The current issue of _Perspectives on Politics_ includes a symposium which I organized (i.e. I invited the participants and wrote the foreword), on UCLA political scientist Timothy Groseclose’s Left Turn: How Liberal Media Bias Distorts the American Mind, which builds in part on his earlier research with Milyo. Links below:
bq. Brendan Nyhan “argues”:http://journals.cambridge.org/abstract_S1537592712001405 that there are measurement issues with Groseclose and Milyo’s model, and prefers Gentzkow and Shapiro’s methodology as a starting point. He also suggests that all research on media bias is plagued by deep epistemological issues – there is no good and widely accepted definition of what unbiased reporting is.
bq. Nolan McCarty “has considerable confidence”:http://journals.cambridge.org/abstract_S1537592712001399 in Groseclose and Milyo’s initial results, and wonders whether media bias may be even greater than they believe. However, he is not convinced by some of Groseclose’s newer arguments, and fears that they may overstate the real world impact of media bias.
bq. Kathleen Hall Jamieson “points to”:http://journals.cambridge.org/abstract_S1537592712001193 substantial differences in how she and Groseclose think about framing effects, illustrating this with a debate over the appropriate terminology for a particular form of abortion. However, she emphasizes that she doesn’t seek to argue that media bias doesn’t exist.
bq. Nancy Rosenblum “asks”:http://journals.cambridge.org/abstract_S1537592712001375 whether it is desirable for voters and reporters to hold the same political opinions, arguing that this is far from the only appropriate means of redressing bias. Instead, she argues that media should play an active role in educating and in helping to shape democratic debate.
bq. Justin Gross, Cosma Shalizi and Andrew Gelman “argue”:http://journals.cambridge.org/abstract_S1537592712001387 that Groseclose’s underlying causal model systematically fails to consider the possibility of bias in other important social institutions. They further argue that his model makes assumptions that fit badly with what we know about opinion formation, pays insufficient attention to problems of aggregation, and is not properly tested against the data. An extended and ungated “version of the paper is available here”:http://www.stat.cmu.edu/~cshalizi/leftturn/.