fox/msnbc parity

A study making the rounds from the Farleigh Dickinson University polling outfit is headlined “What you know depends on what you watch.” It’s a national version of the same group’s earlier study of New Jersey residents, both of which purport to show that

exposure to partisan sources, such as Fox News and MSNBC, has a negative impact on people’s current events knowledge

compared not just to viewers of high-knowledge sources like the Sunday morning talk shows, NPR, and the Daily Show, but even to non-viewers.

Color me skeptical. According to data from Pew, which my colleague Neal Caren thoughtfully pulled out and sent to me, the difference in education between Fox and MSNBC viewers is quite substantial (fully 83% of MSNBC viewers have at least some college, compared to 60% of Fox News viewers). The FDU study

control[s] for the effects of partisanship, age, education and gender, all factors, which commonly predict vote choice.

which means, if I understand correctly, that MSNBC viewers start out smarter and stay that way, while Fox News viewers start out dumber and get a little dumber by watching Fox. (I’ve asked the FDU outfit for the raw data to see if I can model this alternative hypothesis but haven’t heard back yet.)

This all plays into the nice narrative that MSNBC and Fox are the same beast on different sides of the aisle, a claim that reproduces the dueling-viewpoints view of American politics and avoids the obligation to evaluate the actual content of the two. I suspect that a dispassionate consideration of MSNBC and Fox would reveal differences in the degree of factual (mis)information, insinuation, and misleading claims.

Author: andrewperrin

University of North Carolina, Chapel Hill

6 thoughts on “fox/msnbc parity”

  1. I am a little bit skeptical as well, because it seems like they are measuring general political knowledge. Given the general nature of the four international questions and the four domestic questions that they asked, I can’t imagine a reason why people who consume no news would be more likely to answer those questions correctly than someone who watches Fox or MSNBC.

    However, if they were talking about specific types of political knowledge that would be inconvenient to one side or the other, there is enough evidence/theory from other places to make me believe that it would be correct. The Fox News results have been replicated numerous times (for example, the PIPA polls from 2003 documenting gaps in knowledge about the Iraq War based on news source). In theory, MSNBC viewers ought to exhibit the same bias with respect to inconvenient information for leftists: for example, about the looming budget deficits for Medicare, public employee pensions, etc. I’m a fan of Markus Prior’s 2005 AJPS article demonstrating that increased media choice (from cable TV and the internet) widens gaps in political knowledge and participation because it allows people to reinforce existing preferences and reduces incidental exposure to news. I would be shocked if that same principle didn’t apply to different dependent variables, like specific kinds of political knowledge with ideologically-loaded implications.


  2. Local TV news, local newspapers, and evening news are the most consumed media sources, but they aren’t mentioned in the study writeup. (Blogs aren’t discussed either.) Since it’s rare for people to just follow one news organization, I wanted to see more about how Fox viewers, MSNBC viewers, etc. were operationalized as categories. For example, does someone who watches “Just MSNBC” ignore newspapers? Do Fox News viewers ignore local TV news? I see two possibilities:

    1) Categories of “Just MSNBC” “Just Fox News” “Just NPR” and so on are constructed in reference to other kinds of newer media; they ignore traditional media viewership. It could make some sense to treat people who just watch traditional news as a reference category, while people who watch traditional news + NPR as a comparison group. However, this methodology would make it difficult to attribute effects to a particular news organization.

    2) Categories of “Just MSNBC” and so on are literal; people in these categories really don’t consume any other form of news. As the Pew surveys and other studies of selective exposure show, exclusively consuming one form of news media is rare. People opting for all partisan media tend to visit friendly websites and blogs, along with cable. If categories like “Just MSNBC” are literal, they would have much smaller subsamples than the authors suggest.

    Andrew, to tie in to your hypothesis, my understanding of the literature is that there are more conservatives who just consume conservative media than liberals who just go to liberal outlets. Perhaps education predicts both media consumption patterns and how much information people get out of watching, while the news organization has an independent effect?


