Incumbent Rule: A Caution and a Plea

Incumbent Rule Interpreting Polls Legacy blog posts

As many of you may have discovered the “incumbent rule” (and this site) through yesterday’s post on Ohio, I want to add one important point: But be careful not to make the same mistake some made back in August and assume that the current snapshot of preferences will remain fixed over the next 12 days. True, the number of movable voters is very small and getting smaller every day, but it is still possible for Bush’s numbers to change. Forty-seven percent (47%) of the vote certainly signals trouble, but at that level Bush needs to pick up only a few points (or have the polls be off by only a few points) to win. That is why pollsters continue to interview voters through the final weekend. The main point of my post yesterday is not that Bush’s percentage of the vote is now “capped,” but that Bush’s number is the one to watch.

Some will argue that the incumbent rule is moot in times of war or in the post 9/11 era. I am dubious, but it is worth noting that he incumbent rule did not seem to apply to a handful of races in 2002: The actual share of the vote received by incumbent Governors Jeb Bush (FL), George Pataki (NY) and Grey Davis (CA) was 4-5 percentage points higher than what each received on the average of final polls taken the last week of the campaign. Senate incumbents Wayne Allard (CO) and Jean Carnahan (MO) similarly outperformed their final polls (data from the subscriber pages of The Polling Report and this scorecard from Survey USA).

I discount these anomalies for several reasons. First, the historical data and theoretical underpinnings of the incumbent rule are far stronger for presidential races. Second, 2002 was an odd off-year election in a lot of respects, including heavier than usual Republican turnout in many areas and a much different political environment with respect to terrorism and national security.
http://www.mydd.com/story/2004/9/3/22294/96534

Finally, I have seen scattered attempts by pollsters this year to roll up undecided voters from multiple surveys to get a large enough sample size for analysis. Each time the data show overwhelmingly negative views of President Bush’s job performance (here and more recently, here). Now, four or more survey organizations are conducting ongoing rolling average tracking surveys that reach as many as 80-100 truly undecided likely voters each week. It certainly would be helpful to see the Bush job performance and Bush and Kerry favorable ratings among several weeks worth of undecided voters, especially if we can compare the small samples of several different organizations. Any polling directors listening out there?

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.