http://www.outsidethebeltway.com/archives/2007/09/too_many_conservative_syndicated_columnists_/

Underlining how difficult (and subjective!) the coding can be, the study labels Mort Kondracke as a conservative, OTB thinks he should be coded as a progressive, and Andrew Sullivan thinks a centrist (he also calls Kondracke “Fred’s fig leaf”).

When it comes to your rebuttal, consider me unconvinced. For starters, finding one example of a “liberal” position from a columnist (universal health care is widely supported by moderates, BTW) hardly means that columnist is “liberal”. After all, Ron Paul is anti-Iraq-War, but I don’t think we could call him a liberal. Heath Schuler is anti-abortion, but I don’t think we could call him a conservative. Besides, the column published immediately prior to the health care one by that author was a relatively glowing review of Rudy Giuliani’s interpretation of state’s rights.

Similarly, Navarette regularly writes columns in support of George Bush and Alberto Gonzalez. Hell, the Pope believes in global warming, and he sure as heck isn’t a liberal!

Is it possible that MM miscoded unlabeled columnists? Of course. But you haven’t convinced me that they actually had a coding bias.

As for listing the papers used in the study (all 1400 of them), I have to say that when I did scientific studies, I don’t think we ever published raw data when we published results. If you want the raw data, ask! Otherwise, that dog just don’t hunt.

There is one - only one - of their conclusions that omits the limiter “syndicated”. Your contention that their conclusions are flawed because they didn’t consider local, non-syndicated columnists is pretty weak when they indicate in virtually all of their conclusions that they are only considering syndicated columnists.

Assuming no coding bias (which again you haven’t convinced me of), the only conclusion that might be unsupported by their study is this: “In 38 states, the conservative voice is greater than the progressive voice — in other words, conservative columns reach more readers in total than progressive columns. In only 12 states is the progressive voice greater than the conservative voice.” And the truth of that rests on the distribution of local columnists. They may be mostly liberal. They may not. We don’t know.

This study is the most flawed study I have seen in a while, and I understand how to read social science studies.

I give a very detailed analysis below.

Media Matters Spouts its Own Flawed Study as Fact: How They Did It, In Great Detail