At the risk of creating an expectation for edification that we’ll never again approach satisfying, CCP Blog again brings you an exclusive guest post by a foremost scholarly expert on an issue that everyone everywhere is astonishingly confused about! The expert is political scientist Kevin Arceneaux of Temple University. The issue is whether partisan cable news and related media outlets are driving conflict over climate change and other divisive issues by misinforming credulous members of the public and otherwise fanning the flames of political polarization. I’ve questioned this widely held view myself (see, e.g., here & here.) But no one listens to me, of course. Well now Arceneaux–employing the novel strategy of actually bringing evidence derived from valid empirical methods to bear–will straighten everything out once and for all. His post furnishes a preview–again, exclusively for the 14 billion readers of the CCP Blog!–of his soon-to-be-published book, Changing Minds, Changing Channels (Univ. Chicago Press 2013), co-authored with Martin Johnson. (Psssst … you can actually download a couple of chapters in draft right now for free! Don’t tell anybody!)
Kevin Arceneaux:
There is little doubt that the American legislative process has become more partisan and polarized. But is the same true for the mass public? For the most part, it seems that most Americans remain middle of the road. Rather than becoming more polarized, people mostly seem to have brought their policy positions in line with their partisan identification.
Despite the empirical evidence, many—especially pundits—cannot shake the notion that Americans are becoming more politically extreme and divided. Not only do many in the chattering class take mass polarization as a self-evident fact, the culprit is equally self-evident: the partisan news media.
On some level, I understand why this is such a popular conclusion. If political elites are so polarized, and clearly they are, it only seems intuitive that the same must be true for the mass citizenry. What’s more, people tend to overestimate the effects of media content on others, and what is the mass public if not masses of other people.
Nonetheless, in our soon-to-be published book Changing Minds or Changing Channels, Martin Johnson and I challenge the conventional wisdom that Fox News and MSNBC are responsible for polarizing the country.
We must keep in mind that in spite of their visibility to people like us who are politically engaged, relatively few people tune into shows like The O’Reilly Factor or The Rachel Maddow Show. For instance, voter turnout in the 2012 presidential election was roughly 12 times the size of the top-rated partisan talk show audiences on Fox News and MSNBC.
More important, people choose to watch partisan news audiences. The type of person who gravitates to partisan news shows is more politically and ideologically motivated than those who choose to watch mainstream news or tune out the news altogether, partisan or otherwise. People are not passive or particularly open-minded when it comes to political controversies. Not only do they choose what to watch on television, but they also choose whether to accept or reject the messages they receive from the televisions shows they watch.
In short, two forces simultaneously limit and blunt the effects of partisan news media. First, partisan news shows cannot polarize—in a direct sense—the multitude of Americans who do not tune into these shows. Second, the sort of people who actively choose to watch partisan news are precisely the sort of people who already possess strong opinions on politics and precisely the sort of people who should be less swayed by the content they view on these shows.
Wait—you may be thinking—don’t studies conclusively show that Fox News viewers know less about foreign events and express more conservative opinions on important policy issues like climate change?
The fact that people select into partisan news audiences also makes it difficult to study the effects of these shows. If people tune into Fox News because they care more about domestic political debates than foreign events or because they have conservative views, we would expect them to know less about foreign policy and distrust climate scientists even if Fox News did not exist.
What these studies do not and cannot tell us is the “counterfactual”: What would Fox News viewers know and believe about politics if we lived in a world without Fox News?
The counterfactual is, of course, unknowable, and the central goal of causal inference is finding a way to estimate it. It turns out that observational designs do a terrible job at this.
Consequently, Martin and I turned to randomized experiments to investigate the effects of partisan media. By randomly assigning subjects to treatment and control groups, we are able to simulate the counterfactual by creating equivalent groups that experience different states of the world (e.g., one in which they watch Fox News and one in which they do not).
Using randomized experiments to study media effects has a long and successful history.
However, without modifications, the standard experimental design that assigns one group to a control group (e.g., no partisan news) and another group to a treatment group (e.g., partisan news) would not help us understand how selectivity—these choices we know viewers are making—influences the effects of partisan news shows. Forced exposure experiments (as we call them) allow one to estimate the effects of media content under the assumption that everyone is exposed to it. The current media environment, rife with abundant choice, makes it impossible for anyone to assume even a majority of viewers are exposed to a type of program, let alone everyone .
