More on the Incumbent Rule

Incumbent Rule Interpreting Polls Legacy blog posts The 2004 Race

Several readers took strong exception to my discussion of the incumbent rule, the idea that undecided voters tend to “break” toward the incumbentchallenger just before Election Day. Gerry (of Daly Thoughts) was most succinct:

We are a month away from the election. Those who are undecided now are not more likely to break for the challenger. At least, that is what the history of Presidential elections says (lower ticket races may be different). Perhaps those undecided the day or two before the election are, but we are not there yet.

To which I say, true, but that was not my point. The “incumbent rule” is about the “break” toward the challenger between the last poll and Election Day. From a month or more out, Gerry is right, presidential races do not seem to break with much consistency either way. Looking at Gallup’s collection of presidential election polling charts dating back to 1936, I see trends favoring four challengers (Wilkie 1940, Goldwater 1964, McGovern 1972, and Dole 1996) and four incumbents (Roosevelt 1936, Truman 1948, Eisenhower 1956 and Ford 1976). In 1980, Jimmy Carter gained support during September and October, but then Reagan surged in the final week after the debate. In 1992, George H.W. Bush lost a few points in September then gained them back in late October. (See some possibly different interpretations here, here and here).

The point of all this is not predicting trends to Kerry or Bush over the next three weeks – if you can do that, your crystal ball is better than mine – but how to characterize where the race would stand “if the election were held today.” Should news coverage focus so relentlessly on where the race stands at any given moment? Probably not, but there is no question that it does. Thus, if we want to accurately characterize where the race stands right now, the incumbent rule tells us to focus more on the incumbent’s percentage relative to 50% than the margin between the incumbent and the challenger.

It is also a misconception that the race over the next three weeks is only about appeals to the completely “undecided” (and efforts to mobilize base constituencies). Voters may still shift from one candidate to another or to undecided. The just released ABC/Washington Post poll, for example, shows only 3% of likely voters completely undecided, but another 11% who could “change their minds” about the candidate they now support. Similarly, the new Marist Poll shows 6% undecided, and another 4% who support a candidate but say they “might vote differently” on election day.

MyDD’s Chris Bowers also pointed out that I understated one aspect of his data: “86 percent of the total number of undecideds broke for the challenger” in the 28 presidential polls he examined (not 86% of the polls – emphasis added). He also noted that the size of the break is typically modest, only 2-3 points, since the pool of true undecideds is small in the final week. However, if the current standings persist, that 2-3 points could be crucial.

Finally, one important caveat: A number of races defied the incumbent rule in the 2002 off-year elections. Commenter “Dr. X” was right that “many low polling sub-50% incumbents, such as Wayne Allard, Tim Johnson, Mary Landrieux Gray Davis…were re-elected.” Is the incumbent rule moot post 9/11? I doubt it, but we should remember that the rule, while based on solid evidence, is still more art than science.

[Continue with an application of the incumbent rule in Ohio]

Mark Blumenthal

Mark Blumenthal is political pollster with deep and varied experience across survey research, campaigns, and media. The original "Mystery Pollster" and co-creator of Pollster.com, he explains complex concepts to a multitude of audiences and how data informs politics and decision-making. A researcher and consultant who crafts effective questions and identifies innovative solutions to deliver results. An award winning political journalist who brings insights and crafts compelling narratives from chaotic data.