August 08, 2003

Bias in the U.S. Media

Someone at Harvard University did a study of editorial bias in four major U.S. newspapers by examining 510 of their editorial responses to 10 similar political and policy initiatives in the Clinton and current Bush administrations. The results, in a neat table:
PaperAttacks ClintonDefends ClintonAttacks BushDefends Bush
New York Times37%37%68%8%
Washington Post22%34%65%12%
Wall Street Journal83%5%3%75%
Washington Times93%1%10%78%

Summing positive and negative values from the accumulated totals in the study, the "liberal" papers are +6% Clinton and -57% Bush while the "conservative" papers are +70% Bush and -86% Clinton. In other words, the "liberal media" has little to no discernible pro-Democratic bias while showing a strong anti-Republican bias, and the "nonpartisan, unbiased, factual, fair and balanced media" has anti-Democratic and pro-Republican biases each far surpassing the anti-Republican bias of the "liberal media".

Another good resource on media bias is Lying In Ponds which tracks the partisanship of individual newspaper editorialists. I don't know of anything similar for news reports, talk shows, or bloggers. Also worth noting is Fairness and Accuracy in Reporting which is sometimes more an example of biased and inaccurate news reporting than an investigator of such.

Speaking of editorial slants, contemplate the possible undergoing transformation of George Will. As long as I've been paying attention, Will has a staunch partisan Republican even if his prim, professional columns are hardly recognizable as such compared to the idiotic ravings of most partisan Republicans. However, Will's last two columns could be from Paul Krugman. First he strongly condemns the Republican Party's destruction of traditional checks and balances, and then he writes a moderate if not positive highlight of Howard Dean's campaign. Might we soon see another partisan flipflop like David Horowitz or Eugene Debs? Time will tell.

Posted by Warrior Tang at August 8, 2003 09:34 PM


Comments: