Tuesday, September 26, 2006

The Growing Undecided?

By Stuart Rothenberg & Nathan L. Gonzales

We're only a couple weeks into the post-Labor Day sprint to November, and many voters are only now taking the time to see who is running in November.

As a whole, Republican congressional incumbents are starting off in a weaker position this cycle compared to previous cycles, and even unknown Democratic challengers have already solidified their base. That's good news for Democrats.

But we are starting to wonder about some Democratic polling. Do the numbers tell us not only where a race is now but also where it is likely to be in a few weeks? Sure, polls are snapshots. But are they presenting the whole picture? Or are some Democratic polls showing Republican candidates with artificially low numbers by not forcing respondents to choose a candidate, thereby boosting the percentage of undecided voters?

For example, in NY-20, an August 29-30 McLaughlin & Associates poll for Rep. John Sweeney (R-NY) showed him leading attorney Kirsten Gillibrand (D) 53% to 36%, with 11% undecided. An independent August 21-23 Siena Research Institute poll had Sweeney leading Gillibrand 53% to 34%, and 13% undecided. And finally, an August 29-31 Global Strategy Group survey for Gillibrand showed her trailing 47% to 39%, with 14% undecided.

Yet, the DCCC released a September 5-7 poll by Grove Insight that showed Sweeney leading Gillibrand 43% to 31%, with 26% undecided.

In this extremely polarized environment, how could one-quarter of the electorate be undecided? And why is the undecided in the Grove Insight poll almost double the other three polls taken earlier in the race?

We aren't suggesting the numbers are made up, only that the poll's methodology may have produced a result that overstates the percentage of voters in play and understates Sweeney's true standing in the district.

In FL-22, an August 20-24 Benenson Strategy Group survey for the DCCC had Rep. Clay Shaw (R-NY) leading state Sen. Ron Klein 42% to 38%. But 20% undecided seems surprisingly large in a race in which both candidates, and even outside groups, have been spending heavily on media for almost a year.

In contrast, a Democratic poll in Pennsylvania's 4th district, which has been a slow-developing contest, showed the total undecided at eight points, with Rep. Melissa Hart (R-PA) leading Jason Altmire (D) 48% to 44%, with only 8% undecided.

The lesson is to be cautious about believing everything that you are seeing. In some cases, large numbers of voters truly are undecided. But in other instances, a huge undecided may well reflect a poll that seeks to understate the fundamental partisanship of a district.

A great example is in NE-3, where a September 20-21 Penn, Schoen & Berland survey for Scott Kleeb (D) shows him trailing state Sen. Adrian Smith (R) 40% to 31% with a whopping 29% undecided. President Bush won the district 75% to 24% in 2004, so you can bet most of those undecideds won't be voting for Kleeb.

This column first appeared on Political Wire on September 22, 2006.