« Legalizing Torture | Main | Ethical question. »

September 29, 2004

Comments

As I understand it, Sebastian (and I admit that I do not understand the thinking behind interpreting pre-election polling data very well) it's a given that the incumbent should be leading the challenger in the polls - a kind of automatic margin, presumably (I'm guessing) caused by people who are asked "who would you vote for for President"? having a bias towards responding with the name of the current President, even if they are actually thinking of voting for the challenger. Therefore, for Bush to be considered clearly ahead in the polls, he has to have a substantial lead over Kerry, whereas Kerry's narrow margins either ahead or behind of Bush apparently look more significant than the mere numbers suggest. Please understand this is me reporting a theory I have read about, do not understand very well (I did study statistics, but I studied them for scientific purposes, not electoral polling!) and am not sure whether or not to believe.

But, honestly, I don't consider polls to be worth getting worked up about. I would think it more worth your time to write a scathing attack on the media habit of treating polls well over a month before the election (and even further away) as gospel rather than as doubtful data of little value. As far as I can see, all and any of the polls say is that right now, no one knows who's going to win in November. And that is, really, how the media should be reporting it. Any claim that the polls show Bush or Kerry to be ahead (and I've seen both) should be dismissed.

You're focussing on reports that Kerry's ahead, and trying to claim that this proves a "liberally-biased" media. But I think that's a waste of your time: the media as a whole isn't liberally-biased. It's dumb. And where it isn't dumb, it's corporate.

I could, of course (and doubtless others will) respond with anecdotal evidence similiar to yours above that "prove" the media is conservatively-biased. But anecdotal evidence rarely proves anything, and certainly won't in this case.

The media in the US is very largely controlled by large corporations, whether as advertizers or as owners, or both. That makes the media biased towards conservatives.

Journalists, however, in whatever media they work in, they tend to question authority, to speak truth to power. Not always and these days not often, but it is a strong tradition in their profession, a belief in themselves that they report the truth those power don't want you to know. That makes the media biased towards a liberal, anti-conservative viewpoint - especially when the current US government is so strongly right-wing.

The tension between the two - what the big-business corporate owners want their news outlets to say, and what the people whose job it is to report the news feel they should say - is nothing new. My feeling, however, is that in US, more and more, it's the corporate owners who have been winning out.

jesurgislac - "for Bush to be considered clearly ahead in the polls, he has to have a substantial lead over Kerry"

Wouldn't this be predicated on how the folks the polls are questioning form their opinions. If the predominance of responders get their opinions from the evening news or major metropolitan newspapers, there is no doubt in my mind that they have a misinformed view of today’s events.

I think I read some prognostication that the liberal bias should be worth 15 points to the challenger.

Sebastian's argument from Patterico says it all. Given the same numbers, headlines differ greatly. Reading about last night's report on the draft by CBS is as prime example as you'll ever need.


I haven't looked the details, but it's too simplistic to consider just the current margin -- the trend is at least as important. If Arizona was 60-30 for Bush a month ago and is now only 54-38, whereas California had Kerry about 15 points up all along, then I'd say the headlines are justifiable.

But even if this is an apples to apples comparison, it's hardly proof of the silly "liberal media" charge. Anyone can find evidence of "bias" one way or another in any given media source in any given day or week. The "media" is not monolithic, and bias is not so easy to measure.

kenB - As Sebastian suggests, a 16 pt. lead in no way justifies a headline claiming a state is in play. Kerry gaining strength may be justifiable but not really. I often wonder if editors read the article before they supply the headline.

Oh lord.
Arizona has a rising population of people who self identify as democrats. Because it is rising the media decides to mention that growth in their headline.
There is no such trend in California.

Sebastian's misreading of the LATimes' Arizona story is curious.

As evidence of "bias," Sebastian notes:

Not until paragraph 17 does the paper bother to report: "A poll taken for the Arizona Republic and released last week showed Bush ahead of Kerry, 54% to 38%."

However, Sebastian studiously omits the following text:

In contrast, the newspaper's polls in August and June had Bush ahead of Kerry by just 3 percentage points, a lead within the survey's margin of error.

The new poll numbers ricocheted through both parties, with Kerry officials insisting that Bush was simply benefiting from a successful convention, though Republicans said the figures were evidence of Arizona's proud GOP tradition.

Whatever the poll numbers are on a given day, the LATimes' story makes clear that both candidates seem to be devoting a lot of attention and campaign resources to a state that has recently been seen as a lock for the GOP.


Blogbuds: If the predominance of responders get their opinions from the evening news or major metropolitan newspapers, there is no doubt in my mind that they have a misinformed view of today’s events.

They do indeed, especially if they watch Fox News or if they listen to Bush & Co's speeches.

Some input on polling data...

