Philosophy, Film, Politics, Etc.

Thursday, February 4, 2010

The DK/R2000 Poll and O'Reilly's Response

I don't know how accurate the results of this poll are, but it seems plausible that a somewhat large percentage of self-identifying Republicans harbor irrational and hostile beliefs about Obama, and generally unsettling views about homosexuals, creationism, and other staples of conservative controversy. There's nothing surprising here, as disturbing and sad as it is.

It is also unsurprising that prominent Republicans would want to distance themselves from those who, for example, think that Obama should be impeached, or that Obama "wants the terrorists to win." So we should not be surprised to find Bill O'Reilly distancing himself from what he considers a fringe of conservative extremists. But instead of just distancing himself from their wacky beliefs, and defending Obama, homosexuals, and the American education system--instead of talking straight politics--O'Reilly goes on a smear campaign against the poll itself.

First, he tries to poison the well, claiming that the poll is bad because its source is "unreliable" and "insane." Apparenly, if you compare Republicans to the Taliban, you are insane. If you had a poll which did not predict the outcome of the recent Mass Senate election, your polls are not usually accurate. O'Reilly uses these accusations to build up to his stunning revelation: the poll is a fraud! Yet, he offers no evidence or argument to back up this claim. This is a smear campaign, as irrational as the "kooks" O'Reilly is trying to distance himself from.

What is most interesting is that O'Reilly seems intent on figuring out just how many Republicans share the "crazy" beliefs in question. (Though O'Reilly does not say which of the poll's results he thinks are crazy. Many of the results are relatively benign.) So he asks his guest, Karl Rove, who in no way even attempts to answer the question, just how many Republicans are nutty extremists. Instead of answering, Rove goes on a rant against the DailyKos. And what does this criticism amount to? That the poll allows people to "bloviate" by asking yes/no/I-don't-know questions, like, "do you think Obama was born in the United States?" Ahh, right, you cannot trust a poll that asks such straightfoward questions. I see.

O'Reilly claims the poll is meant to "demonize" Republicans and anybody who is not a liberal. Again, he makes no argument and offers no evidence to support this accusation.

I wonder, what does "insane" mean for O'Reilly? He thinks the poll is a calculated attempt to undermine the image of Republicans, and this apparently means that the perpetrators of this alleged scheme are insane. Is this a calculated smear campaign on O'Reilly's part, or an irrational defense reflex? I guess both.


Feb. 6, 2010:

Jake Simpson at The Atlantic Wire draws our attention to reports on two problems with the DK/R2000 poll: sample bias (by Charles Lemos) and acquiescence bias (by Mark Blumenthal) . Though they are well worth considering, I don't think these are damning criticisms.

For one thing, while the sample bias is noticable, it is not outrageous, and it might not be significant. Considering how alarming the results of the poll are, the slight sample bias gives us no reason to ignore them. Still, a survey which was more in line with the age and geographic spread of Republicans might be different, but probably not dramatically so; and not necessarily in a way which would improve the Republican image. Of course, we won't know the significance of the bias unless and until somebody pulls off a better survey. Still, I don't think these results should be ignored on this basis.

As for the other charge, it is certainly worth taking acquiescence bias into consideration when analyzing this poll; however, it does not invalidate the results. The questions in the yes/no/I-don't-know form here--such as whether or not Obama wants the terrorists to win, or whether or not creationism should be taught in schools--are questions which, in today's conservative mindset, offer no shades of grey. While there are many shades of grey when it comes to beliefs about whether men or women are better drivers, for example, it doesn't seem likely that the same applies to beliefs about whether or not Obama wants the terrorists to win. But, again, I guess we won't know unless and until somebody devises a survey which attempts to rephrase the same questions in another form. Maybe something like, "Which of the following do you most agree with: a) Obama wants the terrorists to win more than he wants Americans to win; b) Obama equally wants the terrorists and Americans to win; c) Obama wants Americans to win more than he wants the terrorists to win?"

I doubt it would make a difference.

It is claimed on Blumenthal's blog that "10-20% of respondents tend agree with any statement (likely due to social norms)." Frankly, I don't think societal pressure would influence 10-20% of voting Democrats to agree that Obama wants the terrorists to win. When it comes to surveys of political ideology, acquiescence bias doesn't seem like such a significant factor.

March 27, 2010:

Of course, the issue of acquiescence bias cuts both ways here. Only 42 percent of Republicans answered "yes" to the question, "Do you believe Obama was born in the United States?" If we take acquiescence bias into account, we might suppose that the real number is significantly lower. In any case, I still suspect that, for ideologically charged quesitons, acquiescence bias probably isn't much of a factor.