Back in the fall of 1989, a group of researchers published a paper looking at the ways in which individuals contort the circumstances surrounding them into their preexisting world views.

This particular investigation, titled “Expert Decision Making in Evolving Situations,” gave 11 groups of Army intelligence analysts a realistic battlefield scenario and asked them to assess the most likely avenue for an enemy attack. The scenarios were largely the same, though with slight variations to produce different answers. Each group was given time to study and each expressed confidence in their answers.

The noteworthy stuff is what came next. The groups were given updated intelligence reports and asked to reconsider their assessments. Some reports contained items confirming initial judgements. Others were designed to spur skepticism. The majority were neutral. The process was then repeated two more times.

In the aggregate, the level of confidence should have stayed roughly the same. But what the researchers found was that the groups grew more convinced in their initial judgements the more information they received. Only one of the 11 teams changed its assessment of how the enemy would  attack. Seven of the 11 teams expressed more confidence in their call over time.

Additionally, the subjects gave significantly more weight to information that reinforced their earlier decisions. Not only that, but when presented with contradictory evidence, they were dismissive or downplayed its significance.

Confirmation bias like this had been observed before. What stood out to the researchers was that individuals trained to be open and sober-minded were now exhibiting it.

“The results of this experiment lend support to the general conclusion that trained subjects in an evolving, realistic, decision environment demonstrate performance characteristics similar to those of novices working with less realistic and relatively more static scenarios,” the study read. “Specifically, confidence in an initial hypothesis is generally high, regardless of the hypothesis.”


Clinton isn’t well liked. But she is well known. And that might be more important.

Presidential campaigns are not literal battlefields. And voters are not Army intelligence analysts. But as the 2016 general election comes into focus, the same behavioral patterns observed in this study will play a significant role in determining the next president.

Donald Trump, the presumptive Republican nominee, filleted a field of 17 Republican primary candidates by branding them in uniquely terrible ways: Little Marco Rubio, Lyin’ Ted Cruz and Low Energy Jeb Bush. With his attention shifting to November, the fear among preternaturally panicked Democrats is that he will do the same against his likely opponent: Hillary Clinton. Trump has already begun trying, adding the descriptive “Corrupt” to her first name.

But political scientists and branding experts aren’t so sure that he’ll find much success. And it goes back to “Expert Decision Making in Evolving Situations.” Referencing that specific study, Timothy Calkins, a clinical professor of marketing at the Kellogg School of Management at Northwestern University, noted just how hard it is to mold perceptions when people have already thought through their choices.

“It is very hard to reposition a well-established brand, and what we have here are two really well-established brands,” Calkins said of the election matchup. “There is a whole idea of mental exhaustion. When you force people to really think about something, it is difficult and challenging. And the easy thing to do is to just not think about it. For someone to really challenge and change their beliefs requires a lot of energy.”

When you force people to really think about something, it is difficult and challenging. And the easy thing to do is to just not think about it.
Timothy Calkins, a clinical professor of marketing at the Kellogg School of Management at Northwestern University

At its current juncture, the Democratic primary is boiling down to a fight over electability. Sen. Bernie Sanders (I-Vt.), who trails in pledged delegates, has argued that party insiders should switch their votes to him precisely because his polling numbers are better suited for the general. And that’s true. Sanders does better in mock contests against Trump. His favorability ratings are far superior to Clinton’s.

These strengths, however, are somewhat cosmetic. Though he’s been on the trail for over the year, Sanders is not as known a political figure as Clinton. He’s faced a tiny sliver of the negative attacks. As The Huffington Post reported in mid-April, of the roughly $383 million spent on campaign television advertising in 2016, only about 2 percent was on anti-Sanders ads, much of which just briefly mentioned his name or featured his image.

“People are pointing to his general election numbers as being stronger than Clinton’s, and that’s largely a byproduct of the fact he hasn’t seen incoming fire,” said Brendan Nyhan, a political science professor at Dartmouth College and a columnist for The New York Times.

Clinton, by contrast, presents a surer bet, albeit with less potential upside. Should she secure the Democratic nomination, she would have a favorability rating worse than any general election candidate in history … save for Trump himself.

But she brings advantages to the ticket too.

On the trail, Clinton touts the political battles she’s experienced as proof that she can succeed where Little Marco, Lyin’ Ted and Low Energy Jeb failed. The impression left is that she has the political acumen to navigate a race against Trump. But what she is also underscoring is that she has the longevity to not be defined by her opponent. Like those Army intelligence officers, the voters have studied her resume, and new parcels of information will simply be plugged into their preexisting views.

“People have an amazing ability to reinforce what they believe,” said Calkins.

Consider this: In public opinion polls, a full 96 percent of the public is able to rate Clinton either unfavorably or favorably, the same percentage as Trump. By comparison, 86 percent of the public was able to rate Mitt Romney when he was the presumptive nominee in May of 2012. In July of 2015 — roughly when the Republican primary began — 67 percent of the public was able to rate Ted Cruz and 64 percent of the public was able to rate Marco Rubio.

The public is about as likely to have a strong opinion about Clinton at the start of the general election as they were to have any opinions about Cruz or Rubio at the beginning of the GOP primary. Across recent polls, more than 60 percent rate her at one extreme or the other: either “very favorably” or “very unfavorably.” Since last June, her numbers have moved relatively little, considering all the campaigning and negative headlines. Her average unfavorable rating has drifted from the high 40s to the low- to mid-50s, but that is likely due to dissipating goodwill from her time as secretary of state. Her favorable rating has dropped, but it’s likely to rise again should disaffected Democrats (Sanders supporters) come back on board.

The numbers are relatively static for Trump, too. Despite being the most divisive political figure in the country over the past year, his unfavorable rating remains in the low 60s (the same place it was in June 2015). His favorable rating is the one that’s changed, rising from the mid-20s to the mid-30s, presumably as Republican primary voters have gotten to know him as a politician.

“Look at Trump,” said Nyhan. “With all the stuff that has been said about him, his unfavorables … they barely changed. This whole time. With everything that has been said about him. It is strikingly stable.”

In a year without Trump, the case could be made that Clinton would be a serious gamble for Democrats — voters’ confirmation biases would be working against her were she facing a more-liked Republican nominee. But there are other factors influencing elections beyond a candidate’s favorability rating. Often, in fact, favorability ratings tend to be overstated as a metric. Nyhan has written extensively about this.

While it might seem obvious that people vote for the candidate they like best, that notion often gets the direction of causality backward. In the heat of the campaign, we ultimately tend to find reasons to support candidates who share our party affiliation or seem to have a good record in office (and to oppose candidates who do not).

Certainly, there are exceptions to the rule that party, not personality, is more determinative of election results. Trump could very well be one. The outsized force of his personality overshadowed nearly all the traditional contours of the Republican primary, and the next six months will test whether partisanship is an even stronger motivation.

But by and large, as the general election progresses, the expectation among political scientists is that we will enter a more stable race than the current political commentary foreshadows. Republican voters will warm up to the nominee. Beleaguered Sanders supporters will find a way to Clinton. A brutally negative campaign will be waged, but confirmation biases will once again take hold.

“Public opinion figures tend to converge,” Nyhan noted. “When Al Gore ran, Democrats weren’t enthusiastic about his candidacy but they mostly made their peace with with. [Senator] John McCain had incredible favorability numbers. But to win the nomination he became a classic Republican, and he ended up performing like a general Republican when the election came around. So personal qualities tend to be overstated relative to other structural factors.”