REVIEWS
“Overturning decades of conventional wisdom, Coppock compellingly demonstrates that persuasive messages have similar effects for all kinds of people. This finding will set the agenda for the fields of public opinion and political communication.”
— Brendan Nyhan | Dartmouth College
"Persuasion is hard, but information matters—and it matters similarly to every-one. In Persuasion in Parallel, Coppock demonstrates that people’s existing views about the kind of world they want to live in are a central feature of their future views about the kind of world they want to live in, but he also shows that new information is not lost on people. Where you start has a lot to do with where you land, but making sure people get new information, more information, and factual information can move everyone’s thinking in the same direction. Politics may be polarized, but Coppock shows persuasion is not.”
— Lynn Vavreck | University of California, Los Angeles
"Motivated reasoning theory argues that when people encounter information with which they disagree, they will backlash against it and become more extreme in their prior positions . . . Coppock posits that people update their positions in ways that align with the direction of the information to which they are exposed . . . Whereas motivated reasoning theory has led to despair over the polarization of American politics, Coppock's data gives hope that information can bridge the political divide. Recommended."
— Choice
TABLE OF CONTENTS
Chapter 1: Persuasion in Polarized America - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0001
[backlash;motivated reasoning;persuasive information;group cues]
This chapter is Persuasion in Parallel in miniature. It explains the backlash prediction from the motivatedreasoning literature and foreshadows how the empirical evidence for backlash rests on weak research designs. It introduces the persuasion in parallel hypothesis that the treatment effect of persuasive information on target attitudes is small, positive, and durable for everyone, taking care to distinguish between persuasive information and group cues. The chapter provides evidence from one survey experiment that Republicans and Democrats update their tax policy preferences in the same direction and by about the same amount in response to an op-ed in support of the flat tax. Lastly, the chapter connects the persuasion in parallel hypothesis to the classic parallel publics finding from the public opinion literature, contrasting the survey experiment and rolling cross-sectional survey research designs. (pages 1 - 16)
This chapter is available at:
https://academic.oup.com/chica...
Chapter 2: Reinterpreting a Social Psychology Classic - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0002
[backlash;counter-attitudinal information;replication study]
The 1979 study from CharlesLord, Lee Ross, and Mark Lepper is frequently cited as evidence that people respond with backlashto counter-attitudinal information. This chapter explains how two flaws in the original research design (a lack of random assignment and biased outcome measurement) lead to this incorrect conclusion. When these flaws are corrected in a replication study conducted in 2015, we see that both policy opponents and proponents update their views in the direction of information, by about the same amount. (pages 17 - 30)
This chapter is available at:
https://academic.oup.com/chica...
Chapter 3: Definitions and Distinctions - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0003
[persuasive information;homogeneous effects;group cues;heterogeneous effects;target policy attitudes;nontarget attitudes;affective evaluations;group membership]
This chapter defines each of the words in the persuasion in parallel hypothesis: "treatment effect," "persuasive information," "target policy attitudes," "small," "positive," "durable," and "everyone." Along the way, these terms are distinguished from related concepts like "group cues," "nontarget attitudes," and "affective evaluations." With these definitions and distinctions in hand, the chapter lays out the theoretical predictions about the effects of two kinds of treatments on three outcome variables. These are: persuasive information has mostly homogeneous effects on target attitudes, mostly null effects on nontarget attitudes, and heterogeneous effects on affective evaluations of messages and messengers that depend on baseline views. Group cues have heterogeneous effects on target attitudes that depend on group membership and uncertain effects on nontarget attitudes and affective evaluations. (pages 31 - 51)
This chapter is available at:
https://academic.oup.com/chica...
Chapter 4: Research Design - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0004
[panel survey experiment;MIDA framework;expectancy value model;target attitudes;difference-in-means]
This chapter characterizes the panel survey experiment research design using the MIDA framework, which stands for model, inquiry, data strategy, and answer strategy. The theoretical model of information processing and the survey response is the expectancy value model. The inquiries are the conditional average treatment effects of persuasive information on target attitudes, conditioned on demographic and political traits. The data strategy involves probability and convenience sampling of diverse online survey respondents, random assignment of these subjects to treatment conditions, and post-treatment survey measurement of attitudes. The answer strategy is conditional difference-in-means. The chapter defends this research design against alternatives like mediation analysis and addresses concerns related to the interpretation of the word "parallel." (pages 52 - 73)
This chapter is available at:
https://academic.oup.com/chica...
Chapter 5: Persuasion Experiments: Originals, Replications, and Reanalyses - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0005
[survey experiment;meta-analysis;two-sided messages]
This chapter describes each survey experiment as a case study first, then meta-analyzes all the studies together. The chapter culminates in Figure 5.16, which provides strong evidence for the persuasion in parallel hypothesis by showing how strongly correlated the effects are across various subdivisions of society. If the effect is large and positive for Democrats, the effect for Republicans is likely to be large and positive as well; similar patterns hold for liberals and conservatives, White and Black Americans, men and women, younger and older Americans, and the better- and less-educated. The chapter also shows how the effect of exposure to two-sided messages is well-approximated by the sum of the effects of each message, which is to say they mostly cancel. (pages 74 - 105)
This chapter is available at:
https://academic.oup.com/chica...
Chapter 6: Persistence and Decay - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0006
[persistence ratio;persuasive information;emphasis frames;model;inquiry;answer strategy]
This chapter tackles the question of how long the effects of persuasive treatments endure. It describes how extending the design through time requires changes to the model, inquiry, data strategy, and answer strategy. Our inquiry is the "persistence ratio," or the ratio of the average effect at time two to the average effect at time one. The average estimate of the persistence ratio across 12 panel studies is 34%. Persistence is stronger among treatments that operate primarily through new persuasive information and weaker among treatments that operate primarily through emphasis frames. (pages 106 - 120)
This chapter is available at:
https://academic.oup.com/chica...
Chapter 7: Models of Information Processing - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0007
[motivated reasoning;Bayesian reasoning;nonfalsifiable]
Motivated reasoning theories posit that information processing is guided by a mix of accuracy and directional goals; which set of goals dominate when encountering persuasive information will determine the direction of the treatment effect. Bayesian reasoning theories posit that information processing occurs by combining evidence, prior beliefs, and likelihood functions to arrive at posterior beliefs. The chapter explains that both theories are nonfalsifiable with evidence about the treatment effects of persuasive information, since both can accommodate treatment effects of any sign or magnitude. However, since the effects of persuasive information are apparently always positive (as shown in Chapter 5), we can draw conditional conclusions. If people are motivated reasoners, accuracy goals dominate. If people are Bayesian reasoners, they have "reasonable" likelihood functions. (pages 121 - 139)
This chapter is available at:
https://academic.oup.com/chica...
Chapter 8: Persuasion Is Possible - Alexander Coppock
DOI: 10.7208/chicago/9780226821832.003.0008
[political implications;differential exposure;persuasive information;target attitudes]
This concluding chapter insists on the political implications of this work: despite claims to the contrary, it is not counterproductive to try to persuade the other side. The chapter discusses three cognitive errors that make it seem like persuasion is counterproductive: mistaking levels for changes, mistaking affective evaluations for target policy attitudes, and mistaking differential exposure for differential treatment response. Persuasive attempts work in the sense of changing target attitudes, so we should continue to deliver persuasive information to everyone, not just those who are amenable to our perspectives. (pages 140 - 142)
This chapter is available at:
https://academic.oup.com/chica...
