On November 23, 1951, Princeton University’s football team beat rival Dartmouth in a hotly contested game in which key players on both sides suffered injuries and there were several infractions. The referees saw Dartmouth as the primary aggressor, penalizing them 70 yards to Princeton’s 25. In the aftermath, there was controversy in the press about allegations of overly rough and dirty play.
In 1954, social psychologists Albert Hastorf (of Dartmouth) and Hadley Cantril (of Princeton) put aside their differences and published a study entitled “They Saw a Game.” Two types of data were collected. Samples of Dartmouth and Princeton students were given a questionnaire measuring their recall of the game. Secondly, a smaller sample of 48 Dartmouth and 49 Princeton students were shown a film of the game and asked to identify any rule violations they saw. The results suggested that they saw a different game. For example, on the questionnaire, 86% of Princeton students but only 36% of Dartmouth students thought that Dartmouth had started the rough play. The mean numbers of judged infractions are shown here:
Dartmouth students thought the number of violations had been about equal, but Princeton students saw more than twice as many infractions by the Dartmouth players.
This study is an example of myside bias, which is in turn a special case of confirmatory bias, the tendency to search out, interpret and recall information in a way that supports your pre-existing beliefs. (“Myside bias” is more likely to be used when two competing groups, such as Democrats and Republicans, are at odds.) There are hundreds of studies of confirmatory bias.
For example, Dan Kahan and his colleagues did a study entitled “They Saw a Protest.” Participants were shown a video of a political demonstration. Half were told that it was a protest against the military’s “don’t ask, don’t tell” policy, and the others that it was an anti-abortion protest. As expected, liberals and conservatives differed on whether they had observed free speech or illegal conduct. Liberals were more likely to see the demonstrators as obstructing and threatening bystanders when the demonstration was identified as anti-abortion, while conservatives were more likely to see the anti-military protest as containing illegal behavior.
Inspired by the flagrant misperceptions of President Donald Trump, political scientist Brian Schaffner and Samantha Luks of the YouGov polling organization surveyed 1388 American adults on January 23 and 24. They showed them the two photographs below.
Half the respondents were asked which photo was from the Trump inauguration and which was from President Obama’s 2008 inauguration. The other respondents were simply asked which crowd was larger. Finally, all participants were asked for whom they had voted.
The data on the left show that, consistent with their presumed belief that Trump has broad public support, Trump voters were more likely to misidentify Photo B as his inauguration than either Clinton voters or non-voters. A more surprising result is shown at right. Fifteen percent of Trump voters said that Photo A contained more people!
The finding that Trump voters were more likely to choose B as the Trump inauguration is an example of myside bias. People (mis)identified the photos in way that was consistent with their political affiliation. An alternative explanation is that, since Trump voters are more likely to be what political scientists call “low information voters”—people who don’t often follow the news—they were less likely to have seen the two photos on TV or in a newspaper. It’s unfortunate that the authors didn’t ask respondents whether they had seen them before.
The behavior of the Trump voters who said Photo A had more people is more difficult to interpret. We can assume that they deliberately gave an incorrect answer. The authors interpret this as a partisan attempt to show their support for Mr. Trump, which has been called expressive responding. A related possibility is that they may have suspected the study was an attempt to embarrass Mr. Trump, and their response was an upraised middle finger directed at the researchers.
You may also be interested in reading: