Social Norms and Selectivity: Effects of Norms of Open-Mindedness on Content Selection and Affective Polarization

ABSTRACT

Given the polarizing effects of exposure to like-minded political information and the alleged societal benefits of exposure to diverse and dissimilar perspectives, we examine how to promote the selection of balanced and counter-attitudinal political content, thereby minimizing affective polarization. Two online experiments on American partisans (N = 389 and N = 1378) tested the influence of social norms promoting open-mindedness. Study 1 focused on supporters and opponents of President Trump, showing that highlighting social norm of open-mindedness (vs. close-mindedness) in the general public increased exposure to balanced articles. Study 2, which focused on Democrats vs. Republicans and on the social norms among one’s partisan ingroup, showed that highlighting ingroup open-mindedness (vs. control) enhanced counter-attitudinal exposure and indirectly attenuated several indicators of affective polarization. We discuss theoretical and practical implications for informed citizenship in an era of abundant media choice.

Acknowledgments

The authors would like to thank Lea Kremer, a Research Master student at the Amsterdam School of Communication, University of Amsterdam for her research assistance, and Tim Rogers from Gocardless.com for the website design.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ We see balanced content, which covers both sides of an issue, as distinct from neutral content that does not include any slant, or moderate content, which is slightly biased toward one side.

² In fact, the study employed a 3 (social norms: open-mindedness vs. close-mindedness vs. none) x 3 (expected discussion: ingroup vs. outgroup vs. none) between-subjects design. The second manipulation aimed to test whether expected political discussion with an ingroup or an outgroup member influences selectivity. We only focus on social norms because a randomizer error assigned participants according to a path diagram, which led to unequal distribution between the conditions and very small cell sizes in some combinations. Analyses with the fully crossed design do not allow for meaningful conclusions. However, all the models were re-estimated with the discussion factor. The results were parallel to those presented in terms of directionality, magnitude, and significance (see Appendix D). The unequal cell sizes across the conditions were also due to the randomizer error, such that participants were first assigned to norm vs. no norm groups, and later to close- or open-minded groups, resulting in a larger sample size in the control. Given the relatively large groups, this distribution was not deemed problematic.

³ An ANOVA with the composite norm salience scale found a significant difference between the open-mindedness (M = 4.05, SD = 0.54) and the close-mindedness (M = 3.29, SD = 0.75) conditions, F_{(1, 46)} = 16.09; p < .001, η_p² = .26. The control group in the main study did not receive the manipulation check statements because asking about social norms may have primed their salience and bias the results by indirectly treating the control group with the manipulations.

⁴ Results showed a strong effect of text type on the perception of how supportive the text is of the Trump administration (on a 5-point Likert scale), F_{(1.48, 42.93)} = 350.34; p < .001, η_p² = .92 (Greenhouse-Geisser correction). As expected, pro articles were seen as more supportive (M = 4.71, SE = .10) than balanced (M = 2.76, SE = .07) and contra articles (M = 1.22, SE = .09).

⁵ We estimated all the MANCOVA models without the covariates, finding nearly identical results (see Appendix E for p values for pairwise comparisons). Study 1: number of articles read: balanced F _{(2, 386)} = 2.59; p = .077, η_p² = .01; counter-attitudinal F _{(2, 386)} = .11; p = .895, η_p² = .001; pro-attitudinal F _{(2, 386)} = .06; p = .945, η_p² = .000; time spent: balanced F_{(2, 386)} = 3.12; p = .045, η_p² = .02; counter-attitudinal F_{(2, 386)} = .04; p = .963, η_p² = .000; pro-attitudinal F_{(2, 386)} = 1.10; p = .333, η_p² = .01. Study 2: number of articles read: balanced F_{(4, 1373)} = 1.79; p = .129, η_p² = .01; counter-attitudinal F_{(4, 1373)} = 2.17; p = .070, η_p² = .01; pro-attitudinal F_{(4, 1373)} = .70; p = .590, η_p² = .002; time spent: balanced F_{(4, 1373)} = 1.69; p = .149, η_p² = .01; counter-attitudinal F_{(4, 1373)} = 3.07; p = .016, η_p² = .01; pro-attitudinal F _{(4, 1373)} = 2.20; p = .067, η_p² = .01. We also re-estimated the mediational models without the covariates, finding similar results: in Study 1, the indirect effects were insignificant; in Study 2, we find an additional significant indirect effect of the ingroup open-mindedness treatment, which decreased all three affective polarization indicators compared to the control through time spent on counter-attitudinal articles.

⁶ There is nothing about the change of the title that should have influenced the results of Study 2. Also, because the red color of the mock news site in Study 1 could have created associations with the Republican Party, the colors of the site were changed to purple in Study 2.

⁷ The presented effects are not moderated by whether participants were supporters or opponents of President Trump (in Study 1) or the Democratic or Republican Party (in Study 2).

Additional information

Notes on contributors

Magdalena Wojcieszak

Magdalena Wojcieszak is a Professor of Communication at the University of California, Davis, and an Associate Researcher (PI, ERC EXPO) in the Amsterdam School of Communication Research at the University of Amsterdam.

Stephan Winter

Stephan Winter is a Professor of Media Psychology at the University of Koblenz-Landau, Germany.

Xudong Yu

Xudong Yu is a Ph.D. student in the Department of Communication at the University of California, Davis.