STUMPER: Much has been made of the tightening polls going into the conventions. As the folks at MSNBC's First Read wrote this morning,
"there is no longer this widespread belief among the wise guys and gals
of both parties that we're all just sitting around waiting for this
race to break in Obama's direction. The polls -- as well as the money
race -- suggest otherwise." According to history, is it wise to draw
any conclusions about the eventual outcome from pre-convention polling?
HOLBROOK: Not really. Of course, as you get closer to the election, the polls are going to be
a better predictor of the outcome. My own view, though, is that taking
the pre-convention polls as a predictor of the eventual outcome is a
pretty risky business. You can look back and see, like in 1988, Dukakis
was ahead of Bush before the Republican convention. In 1992, some polls had Bill
Clinton in third place a month before the Democratic convention.
So is there any reason to obsess over pre-convention polls?
Sure.
Why?
Because they have a lot to do with what happens in the polls after the conventions--and
that, in turn, could affect what happens at the polls in November. If
you look at this historically, one of the things that determines the
magnitude of each candidate's post-convention "bump" is where the
candidate is in the campaign prior to the convention--especially
relative to where you might expect him to be. For instance, this year
Barack Obama looks like he's ahead, on average, by two or three
percentage points. But if you think about the kind of year this
is--very low levels of presidential approval, high levels of
dissatisfaction with the direction of the country--you would expect him
to be doing better than that. So in this case, the convention should
provide a sort of corrective. If it follows the predictable pattern, it
should give him a pretty substantial bump and bring him more in line
with where his poll standing should be if the election were to turn out
about the way one might expect it to.
So you'd say that McCain is overperforming, given the climate?
Yes, I'd say so. Although not wildly so.The other thing that seems to
matter here is that the first convention seems to get a bigger bump
than the second. Not always, and it's not always a huge difference. But
you compound that with the fact that Obama is running a bit behind
where he should and I think it's safe to say that he's going to get a
bigger bump than McCain.
Any predictions?
McCain could get a nice four or five point bump. If he does, I would
expect Obama to end up with a six-to-eight-point bump. It's a
little hard to tell right now without more pre-convention data. But I
think something in that range wouldn't be unexpected.
Have we seen conventions act as a corrective on the polling in the past?
Absolutely. Al Gore was running significantly behind expectations
before his convention in 2000, then got a substantial bump that brought
him up closer to kind of victory that most objective observers figured
he would get. But the actual magnitude of the bump is, in my view
anyway, in part a reflection of the conventions as a corrective. They
provide the public with a lot of information. The candidates get out
there and make their case with relatively little interference. That
information gets to the electorate, the partisans come home and that
brings the candidates more in line with where we might expect them to
be on Election Day.
Of course, that doesn't always mean that "he with the biggest 'bump' wins," right?
Right. Gore's bump that dissipated over time--which is one thing that
usually happens with these convention bumps. Much of the time, they
slowly but surely erode. The other thing is that sometimes when a
candidate gets a huge bump, it only brings
them up to where they should be--and they still lose miserably. One of
my favorite examples is Goldwater in 1964. He got a 13-point convention
bump. But that was because he was running at about 22 percent in the
polls before the convention.
Here are a few examples to put this in perspective. In 1972, for
instance, Richard Nixon got virtually no convention bump, but we know that
he won that election in a landslide. One of the reasons he didn't get a
larger bump--by my estimation, it was less than one percentage
point--was that he was running way ahead in the polls before the
convention, and when you're running that far ahead, you're not going to
gain much more. If you go back to, say, 1980, both candidates got, by my
estimation, a 12-point bump. Again, if you look at where they were
standing in the polls prior to the conventions, it makes sense--they
were running behind where you would've expected them to be at that
point. A lot of voters were undecided, and so 12 percent swung to
Reagan after his convention and 12 percent swung to Carter after
his--perfectly offsetting each other. In the end, Reagon won in a landslide, too.
Here's what I'm wondering, though. Given that these bumps
tend to dissipate, is there any reason to think that the immediate
post-convention polling will tell us anything about the outcome in
November?
Yes, and here's why. If Obama does get a nice big bump and ends up
ahead by six points or so, obviously that's good for his
campaign. While it doesn't necessarily predict that he's going to win,
it does
says that he was undervalued going into the convention and that the
ship's finally been righted. The real danger, though, is the "no bump"
scenario. Given that the race is relatively tight now, if Obama doesn't
get a big bump
out of this convention, I think that will say something about how hard
it's going to be for him to increase
his lead in the polls. If he can't do it substantially over a four-day
period when it's all his show, then I think his
campaign should be worried about the months ahead
Should the Obama folks be concerned about conflict with Clinton supporters at the convention? Could that diminish the 'bump'?
Sure, what goes on at the convention probably matters as well. There
are times when the conventions are a mess, and that really ends up
hurting the convening party. Take the Democrats in 1968 and 1972, for
example. In 1972, George McGovern came out of the convention running
two points worse than he was running before it. Most people attribute
that to the fact that the convention was a mess, with McGovern
delivering his acceptance speech in the middle of the night. That said,
I don't think there's going to be much real conflict in Denver. It'll
probably look more like 1988, when there was the whole argument between
Michael Dukakis and Jesse Jackson--would Jackson speak or not? Despite
the clash, Dukakis got a nice bump--just about seven percentage
points--and was vaulted into the lead. Of course, that didn't last. But mild conflict doesn't necessarily correlate with a "bumpless" convention.
This year we have this unique situation where the
conventions are separated by a weekend, as opposed to a full week or
more. They're also relatively late in the season. How will this year's
weird schedule affect things?
In my own research, I've found that the earlier in the summer the
conventions take place, the bigger the bumps will be. I think in part
because people are less settled on whom they're going to vote for and
more open to persuasion. But the thing I'm most concerned about is the
closeness of the two conventions. I think since we've have modern
polling the closest two conventions have ever been is a week
apart--Clinton and Dole were a week apart.
There are two
possibilities here, and both these things could happen. The first one
is that the Democratic Convention ends on Thursday, McCain will
undoubtedly announce his vice-presidential choice on Friday and that
will blunt any post-convention glow that usually translates into a bit
more of a bump for Obama. The other possibility is that the convention
hubbub is really getting started now, and Obama is going to announce
his pick by Friday. So he's got these extra few days of pre-convention
run-up publicity. Now, McCain's not going to have that. He won't the
floor, so to speak, until next Friday. So the compressed schedule could
also blunt McCain's ability to generate a large bump. It's a wrinkle
that will probably have some effect. It might affect them both, or it
might have a stronger effect on one than the other.
Yet another twist in a race that was already pretty unprecedented to begin with.
Exactly.
