Political Animal


July 23, 2012 10:00 AM Burying the “Incumbent Rule”

By Ed Kilgore

One of the hoary assumptions about presidential elections that used to be intoned regularly as an ‘iron law” was the so-called “incumbent rule:” the idea that in contests involving incumbents, undecided voters tend to break sharply in favor of challengers late in the cycle. The “incumbent rule” has been an important prop for the still powerful “referendum rule” holding that such elections are always and invariably up-or-down votes on the record of the incumbent, with the identity, ideology or agenda of the challenger being essentially irrelevant.

Together, the “incumbency” and “referendum” rules paint a picture of an electoral dynamic in which voters make up their minds about the incumbent at some point well before Election Day and then stampede into the challenger’s column. For obvious reasons, this is the picture tht pro-Romney spinners—most notoriously Dick Morris—are busy projecting in every utterance.

I mention all that by way of suggesting you read Nate Silver’s latest FiveThirtyEight post, which (although Nate is as always cautious) pretty much buries the “incumbent rule.” He does find that early polls often understate challengers’ final performance because (a) they are engaged in primary battles that artificially depress their support levels among members of their own party, and (b) they are not as well known early on as incumbents. We’ve seen that phenomenon in 2012, when Mitt Romney was running well behind Obama in general election trial heats back in the early spring before he put away Rick Santorum and began to consolidate Republican support. But Nate suggests that incumbents are as likely to get a late “bounce” as challengers, based on looking at every presidential election since 1968.

It’s important to understand that the “incumbent rule” is the kind of hypothesis that is relatively simple to test empirically. The “referendum rule” isn’t, for the obvious reason that it’s much harder to deduce why voters wound up supporting a challenger as opposed to an incumbent. A swing voter who says he or she decided to “fire” an incumbent for poor job performance in order to “give the challenger a chance” may have actually compared the two candidates and found the challenger’s message—which of course will always involve a recommendation that the incumbent be “fired” for poor performance—more compelling. And lest we forget, the vast majority of voters, particularly at this heavily polarized juncture, are casting ballots on the basis of partisan affiliation and aren’t making any of the calculations that analysts so often attribute to them.

Ed Kilgore is a contributing writer to the Washington Monthly. He is managing editor for The Democratic Strategist and a senior fellow at the Progressive Policy Institute. Find him on Twitter: @ed_kilgore.


  • c u n d gulag on July 23, 2012 10:07 AM:

    Dick Morris?


    Dick Morris makes feckin' "Bloody" Bill Kristol look like 'The Oracle at Delphi!'

  • stormskies on July 23, 2012 10:23 AM:

    And while pontificating this crap he was sucking his toes............

  • jcricket on July 23, 2012 10:43 AM:

    Dick Morris, after all these years, still hasn't been able to reconcile himself to the rejection he received from the Clintons because of his own behavior.

    Look up 'sour grapes' in the dictionary and you will see a photo of Morris.

  • DeirdreTom on July 23, 2012 10:44 AM:

    just as Eva said I am taken by surprise that anyone can profit $7722 in a few weeks on the computer. did you see this page(Click on menu Home more information) http://goo.gl/lkSj4

  • c u n d gulag on July 23, 2012 10:54 AM:

    Good boy, CRAPTCHA!

    Thanks for proving once again, that you're about as useful as mammaries on a male bovine.

    I can't read your feckin' crap, but some 2-bit, internet work-at-home hustler's computer program can.

    Explain to me, once again, oh WaMo Guru's, why we need CRAPTCHA again?

    Oh yeah, to keep out spam!

    Good boy, CRAPTCHA!!!

  • exlibra on July 23, 2012 11:04 AM:

    Funny, gulag... I've always thought of Craptcha as a female (because of the -a ending) and have been wondering why she's - atypically -- so inefficient at weeding out those ads, while giving me hell. But, if it's a male... Probably Republican, too (white and without a college degree)

  • c u n d gulag on July 23, 2012 11:11 AM:

    Whether CRAPTCHA is a he, a she, or an it, there's no question it's DEFINITELY Conservative.

    First, it limits free speech.
    Then - it's into 'comment suppression.'

    CRAPTCHA is NO Liberal or Progressive, THAT we can ALL be sure of!

