People complain that American mainstream media are becoming more and more polarized. There is a tradition in American journalism that the journalist should be objective and report the facts without judgment. Opinion pieces and Editorials are relegated to the back pages.
Nowadays those standards are eroding. Fox News, MSNBC, and CNN have discernible biases but still pander to the idea that they provide objective journalism. Meanwhile there is the perception that this trend is degrading the quality of information.
From a narrow perspective that may be true. I learn less from Fox News if they selectively report information that confirms the preconceptions of their audience. But media bias makes the media as a group more informative, not less.
Suppose I have a vast array of media sources which are scattered across the left-right spectrum. When a policy is being debated I look at all of them and find the pivotal outlet: all those to the left of it are advocating the policy and all those to the right are opposed. Different policies will have different cutoff points, and that cutoff point gives me a very simple and informative statistic about the policy. If the range is more narrow or more sparsely distributed this statistic is simply less informative.
Another way of saying this is that there is social value from having advisors with extreme biases. When I am thinking about a policy that I am predisposed to like, I learn very little from an unbiased source but I learn a lot if a source with my bias is opposed to the policy or a source with the opposite bias is in favor of it. It must be especially good or bad for these extremists to go against bias.
Cheap Talk 
9 comments
Comments feed for this article
April 9, 2012 at 9:08 am
anonymous
Do you know who first explicitly discussed this cool idea about the value of biased opinion? In political science I know Randall Calvert has a 1985 JOP article titled “value of biased information”. Do you know anything older than that?
April 9, 2012 at 11:01 am
twicker
If I’m understanding paragraph 4 correctly, then, if I, “look at all of [the vast array of media sources which are scattered across the left-right spectrum],” then I can, “find the pivotal outlet.”
This assumes:
A) That there’s a full spectrum of sources (and not, say, one cluster on the right and one on the left, with a discontinuity somewhere along the spectrum);
B) That the media sources are acting on the actual value of the issue and its underlying rationales when presenting their cases (and not just saying, “It was proposed by someone on the other side; therefore, it’s bad”), so they’re acting as purely informative, and not as political, players;
C) That there is, in fact, a pivotal outlet;
D) That I, as a normal person, will take the time to look across the entire vast spectrum;
E) That I will further take the time to specifically look for and at the pivotal outlet and the outlets just to each side of it (so I can determine where the pivot point is); and
F) That I’m open to disconfirming information.
Of these, I’m not convinced that A through F hold.
On the argument in paragraph 5, there’s *value* in having extreme advisors; however, that doesn’t mean that you learn very little from having an unbiased source. If the options on Issue X are a, b, c, d, e, and f (ranging from one side of the spectrum to the other), one biased advisor will scream, “AAAAAAAAA!!!!!!!” The other will scream, “FFFFFFFFF!!!!!!!!” The unbiased source will (hopefully) mention options b-e, as well. If I buy some of the philosophical arguments behind a, and some behind f, then knowing that options b through e exist actually gives me quite a bit of information. However, knowing that information does far less for developing the overall political power of the extremes, and is, in many ways, less entertaining (confrontation is almost always more exciting than cooperation, and painting everything as an existential “us or them” fight, even when it isn’t, is far more emotionally impactful, which leads to more viewers and more ad revenues, overall).
April 9, 2012 at 12:49 pm
Sebastien Turban (@PtitSeb)
I think the story changes a bit if we introduced confirmation bias, say a la Rabin and Schrag (1999). If I interpret some signals against my opinion as signals in favor of my opinion, polarization might make the problem worse. Also, Redlawsk (2002) for instance has evidence that when facing “incongruent information” (info against your prior, basically), people become more biased in the direction of their prior, and argue that this might be because they try to find an argument against the signal they’re facing and in favor of their prior.
April 9, 2012 at 12:59 pm
jeff
Comment from Rajiv Sethi:
Jeff, this is a really interesting idea, but it presumes that biases are
observable. In practice, the precision with which we estimate the biases
of others varies widely. Since it’s easier to interpret the opinions of
those whose biases are well-known to us, this sets up an interesting
trade-off between being “well-understood” and being “well-informed”.
When one is seeking information and can observe only a limited number of
opinions, it might be better to observe someone who is more poorly
informed provided that they are sufficiently well-understood.
This has some interesting implications, since past observation itself
increases the precision with which we estimate an individual’s biases.
If one thinks of observation (subscribing to blogs, following on
twitter, etc.) as generating a network, this trade-off can give rise to
interesting network structures displaying various degrees of hysteresis
even if individuals are ex-ante identical.
I’m working on a paper with Muhamet Yildiz in which we model this
process, based on heterogeneous and *unobservable* priors, building on
our earlier work on public disagreement. It’s not ready for circulation
yet, but in case you have interested readers in London I’ll be
presenting it here next month:
http://www.econ.qmul.ac.uk/events/2012/05/spring-2012-workshop-in-economic-theory
Any pointers to related literature would be welcome.
April 9, 2012 at 5:10 pm
econandcarryon
Picking up on assumption D from Twicker’s comment, it seems that individuals will find it difficult to take advantage of the spectrum of media sources to identify the pivotal viewpoint. While information collection is made easier through aggregation services like Google News, it is still the responsibility of the individual to sort through various accounts to find the salient differences to determine where the pivot is located. For many individuals, the sorting process is too costly and it is easier for them to rely on a single source for all of their news/information rather than consulting the range of opinions presented by the spectrum.
April 9, 2012 at 9:45 pm
Noto
OK, I give up. What does the picture (and the hidden text) have to do with media bias? I’m not clever enough to figure it out.
April 9, 2012 at 10:44 pm
jeff
The pictures are just random pictures and the mouseover texst are thoughts that are even too pathetic for Twitter.
April 9, 2012 at 10:56 pm
twicker
And it’s Jeff ftw …
April 10, 2012 at 7:32 am
ubiquitousuk
Schleifer & Mullainathan (2004) make a similar argument in their AER without (if I recall correctly) requiring that there exists a pivotal outlet.