What is it exactly that the NYT poll measures?
The headline is “No Poll Boost for Edwards,” refreshingly honest for a NYT piece. You have to read all the way to the penultimate paragraph to find out that the race is Kerry 49, Bush 45, according to the poll. There are tidbits on Iraq (a majority against the war), hope for Democrats (Cheney’s negatives high), and so forth.
But here’s my question: from the standpoint of electoral politics, does the poll tell us anything at all?
The answer is no.
Broadly speaking, here’s why: (a) this is a poll of the great unwashed, the kind of people who routinely cannot name the Veep or Secretary of State on Sean Hannity’s show: there is no serious attempt to identify actual voters; (b) the poll skews Democrat, and there is no attempt to control for this in the reported results. This CBS/NYT poll’s methodology ought properly to be described as 50% people who care, 50% people who don’t: and we will not sort out one from the other.
Let’s start with the sample, 955 adults, 823 of whom are registered to vote. Good, throw away 132 — but they don’t, even though these people literally do not count in an election. Next, the 823. There’s a lot of lying to the pollster going on. 27% report that the last time they voted was 2004: do you believe that? I don’t. 19% report they last voted in 2003. The most pathetic is the 1% who said they last voted in 2001: what was the election, and how do you even remember it? Here’s a better methodology: unless a person can name the local or special election of 2003 or 2004, throw them out: that’s up to 46%!! So we should be down to 444 respondents or so, and maybe they could tell us something meaningful, but we never get the chance to find out.
Another perspective on our fellow citizens who lie to pollsters: 83% of those polled say they are registered to vote, while 85% say they definitely will vote in 2004, and another 11% say they will “probably” do so. Actual turnout is around 50% of registered voters. Put it this way: the margin of error is supposed to be +/- 3% in the poll, but we know that the margin of error on who will actually vote is around 50%!! Therefore, no accurate conclusions about an election can be drawn from these results, unless you can prove the the 50% who don’t vote have the same views as the 50% who will vote.
And they can’t prove or claim that, cause it’s not true, bringing us to point (b) from above.
The poll skews Democrat and there is no attempt to correct for this. For starters, by 46% to 37%, those polled said they will vote for the Democratic candidate for Congress in their district. Why does this not bring the pollsters to a screeching halt? No one believes — neither Dem nor Rep — that there is the remotest chance of Democrats reversing the 222 – 208 (51%) GOP control of the house, which has been gerrymandered into mostly all safe districts for each party. So a useful corrective to the results would have been to adjust the presidential horse race results by the reality that is the House. They don’t do this.
The poll is not literally dishonest. It reports that the pollees self-described as 29% GOP, 37% Democrat, 30% independent. It just doesn’t do anything useful with the information that Democrats are being oversampled. Similarly, if this stalwart sample of the electorate had had their way in 2000, according to what they say, the 74% of them who claim they voted would have given Bush a 38-33% margin over Gore. This seems to me evidence of what? — that people tell pollsters they went with the winner, that one third of the 74% interviewed about the 2000 election are lying outright to please the pollster?
What it wells me is that this is not an election poll, nor, in an important sense, even a public opinion poll. It is rather a poll which is designed to give results about the way America should think and vote. It is the New York Times version of the Laker girls.
This is sad, and it is the very opposite of the public service which the Times thinks it is delivering. Somewhere between a third to a half of people contacted by telephone lie to representatives of the New York Times, and the Times publishes the intermingled truth and lies with no serious attempt to distinguish the one from the other, all in the interest of Democratic cheerleading.
On November 3, President Bush will in all likelihood be re-elected, and may well carry over forty states. Those who rely on the New York Times for an inkling this will happen are little better off than those couch potatoes who get their opinions from watching the first five minutes of Peter Jennings twice a week.
At that time, there will be all sorts of recriminations against John Kerry as a poor candidate, charges that the Democrats “didn’t get their message out,” and similar fish-wrap tarted up as news. There will be shock and surprise, all of which is completely unnecessary if the NYT and the other outlets of the elite media were less committed to reporting their wishes and hopes as news.