So, we modified the forced exposure experiment in two ways, which I’ll describe in turn.
The first modification involved creating of a research design we call the Selective Exposure Experiment to compare a world where people had to watch partisan news to one that more closely approximates the one in which we live, where people can choose to watch entertainment programming instead. This experimental design starts with the forced exposure experimental design as its foundation. We randomly assigned some people to watch partisan news and some people to a control group where they could only watch an entertainment show.
These conditions allow us to estimate the effects of partisan news if people had no choice but to watch it. To get at the effects of selectivity, we randomly assigned a final group of subjects to a condition where they could watch any of the programs in the forced exposure conditions at will. We gave these subjects a remote control and allowed them to explore the partisan news programs and entertainment shows just as they would at home. They were free to watch all of a show, none of it, or flip back and forth among shows if that’s what they wanted to do.
The Selective Exposure Experiments taught us that the presence of choice blunted the effects of partisan news shows. To take one example from the book, we conducted an experiment in which some people watched a likeminded, or proattitudinal, news program (e.g., a conservative watching Fox) about the health care debate back in 2010; others watched an oppositional, or counterattitudinal, news program (e.g., a liberal watching Fox) on the same topic; others watched basic cable entertainment fare, devoid of politics; and finally, a group of subjects were allowed to choose among these shows freely.
The figure below summarizes the results from this Selective Exposure Experiment. The bars represent how polarized liberals and conservatives are after completing the viewing condition.
Across a number of aspects in the health care debate—how people rate the major political parties to deal with the issue, the personal impact of the policy, and the wisdom of the public opinion, individual mandate, and plan to raise taxes on the wealthy—forced exposure to both pro- and counterattitudinal shows increased polarization. So, it is clear that partisan shows can polarize.
However, subjects in the choice condition were much less polarized. Keep in mind that subjects in the choice condition only had four options from which to choose. Had we given subjects over 100 channels to choose from, as is commonplace in most households today, we can only imagine that these effects would have been even smaller.
Next, we wished to sort out why we observed smaller effects in the choice condition. Undoubtedly, part of the explanation has to be that with fewer people watching, one should observe smaller overall effects. Recall, though, that we also anticipate that those who seek out partisan news—news-seekers as Markus Prior calls them—should be less susceptible to partisan news effects.
It was to investigate this hypothesis that we devised our second modification of the standard forced-exposure experiment.
In a design we call the Participant Preference Experiment, we measured people’s viewing preferences before randomly assigning them to view a proattitudinal, counterattitudinal, or entertainment show. Measuring viewing preferences before exposure to the stimuli allows us to gauge whether news-seekers react differently to partisan news than entertainment-seekers.
The figure below shows the results from one of these experiments. The news programs in these experiments focused on the controversy around raising taxes on the top income earners. Across a number of issue questions on the topic, we find that partisan news shows do more to polarize entertainment-seekers forced to watch the partisan news program than it does among news-seekers who often watch these shows.
Note that the proattitudinal program had almost no effect on news-seekers, while the counterattitudinal show did. If people tend to gravitate toward likeminded news programming and entertainment seekers tend to tune out news, then these findings suggest that the direct effects of partisan news should be minimal.
As an aside, notice that the counterattitudinal news programming across all of these studies, if anything, polarizes those who are forced to watch it. Not only is this finding consistent with our thesis that people are not passive, blank slates (they can reject messages with which they disagree!), but it also undermines the Pollyanna notion that if people would just listen to the other side, the country would be a more tolerant and moderate place.
Finally, let me be clear that Martin and I are not arguing that partisan news shows have no effects. For one, they seem to lead many people to perceive that the country is more polarized, even if it isn’t. For another, they may have indirect effects on politics by energizing viewers (if not changing their minds) to contact their elected officials and vocalize their extreme opinions. Fox and MSNBC may indeed be a polarizing force in politics, but it is unlikely that it is causing masses of people to be more and more extreme.