The New Hampshire polling miscue, whatever its ultimate explanation, has an important lesson: In the current presidential primary race, we should pay much more attention to what surveys say about voter uncertainty than the answers they give when pushed to say which candidate they would support "if the election were held today."
When three candidates have ratings in the stratosphere, pushing voters to make a choice before Election Day will produce volatile results.
Consider what the final polls got right: In New Hampshire, for example, the final poll from CNN/WMUR and the University of New Hampshire (conducted the Saturday and Sunday before the election) showed 21 percent of likely Democratic primary voters "still trying to decide" which candidate to support and 26 percent just "leaning" toward a candidate (even though only 6 percent were completely undecided). On the exit poll, 17 percent said they made up their minds on Election Day, and a total of 38 percent made up their minds over the last three days.
Why the uncertainty and late decision-making? Both the final polls and exit polls had a consistent answer: All three Democratic front-runners finished with favorable ratings of 75 percent or more among Democratic primary voters.
Ask pollsters what they expect to happen when one candidate closes out a campaign with a 75 percent (or higher) favorable rating, and they will likely predict a win for that candidate. So when three candidates have ratings in the stratosphere, voters are going to have a hard time making up their minds. Under those conditions, pushing voters to make a choice before Election Day will produce volatile results.
The same dynamic that existed among New Hampshire's information-saturated Democrats is now playing out in a different way nationwide. As of this writing, we have seven new national surveys conducted since last week's New Hampshire primary. Four show Hillary Rodham Clinton leading Barack Obama by 12 percentage points or more. Three show Clinton leading by 5 points or less.
With 25 nominating contests looming on Feb. 5, the national polls are now worth watching. So what gives?
The latest Diageo/Hotline poll out today shows the same pattern of uncertainty. Both Clinton and Obama, as pollster Ed Reillyobserves [PDF], "are extremely well-regarded by the vast majority of Democrats." Clinton's favorable rating among likely Democratic primary voters is 88 percent; Obama's is 76 percent. Each candidate is the lead second choice among each other's supporters (53 percent of Obama's supporters name Clinton as a second choice; 52 percent of Clinton's supporters choose Obama).
Again, in this environment, very small differences in timing, question wording, interviewer procedures and sample composition may loom unusually large. Consider also the big demographic differences: For months, virtually all the surveys have shown Clinton doing better among older and less affluent Democrats, with Obama doing best among the younger half of the electorate and those with more education and income.
Now, the surveys consistently show Obama doing better than before among African Americans, and today, the latest Pew Research Center survey ads a new wrinkle: "Obama now runs even with Clinton among liberals; he trailed by more than 20 points among liberals in late December (49 percent Clinton vs. 27 percent Obama)." So more than usual, the kinds of people that qualify as "primary voters" on the national polls may determine which candidate does better or worse.
On the Republican side, polls this week have shown less variation, but Republican voters express even more uncertainty than Democrats. Although John McCain has received a significant post-New Hampshire "bump" to between 28 percent and 34 percent of the Republican vote in the seven national polls, a whopping 59 percent on the Diageo/Hotline poll say they could "still change their minds."
Here the uncertainty has a different source. McCain's favorable ratings now easily lead the Republican field (75 percent among Republican primary voters in the Diageo/Hotline poll). McCain also benefits from newly perceived viability in the wake of his New Hampshire primary win. Four out of five Republicans in the Diageo/Hotline poll (82 percent) now say McCain is likely "to be able to win the Republican primary nomination," far more than Mike Huckabee (56 percent), Mitt Romney (53 percent) or Rudy Giuliani (45 percent).
The big challenge for McCain, according to the Pew survey, is that his gains have come mostly from moderate to liberal Republicans (41 percent support him, compared with 20 percent for Huckabee and 8 percent for Romney). He trails Huckabee (25 percent to 33 percent) among evangelical conservatives and Romney (22 percent to 29 percent) among "other conservatives."
Thus the root explanation for the uncertainty: Many Republicans see McCain as less conservative than they see themselves and candidates Romney and Huckabee. According to the Pew survey, roughly two-thirds of Republicans (66 percent) describe themselves as conservative, but only 50 percent say the same for McCain. Roughly two-thirds describe Romney (68 percent) and Huckabee (65 percent) as conservative.
So the lesson? The national "horse race" results may vary from survey to survey, or -- as perceived "momentum" shifts between primaries -- from week to week. Uncertainty is the one constant in both races, at least for the moment. As long as it persists, we are in for a very interesting campaign.
-- Mark Blumenthal is editor and publisher of Pollster.com. His e-mail address is firstname.lastname@example.org.