Posted 2:13 PM by Robert

The Very First Time

Many Democrats - including Arianna Huffington - hope that first time voters and their cell-phone-only ways will save Kerry-Edwards. This Pace University Pace Poll/Rock the Vote Survey Research Study opens with words of wisdom from James Carville addressing such hopes:

You know what they call a candidate who's counting on a lot of new voters? A loser.

Or, perhaps, "pathetic ... and bound to lose?" Just asking.

The Pace study also includes these observations, which should warm each Caddell cockle:

Senator Kerry's unfavorability rating [among prospective first-time voters] has risen 10 points since July to 46%. During the same time period, the President gained 8 points on Senator Kerry in the head-to-head ballot question, moving from 40% in July to 48% today. ... [T]he President leads Kerry among new voters 48% to 44%. But first-time voters who remain undecided may wind up supporting the challenger.

Of course, we won't know until after the last poll and the election count is completed. But already there are very few "undecideds" left - Zogby, for example, says that the pool of "undecideds" has shrunk to 3% - 4% at this point.

If Zogby is right, who will be left to "break to the challenger" once the Incumbent Rule kicks in? Setting aside margins of error for the moment, suppose 60% of the "undecideds" break to Kerry-Edwards at that point, and they then make up 2% of the voting public. That means Kerry-Edwards could expect a boost of about .2% from the Incumbent Rule.

Whoopee! Bigger effects are probably caused by bad lighting in polling places (mine is in a neighbor's garage) or voters whose arthritis or tremors causes them inadvertently to pull the wrong lever. Perhaps pollster/pundits should spend more of their time analyzing those factors.

POSTSCRIPT: It is worth a minute to peruse a certain feature of the Incumbent Rule that makes it partially self-fulfilling where the incumbent leads in the final poll, and therefore partially meaningless:

The formulation of the Incumbent Rule means where two candidates are arbitrarily close to each other in the final poll with the incumbent leading, it is perfectly possible for the challenger to "benefit" from the Incumbent Rule but the incumbent sometimes still win the election.

That's because the Incumbent Rule considers any split of the undecideds which is less favorable to the incumbent than the incumbent's ratable share of the determined vote in the final poll to be a "break for the challenger." In particular, if the incumbent leads by any amount in the final poll, a 50%-50% split of the undecideds is a "break for the challenger" even though in such a case such a split can never tip the election to the challenger.

For example: Suppose the final poll shows the incumbent drawing 45% of the vote with the challenger at 40% and the remaining 15% of the voters "undecided" at the time of the final poll. If the "undecideds" split 50%-50%, the incumbent will obvously win with 52.5% and the challenger receive 47.5%. The Incumbent Rule calls that result a "break to the challenger." But so long as the incumbent was leading in the final poll, the incumbent will always win the election if the "undecideds" split evenly.

In other words, the Incumbent Rule relies on some of the wisdom of an old Abbott & Costello routine in which Abbott first asks Costello about a 5-year-old girl whose father is 25 years old, or 5 times his daughter's age. "OK," says Costello. Abbott then points out that in 5 years the girl will be 10 years old and her father will be 30, or only 3 times her age. "OK," says Costello. Abbott observes that after 10 more years the girl will be 20 and her father 40, or only twice her age. "OK," says Costello. Then Abbott pops the key question: How long will it be before the girl and her father are the same age?

Or, correspondingly, how many elections have been won by challengers on account of the Incumbent Rule where the incumbent leads in the final poll and the undecideds break to the challenger by splitting 50%-50%?

Now there's a topic for some intense pollster/pundit research.