UPDATE UP HERE: I need to clarify something that's created controversy below.
I recognize that they clearly did interview some people (and in the case of region, perhaps a decent amount of them) outside of the categories of "White, 50+, College-educated, Southerner, and Not Liberal." Implying that they interviewed zero people in those categories was hyperbolic on my part and invited responses of simply proving that that straw man wasn't true. I hope that that concession puts some of that argumentation to rest.
My argument (and my anger) involves their having apparently interviewed so relatively few people in those categories as to make generalization to a statement about the voting population absurd and misleading -- and then not informing viewers and readers about how poorly generalizable that data was.
I believe that they did that deliberately to drive a storyline that Obama had not simply "lost" -- but lost badly, amazingly badly. That could have been "pro-Romney winning or, in my opinion more likely, to make sure that the race continues to be interesting to viewers so that they keep watching CNN and reading CNN.com. Either is reprehensible.
Various comments below suggest how they could have knowingly created a situation where they would know that a poll would skew sharply in a given way (through, for example, their use of their panel, recruited in some way unknown) without actually cooking the data. Doing that is still wrong -- and, in the church of Science -- a significant sin, unless one informs the consumers of the information.
So, yes -- I use the term "zero out" certain groups when I should have used a term like "substantially undersample." But the evidence that persons under 50 either were barely touched or unexpectedly turned out to be just the same as their elders is clear from the marginals: 25 to 67 overall, versus 24 to 67 for the 50+ crowd. College vs. No College is similar. Extrapolating from the standard errors, we can see that there were very few liberals sampled; CNN could of course clue us in on exactly how many whenever it chose.
Whether one contrived to arrive at such a bias sample, or whether it was just a "happy accident," discussing it (especially in headlines) as representative of the U.S. voting public when one has good reason to know otherwise is simply dishonest -- and that's in the best case. I stand by my call for action against CNN: taking away Crowley's moderator spot.
This excellent diary riled me up so much, once I got through reading the report, that I went out and created
this Facebook page:
"Punish CNN for Distorted Poll; Remove Candy Crowley As Debate Moderator." The thesis is that what CNN has done in seeking out a completely biased poll and then presenting in a way that most people will think that it is
representative of the viewing population is
so awful,
so odious, and
so antithetical to fair journalism that the network itself needs to be punished.
How do you punish a network? Well, you can take away one of its honors -- in this case, the honor of having one of its hosts and correspondents moderate a Presidential debate.
Candy Crowley should step down from her post as moderator. If she won't, the Commission should remove her. That's the only way I can think of to make it clear to news organizations that the Older White College-Educated Illiberal Southerners poll, when presented to the public simply as representative of voters who watch the debate, is such an infraction against journalistic principles that it will not be tolerated.
If you agree, please like, share, and comment upon that page. Let's take it viral.
Here's a graphic I made up that sets all of the "zeroed out" categories (except the exclusion of rural voters, for which I didn't have space) next to each other in stark and readily comprehended form. (Free for you to share.)
From my first full entry:
CNN did something absolutely monstrous in covering the public reaction to the Oct. 3 Presidential debate: it polled only white college-educated Southerners over age 49 who self-identified as either "moderate" or "conservative" -- but did not clearly inform the public about its ideologically selective sample. This is the equivalent of polling only Blacks under age 25 from the Northeast and Midwest who had not attended college and self-identified as "liberal" -- and not informing its viewers and readers about the sample composition. The poll showed Romney preferred 67-25 over Obama -- a result that, taken out of context as it was, helped to shape the interpretation of who won the debate.
CNN must now be penalized by losing the honor of having one of its correspondents, Candy Crowley, moderate one of the remaining debates. Otherwise, we as a nation are saying that this is OK.
You need a concrete goal to organize people quickly and well. Refusing to let Candy Crowley further CNN's agenda -- that of "making it a more interesting race" through fraud on the viewing and reading public -- from the moderators chair is concrete. I'll bet that we can boot her out of that position, to be replaced by another female journalist -- and in so doing bring public attention to CNN's heinous action.
Join me, will you?
3:23 AM PT: I realized that I left out some of the most important part: CNN has been representing the "67 to 25" result simply as a poll of a group of voters who watched the debate (to paraphrase), rather than a poll of "a group of white, 50+, college-educated Southern moderates and conservatives who watched the debate." This is most damaging when it comes to headlines. And no, this is not at all accidental.
3:53 AM PT: Link to the survey and report of internals.
8:07 AM PT: Here's some of what CNN had said on its website in its story, to provide a sense of how they're presenting this material (i.e., without any nod towards the bizarre apparent sample composition):
Denver, Colorado (CNN) – Two-thirds of people who watched the first presidential debate think that Republican nominee Mitt Romney won the showdown, according to a nationwide poll conducted Wednesday night.
According to a CNN/ORC International survey conducted right after the debate, 67% of debate watchers questioned said that the Republican nominee won the faceoff, with one in four saying that President Barack Obama was victorious.
"No presidential candidate has topped 60% in that question since it was first asked in 1984," says CNN Polling Director Keating Holland.
...
"This poll does not and cannot reflect the views of all Americans. It only represents the views of people who watched the debate and by definition cannot be an indication of how the entire American public will react to Wednesday's debate in the coming days," cautions Holland.
Oh I guess that last paragraph makes it all right! (I don't recall seeing it last night, by the way.)
