"Proving" Media Bias - The Right Pretends To Do ScienceCourtesy of my good friends at NewsBusters, I see that the crack team of Groseclose and Milyo are once again touting their pretend science project that purports to identify and measure Liberal bias in the news media. Basically, the paper claims to scientifically prove that The Drudge Report is neutral, Fox and the Washington Post are to the Right, and everything else is to the Left.
I was first introduced to this work roughly a year ago, when I was sent a link to their paper by a Right-leaning friend of mine. I actually took the time to read the 29 pages in the hope that I might learn something - instead, it simply confirmed my hypothesis that the Right is science-challenged.
Here are my comments from my original read of this work a year ago - I'll reread and update my comments if I get a chance:
Some of their methodology is interesting, some is way off, and its conclusions are extremely unsubstantiated by the findings. Based on the quality of the work, if this was an article in a medical journal it would end up in a fourth-rate journal, next to the article that "proves" that amalgam fillings are killing us by demonstrating that mice die when their little heads are submerged beneath the surface of a pool of mercury for an hour.
For starters, the authors didn't even attempt to justify their methodology of somehow mapping news outlets to the ADA rating for members of Congress. There are easier and better ways to measure bias, so why was this chosen? Given that the authors are from the fringe-right (betrayed by their adoption of the bizarre affectation of substituting "Democrat" when "Democratic" would be correct), I'll bet we know why - it simply "proved" their point better than any reasonable methods they tried.
Second, they don't include any data that tells us whether the differences observed between media outlets were statistically significant or not. Even when when one is measuring what one purports to be measuring, a test of statistical significance is needed to know if any measured differences were likely due to a real affect or dumb luck - this measurement is a fixture of all actual scientific papers. The really stupid thing is that their probably was statistically significant, but they probably couldn't figure out how to do the math.
As far as I can tell, here are the only conclusions that one might possibly draw from this work:
1. Most media outlets studied are pretty close to one-another, bias wise. (Seems correct)
2. Drudge is significantly to the right of "the pack". (Seems correct)
3. Fox is significantly to the right of the Drudge. (Seems correct)
4. So, either all outlets are biased, or only one of the above three groups is unbiased.
5. Based on ADA ratings, Republicans in Congress are far further to the Right than Democrats are to the Left. (This supports the observation that the Republicans have become an extremist fringe element.)
The article utterly flys off the rails when it somehow claims that it can measure absolute bias (i.e., deviation from reality) rather than relative bias (i.e., deviation from each other). Unless I'm missing something, there's no way to measure absolute bias without having a measure of the absolute truth to compare to. Being from the fringe-right, these fellows won't let reality get in the way (no sir!), and they proceed to wander off into Wonderland.
A very simple example will show how far off these muttonheads are. Suppose that, in aggregate, the "Liberal" think tanks issue twice as many press releases per year as do the fringe-Right think tanks. One would expect that the responsible news media, whose duty is to simply report the news, will have something like twice as many references to "Liberal" think tanks as compared to Right-wing think tanks. This would simply be accurate reporting. However, based on this "scientific" paper's methodology, the press would be rated as extremely "liberal" for just reporting the straight truth.
Feh.
(from my site,
http://www.blueworksbetter.com)