SurveyUSA shows Dayton with a one point lead, in the margin of sampling error. Public Policy Polling found a three point lead for Mark Dayton. Polls from Star Tribune/PSRA, St. Cloud State U. and MPR/HHH in the last week show much larger leads for Dayton. My hunch? This is a close race, with a narrow Dayton lead. But what stock should we place in the last two robopolls versus the traditionally dialed results that show a wider Dayton lead? And what about SurveyUSA including some cell phone users in their last poll?
Both PPP and SurveyUSA conduct automated recorded polls. The difference is that SurveyUSA included cell-phone only (CPO) households for the first time in their last Minnesota Governor poll. While including CPO households is an emerging best practice in the polling industry, inclusion of CPO households has significantly changed the demographic composition of the SurveyUSA sample. The biggest change? Age.
The most recent (10/24-27) SurveyUSA sample (including CPO households), showed 58% of likely voters under age 50, only 42% over age 50. This is a big shift from SurveyUSA 10/14 (excluding CPO households) which was 52-48 on the under/over 50 composition. If we go further back, in September (9/12-14) SurveyUSA's sample was 50-50 under/over 50. I doubt the electorate really has skewed from a dead even split at age 50 in September to a 16 point lead for under 50 voters near Halloween. Since SurveyUSA does not weight its landline only results to accommodate CPO users, a big chunk of this change can be attributed to the addition of these households.
Mark Dayton has done better among older voters. He focused his primary campaign on turning out seniors. Every general election poll has shown him leading older voters, going as far back as you care to look. That's part of what makes Minnesota a bizarre aberration in 2010 compared to the rest of the country. While national commentators have focused on the dropoff in young voters as a reason for waning Democratic prospects, in Minnesota older voters have been a mainstay of DFL support.
In terms of age, Emmer's support base has consistently been voters 30-50 years old. Depending on the pollster, you get very diverse results with young voters. For example, PPP shows a healthy Dayton lead among 18-29 year olds, SCSU shows strong Dayton support among young voters, but SurveyUSA and the Star Tribune show Emmer leading among young voters. There's no doubt that CPO households are younger, but the question remains whether these young voters lean toward Emmer or Dayton. That has to be a point of concern for the DFL.
This might explain, in part, another dichotomy between Minnesota and the nation as a whole. The widespread suspicion has been that polls that exclude CPO households understate support for Democrats.
CPO respondents in the SurveyUSA poll split evenly between Emmer and Dayton (35-35). But CPO's are also the most undecided voters; SurveyUSA respondents who were reached on a cell phone were 5 times more likely to be undecided (15%) than respondents reached on a landline phone (3%).
Despite the wonkish nature of this discussion, there are a couple of items to be attentive to in the last few days of the campaign and beyond. First, the polls have little or no agreement about the size and candidate preference of young voters. Whoever gets it right will likely win the polling contest. Second, the way these voters break at the end will be a crucial part of the post-election narrative.
Follow me on Twitter @aaronklemz