OUTSIDE THE MARGIN OF ERROR

InstantMan points to a Time press release about a new poll. The poll shows the race to be 52/41/3, Bush/Kerry/Nader among likely voters if the election were held today. That obviously leaves undecided and "other" at 4%. The margin of error is +/-4%. That means that the survey puts Bush's lead outside the margin of error, and that's a first in this race.

What is the change in voter preferences among the candidates from the last Time poll? 46/44/5 Bush/Kerry/Nader.

There's more in the press release, but not enough to answer the many questions that occur to me, such as:

Did Bush take votes from Kerry, Nader, or
some combination of the two (and what combination)?

What are the numbers without Nader in the race? (Support for third party candidates typically fades as election day approaches, and Kerry is more likely than Bush to benefit from Nader defections. Also, Nader is having trouble getting on the ballot in some states, and he obviously won't be a factor where he is not on the ballot.)

Is there a state by state breakdown (and accompanying totals for the electoral college?

If not, is there at least a breakout of the potential swing states like Ohio, Florida and Michigan?

What percentage of likely voters is undecided, and how did that change from the last Time poll?

Standard warning language: The election is not being held today. Horse-race polls may or may not reflect reality on the day they were compiled. They sure as hell do not provide a reliable indication of what will happen two months from now.

Still. I like the numbers.

UPDATE, 9/4/04: Newsweek comes up with similar figures. They call it 54/43/3, Bush/Kerry/Nader, with (apparently) no undecideds and a margin of error of +/-4%. Perhaps they just left out the undecideds and figured the percentage of voters for each candidate from the pool of decided voters. The Newsweek survey was taken on Thursday and Friday, 9/2-3 (and therefore at least partially reflects the results of Bush's acceptance speech), compared with the Time survey, which was taken on Tuesday, Wednesday and Thursday, 8/31-9/2 (and therefore could not reflect any effect from the speech).

Bush's reaction to the polls is the right one: a shrug and "I've got a lot of work to do."

Assuming that both polls are accurate, there is still no information on the effect the changed voter alignment will have on the electoral college.

Also, Glen Reynolds points to this table, breaking down the Newsweek survey by day to show the effect of the acceptance speech:

9/3 only 54 38 4 4
9/2 only 49 43 3 5
That is a helluva bounce from a single speech.

Glen speculates that the numbers may also reflect the disaster that was Kerry's midnight rally after the Republican convention. (Approximate quote: I will not have it. I will not allow Dick Cheney and his five draft deferments to question my patriotism. Of course, Cheney did no such thing. Cheney questioned his policies and his judgment. The absence of any substantive response to Bush and Cheney was striking.)

This is not good news for Kerry. Those are Dukakis like numbers.