I was reading the latest Language Log thing about Groseclose and…: jcreed

I was reading the latest Language Log thing about Groseclose and Milyo's "A Measure of Media Bias". The accusation is that there is an asymmetry in their model that falls out of the way the arithmetic is done: there may be, Lieberman worries, an assumption implicit in the model that people on one end of the left-right political spectrum pay more attention to the left-right status of the works they cite than the other. Specifically, that conservatives cite equally liberal and conservative sources, and that liberals cite more liberal sources than conservative.

I was thinking that surely the numerical representation of left-rightness goes from -1 to 1 if they multiply citer ideology by citee ideology but as far as I can tell it goes from 0 (conservative) to 1 (liberal). Apparently even the LL post has been updated with other people suggesting the same thing...

(no subject)

(no subject)

(no subject)

Error