Over the last decade, a heated debate has emerged in political science and political economy about whether the US and Europe have become more polarised. For the US, some have argued that the overall distribution of ideology has not changed much since the 1970s (Bertrand and Kamenica 2018, Desmet and Wacziarg 2020, Fiorina et al. 2006, Fiorina 2008), while others contend that ideological differences between partisans have dramatically increased (Abramowitz and Saunders 2008, Abramowitz 2018, Boxell et al. 2020, Gentzkow 2016, Iyengar et al. 2012). But one point scholars seem to agree on is that beliefs about other people’s political attitudes are substantially mis-calibrated, and perceptions of partisan differences systematically exaggerated (Ahler and Sood 2020, Bordalo et al. 2016, Westfall et al. 2015).
The extent to which voters hold distorted beliefs about partisan differences may have profound social and political implications. Larger partisan differences are associated with higher political engagement (Westfall et al. 2015). They can spill over into political attitudes, inducing individuals to adopt more extreme positions and creating a link between perceived and actual polarisation (Bonomi et al. 2020). Moreover, exaggerated partisan differences have the potential to reduce social cohesion, increasing ‘affective polarisation’ and social conflict (Iyengar et al. 2018, Mason 2014).
Our recent work (Bordalo et al. 2020) examines what drives the distortions in beliefs about others’ political attitudes in the US electorate. We build on two empirical regularities already documented elsewhere. First, voters who identify more strongly with one party are more likely to exaggerate partisan differences (Westfall et al. 2015). Second, individuals exaggerate partisan differences more on issues for which extreme attitudes are more stereotypical of each party (Bordalo et al. 2016). These facts indicate that individual and issue characteristics both influence people’s beliefs about partisan differences, but they miss a key intuition – individuals are not equally interested in all issues. Since issue salience varies over time and across people, it may affect the process of belief distortions among voters and have important political and social implications.
Using nationally representative survey data from the American National Election Studies (ANES), we document two stylised facts. First, we show that the gap between perceived and actual differences is larger on issues that respondents consider more important and pressing for the country. That is, individuals hold more distorted beliefs on dimensions that are more salient to them – a pattern that holds regardless of the strength of partisan affiliation and other demographic characteristics. Second, we document that both political engagement and affective polarisation are associated with perceptions of partisan differences on the most salient issues. This suggests that issue salience predicts not only belief distortions, but also individuals’ political behaviour and their attitudes towards others.
Motivated by these empirical regularities, we formally study the role of issue salience in belief distortions by embedding it in a model where distortions can come both from individual characteristics, such as partisanship, and political stereotypes (Bordalo et al. 2016). Stereotypical beliefs about a group emphasize their distinctive or representative types (i.e. types that are not shared with other groups), and de-emphasise types that are common across groups. This leads to an overweighting of extreme types, and thus an exaggeration of true differences across groups. Because individuals are more attentive to salient issues, we allow – in line with evidence on sensorial perception (Nosofsky 1988) – issue salience to modulate the strength of stereotyping. As a consequence, stereotyping is stronger for more salient issues. For less salient issues, differences across groups may be less noticed or accessible, and stereotyping gets diluted.
Our framework generates testable predictions about the link between true partisan distributions of attitudes across issues, the salience of those issues, and beliefs. We begin by showing that shocks to issue salience alter beliefs about partisan differences, above and beyond any change in actual differences. We exploit a major shock to issue salience – the 1991 end of the Cold War – that, from the US perspective, represented a dramatic and unexpected reduction in the salience of external threats. As shown in Figure 1, the share of US voters who considered national defence-related issues more pressing sharply decreased after 1991. Conversely, those who viewed domestic issues such as social welfare and race relations as most pressing nearly doubled after 1991, moving from 45% to almost 80% in less than four years.
Figure 1 Issue salience among ANES respondents, 1980-2004
Notes: Share of respondents who consider external threats and diplomatic issues as the most pressing issue facing the US (solid black line) and share of respondents who consider domestic issues such as social welfare and race as most pressing (dashed grey line). The sample is restricted to ANES respondents during the 1980 and 2004 waves who were either Republicans or Democrats.