  3. Andrew,
    Dan Cassino – the PI on the study – here. We did control for differences in education, and there is a significant effect of education on the ability to answer political knowledge questions (no surprise there). We did this essentially because we’re not trying to figure out if one group is more informed than another, but to try and isolate the effects of media sources. For instance, people who listen to NPR do way better on the questions, but they’re also more educated and more white, so we have to control for all of the demographics we can to isolate the media effects (the demos also do a good job of controlling for the duration of media exposure, another potential confound).
    As for as the ability to answer the questions, overall, Fox News viewers were almost exactly at the average, as were MSNBC viewers. If we don’t control for anything, there’s no difference. This actually makes sense: FN viewers are whiter and male(r?) than average, and both of these factors help them. The dirty little secret is that there’s a fairly strong effect of partisanship or political ideology on the ability to answer these questions: Republicans/conservatives generally do worse than moderates and liberals. Not something we play up (again, we’re trying to compare the effects of the media sources), but it’s in there.
    This is also where the “only Fox” or “only MSNBC” categories come in. Almost no one reported just one media source – so to isolate the effects of media source, we run a logit with all of the demos and dummies for all of the reported media as IVs, with the number of knowledge questions answered correctly as the DV. The coefficients on the individual media sources are, therefore, the effect of consuming just that media source. We also had interaction effects for political ideology with the ideological media sources (Fox, MSNBC and Talk Radio), all of which were significant. Traditional media sources (newspapers, blogs, local tv news, national news broadcasts) were included, but had no significant effect (I’d guess that they were swamped by age, which is pretty highly correlated).
    As for parity between MSNBC and Fox, it isn’t quite what we’re claiming. For moderates, MSNBC had no significant impact on knowledge; for liberals, it helped a bit, and for conservatives it hurt. As for Fox, it had no effect on conservatives, but a negative effect for moderates and liberals. On the whole, MSNBC is better, but both have these strong ideological effects, and they’re similar on that basis.
    As for the micro-data, I’m not sure if I’ve seen the request, but we’re working on the peer-reviewed version now, and we can send it out once it’s in the pipeline (PR also gets very twitchy about us sending things out on high profile studies before we have to). I’m happy to send over the .log files, or run an additional analysis or two if you’re still interested.


  4. Dan, thank you for the thorough and thoughtful reply here. The report details the fact that the study controlled for education — but my point is that that control specifically masks an important part of the story. (I am NOT saying that you or the study did anything wrong, just that I suspect an alternative story explains the findings.)

    Your explanation above fits with my suggestion. To be blunt: I suspect that Fox takes dumb people and makes them a little dumber, and that MSNBC takes smart people and fails to make them any smarter. By controlling for education and for partisanship/political ideology, these effects end up looking quite similar, when the difference in knowledge is really essentially a selection effect. Viewers select into audiences based on education and partisanship, and become roughly as informed as their peer group because their peer group all selects into the same audiences.

    That’s why the suggestion of parity bugs me: it implies that viewers of MSNBC and Fox are equivalently uninformed because it controls away the selection factors that mean they’re not. The very first paragraph of the report highlights this parity by referring to “partisan sources such as Fox and MSNBC.”

    Finally, on the question of the data: On May 24 I sent an email to Dr. Woolley requesting the data. Of course I understand that there are priorities higher than my interest!


  5. Andrew,
    So, I’ve taken a look at the raw data, to compare the actual knowledge scores of individuals who say that the consume the various news sources. Here are the mean results (out of 9 possible correct answers, overall mean 3.4, sd 2.4):

    Local TV News: 3.4 (2.3)
    MSNBC: 3.4 (2.3)
    Fox News: 3.4 (2.3)
    Local Newspapers: 3.5 (2.4)
    CNN: 3.5 (2.3)
    National Evening News: 3.6 (2.3)
    Talk Radio: 4.0 (2.3)
    National newspapers: 4.1 (2.4)
    Sunday Morning News Shows: 4.1 (2.2)
    Websites: 4.3 (2.3)
    Daily Show: 4.4 (2.4)
    NPR: 4.9 (2.1)

    We’re not seeing any significant differences between Fox, MSNBC and CNN at all; nor are any of them significantly different than the overall mean. The high number for Talk Radio is a little surprising, but those individuals also have a higher than average number of other reported news sources. The websites (and I wish we could dig deeper into that category) result is largely driven by education/income.

    Let me know if you want me to look at anything else in the data, and I’ll make sure that Dr. Woolley sends the request through.



    1. Dan, this is compelling, thank you for providing it. I do suspect that clustering of news sources makes the independent effect of each be difficult to interpret, but these results definitely make my alternative hypothesis much less likely. Again, thanks.



Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.