years ago, which nobody expects will happen. And if they focus solely on winning swing voters, there will not be nearly enough of them to give the Democrat another six-year term. Instead, they will need to turn out their own identifiers in high numbers, get a large share of their identifiers to vote for them, win swing voters, and either depress turnout on the other side or persuade some of the other side’s supporters to vote for them.
As Mark Mellman, (2014) the pollster who worked on two of the closest 2012 Senate races, has argued, turnout is crucial, but turnout alone is often not enough to win an election. Enten (2014b) also points out in another post that if Democrats follow Issenberg’s advice in making Alaska their top target in 2014 and focus on voters who drop off in non-presidential election years, they would need to produce an additional 69,000 Democratic votes in a state where the total drop-off between 2012 and 2010 was just 45,000 votes – a near-impossible task.
Yet, contrary to what Enten (2014a,b) argues, the composition of the electorate is also phenomenally important in deciding election outcomes – depending on what you mean by “composition.” Using turnout data from the Current Population Survey, Enten argues that if the age and racial composition of the electorate in 2012 was similar to 2010, Obama still would have won. Put another way, white turnout did not cost Romney the election, and young adult turnout did not win Obama the election.
True enough, and demographics can correlate fairly strongly with vote choice and are sometimes used for targeting by campaigns. But partisan affiliation is a more proximate cause, more strongly correlates with voter choices, and is the metric used for targeting by the campaigns themselves. If the partisan composition of the 2012 electorate was similar to 2010, the White House would have been up for grabs.
The battles for Congress and president are of course not decided by a national vote, and you could do the following analysis state by state. But this simple analysis makes the point that the composition of the electorate is crucial. The first table contains the national party share and performance data in the vote for president from the 2012 exit polls. If you do a little math and multiply the proportion of the electorate by each candidate’s performance, you get a three-point Obama victory.
And here is the party share and performance data from the 2010 exit polls. If you do the math here and multiply share by performance, Republicans won the national House vote by a little less than seven percentage points. Two factors contribute to a 10-point swing between Obama’s 2012 performance and House Democrats’ 2010 vote share: The electorate was more Democratic in 2012 than it was in 2010, and while Obama lost among independents, he performed better among them than House Democrats did in 2010.
Now, what happens if we replay the 2012 election with 2010 turnout, assuming 2010 composition with 2012 performance? Do the math and multiply share by performance for each cell. Contrary to what Enten finds, turnout did matter in 2012. If the composition of the 2012 electorate looked as it did in 2010 and Obama’s performance was what it was, Obama’s three-point victory (in the exit polls) in the 2012 national vote count would have become a two-point loss. We could do the same exercise in all states, or in all battleground states, but it is virtually certain that a two-point loss in the national vote would have translated into enough Romney victories in swing states to win Republicans the White House. The key point here is that election outcomes in competitive races rarely boil down to a single explanatory factor.
The Case of Nevada
Consider a couple examples from recent elections in Nevada that show how persuasion and mobilization combined to help determine the outcomes. In 2012, Barack Obama beat Mitt Romney by nearly seven percentage points, but Democratic Senate candidate Shelly Berkley lost to Republican Dean Heller by a percentage point, as over 85,000 Obama voters (16% of his total) voted for the Democrat for president but not for Senate. Obama handily won Nevada because Democrats turned out in sufficient numbers and overwhelmingly voted for him, while the so-called independents – voters who did not identify themselves as either Republicans or Democrats – gave him enough support. Berkley lost because she failed to persuade Democrats and independents to support her at the same levels that they voted for Obama. With a smaller turnout, Obama might still have prevailed, but by a much tighter margin. So turnout clearly mattered, yet persuasion was also crucial, spelling the difference between Obama’s victory and Berkley’s loss.