  • Mimikatz on July 23, 2012 11:27 AM:

    The thing the pontificators of those "rules" seem to forget is that you can't beat something with nothing. Romney projects an aura of business competence, but the more people find out about him, the more that is tarnished, and he can't really explain why vulture capitalism is really so beneficial. And he and his lady wife just llitale ACCOMPANY,get stranger the better people get to know them.

    So it's "better the devil you know than the devil you don't know" for everyone for whom Obama isn't really the devil.

  • Matt McIrvin on July 23, 2012 1:22 PM:

    Sam Wang at election.princeton.edu learned this lesson the hard way in 2004. He was aggregating polls in much the manner of Nate Silver, only with a simpler and cleaner model. Late in the campaign, he started posting modified electoral-vote diagrams that took into account a theoretical break of undecideds toward John Kerry.

    But it didn't happen; the unmodified poll aggregates did a much better job of predicting the actual 2004 result. (There's a popular legend that Bush mysteriously outperformed his polling, but that was based on early, raw exit-poll results, which are actually much less reliable than ordinary polls taken in the last week before the election; these were collectively pretty much spot on.)

    Then in 2008 Sam Wang decided the incumbent rule was not true, and just based his projections on unmodified poll aggregates. He did startlingly well, better than Nate Silver in fact, with a far less complicated analysis. It's possible to be too clever for your own good.

  • unclemilton on August 07, 2012 10:51 AM:

    Two points.

    1. "It’s important to understand that the “incumbent rule” is the kind of hypothesis that is relatively simple to test empirically."

    Correct. Unfortunately, your friend Nate Silver can't seem to pull it off. A quick read of his article shows that he did not test presidential incumbents, rather he tested presidential incumbent parties. For an actual analysis of the issue google - Incumbent Races: Closer Than They Appear by Nick Panagakis.

    2. "the vast majority of voters...aren’t making any of the calculations that analysts so often attribute to them."

    Analysts aren't attributing these calculations to the vast majority of voters. They are attributing them to the swing vote.

  • Nick Panagakis on August 23, 2012 1:30 PM:

    Regarding Nate Silver: His analysis of incumbency effects includes months and months of campaigning. Baffling. Last two weeks of campaigning are the only polls that count when comparing with election results. Don’t late campaigning have most effect on final election day results? Last two weeks are the only polls that count in my book.

    My Polling Report “article” at PollingReport.com is still posted: “Incumbent Races Closer Than They Appear”. Timeframe: “The 155 polls we collected and analyzed were the final polls conducted in each particular race; most were completed within two weeks of election day.”

    Those polls were all state races including CBS, Gallup, Gordon S. Black Corp., Market Opinion Research, Tarrance Associates, and Mason-Dixon Opinion Research, my own (MSC) and more dating from 1977 to 1988. Results? 82% went to the challenger, 12% to the incumbent and 6% split. The only incumbents included were full term incumbents. Xls available on request.

    Update. In our final October 2010 IL poll for the Chicago Tribune we had three full-term incumbent officers: 2- term AG Lisa Madigan, poll 69% election 65%; 3-term Sec of State Jesse White, poll 72% election 70%; former 3-term State Treasurer Judy Baar Topinka, poll 57% election 53%.

    I rest my case.

    Nick Panagakis

  • Dennis D on September 27, 2012 3:57 PM:

    Why does everyone want to quote Nate Silver? He was a baseball analyst. He predicted 52 GOP House gains in 2010. They gained 63. Nate was off by 20%

  • PatTheRat on October 12, 2012 7:27 PM:

    The Incumbent Rule is already kicking in. The surge seen by Romney cannot be simply due to just his debate performance - although it was by all accounts a "home run". Instead, people have been holding back their decision on whether to support Romney or just not vote at all are deciding in favor of voting for Romney, and the polls have reversed in just the last week. Florida, North Carolina and Virginia have already been called for Romney by pollsters, and no further polling will be done in those 3 states by several. It always seems that the incumbent early in a race tries to discount the Incumbent Rule's authenticity, but it always is proven in the last several weeks of the election cycle. This cycle is no exception. Romney is now projected to win with 331 to Obama's 227. Another projection has Romney winning 348 to Obama's 211. All the money in the world cannot force people to change their convictions and vote for someone they hate. Romney has this election wrapped up.