The most helpful critical customer review:
I have read the book. Such a work is going to be polarizing. Those on the right will put it on a pedastal. The left will cringe like vampires in an Italian restaurant. The right will say the numbers are fine to support their charge of media bias. The left will attack it from every angle they possibly can in a desperate attempt to deny any media bias.
The book's narrative style is relatively engaging. It is peppered with quotes and anecdotes, and has a sort of tongue-in-cheek rolling style that makes it a decent read for what is ostensibly a rather dry topic (math). People on the right who read it will be nodding and chuckling as they go. Those on the left may not agree with his positions, but may at least be entertained by his style.
The problem I had as I was reading was in the way the entire argument was framed. Quite often in the book you the author will step in and say, "It is useful to illustrate the point with a thought experiment..." Or at other times, "If we assume that..." Too many of the points he brings up are supported by what can only be called inferences and assumptions rather than hard statistics. I am a statistician, and I spend the entire book waiting and waiting for hard numbers that never materialized. I'm certain the author has his p-values and correlation coefficients somewhere - but they aren't in this book.
So the book - IMO - fails at 'proving' the case for media bias because it ultimately does not use hard statistics. It uses 'soft' statistics which are based on inferences. Now, those inferences may indeed be based on numerics (PQs and SQs) but when the numbers that CREATE the numbers are inferred then it cannot really be said to be conclusive evidence. Evidence of a sort? Sure. But proof? Not so much.
Do I think that the media is actually biased? Oh - of that I have no doubts at all. Is the media biased to the left? Again - I think that any rational person who conducts even a casual observation of the media marketplace can only come to the conclusion that the news media is rife with left-wing political slant. The chapters that were most interesting (and conclusive) to me were the ones that discussed "Words that aren't cheap" (showing that journalists overwhelmingly donate to leftist politics) and the discussion of 1st and 2nd order bias in the newsroom (what environment does it create to have a 85+% liberally slanted newsroom population?).
As to whether the biased news media effects the voting habits of the population at large? I think he raises some interesting points, and to a certain extent (using OTHER research than his own) he makes at least a preliminary case that the media does in fact influence voting habits. But I do not think his SQs definitively measure actual media slant. Nor do I think that his measures of voter/media/politician central tendency is necessarily accurate.
The "PQ" of congress is a measure of voting record. It INFERS conservativeness or liberalness. I think the PQ is a good finger-in-the-wind (so to speak) but it is not what a statistican would call "Interval" or "Ratio" data which can be used to generate a meaningful mean or median. Likewise, the "SQ" slant quotient are like the PQ numbers - which is to say they are ordinal measurements at best and suitable for simple analysis but not hard stats.
What do I mean by that? In a social survey, a person may be asked a question such as "How much do you like Fruity Pebbles cereal?" and be presented with choices such as "Like strongly, Like somewhat, Neither like nor dislike, dislike somewhat, dislike strongly". To me, the PQ voting records of Congress seem more to suit this style. Voting for a bill doesn't automatically make you a liberal, conservative, Democrat, or Republican. It is a specific act (Yes/No) but its implication is general. Just like you can't statistically 'prove' how much someone likes Fruity Pebbles from a Lichert scale, you can't 'prove' how liberal or conservative someone is from the PQ. It is an indicator, not a scientific measurement.
The same applies to the SQ numbers. They seem far too rooted in generic at udinal measures to suitably fill a statistical role that is based on central tendencies. Even as he was making his case in his final chapters, the language the author used was filled with conditionals that (to me) were tacit admissions that the results were not so much 'statistical proof' as they were 'estimators'. Particularly when he was stating "Hey - I think the SQ is over .5 and under .8, but what the heck let's just call it .7!" That isn't hard statistics. That isn't conclusive proof. That is a finger in the wind.
Is he right? I think in a broad sense he probably is correct. I think the media is liberally slanted, and that it does pull voter opinion along with it to a degree. Is it as much as he states? I have no idea, but this book does not satisfy my that he has definitively answered 'how much'. He's probably in the ballpark, but that's all he is. He hit it in the ballpark (a general area) but we can't say he put his arrow in the gold (a specific target).
Reply With Quote
