This is a very interesting situation developing among pollsters, what should the “likely voter model” look like? I don’t think pollsters have ever seen a situation where people are actually terrified that we may lose our democracy if they don’t show up to vote in what might be our country’s last “partially” legitimate election for a long time.
One of the hottest Governor races in the country is in Wisconsin. Marquette University is considered the gold standard for Wisconsin elections and has just finished a poll there. It happens that NBC has also finished a poll and the results are considerably different. In fact, the results are so different, NBC actually posted something to explain how they came up with different results.
So first let’s look at the Marquette results and their explanation of their methodology:
Table 1: Vote for governor, by alternative likely-voter models
|
Evers |
Walker |
Anderson |
All Registered |
43 |
47 |
7 |
90% of Registered |
45 |
47 |
5 |
80% of Registered |
47 |
47 |
4 |
Standard Likely Voter |
46 |
47 |
5 |
70% of Registered |
48 |
47 |
3 |
The estimates above for likely voters are based on the definition that the Marquette Law School Poll has used since 2012: those who say that they are certain they will vote in November. Alternative models of likely voters could be broader, including those less than certain to vote, or could be narrower, including enthusiasm and attention to politics. Table 1 shows the vote for governor by alternative measures of likelihood of turnout based on an average of responses to certainty of voting, enthusiasm and attention to politics. The vote for Evers generally rises as the set of criteria for a likely voter requires higher levels of certainty of voting, enthusiasm and attention, implying lower turnout but a more enthusiastic and attentive set of actual voters, while Walker’s vote does not vary by alternative likely-voter definitions.
Something that sticks out right away is that lower turnout helps the Democrat. It’s pretty unusual that lower turnout helps a Democrat. If they are right, it means those of us who understand the danger our country faces are confined to the Democratic base and a lot of people who lean Democrat still have their heads buried in who is going to win dancing with the stars rather than paying attention to the smoke from Rome burning. We need to reach those people in the next 3 weeks.
Now the NBC poll and their interesting explanation of the difference between theirs and Marquette’s:
Our NBC News/Marist poll of Wisconsin shows Democrat Tony Evers leading Republican Gov. Scott Walker by 10 points in a head-to-head contest, and by 8 points in an expanded ballot that includes the Libertarian and Green Party candidates.
A day earlier, however, a Marquette Law School poll had Walker ahead by 1 point. (The polls were on the same page showing Sen. Tammy Baldwin, D-Wis., ahead by double digits.)
So what explains the difference between the two polls? We asked Marist pollster Barbara Carvalho. Here's her answer:
1. The polls use different samples: Marquette uses a voter list; Marist uses random digit dialing.
2. The polls use different likely-voter models: Marquette's model is based exclusively on voters certain to vote; Marist's uses a probability model based on interest, chance of vote and past vote.
3. Voter ID: Marquette's sample among likely voters was R+3; Marist's was D+1.
One other observation from Carvalho:
"Finally, one number to look at in our poll is the 80% of Walker supporters who are strongly committed to him (an extraordinarily high number). This intensity would be magnified in a cutoff 'certain to vote' likely voter model which Marquette uses. Of course, it is something to watch and a reason not to count Walker out."
So what these two polls are saying is, we have two different models of who is going to vote. Clearly, the Democrat base enthusiasm is off the charts. The enthusiasm of women is off the charts. The disgust of many Independents is off the charts.
Normally you can simply look back at past midterms and base your likely voter model on what happened in the past. But this time it really is different. The question becomes, are the emotions that most of us here feel confined to a relatively small group of voters who are paying attention, or are people who normally don’t vote in midterms getting the message?
The Republicans win ALL their elections using lies, fear, and hate. Democrats try to reason with the uninformed using complex policy arguments. Maybe it’s time to pull a page from the Republican playbook. If ever there was a real time to vote out of fear, this is the time!
VOTE and GOTV like our democracy depends on it!