Jesurgislac,
I've been reading his posts regularly and I've come to the conclusion that Sebastian is biased.


Ha ha ha

Carsick, we're all biased.

The blogging gestalt at Obsidian Wings has an intentionally distributive bias, from Sebastian at one end of the political spectrum to Hilzoy (or possibly Edward) at the other end. What they all have in common is that they all write well and they are all believers in courtesy in discourse. This is what makes Obsidian Wings one of my favorite blogs, if you want to know...

zzzzzzzzzzzzzz

Ken B, The LA Times report on Arizona showed it trending away from Kerry.

Jesurgislac
Obviously my attempt at humor failed miserably.

Either that or my sense of humor inexplicably and unexpectedly went kablooie on me, which happens more often than I'd like, especially when I'm tired. Which I am.

I'm not going to near the bias claim, but I can say that poll coverage is not a good place to start making that case. The press when dealing with polls or statistics on issues - whether they be left, right, up or down - are wholly and completely incompetent at interpreting the data in a reasonable way. Instead, numbers are clumsily hammered into a narrative that makes an interesting story - in this case, a likely imaginary Kerry rebound - which is only interesting in the previous narrative context of Bush's small bounce.

carsick - what I'm finding is as clever as I think I am, if it isn't written skillfully, Lord knows how it will be taken. I'm learning a lot from the various blogs I stalk, and expressing myself better tops the list.

"I'm not going to near the bias claim" should read "I'm not going near the bias claim." Ugh.

Sebastian, where does the claim of CNN's liberal bias fit with CNN's use of a set of LV polls (Gallup) that in their last two iterations sampled 7 and 12 percent more Republicans than Democrats?

blogbud
Does that mean I'm going to have to start using those little computer shorthand things I so dislike?
;) or lol?

There was a nice story in the WSJ a couple days ago talking about how various technical issues could lead polls to differ by a pretty significant amount. This had to do with things like how you adjust your results to account for, say, getting 130 women and only 70 men in your sample.

Polls provide useful information, but they're not as precise as you might like. And that's outside any partisan axes anyone might care to grind.

--John

CNN says, Bush "apparently" leads Kerry in a new GALLUP poll, for the obvious reason that Gallup has had its head handed to it, for a series of faulty polls, oversampling Republican voters. Hello! CNN is embarassed. Yesterday, they made Gallup's chief make a five minute apologia on Inside Politics.

Sebastian needs to get out more, lose his snarky, misplaced arrogance, and try -- TRY -- to keep up!!

"Sebastian, where does the claim of CNN's liberal bias fit with CNN's use of a set of LV polls (Gallup) that in their last two iterations sampled 7 and 12 percent more Republicans than Democrats? "

I'm not sure which polling technique you are talking about. Were more Republicans intentionally asked, or did more show up in the random sample?

And I'll agree that poll reporting is shoddy. Reporters apparently know very little about polling techniques and they don't bother to report on half of the important issues surrounding polling questions.

But that doesn't offer an excuse for reporting the same poll results in dramatically different ways.

zzzzzzzzz.

Oh wait. . I'll wake up for this one
"more show up in the random sample?"

More showed up in the random sample. Which is an extremely shoddy way to do your sampling, because your sample population (people who answer telephone politics polls) is a highly distorted subset of your target population (US voters). Solving for this problem basically consumes all of the free time of pollers.

Except Gallup, who decided they'd just ignore it.

I remember reading a critique of a poll about women's views of a particular issue and it turned out that everything was kosher until they asked to review the phone logs... the phone calls were made to people's homes in the middle of the day which obviously excluded a vast amount of diversity.

More showed up in the random sample. Which doesn't exactly prove malice on Gallup's part, though it does make one annoyed about all the articles proclaiming Bush's huge lead on Kerry, without mentioning that the poll showing said lead oversampled Republicans. I would have considerably more sympathy/respect for Gallup if their _next_ poll hadn't oversampled Republicans by an even bigger margin. I don't necessarily think that it proves conservative bias on Gallup's part (though one does wonder what the heck is going on over there), but I think for CNN to run with the results of that poll, flawed as it is, is something of a rejoinder to charges of liberal bias on CNN's part.

Or to put it another way, the headline could easily have said "apparently leads" because the polling data that produced that 8-point lead was out of whack with any number of other polls taken around the same period.

"Or to put it another way, the headline could easily have said "apparently leads" because the polling data that produced that 8-point lead was out of whack with any number of other polls taken around the same period."

And if they were reporters they could report that instead of making us guess.

I'll cheerfully admit that CNN sucks. I don't, however, think, that the examples you provide, given the context (especially the issue of the Gallup poll) demonstrates liberal bias.

That's ok, I'm perfectly aware that almost nothing short of a direct admission could convince some people that liberal media bias is a significant factor.

