On Mon, 2 Aug 2004, Doug Henwood wrote:
> I'm going to be interviewing Frank Newport, editor in chief of the
> Gallup Poll, on my radio show on Thursday. Suggestions for questions are
> welcome.
Here's a few:
1) About a month ago, Frank Newport said "None of the five presidents who won re-election [since 1956] were behind their eventual opponent in *any* trial heats after January in the year prior to their election." [Emphasis mine]. And "based on historical patterns, Bush's [under 50 percent] job approval rating is thus underperforming the pattern of presidents who have won re-election."
I'd ask him how seriously he thinks we should take that correlation. Is the fact that Bush has been under 50% in almost all trial heat polls for the last five months the real news that maybe allows us to cut through the noise of Nader and LV models and undecideds? Does he think that at this early stage it's the best available predictor of whether he'll get reelected? Or does he think it's more like the correlation between short skirts and bull markets -- an intriguing factoid we shouldn't bet the bank on?
2. I'd ask him about the Likely Voter vs. Registered Voter models. In his May 19th newsletter, Ruy Teixeira argued that there is little evidence that LV screens work this far out -- that they only work in the Fall, closer to the election when the "interest in the election" variable is stable. He also said that most organizations don't use them this far out. He specifically mentioned Gallup as the main exception, and said he thought the reason Gallup was supplying them so early was so they wouldn't have to explain to their commercial customers why there was an abrupt change in the horserace figures in the fall, but that professional psephologists place less emphasis on them now.
I'd be interested in any part of that he'd like to comment on. I'd also be interested in anything he had to say about how Likely Voter Models work. (It seems every organization has their own secret receipe, although I assume there are accepted principles.) Also whether, because of their assumptions, LV screens work better some years than others, and whether he has any reason to think this year might be an exception.
3. I'd ask him about swing voters, who they are, and about the difference between swing voters and undecided voters. My understanding is that 20% of the electorate presently says that it is too early to have made up their mind. And that this is about 10% less than is usual at this point in the campaign. So that the swing voter population is 20%, smaller at than the 30% that is normal at this point in the election -- in fact smaller for this point in time than any previous presidential election -- but still rather sizeable, and way bigger than anybody's margin.
What is confusing is that the same polls that have a tracking question measuring whether it's too early to make up your mind at the same time only list about 10% as Undecided on their trail heat breakdowns (and sometimes even less that, and sometimes even zero in three way polls). So does that mean that roughly half the swing voters have a preference but might change it? Has that generally been the case?
And lastly, I believe I read somewhere that the incumbent never gets more than a third of the late undecideds because they already know him. Any truth to that? Are there any comparable trends among the swing voters who aren't listed undecideds (assuming I'm not wrong about this categorization)?
4) The question underlying the preceding question is: is the dynamic of this election fundamentally different from all other elections? Normally all parties aim more of their pitches towards the middle because swing voters are more numerous, more fickle, and more costly to lose (since a switch voter hurts you twice). Some people claim that there are so few swing voters this election that that logic doesn't hold -- that each party has more than twice as many people to be gained on the wing as in the center and so should emphasize polarizing claims rather than moderating ones. Does he think that's true?
And aside from whether it's true in fact, does he think the Republicans think it's true?
5) Does he think this will be a high turn-out year or a low-turnout year? (And do polls have any light to shed on that question?)
6) What does he think about the Nader vote? Does he think it will fade to under a percent, the way the Buchanan vote did in 2000? Has Gallup done any polls of Nader voters trying to discern what they'll do if it's still close in November?
7) How does he feel about answering long involved question schedules? :o)
Michael