Source: adapted from Bordalo et al. (2020).
Consistent with the predictions of the model, we find that as external threats faded away with the end of the Cold War, exaggeration of beliefs on partisan differences dropped on external issues and increased on domestic ones. Both patterns are statistically and quantitatively significant, and are driven by movement in beliefs about the attitudes of party members rather than by changes in actual partisan differences. After 1991, perceptions of partisan differences (relative to actual differences) fell by about 25% on external issues and rose by more than 15% on domestic ones. We present these patterns in Figure 2, plotting the trend in actual (grey dotted line) and perceived (black solid line) partisan differences, normalised by their 1984 values, for external and domestic issues in the top and bottom panels, respectively.
Figure 2 Perceived and actual partisan differences
a) External issues
b) Domestic issues
Notes: Perceived (black solid line) and actual (grey dotted line) differences between Democrats and Republicans on the issue of defence spending (top panel) and domestic issues (bottom panel). The sample is restricted to ANES respondents during the 1980 and 2000 waves who were either Republicans or Democrats. Actual and perceived partisan differences are normalized by their 1984 level.
Source: adapted from Bordalo et al. (2020).
We then show that issue salience impacts beliefs by modulating the strength of stereotyping. Again relying on the Cold War shock to issue salience, we document that domestic issues for which extreme attitudes were more representative of the respective parties before 1991 – and thus, latent stereotyping distortions were stronger – experienced a larger rise in belief distortions once their salience increased after 1991. We complement this evidence by leveraging within-issue variation in beliefs across individuals, showing that the interaction of self-reported issue salience and the representativeness of extreme attitudes significantly predicts individuals’ belief distortions. These patterns hold across a variety of subsamples (including independent voters) and are influenced by the attitudes of those who are demographically closer to an individual.
Notably, our results are not specific to the end of the Cold War, but apply more broadly. Indeed, we observe the exact opposite patterns after 2001, when the salience of external threats re-emerged after the 9/11 terrorist attacks and the subsequent Afghanistan and Iraq wars. Holding constant actual partisan differences, we show that not only did Americans shift their attention away from domestic issues and towards external ones after 2001, but they also significantly lowered (raised) perceived partisan differences on domestic (external) issues. To further assess the external validity of our findings, we estimate the model on the basis of the dynamics of aggregate beliefs observed around the Cold War shock. This exercise confirms that issue salience plays an important role, altering belief distortions by as much as 35%. The model also performs well when predicting changes in beliefs out of sample after the 9/11 shock. The estimated values obtained in our work are comparable to those from other settings, including laboratory evidence on beliefs about gender ability (Bordalo et al. 2019).
Our evidence indicates that beliefs about partisan differences reflect stereotyping modulated by issue salience. The fact that beliefs are more distorted towards stereotypical extremes for issues that are salient provides a novel, individual-level test of the stereotyping mechanism. Yet, since beliefs are inherently ‘equilibrium outcomes’ – shaped both by voters’ cognitive processing and by politicians’ and media’s messaging – our framework can also be interpreted as reflecting successful political entrepreneurs or media outlets communicating effective messages on partisan differences (Murphy and Shleifer 2004). Shocks to issue salience due to the end of the Cold War might have been accompanied by supply side responses in which party leaders or the partisan media took more extreme positions on domestic issues, as parties sought to establish new dividing lines (Glaeser et al. 2005). The truth is likely somewhere in between. However, our evidence suggests that for political persuasion to be successful, it should be aligned with and catered to voters’ latent stereotypes.