I still don't see what issue in the Gallup poll mitigates anything I discuss. Gallup polls were used in both examples. In the Kerry example it was used to cite 'leading'. In the Bush example it was used to cite 'apparently lead'. In the LA Times a 16 point lead for Bush is 'in play' while a 15 point lead for Kerry is 'safe'. Same statistics, but the media applies a different spin to the statistics based on who is ahead. And interestingly the spin direction is consistently against Bush.

That's ok, I'm perfectly aware that almost nothing short of a direct admission could convince some people that liberal media bias is a significant factor.

Nice snark. I, for one, freely admit that bias is a factor in the media (as are incompetence, smugness, mediocrity, and many other undesirable human attributes). I wouldn't be surprised if overall, having found some reasonable way to quantify said bias and some way to total all instances of bias across all mainstream media outlets, the needle ended up pointing more to the left than to the right, especially regarding domestic social policy.

However, no one has come up with such a metric, instead we're just offered a parade of scattered examples. These do nothing to prove your point, because (a) it's ridiculous to expect an absolute lack of perceived bias in every single word of every single newspaper issue or tv segment, and (b) you're obviously more sensitive to bias that goes against your "side" than bias that goes the other way. Have you ever spent any time at Atrios or Kos? For every instance of supposed liberal bias you can offer, folks on the left can give you one of supposed conservative bias in the "SCLM".

Anyway, you can believe what you like. Absent the existence of systematic research with some sort of reasonable metric, I'll continue to not take the "liberal media" accusations any more or less seriously than the "conservative media" charges from the left, and I'll continue to see such charges as a sign that the person making them is excessively partisan.

Funny how you mock my snark and then immediately confirm that it is pretty much correct.

"(a) it's ridiculous to expect an absolute lack of perceived bias in every single word of every single newspaper issue or tv segment, and (b) you're obviously more sensitive to bias that goes against your "side" than bias that goes the other way. Have you ever spent any time at Atrios or Kos? For every instance of supposed liberal bias you can offer, folks on the left can give you one of supposed conservative bias in the "SCLM"."

Absolute lack of perceived bias is what I'm asking for. I'm asking for sets of facts (like polls) to be treated the same way in the same media organization instead of changing their meanings depending on who is ahead. That isn't slight bias. Atrios and Kos mostly whine about Fox News, the very newest and least established of the media. There are studies with good metrics showing liberal media bias, but I suspect that you won't find them reasonable.

Sebastian, what I already said was that there is evidence for both liberal and conservative bias in the media - and that this is nothing new. It's the tension between the corporate owners, who naturally prefer the news outlets they own to support them (plus pressure from advertizers, which doesn't have to be overt to be a real factor), and the journalistic tradition of questioning/challenging authority. The first gives a conservative bias: the second tends towards a liberal bias.

But your example was bad in the first place, because I'm beginning to be convinced: it's not that I'm stupid about polling data, it's that the media, without exception and without bias, interprets all polling data stupidly. Now if you want to attack that, I'm a receptive audience.

Funny how you mock my snark and then immediately confirm that it is pretty much correct.

Thanks for demonstrating your standards for proofs -- a single example is enough to confirm a hypothesis that you're already inclined to believe. :P

There are studies with good metrics showing liberal media bias, but I suspect that you won't find them reasonable.

Dude, this is the internet, you can't throw out a statement like that and not provide the cites. The only one I'm aware of is the one Dave Schuler mentioned a while back, which was based on the relative number of citations of various thinktanks. Forgive me if I find that less than convincing.

Atrios and Kos mostly whine about Fox News

This is not true. Well, that is, they certainly whine about Fox, but they don't consider it part of the "So-Called Liberal Media". The latter is blamed for Gore's loss in 2000, cheerleading for the Iraq war, recitation of GOP talking points in interviews, and on and on. I guess the one difference is that they tend to single out individual people (e.g. Judith Miller of the NYT, Nedra Pickler of one of the wire services, etc.) instead of broad-brush attacks on "the media" as a whole.

So, are there instances of bias? Yes. If you lined them all up, would there be more to the left than the right? It's certainly possible. Are you justified in referring to a monolithic "liberal media" that attempts to manipulate the American people into supporting liberal causes and Democratic candidates? No way.

Absolute lack of perceived bias is what I'm asking for.

Good luck with that. The best we can hope for would be for reporters and media outlets to be up front with their potential biases while trying their best to avoid having them unduly affect their product.

If I can provide a scattershot sampling of mainstream news articles with a verifiable lean to the right, can I use this to form the substance of a post implicitly decrying the horrendous "conservative bias" which prevents us poor liberals getting our point of view across?

The comments to this entry are closed.

Blog powered by Typepad