top of page

Those with higher cognitive ability are more polarized in how they consume/interpret media(part 3/3)

Today’s blog is Part 3 of a 3-part series explaining some of the background and results for a paper recently accepted in the journal Intelligence. In it, we examine ability-based polarization in COVID-19 and in non-COVID topics. (Click for Part 1 and Part 2)

Shoots-Reinhard, B., Goodwin, R., Bjälkebring, P., Markowitz, D., Silverstein, M.D., & Peters, E. (forthcoming). Ability-related political polarization in the COVID-19 pandemic. Intelligence.

You can download the paper for free before October 9, 2021 by using this link.

In the UO-EPIDeMIC study, we found that political polarization in reported negative emotions and risk perceptions was more pronounced among those higher vs. lower in verbal ability. After controlling for it, numeracy-related political polarization disappeared.

We were interested in what types of processes might contribute to this polarization.

In several survey waves in 2020, we asked participants to indicate how much they get information about the coronavirus from four news sources that liberals trust (i.e., New York Times, MSNBC, Washington Post, and NPR) and two sources that conservatives trust (i.e., Fox News and Breitbart). As expected, people spent a greater proportion of time on sources that matched their political ideology. We tend to read and listen to people with whom we agree. This effect is the well-known selective exposure to information that can underlie motivated thinking. People higher in verbal ability did it more!

This greater political polarization in media consumption by those higher in verbal ability might underlie why higher-ability people became more polarized. By exposing themselves to different knowledge, they can might use it in turn to justify what they want to believe, their polarized beliefs and reactions. But selective exposure to information isn’t the only possible reason for polarization.

We were also interested in what would happen if people had access to the same information and had to interpret it. We thought that in addition to selectively exposing themselves to different information, people higher in ability might also interpret the same information to confirm their political views.

For example, we asked people whether or not they agreed that “If there are only a very few cases of coronavirus appearing in my community now, the risk in the future can be considered to be small.” Conservatives were more likely to agree than liberals, and especially when they were higher in ability.

These interpretations also may have contributed to the polarized responses to coronavirus, just as polarized media consumption did.

We found the same pattern when we asked people to interpret lower-than-expected COVID-19 deaths. Here, conservatives were more likely than liberals to attribute the reduction to the CDC overstating risks rather than social distancing reducing the risks. Similar results emerged in people’s confidence that they could go back to normal if they had a positive COVID antibody test (liberals were less confident than conservatives). In all cases, polarization was more pronounced for people higher in verbal ability, and the polarized interpretations accounted for the polarized emotional responses and risk perceptions.

To summarize, people higher in verbal ability both selectively consume news and interpret information in ways consistent with their political ideology. These political polarized ways of consuming and processing information then seem to contribute to their polarized beliefs.

The results point towards the notion that we can’t reduce political polarization simply by providing more information—people will avoid information they disagree with and interpret evidence to be consistent with their existing views. Instead, we need to investigate ways to encourage people higher in ability (and verbal ability, in particular) to accept information that conflicts with their views.

This research was supported by grants from the National Science Foundation (SES-20010000 and SES-2017651). Portions of these data will be published in the journal Intelligence ( can download the paper for free before October 9, 2021 by using this link.


bottom of page