Why do the results of the national opinion polls on presidential candidate preference differ from the results of the state-by-state polls? That’s the question Sean Trende addresses in a recent RCP post which I commend to your attention. The national polls indicate a slight advantage for Romney in the popular vote while the state-by-state polls indicate a slight but statistically significant advantage for Obama.
Using a variety of approaches Mr. Trende attempts to square the circle. Here’s his conclusion:
How do we resolve this? Which will be correct? My best answer is “I don’t know; it is a source of uncertainty in projecting the election.†I suspect one group of polls will converge upon the other in the next week, and we’ll get a better idea.
Historically, the national polls have actually been more accurate than the state-by-state polls. Next Tuesday we may know.
My guess at this point is that Romney will win the popular vote while Obama takes the electoral vote and, consequently, is re-elected. But it’s no more than a guess. It’s just too close to call. I just don’t believe the folks who are predicting a Romney landslide or an Obama electoral landslide and there are both.
There are several problems with that prediction. Among them are that no president has ever won re-election with the economic trends—not just the raw statistics but the trends—as weak as they are now. Presidents re-elected with high unemployment rates had unemployment rates that were notably trending in the right direction. Another is that there is a general tendency for presidents to be re-elected with larger numbers of electoral votes than they had the last time around. If they get smaller number of electoral votes than they did when first elected, they tend to be defeated.
I assume the discrepancies are multi-causal. The major one I foresee is the unique turnout of 2008, and to a lesser extent 2004. Its hard not to believe that many parts of the country, African-Americans came to vote in places like Alabama for the singular pleasure of a lifetime of casting a vote that would garner no electoral votes for the first African-American President. At the least, this suggests Obama’s 2008 nationwide popular vote was “juiced,” and if he had been white, his victory would probably have looked more like Clinton’s in ’96 (who won with less than 50% of the popular vote). Obama’s 2012 victory will probably look more normal (i.e., more like ’96).
The related questions that might favor Romney are how the individual polls are constructed to get a fair sample in this changing environment, and whether its possible that less minority intensity in swing states is not getting picked up also?
The other discrepancy I suspect comes from Obama’s substantial negative campaign against Romney int the swing states before he could mount a defense. Romney’s post debate bounce appears to be strongest in areas in which Romney brand was not permanently damaged.
PD: In 1992, 1996, and 2000 we have third party candidates who got significant votes. That wasn’t the case in 2008 and is unlikely to be the case this year; Gary Johnson is an afterthought. So, Obama would surely have exceeded the 50 percent threshold in 2008 even without the novelty factor.
James, good point. I had forgotten Perot still had a significant popular vote draw the second time around.
I had also thought about whether third-party candidates are a potential source of discrepancy, but don’t know enough about how/whether the polls might be measuring that differently.
A good libertarian showing would be 0.50% of the popular vote (though Ed Clark got 1.06% in 1980). That might be important in states like CO, FL or VA that Silver’s model suggests are within 1%. My gut instinct is that none of the polls are probably good at measuring the impact of such small third parties, though it looks to me like the libertarians might have cost McCain Indiana and North Carolina four years ago.
I just noticed something. Obama did worse in 2008 than GHW Bush did in 1988, both in terms of popular vote and in terms of electoral votes. Bush had an obvious advantage running on the tail-end of the Reagan era, but also an obvious disadvantage running as a sitting VEEP. Traditionally that just doesn’t work very well: only John Adams, Thomas Jefferson and Martin van Buren have pulled it off in the past, and Adams and Jefferson did it under the old electoral system. I’d be tempted to call it a wash, but let’s face it, Reagan was just that damned popular.
Obama did better than Reagan in 1980 in terms of popular vote, but not in electoral votes. Reagan’s total popular vote was impacted by the John Anderson candidacy. Gotta think that of the 6.6% of Americans that voted for Anderson that Reagan would have won enough to make up the 2.1% popular vote gap with Obama if Anderson hadn’t run. Or maybe Reagan could have borrowed from the 58.8% popular vote total he got in 1984. That’s up there in the LBJ, FDR, Richard Nixon and Warren G. Harding levels! Not bad for a senile old man!
Looking back one can see that third party runs are actually fairly common. Although I’m not sure what to make of the 1824 election. Only one party ran a serious candidate for office that year, the Democratic-Republicans. But I should say candidates, as there were actually four of them that got electoral votes! Yikes! Say what you will about a two party system, but that wasn’t working so well either.
…
It’s a heckuva lot more fun looking at old election results than contemplating the one coming up on Tuesday.
Great stat today: Almost 76 people have been added to the food stamp roles for every net job created during Obama’s Presidency.
THAT’S something to hang one’s hat on!
Vice President Gaffe-o-matic is at it again.
‘There’s Never Been A Day In The Last Four Years I’ve Been Proud To Be His Vice President’
I’m sure he feels the same about you, Joe.