The New York Times Poll and (Slight) Rhetorical Overstatement
It occurs to me that I may have slightly overstated the case in talking about the “corruption” of various polls. As long as they fully release the meaningful internals of the poll, there is no deception in the poll itself. What is deceptive is the reporting of the poll, and we see that clearly in the Newsweek poll in a way that is more egregious than most — truly dishonest.
Take for example the NYT poll this morning. The headline of the piece is very reasonable and supported by the facts: “Poll Finds Kerry Assured Voters in Initial Debate.” Inside the poll (the pdf is only accessible from the main poll page) we find that, while Bush’s “strongly favor” number increased from 60% to 70%, the Kerry number increased more significantly from 42% to 48%. Same with “strong leader”: 62/36.
Joe Lockhart’s spin seems a little over the top (maybe he got it from Bill Burkett):
“The public took a measure of John Kerry standing next to the president, and came to the conclusion that he had the strength, judgment and experience to be the commander in chief,” said Joe Lockhart, a senior strategist for Mr. Kerry.
It is certainly undeniable that Kerry did himself some good, particularly with his base (and GOPer’s fear, with weak-minded undecideds who are looking for the “strong horse.”
Some other results are interesting. For example, Bush has a favorable 57/37 in how he is dealing with terrorism, but at the beginning of the year that number was in the 90/10 range — complacency is a by product of the administration’s success in preventing another terrorist attack in the US. Similar terrorism and homeland security questions get the sam result: a strong Bush endorsement, but the same fade — which I attribute to complacency.
The Invisible Hand of Bias
While the Times gave a fair headline to its piece, other elements of the poll are misleading — and in a more serious way.
It is significant that the poll has the lowest level of self-described moderates (36% versus normal low 40′s) and the third highest number of liberals (24% versus around 20%) in the 14 years of statistics that the NYT shares with us (p. 31 of the pdf). It is obvious that if the Times weighted its poll by its normal distribution among lib/mod/con, the results would be a Bush win instead of the 47/47 tie they forecast.
The change in the mix of respondents to include more liberals and fewer moderates is evident in what may be the most significant numbers in the poll. Two weeks ago, Kerry had a 44/32 Unfavorable Rating, meaning he was unelectable here or on any planet of his choosing (LGF). Now, he has a 41/40 Unfavorable, meaning that he is still unelectable in America’s Global Test, but could probably become prime minister of Canada.