Examining, Correcting, and Failing to Correct Politically Biased Judgments and Memories in Real-World Contexts
- Author(s): Grady, Rebecca Hofstein
- Advisor(s): Ditto, Peter H
- et al.
This dissertation brings together three projects at the intersection of politics, bias, memory, and decision making. The five studies in total explore how people come to political judgments, when they might be biased in those judgments, how their memory might be affected, and what real-world consequences this can have.
The first chapter describes a longitudinal study conducted over the course of the 2016 presidential election. The study collected data from 602 U.S. citizens on Amazon Mechanical Turk during the presidential primaries in May 2016, with questions focused on voters’ feelings towards, and preferences for, the candidates seeking each major party’s nomination. A follow-up survey was sent to participants again in October 2016 just before the general election to see how their attitudes changed and what they remembered about their past preferences. Participants were re-surveyed once more in January 2017 just before the inauguration to look at their reactions to the election, feelings towards the former candidates, and memory of past attitudes. Across a variety of questions, people’s memory for their past attitudes was strongly influenced by their present attitudes; more specifically, those who had changed their opinion of a candidate remembered their past attitude as being more in line with their current feeling than it actually was. This was especially strong among people who were more confident in their memory and their own rationality. I also investigated the presence of false memory, and found high rates of participants remembering news events that didn’t happen and misremembering what candidate they had supported.
In the second chapter, three studies are presented in the domain of political polling, looking at how the placement and order of questions can create contexts that appear to increase or decrease political polarization. In all three studies, each with over 800 people, participants had to make judgments about similar accusations against two politicians, one Democrat and one Republican, with each rating on a separate page. On the first judgment, partisans provided strongly disparate ratings, with both Democrats and Republicans giving low credibility to the accusations against their in-group politicians and high credibility to the accusations against the out-group. On participants’ second ratings, however, the groups were much closer together. This is because, after making their first rating, participants’ second rating was more in line with their judgment of the first, similar accusation. In one study (but not others), this consistency was increased by first having participants commit to relevant, general principles about whether people should believe this type of accusation in general.
Finally, Chapter 3 takes a look at the growing problem of politically-oriented “fake news,” or completely false stories or headlines. In the real world, such fake news stories are designed to appeal to specific groups of people and be shared widely within these groups. Previous studies found that adding a warning tag to false headlines has minimal effect in reducing belief in these falsehoods. The chapter starts with a secondary analysis of prior data showing a partisan congruency effect, where warnings may help reduce belief in news that goes against a person’s political orientation, but shows less effect when the news supports a person’s political orientation. Following up on that, I conducted a study with over 400 online participants, testing a modified warning that comes before people read false headlines and comparing it to warnings presented at the same time as or after the headline. Participants were shown a series of news items, with the false ones labeled as such with one of the three warning types, and then asked two weeks later how true they thought each headline was. While the before-headline warning was initially very effective, and showed some stronger effects than the other warnings, the follow-up survey two weeks later showed both high levels of belief in the articles and a partisan congruency effect where people were more likely to believe the news that fit with their political orientation, even when they had known just two weeks ago it was false.
Across these three chapters, I look at the effects of politics, bias, and motivation on our memory and judgment in real-world political contexts. The effects of political allegiance are apparent not only on our opinions of politicians, but also on our memories of ourselves and the world around us. These effects are incredibly strong, and far too often we are unaware of them. Some manipulations were effective at reducing differences between ingroup and outgroup judgments, while others were not. These studies add to our understanding of the role of political allegiances on our judgments and memories by identifying some sources of these effects, showing how they play out in real-world contexts, and testing solutions that are more and less likely to show long-term impact in these divisive contexts.