Discussion Paper

No. 2019-28 | April 05, 2019
Media-driven polarization. Evidence from the US


Using US data of media’s coverage of politics and individual survey data, the authors document that in the states with a greater coverage of politics, citizens especially exposed to newspapers have more polarized preferences, partly coming from better political knowledge, and resulting in a higher political involvement measured as contributions to political parties and candidates.

Data Set

JEL Classification:

K4, H0


  • Downloads: 192


Cite As

Mickael Melki and Petros G. Sekeris (2019). Media-driven polarization. Evidence from the US. Economics Discussion Papers, No 2019-28, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2019-28

Comments and Questions

Michel Habib, Department of Banking and Finance, University of Zurich - Invited reader comment
April 23, 2019 - 08:24

This very interesting paper provides evidence that media, specifically newspaper coverage contributes to political polarization: coverage favors the transformation of political interest into affiliation with positions distinct from the center, whether liberal or strongly liberal on the left, or conservative or strongly conservative on the right. The paper addresses ...[more]

... the natural concern that polarization might induce coverage rather than coverage favoring polarization by means of a placebo test: the relation between political interest and polarization in 2000 was unaffected by media coverage in 2008.
As befits a good paper, the paper raises as many questions as it answers. How exactly does newspaper coverage contribute to transforming political interest into polarization? Why are only liberal and strongly liberal affiliations affected by media coverage, but not conservative or strongly conservative? In view of Martin and Yurukoglu’s (AER, 2017) finding that television favors conservative affiliation, it is perhaps tempting, but undoubtedly premature, to conclude that liberals read newspapers whereas conservatives watch TV.

Petros Sekeris - Reply to prof. Habib
May 15, 2019 - 17:23

We thank prof. Habib for these encouraging and constructive comments. Below we attempt providing a convincing answer to the two questions raised in this comment.

1. What is the channel through which newspaper coverage contributes to transforming political interest into polarization?
Our understanding, although we have not been so ...[more]

... far able to actually provide empirical evidence on it, is that individual political preferences become more extreme when people are confronted to news. This is a kind of self-selection mechanism whereby out of the several existing sources of information individuals select the ones more in line with their prior beliefs, and that eventually radicalizes their priors. Several studies back this reasoning. Campante and Hojman (JPubE 2013) clearly hint at such a channel when stating that "it is now often argued that new media such as cable TV or the Internet have increased polarization by enabling individuals to select outlets that conform to their prior ideologies as in an “echo chamber” (Bishop, 2008; Sunstein, 2009)’’. Lelkes et al. (AJPS 2017) provide further evidence in support of this argument when demonstrating that broadband access increases political polarization more through the volume of information consumed rather than through the content/partisanship of the news. In other words, on social networks, one tends to be exposed to news that reflect your own political beliefs, and more consumption of such information radicalizes one’s own beliefs. Lastly, Halberstam and Knight (JPubE 2016) show that news spread quickly in networks of the same political side, thus bringing further elements in support the self-selection mechanism we allude to.

2. Why are only liberal and strongly liberal political affiliations affected by media coverage, but not conservative or srongly conservative?
This is both a very relevant and interesting point, and we are very thankful for drawing our attention on this point. It does seem that your suggested channel, namely that liberals read more newspapers, is the driving reason. By running additional tests, we find some evidence that more conservative individuals pay less attention to newspapers while more liberal ones pay more attention. Interestingly, in terms of interest paid to the campaign, or whether they watch TV news, liberals and conservatives seem to behave in quite comparable ways. Lastly, as further evidence confirming the channel you are suggesting, we show that the effect on polarization is driven by the interaction between the presence of media and the “read newspapers” variable. We will certainly include all these elements and elaborate on it in a revised version of our paper.

With kind regards,

M. Melki & P. Sekeris

Anonymous - Referee report 1
May 27, 2019 - 13:06

I like the paper. It picks up on a hot topic in political economy, namely the relationship between media
coverage and polarization. They note that there have been contradictory results, but this paper find
evidence on the side of media coverage contributing to polarization. The paper sensibly acknowledges
the possibility ...[more]

... of reverse causality and rules this out with a placebo test which I am happy with. The
result is perhaps surprising. My prior would be that since the variable for media coverage is newspapers
(and that newspapers require more time and intellectual effort relative to TV or social media) that
higher newspaper coverage would lead to more moderation. The idea of greater newspaper coverage
requiring more intellectual effort seems borne out by it being positively linked to political knowledge. So
what is going on with the positive link to polarization? It seems that this is being driven by liberal
respondents. This is interesting because in the liberal world which academics like myself mainly inhabit,
every day discussions about polarization tend to view it as a bad thing (driven by ignorance) and tend to
think first of polarization as more salient on the right than left. This paper is a useful reminder that
polarization requires a left as well as a right and that it may correlate more with knowledge rather than

Anonymous - Referee report 2
May 27, 2019 - 13:09

see attached file