Junk Science

What is it with some people's almost pornographic obsession with Fox News Channel?  (Disclosure: I almost never get my news from television.  I don't much care for it.)

The Washington Post reports today on a "study" that purports to show that George Bush might "owe" his 2000 election victory to FNC:

"Our estimates imply that Fox News convinced 3 to 8 percent of its audience to shift its voting behavior towards the Republican Party, a sizable media persuasion effect," said Stefano DellaVigna of the University of California at Berkely [sic] and Ethan Kaplan of Stockholm University. In Florida alone, they estimate, the Fox effect may have produced more than 10,000 additional votes for Bush -- clearly a decisive factor in a state he carried by fewer than 600 votes.

My, what a modest claim to make!  But it sounds to me like a textbook case of a "Post Hoc Fallacy."  Event B occurs after Event A; therefore, Event A must be the cause of Event B.

What kind of bullshit science is this, anyway?  Actually, it is a 51-page piece of bullshit science called "The Fox News Effect: Media Bias and Voting."

Admittedly, I have not yet read the whole paper.  And maybe the authors think I am supposed to be impressed and/or intimidated by things like this:

Or perhaps they want to lull me into submission with stultifying passages like: "The Fox News effect could be a temporary learning effect for rational voters, or a permanent effect for voters subject to non-rational persuasion."

But the fact remains that, nowhere in the "study" (to my reading) or in the related media reports have they established cause, only contemporaneousness.

The argument is that Fox News was the reason that people voted more conservatively.  But couldn't the opposite be just as true?  That is, couldn't Fox owe its existence to a rightward political trend that was already in progress?

Because FNC was created in 1996, why should I not claim that the "Republican Revolution" of 1994 was the "cause" of Fox News?  Only two years separated those two events.  What explains the electoral bath the Republicans took in 1998, only two years after the creation of Fox?

And how would the authors explain the countering effect of the measured left-leaning bias of almost every other media outlet, whose combined reach is infinitely greater than Fox?  The answer is, they don't, and their failure to do so makes their agenda all the more transparent.

Admittedly, it was a great way for DellaVigna and Kaplan to get publicity, especially among the vast numbers of reporters who loathe Fox.  But if a guy like me with only one college-level statistics course under his belt can see through them, why can't the WaPo?

Comments

[Yum Yum]

What I think would be better to study is how CBS, ABC and NBC convinces their audience to shift its voting behavior to the Democratic party.

I think it was actually done before and it was shown to be about 15%.

A bunch of hypocritical whiners if they're complaining about Fox News.

[Yum Yum]

And besides, so what if Fox convinces its viewers to change voting behavior? Isn't that how democracy works? If you want change, you have to convince your fellow countrymen to agree with you. And if you are successful in persuading your fellow citizens, then so be it. I don't like the implied assumption that Americans can't think for themselves and are herded like cattle by the "evil corporate media." When those kinds of arguments gain acceptance, what invariably follows are restrictions on free speech.

[Aatom]

wait, you mean the press has an affect on voters??? that is CRAZY!

isn't that like, the whole fucking point of a free press??

seriously retarded.

[Aatom]

wait, you mean the press has an affect on voters??? that is CRAZY!

isn't that like, the whole fucking point of a free press??

seriously retarded.

[North Dallas Thirty]

It's the same reason that Democrats scream "voter fraud" every time they lose an election.

The rest of the media doesn't understand why Fox News generally wipes the floor with them from a ratings standpoint; "mind control" is an answer that keeps it from being anything wrong with them and points the finger of blame solely at Fox News.

[anapestic]

Certainly the press, and information generally, has an effect on voters, but a news outlet is not supposed to have an agenda, especially a hidden agenda.

It does seem more likely to me that people who were going to vote Republican gravitated to Fox than that Fox caused anyone to vote Republican. There may, however, be ways to control for that phenomenon when you're analyzing the data, so without a stronger understanding of the statistical methods involved, you can't really dismiss the study out of hand.