Now take the case of Harry and Rory Reid. According to voter records, in 2010, the composition of the Nevada electorate was less friendly to Democrats. They made up an ever so slightly smaller proportion of the electorate compared to the 2008 and 2012 presidential electorates (41%, compared to 42% and 43% in the presidential years), but the smaller proportion of independent others in the electorate was reflected in a higher proportion of Republicans. Indeed, contrary to the wide advantages Democrats had in the partisan composition of presidential years, 42% of the electorate in 2010 was registered as Republican and 41% was registered as Democratic. Both the two major statewide races in 2010 were won by comfortable margins, one by a Democrat and one by a Republican. Democrat Harry Reid beat Republican Sharron Angle in his bid for re-election to the US Senate 50.3%–44.5%, and Republican Brian Sandoval beat Reid’s son Rory by 53.4%–41.6% to become Governor.
What happened? According to exit polls, in the US Senate race, Reid won self-identified Democrats 91% to 5%, and lost self-identified independents/others 44% to 48%, while, importantly, managing to win 11% of the self-identified Republican vote (compared to 84% for Angle). Reid’s ability to persuade Democrats to overwhelmingly support him, along with many independents and a disproportionate number of Republicans, was not matched by his son. With the same electorate, Rory Reid managed to win 86% of self-identified Democrats (compared to 10% by Sandoval) and 32% of self-identified independent/others (compared to 60% by Sandoval). In contrast to his father, Reid was only able to draw the support of 4% of Republicans (compared to 92% by Sandoval), a figure that was even more important in 2010 than in presidential year elections because Republicans were accounting for a greater proportion of the electorate.
These examples show how candidates of the same party, with the same electorates (and sometimes even with the same last names!) can draw vastly different vote shares. The motivation of citizens to vote for candidates of different parties comes from evaluations of those candidates, and voter judgments come in no small measure from the persuasion activities of the campaigns. The exit polls from Nevada show something that is also often overlooked amid this overly simplified debate. The difference in performance is partly due to how candidates are supported or opposed by non-partisans. But it is also a function of how well they do with their own (and opponents’) partisans.
This is particularly important in non-presidential elections, where independents make up a smaller proportion of the electorate, but must be understood by those advocating that campaign woes can be solved with higher partisan turnout. Higher partisan turnout can clearly help, but it should not necessarily be assumed that partisans will uniformly or universally support candidates of their preferred party, something that is made even harder when structural difficulties (such as lack of party registration) makes finding likely partisans even trickier.
In the case of the 2014 midterm elections, at the margin, Democrats obviously and certainly would improve their chances of holding the Senate by getting more of their drop-off voters to the polls in the most closely-contested states. And they will need to do that in an election where they will likely not enjoy the huge resources that funded Obama’s efforts in 2008 and 2012. That said, they also need to attain more support from voters who turned out for Romney two years ago. They need to perform better among Democrats, Republicans, and independents, and they need more of their identifiers to vote on Election Day. This is hardly a binary choice, as some analysts would have us believe.
The multiplicity of factors that account for campaign victories has implications for those of us hoping to study the effect of campaigns. Figure 3 below shows that campaign effects are going to be heterogeneous. Sometimes we will be looking for persuasion effects, sometimes for mobilization effects, and sometimes one side will have complete control of, say, the megaphone. The point is that we need to look for different sorts of effects in different quadrants if we want to understand election outcomes. Some quadrants will have one-sided flows, some of them will have equal flows, and even when political scientists think they have equal flows, the appearance of equality might be misleading.
For example, political scientists in the main argue that the equal nature of most general election campaigns cancels out any potential effects that advertising might have. Thus the vast expenditures of a modern presidential race come to look like the campaign equivalent of the First World War – each side, entrenched, has equal firepower, with no side winning on that particular battlefield. But the notion that competing campaign communications cancel one another out is unpersuasive. Even though Republicans outspent Democrats in the 2012 presidential race, better targeting, better buying, and arguably a more relevant, effective message allowed the Obama campaign to land more blows (Goldstein 2014).
In the end, the three schools – persuasion, mobilization, and fundamentals – all help to explain how candidates fight, win, and lose American elections. But none of these camps by themselves allows for a wide-enough telescope that sufficiently captures the complex angles and dynamics that constitute national election campaigns. Campaigns are targeting their own voters, seeking to persuade a small minority of swing voters, and trying to depress turnout or win votes from the other side. Their hope is to win an additional 1 or 2% of the electorate – a small margin at first blush, but as Al Gore or John Kerry and countless losing congressional candidates well know, even a tiny margin can mean the difference between the Oval Office or a House or Senate seat – and nothing.
Bartels, Larry M. 1993. “Messages Received: The Political Impact of Media Exposure.” American Political Science Review 87: 267–285.
Campbell, Angus. 1960. “Surge and Decline: A Study of Electoral Change.” Public Opinion Quarterly 24 (3): 397–418. [Crossref]
Campbell, Angus, Phillip Converse, Warren Miller, and Donald Stokes. 1960. The American Voter. New York: Wiley.
Campbell, James E. 1987. “The Revised Theory of Surge and Decline.” American Journal of Political Science 31 (4): 965–979. [Crossref]
Enten, Harry. 2014a. “Mid-Term Election Turnout Isn’t So Different from Presidential Year Turnout,” 538.com, May 7, 2014.
Enten, Harry. 2014b. “The Problems with The New Republic’s Senate Rankings,” 538.com, May 1, 2014.
Finkel, Steven E. 1993. “Reexamining the ‘Minimal Effects’ Model in Recent Presidential Campaigns.” Journal of Politics 55 (1): 1–21. [Crossref]
Gerber, Alan S., and Donald C. Green. 2000. “The Effects of Canvassing, Telephone Calls, and Direct Mail on Voter Turnout: A Field Experiment.” American Political Science Review 94 (3): 653–663. [Crossref]
Goldstein, Kenneth. 2014. “Gaining Separation in Presidential Elections.” Manuscript. http://www.usfca.edu/facultydetails.aspx?id=6442451088.
Green, Donald P., and Alan S. Gerber. 2008. Get Out the Vote: How to Increase Voter Turnout. 2nd ed. Washington: Brookings Institution Press.
Hill, Seth J. 2014. “How Candidates and Election Context Swing Voters and Mobilize the Base: Evidence from Florida.” Working Paper. http://www.sethjhill.com/SJH_CandidatesAndContext.pdf. Accessed July 1, 2014.
Hillygus, Sunshine, and Todd Shields. 2008. The Persuadable Voter. Princeton: Princeton University Press.
Issenberg, Sasha. 2014. “How the Democrats Can Survive Going Down This November: The new science of Democratic survival.” The New Republic, April 27, 2014.
Key, V. O. 1966. The Responsible Electorate. New York: Random House.
Lazarsfeld, Paul, Bernard Berelson, and Hazel Gaudet. 1948. The People’s Choice: How a Voter Makes up His Mind in a Presidential Campaign. New York: Columbia University Press.
Martinez, Michael D., and Jeff Gill. 2005. “The Effects of Turnout on Partisan Outcomes in U.S. Presidential Elections 1960–2000.” Journal of Politics 67 (4): 1248–1274.
Mayer, William G. 2007. “The Swing Voter in American Presidential Elections.” American Politics Research 35 (3): 358–388. [Crossref]
Mellman, Mark. 2014. “Turnout Alone Won’t Mean a Win.” The Hill, May 13, 2014.
Mutz, D. C. 2012. “The Great Divide: Campaign Media in the American Mind.” Daedalus 141: 83–97. [Web of Science]
Nagel, Jack H., and John E. McNulty. 1996. “Partisan Effects of Voter Turnout in Senatorial and Gubernatorial Elections.” American Political Science Review 90 (4): 780–793. [Crossref]
Popkin, Samuel. 1991. The Reasoning Voter. Chicago: University of Chicago Press.
Shaw, Daron R. 2006. The Race to 270: The Electoral College and the Campaign Strategies of 2000 and 2004. Chicago: University of Chicago Press.
Vavreck, Lynn. 2014. “The Myth of Swing Voters in Midterm Elections,” NYT.com, The Upshot, April 22, 2014.
About the article
Ken Goldstein is Professor of Political Science in the Department of Politics at the University of San Francisco and Director of the University’s Washington program.
Matthew Dallek is Assistant Professor of political management in the Graduate School of Political Management at George Washington University.
Joel Rivlin is Senior Vice President of the Pivot Consulting Firm and a PhD candidate in political science at the University of Wisconsin-Madison.
Published Online: 2014-08-08
Published in Print: 2014-07-01