"There is always a philosophy for lack of courage."—Albert Camus

Wednesday, January 11, 2006

More on Media Bias: I Mean, It's Serious.

Came across a fascinating article in Investor's Business Daily on dead tree (see, you accomplished something with that free trial).

The IBD staff compiled evidence of media bias via the various studies on the subject. Here are some of the results:

1. 2005 Study -- Every major media outlet except The Washington Times and Fox News Special Report leaned to the left. The furthest left (by a long shot) was the Wall St. Journal news division (as distinguished from their conservative op-ed group). Other groups leaning left were all the network morning news shows (with one exception), NPR Morning Edition, and the major newsmags. The closest to the center were Aaron Brown's Newsnight (now canceled, hmmm), PBS Newshour with Jim Lehrer, and ABC's Good Morning America. The authors of the study hailed from UCLA and Univ. of Missouri.

2. A 2004 study showed that nine of 10 major newspapers are more likely to portray economic data as negative if a Republican is president.

3. A 1996 study showed only 7% of Washington correspondents voted for George H.W. Bush in 1992, nearly half as many as voted for Bush in ultra-liberal Berkeley, CA.

4. A 2004 NYT story found only 8% of Washington correspondents thought Bush would be a better president than John Kerry.

UPDATE: I should have mentioned that I found item 2 particularly interesting. For years, I have had the sense that Republican economic data has been spun in a negative light relative to when Democrats are in office. Looks like the old intuition was right.

11 comments:

Nidsu said...

To be honest it really doesn't surprise me that journalists are typically liberal. They start off at a liberal arts school where they are taught to dislike capitalism and evil corporations.

Then the media outlets send them on stories of corporate corruption and evil millionaires that aren't sharing with the poor.

Also, many people who tend to believe in the capitalist ideal tend to be working in jobs that promote that ideal. Either that or starting their own businesses.

Tlaloc said...

the 2005 study used an embarrassingly bad system to measure. Specifically they counted the number of times a given think tank was mentioned by a source. If they mentioned say the ACLU it was considered a liberal point regardless of whether the paper was accepting or countering an ACLU statement.

By that measure Powerline would be extremely left wing as they are constantly talking about the people they disagree with.

Did you read the methodology? There is no way after looking at how they did the study that you can conclude it was anything but meaningless.

As for the others can you provide links? I'd like to check their methods as well.

The media certainly has a bias but it is neither left nor right overall. It is self centered. When moving to the right will sell more copy they do it (see the period from 2000 to 2002 for a drastic example). When the opposite is true then they move a bit left.

Hunter Baker said...

As Letterman said to O'Reilly (in a nice display of media bias), I think you're about 60% (actually 90%) full of it. The study is well-constructed. It won't be perfect, but I think knowing that a particular outlet cites the ACLU in a story more often than they do Heritage, then we can tell they are probably more liberal.

At both ends, the study works well. Think about it. If someone cites Heritage all the time, you know they are generally going to lean right. Ditto someone who cites the type of liberal think tanks you like to roll out as a left-leaner.

The methods are really a pretty ingenious way of studying the issue. There are other interesting variations available, but this sounds pretty good to me.

Here's a description:

http://www.newsroom.ucla.edu/page.asp?RelNum=6664

Tlaloc said...

"The study is well-constructed. It won't be perfect, but I think knowing that a particular outlet cites the ACLU in a story more often than they do Heritage, then we can tell they are probably more liberal."

Interesting. When I use google to search powerlineblog.com for "ACLU" I come up with 80 hits. When I searh it for "Heritage foundation" I get 49.

So as I said before by this method (and using the key words you yourself picked) Powerline is a liberal blog.

Obviously that's an insane method that leads to a ridiculous result. Powerline is one of the most right wing blogs.



"At both ends, the study works well. Think about it. If someone cites Heritage all the time, you know they are generally going to lean right."

No you don't. I cite the National Review a lot more in my blog than I do the ACLU. You know why? Because I'm mocking the NR. Citing a source means nothing when it is removed of the context in which they are mentioned.

