Since my email box was chock full of forwarded copies of Andrew Kohut’s op-ed piece on polling in this morning’s New York Times, I thought I would blog it. Among other things, he chimes in on the debate over the incumbent rule and alternative theories:
One piece of received wisdom is that the swing voters will ultimately break for Mr. Kerry. The argument goes something like this: These voters know George Bush and are unhappy with his performance and the state of the nation. It is therefore hard to imagine that they will stick with him. Well, there is a wealth of data showing that swing voters – who still constitute 14 percent of likely voters – think all these things and are still lukewarm about Mr. Kerry, particularly his ability to handle the war on terrorism.
Another common assumption is that voters will not switch horses when the country is facing an overseas threat and has troops are in the field. Despite his anemic approval rating of 47 percent, Mr. Bush will hang on because uncommitted voters will not put the country at risk. No wartime president has lost a re-election bid, those who argue this position add. This just might be so, but we’re not living through World War II or Vietnam. Right now, there are plenty of swing voters who give higher priority to domestic issues (where Mr. Kerry is strong) than to security issues.
It’s hard to argue, as seems to feel strongly both ways.
Kohut’s lead will provoke a bit more debate. He explains that voter volatility is the reason “polls are not going to give us a clear picture” of the outcome. “This is not because polling no longer works – it’s because voter opinion is highly unstable.”
While I agree that polling still works, that voter opinion can be unstable and that cross-pressured swing voters still exist, I am not sure we can ignore the volatility in some likely voter models. Emory Professor Alan Abramowitz emails with a valid counterpoint:
The problem is that poll results are volatile due to the effects of sampling error, differences in weighting procedures, and different methods of identifying likely voters. As a result, even polls conducted at exactly the same time can produce divergent results. Nor is this situation unique to the 2004 election. The same volatility was evident prior to the 2000 election when, for example, the Gallup tracking poll released on October 26 showed George Bush leading Al Gore by 13 points while the Zogby tracking poll released the same day showed Gore leading Bush by 2 points. [Note to Abramowitz: You really should blog this stuff!]
Another good tease to the upcoming discussion of likely voter models. [If I were a certain well known blogger, a mysterious yet imaginary “editor” might now suddenly appear and ask, Didn’t you promise that five days ago?—ed. I’m working on it. I’m working on it!!]
Mark, this won’t add to the discussion, but I just want to thank you for blogging on the art and science of polling. Your insights are illuminating. I look forward to each new post!
Kohut is missing the point. Of course, swing voters are still lukewarm about Kerry. That’s why they’re “swing” voters, and not “swung” voters. The question is not how they feel now, but how they’ll feel on Election Day when they don’t have the luxury of remaining undecided. Will their distaste for Bush trump their lack of enthusiasm for Kerry? History says that, yes, it probably will.
I wanted to bring a bit of a British perspective to all of this. The pollsters for the 1992 General Election got it spectacularly wrong and that election is considered the low water mark of UK polling:
http://www.alba.org.uk/polls/accuracy.html
The post mortem found that most of the 8% innaccuracy came about simply because many Conservative Party voters were too shy/embarrassed/ashamed to admit who they were going to vote for. I feel many Bush voters may feel the same way in this election.
Another difference between UK and US polls is sample size. Most polls published in the UK use sample sizes of 2-3000 whereas polls using only 500 people seem to be acceptable in the States.
Fred App speculates that swing voters
will eventually “pull their lever”|”mark
their ballot”|”punch their chad”|”press
the screen for Kerry” because history
says they probably will.
An equally strong – and just as dubious –
counter argument would be “this time
it’s different.”
Frankly, I don’t think the pollsters
have a clue about what is GOING to
happen on November 2nd. In fact,
I’m sure each and every one would
say that their numbers are only good
“if the election were held today”.
This one seems to be close now. It
may come down to a “get out the vote
effort” by the Democrats|Republicans.
Or it could be a landslide for the
President|Senator Kerry.
How’s that for a straddle? :-}
I think Fred App has it precisely backwards.
Those “swing voters” have had 4 years of BusHitler, “selected not elected”, MSM beating up on Bush whenever they can, and they still haven’t committed to Kerry.
If they were ABBers, they’d already be with Kerry. So Kerry has to win them, not just inherit them.
Kerry’s had 18 months to win them, and hasn’t. Kerry’s sole message is “trust me, I have a plan, and I’m not George Bush.” He won’t tell us what the plans are, he just wants us to take it on faith that they’re great, and will work.
If you believe in Tinker Bell, that may work, for the rest of us, it’s a rather hard sell.
IMHO, the vast majority of the “undecideds” are going to vote for Bush. We’re at war, and if Kerry wants people’s votes, he’s going to have to earn them.
He hasn’t done that.
Greg says Kerry’s had 18 months to win over these swing voters and hasn’t, so they’re likely to vote for Bush. How does that make sense? Bush has had FOUR YEARS to win them over and he hasn’t done that.
Yes, if they were Anyone-But-Bushites, they’d be in the Kerry column already. But they’re not. They’re just Not-Bushites — meaning they won’t vote for Bush, but they don’t reflexively vote for any opponent. Most of them have already decided they don’t like Bush; what they’re struggling with is whether they like Kerry. given that, isn’t it more likely they’ll vote for the guy who failed to impress them in a year than the guy who hasn’t impressed them over an entire presidential term?
That October 26th 2000 Gallup result is a thing of beauty. I have it framed on my wall, next to my Gary Hart poster.
Is there any comparative work on sample sizes across different countries? In Australia, with only 20 million people, our national polls typically use samples of 1200 and up. For the record, Australian polls were as erratic before the 9 October federal election and as useless in predicting the result. It might be necessary to look for a cause to this that reaches beyond one country.
Neil: In the 1992 UK election the last minute swing was predominantly traditional Labourites who changed at the last minute. Unlike in the US, in the UK the winning party can pretty much implement its agenda unimpeded. And the 1992 Labour platform included tax increases that would have devastated many of the middle class people who had become homeowners under Thatcher’s policies (prior to Thatcher’s arrival only a tiny minority of UK citizens owned their home). It was these policies that caused much of the last minute swing.
Fred,
1: “Not Bushies” who haven’t been won by Kerry are likely to just not vote.
2: But if they do vote, when the US is at war, who are they more likely to vote for: the “unknown” who hasn’t impressed them, or the “known” who’s done an ok, but not great, job (if they thought he’d done a lousy job, they’d be ABBers)?
It’s been over threee years since 9/11/01, with no major attack on US soil. You think they really want to bet that Kerry will do as well?
Just saw this link on Slate — Pat Caddell says the incumbent rule is all hogwash (as of September ’04, anyway) — see for yourself. He doesn’t have any data to back this up at all.
http://homepage.mac.com/mkoldys/iblog/C168863457/E477987530/index.html