Just from looking at the abstract, it would appear that the authors are comparing locations where Fox News was available as of 2000 with locations where it wasn't. This methodology would appear to provide a reasonable control, rather than a simple temporal analysis, as you seem to be implying.

[Josh]

Like most lazy social science, the authors confuse correlation and causation. While the two events—a general swing to the right, AND and increased viewership of Fox News—occurred at the same time, they were unable to prove a causal relationship between viewership of FNC and a change in voting behavior.

Anapestic, simply comparing locations where FNC is available doesn't work—a real study of this nature would require a frighteningly complex regression analysis with a huge N to have any degree of rigor. Unless the authors interviewed a statistically significant number of viewers of Fox News, and determined in each that their watching is what changed their voting behavior (as opposed to other causal factors, like Al Gore being a douchebag), they have zero case. It's not a "reasonable control," it's "baseless supposition."

And "agenda free news?" What rock have you been living under? In the very first paragraph of the article, the authors proclaim that media bias is endemic to supposedly neutral sources all over the map. Media bias is more or less a given in most studies.

Oh, and their formulas for estimating voter bias are made up... none of the other democracy literature I've read has used a similar methodology. I think this is because the two authors are economists who are trying to apply rational theories to sociology and political behavior, two things economic theory is particularly ill-suited for. Vernon Smith won his Nobel in Economics in 2002 through his study of non-rational economic behavior and alternative market mechanisms, a methodology which would be better suited to a study of media bias and its impact on voting behavior. What DellaVigna and Kaplan resort to are four decade old Keynesian models of economic behavior, which have since been discredited.

So yes, Mal, it is junk science.

[Tommy]

Mal - you should try "NewsHour With Jim Lehrer." I love it, at least at those times when I can devote a whole hour away from internet porn.

Tommy - Athens Greece

[Aatom]

i don't think there's anything reasonably 'hidden' about Fox's agenda, Anapestic.

i also think the quaint notion of media "objectivity" has been finally shown as the naked Emperor it always has been. the rise of blogs and opinion journalism, and most notably in the perversely satirical Fox slogan "Fair and Balanced", has all amounted to pulling the curtain back on the Wizard so to speak. So, if anything, I would CREDIT Fox for correcting some dangerously outdated ideas about what the role of journalism has been and should be in this country.

[anapestic]

I'm not aware that Al Gore was any more or less of a douchebag in markets where Fox News was available than in those where it wasn't. I'm skeptical that they can give a reasonable estimate of the number of votes that are caused by any one factor, but if their claim that the availability of Fox News is not associated with some other causative factor is valid, then they wouldn't need a complicated survey to demonstrate an effect.

Also, there is a clear distinction between bias and agenda. The statement that I was reacting to (and please note that I use quotations to indicate what was actually written, not my interpretation of what was written) was "And besides, so what if Fox convinces its viewers to change voting behavior? Isn't that how democracy works? If you want change, you have to convince your fellow countrymen to agree with you." I certainly don't expect Fox News or any news outlet not to have a bias, but agenda is something else. If Fox was actively trying to convince its viewers to vote for Bush (something I'm not willing to accuse them of, by the way), then it really should have disclosed that intention up front.

[Dan]

It does seem like a pointless exercise. Spending all that time and money to prove that, yes - the media does have an impact on voting patterns (no!) and yes - FNC is a mouthpiece of the right (shocking!).

It's why I've always loved their whole "Fair and Balanced" tagline. It elevates irony to an artform.

[Queer Conservative]

Next thing you know they'll try and tell us the New York Times actually checks its facts before going to press...

[Josh]

Anapestic, do you have anything more than vague, undefinable doubts and skepticism? What is the basis for what you're saying? Because it looks like you're searching for a hypothetical on which to validate the study, when the study itself uses highly suspect methods and assumptions.