And if you needed more proof that the method is flawed all you have to do is wonder what would happen in a year in which the ACLU filed a number of big name lawsuits. All the news sources would certainly report on it and consequently it would look (according to this method) like they all swung to the left.

Come on Hunter, I know you are smarter than this.

Hunter Baker said...

Your method is no good. You're only measuring one think tank versus a liberal org. If you do like the authors and measure ALL liberal groups versus ALL conservative groups cited, you might get a different result.

In addition, your choice of Powerline is spurious because it is a legal based blog and would surely mention the ACLU more than Heritage.

Tlaloc said...

"Your method is no good."

Funny that's what I've been saying all along.



"If you do like the authors and measure ALL liberal groups versus ALL conservative groups cited, you might get a different result."

The point is clear that the method is itself no good. Simply mentioning a source in no way indicates support for or from it. No matter how many keywords we give it the underlying methodlogy is simply bad.



"In addition, your choice of Powerline is spurious because it is a legal based blog and would surely mention the ACLU more than Heritage."

Then pick a conservative legal group. How about "federalist society"? 49 hits, still less than ACLU.
Come on Hunter you have to admit that the context of how you talk about a source is more important than simply counting the number of times a source is mentioned. I can't believe we are even debating this.

Hunter Baker said...

I can't believe it either, because it's just too clear "all things considered."

Tlaloc said...

By the way Reform club is also a liberal bastion:

ACLU hits: 13
Heritage foundation hits: 2

getting the point yet? A model that fails to predict reality is a problem. Is it the model or the reality that has the fault?

Hunter Baker said...

No, you're still screwed up. Blogs like ours basically exist to critique groups like the ACLU. Besides, I'm a legal type and thus this blog is legal like Powerline is legal. Much more likely to mention ACLU than Heritage. Plus, you continue to make the mistake of pitting one group against one group rather than universe of groups versus universe of groups. Replicate the authors's study, analyze the data for the problem you claim they have, and THEN I'll listen to you.

Tlaloc said...

"Blogs like ours basically exist to critique groups like the ACLU."

Yes indeed, and news papers exist to report the news. In both cases simply counting citations mean exactly nothing.

I'll try one more time Hunter but I'm really getting the feeling you know the study is garbage but you refuse to admit it.

Let's try a Gedanken:
Imagine a year in which absolutely nothing else changed except that the ACLU filed a large number of high profile law suits. The news papers (all of them) would end up citing the ACLU more often that then usually did. By the method you are defending that would register as an across the board swing to the left. However there would not be any real shift n political center of the news world.

The model in other words would deviate from reality because it fails to correctly measure what it tries to measure.

It simply does not get any clearer than this hunter: a model that cannot accurately predict is a failure.

James Elliott said...

Hunter, Tlaloc is completely in the right, here. (Oh, the pun!) The methodology is completely flawed. By your same methodology, an NPR ombudsman study of two years ago found that 60% of NPR guests were conservatives. Your contention would make NPR a conservative station, quite the contrary to the whining of conservatives.

The data collected by your methodology doesn't allow for context. It relies upon ill-formed, researcher-defined definitions of "liberal organization." For example, the ACLU is not a "liberal" organization in the same sense as NOW. The ACLU, as a filer of high-profile lawsuits, will be all over the news, skewing your results.

The methodology is flawed. Conservative news sources lambasting the ACLU will, under that methodology, be rated as liberal, while a liberal source dedicated to countering say, Hudson, will be rated as conservative. As the NPR example above notes, the data will not support the hypothesis.

You've failed to reject the null, man.

And Nidsu? Your first comment is easily the dumbest thing I've read in weeks. Thanks. As a graduate of one of those ultra-liberal universities living in one of those ultra-liberal states, I can tell you that no professors EVER questioned capitalism or inserted their own political opinions into lecture or grading. (Now, TAs? That's another story entirely...) A good professor wants people to make their own conclusions, not parrot his or her own.