. . . muttered the ogre

Robin Hanson reports:

I just produced the following draft (PDF), which tries a new statistical approach on the question of which side is “right” in a media controversy. I applied it to the coverage of PAM, but it might also apply it to other controversies.

The Informed Press Favored the Policy Analysis Market
The Policy Analysis Market (PAM), otherwise known as “terrorism futures,” burst into public view in a firestorm of condemnation on July 28, 2003, and was canceled the next day. We look the impression given of PAM by 396 media articles, and how that impression varies with six indicators of article information: mentioning someone with firsthand knowledge, time since the firestorm, article length, a news versus an opinion style, and periodical prestige and period. All six indicators significantly and substantially predict more favorable impressions of PAM. A multiple regression predicts that a two thousand word news article in a prestigious monthly publication one hundred days later that mentioned an insider would give a solidly favorable impression of PAM.