Painted tabletop models of the Republican elephant and the Democratic donkey shown in 2002 in Washington. (Jacqueline Roggenbrodt/AP)

Elections can be seen as games — expensive and consequential ones. They involve offensive and defensive maneuvering, a set of rules governing play that can be controversial, and a clearly defined goal for victory: Win more votes than the other team.

And according to our research, people who vote for the losing side don’t feel that government — the winning team — represents them very well.

Is that true as well for voters whose candidate loses a primary election?  We don’t believe so. Most of those who turn out for the primary are dedicated party voters. And those “fans” are under extreme pressure — both internal and social — to stick with their teams. Some Republicans may treat the November election as a referendum on Donald Trump. But the president holds veto power over legislation, a capability worth controlling for a party. And at least one Supreme Court seat is up for grabs. Defecting would bring pretty severe consequences for anyone who cares about their team — yet popular commentary often alleges that these divisive primaries hurt parties’ nominees during the November general election.

In November, will Democrats who supported Bernie Sanders and Republicans who supported Trump’s rivals vote for their own party’s candidate — or someone else?

The scholarship is divided. Some scholars say hotly contested primaries encourage voters to support the final candidate in a general election; others say that bitter partisans may support alternative candidates. To explore this question, we looked at vote choice in the 2008 presidential election, which, as is true in 2016, had no incumbent in the race — and included a hard-fought primary. Democrats were fiercely divided between Hillary Clinton and Barack Obama. Clinton won victories in important states. But by halfway through the primary season, Obama was pulling in more votes and delegates, soon followed by superdelegate support.

When Clinton eventually conceded, did her voters fail to vote in November, or defect to the Republican candidate or a third party?

We looked at the 2008 election to find out.

To find out, we examined data from the 2008-2009 ANES Panel Study.

Of all those who voted for Clinton in the primary, only about 5 percent said they did not vote in the general election. By comparison, among those who voted for one of the two main Republican primary challengers, Mitt Romney or Mike Huckabee, roughly 7 percent did not cast a vote in the general election.

In other words, the sore primary losers who sat out the general election for each party roughly canceled each other out.

However, as you can see in the graph below, some primary voters whose candidate lost in the primary actually did end up voting for the other party’s candidate. Among Republican primary voters who did not vote for John McCain — the “winner” of the primary season — relatively few Republicans defected to vote for Barack Obama (about 8 percent). On the other hand, roughly 17 percent of Democrats who did not vote for Obama during their party’s primary voted for McCain.

DAVIS & HITT 1

Does this mean that Clinton may lose Sanders supporters, who will defect and vote for the Republican nominee?

Probably not, for three reasons.

First, not very many voters actually turn out for primaries. Many independent voters or registered Democrats surely plan to vote in the general election but didn’t turn out for the primary.

And even if a similar percentage of Sanders voters were to vote Republican in the fall, as did Clinton supporters in 2008, that would amount to a relatively small number of total votes. Indeed, in the 2008-2009 ANES Panel Data we analyzed, more than five times as many respondents voted for Obama without having voted in the primary than respondents who voted for Clinton in the primary but McCain in the general election.

Apparently “sour grapes” affects election results only slightly.

Second, Trump faces a dilemma that’s similar to the one Obama faced in 2008. Ted Cruz, John Kasich and Marco Rubio have openly wrestled with the prospect of supporting Trump. But while these stories make good newspaper copy, congressional Republican leadership has thrown its support behind Trump, although Speaker Paul Ryan (R-Wis.) has criticized Trump over his attacks on a federal judge, and some have argued that delegates are not necessarily bound to Trump if it violates their “conscience.”

Still, recent polling suggests that both Republican and Democratic voters plan to support their party’s nominees at similar rates.

Add all this up, and defections by disappointed primary voters should have a vanishingly small effect in the general election.

In fact, fewer voters will probably switch to the “other” party’s nominee in 2016 than did so in 2008.

Political scientists find that citizens aren’t likely to vote on political attitudes; rather, they vote based on tribal attachments more than on political attitudes.

Most people’s policy attitudes are a grab-bag of contradictory ideas — which means they’re not voting according to rigid ideological standards. That’s why, for instance, Trump’s ever-shifting position on issues like abortion never dented his support during the primary. And according to our research (Davis’s), “sorted voters” — those voters who strongly identify, for example, as both liberal and as a Democrat, or as both conservative and a Republican — almost never defect.

As a result, Sanders supporters who are otherwise liberal and lean Democratic are highly unlikely to switch sides. And although Trump has generated an unprecedented amount of existential angst among conservatives, recent polling indicates that he can win the type of support that John McCain earned in 2008, despite his poor “favorability” ratings.

When push comes to shove, we expect that defections might be less likely this time around for two reasons.

First, for those voters who do pay attention to policy, Sanders supporters will stay closer to their preferred policies by voting for Clinton more than for Trump. Similarly, Trump has tapped into real anxiety over the disparities in economic recovery, which appears to reach voters who once preferred other Republican candidates. Further, for Republican voters, achieving any number of their preferred policies will be much more difficult if Clinton is elected president. The enemy of my enemy is my friend, as the maxim goes.

Second, for those who vote based on identity, a Sanders supporter who identifies as progressive will still do better with Clinton than with Trump. And Trump, in turn, may attract Republicans of varying degrees of conservative identity, who at least share extremely unfavorable perceptions of Clinton.

Each party’s base will likely turn out to vote for the party’s nominee — or at least, against the other party.

Research has shown that getting the support of one’s party’s base makes a big difference in the general election. Trump may do better by consolidating Republican voters who share this intense distaste of Clinton than by trying to broaden his appeal, especially since undecided voters are such a small slice of the electorate.

In other words, the recent past suggests that most Sanders supporters will eventually vote for Clinton. The numbers of voters who either fail to vote or switch sides will be dwarfed by the number of Democrats and left-leaning independents who sat out the primary but do vote in November.

Similarly, many people who didn’t vote for Trump are now endorsing his candidacy — which gives us a hint of what will likely happen in November. Despite concern about Trump’s rhetoric and intermittent calls to replace him at the Republican National Convention, popular support for  Trump has remained remarkably steady (although some appear to be holding out hope that his candidacy will implode).

Rare are the ideological voters who betray their ideals by defecting. Although it is too early to predict November’s election, prospective voters usually realize that their party’s nominee fits their identity better, even if they don’t love the result the way that the backers of winners do.

Nicholas T. Davis is a PhD candidate in American politics at Louisiana State University whose dissertation research analyzes the causes and consequences of political sorting in the mass public. Follow his occasional blog.

Matthew P. Hitt is an assistant professor of political science at Louisiana State University; he will join the department of political science at Colorado State University in the fall. Follow him on Twitter @matthewhitt.