Our paper offers two broader insights into the drivers of perceived polarisation. First, it indicates that if issue salience changes, then beliefs about political groups can shift dramatically even when the underlying fundamentals undergo little change. Perhaps surprisingly, the malleability of beliefs is most pronounced among partisans who do not strongly identify with their party, and is evident even for independent voters. The link between beliefs about political attitudes and voting turnout raises the possibility that belief distortions are strongest among swing voters, who often decide electoral outcomes. Second, our findings suggest that the sudden appearance of an ‘external threat’ can unite a country by prompting citizens to perceive one another as more similar – a mechanism that social scientists have long suspected (see also Levendusky 2018). In our paper, we show that the removal of a common threat can have the opposite effect, inducing citizens to perceive each other as further apart on domestic issues, and possibly undermining a country’s social cohesion.
As we demonstrate, the appearance or disappearance of external threats need not change actual political attitudes and ideology in order to create shifts to national unity and citizens’ beliefs about each other. We suspect that if domestic issues become more salient at a time when actual polarisation also increases, the perception of polarisation could rise dramatically through a multiplier effect induced by the increase in issue salience.
Abramowitz, A I (2018), The Great Alignment: Race, Party Transformation, and the Rise of Donald Trump, Yale University Press.
Abramowitz, A I and K L Saunders (2008), “Is Polarization a Myth?”, The Journal of Politics 70 (2): 542–555.
Ahler, D and G Sood (2020), “Typecast: A Routine Mental Shortcut Causes Party Stereotyping”, Available at SSRN 3550117.
Bertrand, M and E Kamenica (2018), “Coming Apart? Cultural Distances in the United States Over Time”, Working Paper: 1–81.
Bonomi, G, N Gennaioli and G Tabellini (2020), “Identity, Beliefs, and Political Conflict”, Working Paper.
Bordalo, P, K Coffman, N Gennaioli and A Shleifer (2016), “Stereotypes”, The Quarterly Journal of Economics 131(4): 1753–1794.
Bordalo, P, K Coffman, N Gennaioli and A Shleifer (2019), “Beliefs about Gender”, American Economic Review 109(3): 739–773.
Bordalo, P, M Tabellini and D Y Yang (2020), “Issue Salience and Political Stereotypes”, NBER Working Paper 27194.
Boxell, L, J Conway, J N Druckman, M Gentzkow et al. (2020), “Affective Polarization Did Not Increase during the Coronavirus Pandemic”, Technical Report, National Bureau of Economic Research.
Desmet, K and R Wacziarg (2020), “The Cultural Divide”, Working Paper, February: 1–42.
Fiorina, M P and S J Abrams (2008), “Political Polarization in the American Public”, Annual Review of Political Science 11(1): 563–588.
Fiorina, M P, S J Abrams and J Pope (2006), Culture War? The Myth of a Polarized America, Longman Publishing Group.
Gentzkow, M (2016), “Polarization in 2016”, Working Paper, January, 1–23.
Glaeser, E L, G A M Ponzetto, and J M Shapiro (2005), “Strategic Extremism: Why Republicans and Democrats Divide on Religious Values”, The Quarterly Journal of Economics 120(4): 1283–1330.
Iyengar, S, D R Kinder, G Sood and Y Lelkes (2012), “Affect, Not Ideology”, Public Opinion Quarterly 76(3): 405–431.
Iyengar, S, D R Kinder, Y Lelkes, M Levendusky, N Malhotra and S J Westwood (2018), “The Origins and Consequences of Affective Polarization in the United States”, Annual Review of Political Science 22(1): 129–146.
Levendusky, M S (2018), “Americans, Not Partisans: Can Priming American National Identity Reduce Affective Polarization?”, The Journal of Politics 80(1): 59–70.
Mason, L (2014), “‘I Disrespectfully Agree’: The Differential Effects of Partisan Sorting on Social and Issue Polarization”, American Journal of Political Science 59(1): 128–145.
Murphy, K M and A Shleifer (2004), “Persuasion in Politics”, American Economic Review 94(2): 435–439.
Nosofsky, R M (1988), “Similarity, Frequency, and Category Representations”, Journal of Experimental Psychology: learning, memory, and cognition 14(1): 54.
Westfall, J, L Van Boven, J R Chambers and C M Judd (2015), “Perceiving Political Polarization in the United States”, Perspectives on Psychological Science 10(2): 145– 158.