Drivers of US political polarisation: Three stylised facts and their implications

Levi Boxell 25 August 2021



The US is becoming increasingly divided. While there are different measures and concepts of political polarisation, a similar pattern emerges across many of them – polarisation among the US electorate is the highest it has been in recent decades. 

The increase in polarisation is concerning because polarisation can increase social group homophily, alter economic behaviour, and reduce government effectiveness (Hetherington and Rudolph 2015, Iyengar et al. 2019). More recently, polarisation has influenced behavioural and public policy responses to the Covid-19 pandemic (Akovali and Yilmaz 2020, Ajzenman et al. 2020, Milosh et al. 2020).

The growing divisions in American politics have also created an extensive debate on the causes of mass polarisation. Shifts in the media environment, such as the social media and the internet (Lelkes et al. 2017, Sunstein 2017, Zenou 2019) or partisan cable news (Martin and Yurukoglu 2017, Duca and Saving 2017), are one set of potential drivers. Socio-economic factors, such as economic inequality (McCarty et al. 2008) or racial divisions (Valentino and Zhirkov 2017) are another. Growing divisions among political elites may also drive divisions among the electorate (Levendusky 2009).

When comparing the plausibility of these different explanations, it is useful to keep three stylised facts in mind: 

  • First, polarisation in the US has been growing since at least the 1980s (Iyengar et al. 2012, Boxell et al. 2017). 
  • Second, since 1996, polarisation in the US has grown at least as quickly for older age groups as for younger age groups (Boxell et al. 2017).
  • Third, polarisation has grown more quickly in the US than in many other OECD countries (Iyengar et al. 2012, Boxell et al. 2021). 

Below, I review each of these stylised facts in more detail along with their implications for what is driving mass political polarisation in the US. 

Stylised fact #1: Polarisation has been growing since the 1980s

Boxell et al. (2017) construct an index of eight different polarisation measures in the US. These measures include concepts of partisan animosity, ideological extremity, and partisan-ideological sorting. 

Figure 1 shows that this index of mass political polarisation has been increasing steadily since the 1980s. Many of the individual measures used to construct the index follow similar patterns—increasing since the 1980s or earlier. Iyengar et al. (2012) also find growing trends in partisan animosity, or affective polarisation, since the 1970s. 

Figure 1 Trends in polarisation index in the US

Source: Boxell et al. (2017).

Stylised fact #2: Polarisation has been growing at least as quickly for older age groups

Boxell et al. (2017) also compare trends in the polarisation index across different age groups. Figure 2 shows that, since 1996, political polarisation has increased more quickly among older age groups (65+) than among younger age groups (18-39). Again, these patterns are similar across many of the individual measures of polarisation used to construct the index.

Figure 2 Trends in polarisation index in the US by age group

Source: Boxell et al. (2017). 

Stylised fact #3: Polarisation has been growing more quickly in the US

One particularly important measure of polarisation is affective polarisation. Rather than focusing on partisans taking more extreme positions on policy issues, affective polarisation focuses on the extent to which partisans have negative feelings towards or dislike members from the other party. Affective polarisation is often measured on a ‘thermometer’ scale that asks how ‘warm’ or ‘cold’ the respondent feels towards a given party. 

Figure 3 reports trends in affective polarisation across 12 OECD countries (Boxell et al. 2021). The US has exhibited faster growth in affective polarisation than any of the other countries, and this difference is statistically significant for all countries except Switzerland. Iyengar et al. (2012) also show that respondents in the US have experienced greater increases in displeasure at the prospect of inter-party marriage of progeny between 1960 and 2010 than respondents in the UK.

These findings suggest the US is distinct in the pace of polarisation, and that polarisation is not a global phenomenon.

Figure 3 Trends in polarisation across 12 OECD countries

Source: Boxell et al. (2021). 

Implications for drivers of polarisation

What hypothesised drivers are most consistent with these stylised facts? If an explanatory variable is a major driver of polarisation, the simplest models would predict it to:

  1. be growing since the 1980s;
  2. have equal or greater exposure to older age groups as younger age groups; and
  3. be increasing in countries with growing polarisation and decreasing (or increasing more slowly) in countries with declining polarisation.

The set of models suggested by the above implications are restrictive. If important treatment effect heterogeneity or unobserved confounds are present, these implications need not hold.


Figure 4, which plots trends in online news consumption for 12 OECD countries against the corresponding trend in affective polarisation, shows no evidence for a positive correlation between trends in online news consumption and trends in polarisation. Moreover, internet and social media use have only been a significant portion of news diets since the 2000s, and younger age groups use these technologies more heavily than their older counterparts (Boxell et al. 2017). The three stylised facts above are inconsistent with each of these points. 

Figure 4 Trends in online news consumption

Source: Boxell et al. (2021).

In contrast, the broadcast media hypothesis is more consistent with these stylised facts. First, cable TV market shares have been rising in the US since the 1970s (Duca and Saving 2017). Second, consumption of cable TV is higher among the older age groups that have polarised at least as quickly in recent years (Martin and Yurukoglu 2017). Third, Figure 5 shows a positive, but insignificant, relationship between the growth of private 24-hour cable TV news channels and affective polarisation. Duca and Saving (2018) also show that the share of households with cable TV is roughly 70% higher in the US than the average across a set of five European countries.

Figure 5 Trends in private cable TV news consumption

Source: Boxell et al. (2021).

Other important differences exist in broadcast media internationally. The FCC Fairness Doctrine in the US, which required equal treatment of competing views on broadcast media (radio and TV), was abolished in 1987. In contrast, the UK still requires broadcast news to be ‘reported with due accuracy and presented with due impartiality’ (Ofcom 2017). And, in general, countries with declining polarisation spend more on public broadcast media than countries with growing polarisation, with the US spending the least on public broadcast media (Benson and Powers 2011). 

Overall, these facts suggest important differences in broadcast media markets that are consistent with some of the observed stylised facts for affective polarisation.


A second set of explanations focus on social and economic trends. Boxell et al. (2021, Figure 3) show that trends across countries in inequality, openness to international trade, or immigration are not consistent with these being major drivers of polarisation, as they have been increasing similarly in countries with rising and falling polarisation.

On the other hand, Figure 6 shows that the non-white share of the population has risen more quickly in countries with rising polarisation than those with falling polarisation. At the same time, partisan sorting by race has grown in recent years in the US (Mason and Wronski 2018). Divisions along racial lines are also consistent with splits among elites in the US that date back to the civil rights era (discussed below).

Figure 6 Trends in the non-white share of the population

Source: Boxell et al. (2021).


Elite polarisation, as measured by ideological differences in roll-call voting in the US, has grown dramatically since the 1960s (McCarty et al. 2008). While elites may be responding to shifts in the electorate, some decisions by elites, such as the southern realignment after the civil rights period, appear to be unique to the US and potentially lead rather than follow shifts in the electorate (Levendusky 2009, 2010). 

Consistent with this reasoning, Figure 7 shows that trends in elite polarisation are highly correlated with trends in mass affective polarisation. Overall, a model where the electorate takes cues from increasingly divided elites is consistent with the stylised facts.

Figure 7 Trends in elite polarisation

Source: Boxell et al. (2021).


Absent a precise theory for how a particular explanatory variable can be a primary driver of political polarisation while simultaneously conflicting with the stylised facts, we should be sceptical of claims linking such a variable to the rise in affective polarisation in the US. The above stylised facts suggest potentially fruitful avenues for future research examining the determinants of polarisation in the US (e.g. racial divisions, elites, cable TV) and avenues that may be less fruitful (e.g. internet, inequality, trade).


Akovali, U and K Yilmaz (2020), “Polarised pandemic response and Covid-19 connectedness across US states”,, 10 December. 

Ajzenman, N, T Cavalcanti and D Da Mata (2020), “Leaders’ speech and risky behaviour during a pandemic”,, 02 May.

Benson, R and M Powers (2011), Public media and political independence: Lessons for the future of journalism from around the world, Free Press: Washington, DC.

Boxell, L, M Gentzkow and J M Shapiro (2017), “Greater internet use is not associated with faster growth in political polarization among US demographic group”, Proceedings of the National Academy of Sciences 114(40): 10612—10617.

Boxell, L, M Gentzkow and J M Shapiro (2021), “Cross-country trends in affective polarization”, NBER Working Paper.

Duca, J V and J L Saving (2017), “Income inequality, media fragmentation, and increased political polarization”, Contemporary Economic Policy 35(2): 392—413. 

Duca, J V and J L Saving (2018), “What drives economic policy uncertainty in the long and short runs: European and U.S. evidence over several decades”, Journal of Macroeconomics 55: 128—145.

Hetherington, M J and T J Rudolph (2015), Why Washington Won’t Work: Polarization, Trust, and the Governing Crisis, Chicago, IL: University of Chicago Press.

Iyengar, S, G Sood and Y Lelkes (2012), “Affect, not ideology: A social identity perspective on polarization”, Public Opinion Quarterly 76(3): 405—431.

Iyengar, S, Y Lelkes, M Levendusky, N Malhotra and S J Westwood (2019), “The origins and consequences of affective polarization in the United States”, Annual Review of Political Science 22: 120—146. 

Lelkes, Y, G Sood and S Iyengar (2017), “The Hostile Audience: The Effect of Access to Broadband Internet on Partisan Affect”, American Journal of Political Science 61(1): 5-20.

Levendusky, M S (2009), The Partisan Sort: How Liberals became Democrats and Conservatives became Republicans, Chicago, IL: University of Chicago Press.

Levendusky, M S (2010), “Clearer cues, more consistent voters: A benefit of elite polarization”, Political Behavior 32(1): 111–131.

Martin, G J and A Yurukoglu (2017), “Bias in cable news: Persuasion and polarization”, American Economic Review 107(9): 2565—2599.

Mason, L and J Wronski (2018), “One tribe to bind them all: How our social group attachments strengthen partisanship”, Advances in Political Psychology 39(S1): 257—277.

McCarty, N, K T Poole and H Rosenthal (2008), Polarized America: The Dance of Ideology and Unequal Riches, Cambridge, MA: MIT Press.

Milosh, M, M Painter, K Sonin, D Van Dijcke and A L Wright (2020), “Political polarization impedes the public policy response to COVID-19”,, 23 December. 

Ofcom (2017), “Section five: Due impartiality and due accuracy”. 

Sunstein, C R (2017), #Republic: Divided Democracy in the Age of Social Media, Princeton, NJ: Princeton University Press.

Valentino, N A and K Zhirkov (2017), “Blue is black and red is white? Affective polarization and the racialized schemes of U.S. party coalitions”, Working paper. 

Zenou, Y (2019), “Social media polarization”,, 09 August.



Topics:  Politics and economics

Tags:  polarisation, news consumption, media, US politics

PhD candidate in Economics at Stanford University


CEPR Policy Research