[anapestic]

I have taken courses in statistics, but I am not a statistician. Neither are you, Josh, and neither is Mal. It is clear to me from the abstract that they have attempted to separate causation and correlation, and it is also clear to me that there are valid statistical ways to achieve that separation. Whether they've successfully achieved their goal is something that I lack the expertise to determine. I am not willing to state that something is bullshit science when I don't have the expertise to demonstrate that it's bullshit science. I don't believe either of you has that expertise either, but it doesn't seem to stop you.

[Malcontent]

Anapestic: It doesn't take a PhD to know that 90 percent or more of what emerges from "academia" these days is, indeed, bullshit.

[Tommy]

Is that new?

Tommy - Athens, Greece

[Josh]

Anapestic, this has nothing to do with statistics, it has to do with political science. The authors of this study are neither sociologists (who are trained to study group behavior) nor political scientists (who study political behavior). They are economists trying to to use an outdated methodology to prove a non-existent point. In the process (and I actually read the study) they were totally unable to separate correlation and causation, precisely because they couldn't isolate viewership. You should know that more than what channel one watches goes into the decision to vote for one candidate or another. While the authors showed a correlation between republicans and viewers of Fox News, that isn't exactly surprising, and, as Mal said, does do anything to demonstrate the relationship. Which one caused the other?

My most serious point, though, is that this study was just damned sloppy.

[Dan]

Anapestic: It doesn't take a PhD to know that 90 percent or more of what emerges from "academia" these days is, indeed, bullshit.

Such venom, mal. I shouldn't be suprised, given the animus most republicans have towards anybody or anything that's the least bit enlightened.

I know it's a snappy one-liner, but how many academic journals have you read recently? When's the last time you've even stepped foot on a campus? When is the right wing gonna learn this hatred for all things educated is, ultimately, self-defeating?

I'm not saying there isn't some high-grade BS going on in academia. Like any field, it has it's fair share of whackos and nutjobs. But ATC, I think their rep's a little better than that of your former profession.

[Malcontent]

I know it's a snappy one-liner, but how many academic journals have you read recently? When's the last time you've even stepped foot on a campus? When is the right wing gonna learn this hatred for all things educated is, ultimately, self-defeating?

It's not easy to be both so condescending and presumptuous in such a short paragraph, but you managed nicely.

[blewsdawg]

What I think would be better to study is how CBS, ABC and NBC convinces their audience to shift its voting behavior to the Democratic party.

I think it was actually done before and it was shown to be about 15%.

It has been done, and Newsweek's Evan Thomas actually cited the 15 percent figure.

The election coverage from Big Media has been unusually partisan this time around. As Newsweek's Evan Thomas famously remarked: "Let's talk a little media bias here. The media, I think, wants Kerry to win. . . . They're going to portray Kerry and Edwards as being young and dynamic and optimistic and there's going to be this glow about them . . . that's going to be worth maybe 15 points."

[Aatom]

"But ATC, I think their rep's a little better than that of your former profession."

OH, SNAP!! girl fight!!!

[Josh]

Dan, in Mal's defense, I read quite a few academic journals, in political science no less, and they are indeed 90% bullshit. In more than one class, criticizing the bad assumptions, methods, and findings of these articles is a treasured pasttime. Being able to critically read subject literature critically, without being in awe of what "an academic" said is a useful life skill. A professor is not infallible by any stretch of the imagination, and is more often wrong than right.

It is also significant that this paper has not been published. I would be surprised if the authors could get this kind of work past any peer review at any respectable polisci journal. Certainly not anything by the APSA and its associated guilds.

[Tommy]

"It is also significant that this paper has not been published."

Exactly, so why the kerfuffle?

[Aatom]

yay! kerfuffle! what an awesome word.

[Josh]

Oh, the kerfluffle because a major newspaper is trumpeting it as proof of something nonexistent. Because way more people care what the Washington Post says than what Perspectives in Politics or The American Political Science Review refuses to publish, their praise of the study (or even their mention as if it were credible) is a very big deal.

Without the Post's input, it would be a meaningless academic debate. With the Post's input, it is a public issue with consequences.