Running numbers can reveal patterns. But figures can lie, especially when they come from Media Matters.

Every now and then we are asked why we include Media Matters in our "disinformation" column. After all, they back up everything they say--they're beyond criticism! But then they get caught substituting their opinion for the facts, and even the gullible have to doubt. Now MM has published what it deems to be a definitive analysis of The O'Reilly Factor and, not surprisingly, concludes that it is horribly slanted to the right. On close inspection, it proves to be another one of their numbers games:

Each guest was coded by his or her general partisan affiliation or ideological orientation, not for what he or she said that day on the program.

In other words, the MMers slapped labels on each guest that had nothing to do with the actual content of the discussion at hand. Why would they do that? So that they could take a discussion with Michelle Malkin and Linda Chavez and claim it "tilted right". In fact, the two took opposite positions in an immigration debate, but that's ignored in a deluge of numerical data. We were going to call in our expert witness, Prof Charles Eppes of Cal Sci, to analyze the math, but that proved to be unnecessary. Because what's being manipulated is the labeling.

The MMers use five labels for their "analysis": (R)epublican, (C)onservative, (N)eutral, (D)emocrat, (P)rogressive. They claim that over the four months of the survey, (R) and (C) guests amounted to about 39% of the total, (D) and (P) roughly 18%, and the remainder (42%) were neutral. It didn't take long to see how easy it is to make such a survey say whatever you want it to.

We find Larry Korb marked with a (C). What is this based on? Yes, he worked in the Reagan administration decades ago, but he is today with the Center for American Progress and is no more a conservative than David Brock himself. Ellis Henican, the liberal columnist from Newsday, gets tagged with an (N), while Fox's Eric Burns gets a (C). Is there any logic here, other than getting it wrong? Multiple appearances by Lis Wiehl are labeled (N), but she is a Democrat partisan who has given money to Al Gore.

Legal eagle Jonathan Turley is repeatedly described as (N), even though he has been leading the Bush-Is-a-Criminal charge. Most reporters have an (N) after their names on Brock's list, but not Carl Cameron. They slap a (C) on him. Wendy Murphy is another (C), and she makes at least eight appearances in the survey. Murphy is a longtime Democrat who backed Al Gore, so you can subtract 8 from the (R)/(C) column, and add 8 to the (D)/(P) column right there. Dick Morris, a former Clinton advisor, made almost 10 appearances, giving MM ten more (C)s to mix into their stew. But Morris, the quintessential political technician, holds views that are all over the ideological map, and is clearly an (N).

The biggest thumb on the scale is the ludicrous labeling of Judge Andrew Napolitano as (C). Hasn't Brock read any of his books? The Judge is against Patriot Act I, Patriot Act II, NSA surveillance, and collecting phone records. He opposes the war on drugs and favors legalization. He says illegals have a right to collect social security benefits. The Judge also holds more traditional views, but his mix of left and right doesn't make him a conservative. It makes him a libertarian (like John Stossel, also mislabeled). So you can switch more than a dozen entries in the Media Matters spreadsheet from (C) to (N).

Kirk Cameron and Bo Derek, described as "media personalities", get slapped with the (C), but other celebs get a discreet (N). Michael Mello has written in favor of same-sex marriage and against the death penalty. You guessed it: he's an (N). Liberal and Dem contributor Jayne Weintraub is also camouflaged with an (N). And on it goes.

We all use math every day, to forecast the weather, to tell time...but Media Matters uses it to create deceptive charts and tendentious pseudo-analysis. When the aim is to smear Fox News, anything goes, including fuzzy math.