Wisconsin SD-10 Special Election - Lessons Learned
February 2, 2018
ALG conducted the polling for the Wisconsin State Senate Democratic Committee in the recent special election (SD-10) where Patty Schachtner flipped a historically Republican district and ran 27 points ahead of Donald Trump's 2016 margin. It's part of a pattern of dozens of strong Democratic showings in elections under the Trump presidency, including special elections at the state and federal level. ALG has also worked in three other special elections this cycle where our clients have done far better than historic Democratic performance in their districts/states:
  • Doug Jones (Alabama Senate), on behalf of the DSCC
  • Annette Taddeo (FL Senate-40, a South Florida legislative flip from R to D)
  • Archie Parnell (South Carolina Congress-5)
  • Jon Ossoff (Georgia Congress-6)
There are lessons to learn from all of these, but Wisconsin SD-10 holds particular evidence that Democrats can win in the white working-class parts of America where we were overwhelmed in 2016. The following were lessons we learned from the Wisconsin election we feel Democrats can benefit from, all in further detail below:
  1. Grounding Patty Schachtner with voters did not necessarily mean running outside the party on guns, choice, or immigration
  2. We talked predominantly about skills training, health care, and taxes-two national issues and one local
  3. Here, we didn't mention Trump once
  4. We used the same message to turnout targets and persuasion targets
  5. The "establishment vs Tea Party" dynamic on the other side really didn't affect the race we ran or its outcome
  6. We can win while being outspent
  7. Turnout and persuasion both mattered, probably in equal proportion
  8. We need to be more aggressive in thinking about districts to target
  9. Polling needed changing, but it isn't broken
     
Further details
  1. Grounding Patty Schachtner with voters did not necessarily mean running outside the party on guns, choice, or immigration. This a district with some Minneapolis/St. Paul suburbs and two UW colleges (though both colleges were out on winter break during the election). The majority of voters are white non-college in rural areas and small towns. Trump won it by 17 points, but Romney also won it by 6. It is not the type of district where we can win only by appealing to voters who hold every position of the Democratic platform.

    Patty Schachtner did not compromise her values or paint herself as a die-hard conservative, though. Nor did she shy away from support from progressive groups like Planned Parenthood. We focused on how Patty was "one of us":
    • An EMT who went to technical college (similar to a community or junior college in many states)
    • Someone who grew up on a dairy farm and worked hard starting young
    • A gun owner and bear hunter. This is a gun-owning district: in our polling, 59% of Democrats report having a gun at home. However, we did not accompany with promoting a down-the-line NRA agenda.
More than making sure voters thought Patty was with them on every issue, we had to communicate that Patty understood what voters' lives were like. We fought our battles on which candidate was more connected to people rather than on narrow hot-button issues. Our polling here and for other candidates suggests that, all things being equal, that is a more productive fight to have.
 
  1. We talked predominantly about education and skills training, health care, and taxes-two national issues and one local. We did not shy away from the national debate in this Republican district, engaging on:
    • Health care, especially the GOP plan to take away coverage from millions, deny care to people with pre-existing conditions, and pass an age tax on older Americans). In Wisconsin this included Medicaid/BadgerCare, which Scott Walker and our opponent Adam Jarchow refused to expand.
    • Taxes, highlighting Patty's plan to reverse GOP tax giveaways to the wealthy and give the middle-class tax relief. 
We also talked a lot about education, particularly skills training at high schools and technical colleges. Affordable universities occupy a lot of territory in the Democratic message, but often not enough time is paid to talking about education for people who aren't going to college. Patty's background as someone who has a technical degree and also teaches EMT classes at the technical school was a great entry point into this discussion, and we feel that every Democrat should talk about this more.
 
  1. Here, we didn't mention Trump once.Trump is not popular in the district, but in every special election we've seen that we should not be running campaigns as referendums on him. We can nationalize local and federal races on tough terrain, but we should do that via issues that matter to people rather than using Trump as a cudgel. This point won't apply to every election-for example we invoked Trump often in the Florida Senate District 40 special election, and we are also doing so in primaries around the country. The takeaway is to do so thoughtfully rather than blindly charging in.
 
  1. We used the same message to turnout targets and persuasion targets. We We had a dedicated track of process-oriented turnout communications (digital, phones and mail) to turnout targets. We made a deliberate decision not to bifurcate our message, though, avoiding saying one thing to irregular-voting Democrats and another to a group of independents that is fairly conservative in Western Wisconsin. The polling indicated that a message about health care, taxes, and job training could reach both sets of voters, and we believe it did. We made some slight tactical changes between our communications (e.g. we highlighted the election's date to less-likely voters), but we said fundamentally the same thing to both sets of voters.

    This can be a huge asset to Democrats in the approaching midterm.     Within the party there's been a debate Democrats should rally the base by staking out bolder positions or moderate themselves to appeal to independents. We chose to avoid the public option vs single-payer debate and others, instead focusing on providing access to health care and making taxes more fair which are positions Independents and Democrats share.

    This will not hold for every district, and some primaries will inevitably talk to base voters at the expense of the middle. But broadly speaking, it's great that Democrats' core argument right now appeals to both progressives and moderates.
  1. The "establishment vs Tea Party" dynamic on the other side really didn't affect the race we ran or its outcome. The Republican primary was a fight between a more establishment-type Republican Assembly member from a more swing district and the eventual nominee, a Tea Party Assembly member from a deep red district. In our initial poll, we probed on which one we'd rather face. All evidence suggested it didn't make that big of a difference: voters were ready to punish Republicans locally and nationally creating an opening regardless of which candidate we faced. This has real impacts on targeting for 2018: as a party we should not assume that a Republican is safe just because they have Republican voted relatively conventionally.
 
  1. We can win while being outspent. We had an overwhelming resource advantage in Alabama for Doug Jones, something we think we'll have in approximately 0 elections in 2018 as a party. We had the resources to communicate a message due to the investment from the Wisconsin State Senate Democratic Caucuses-they raised the bulk of the money spent and invested people on the ground. But at the same time, we were badly outspent. It didn't matter here: by our best guesses, Republicans and their allies including the Koch Brothers spent probably $1.5 to 2 million to our less than $500,000. We should not write off seats just because we'll get outspent: if we have the resources to communicate a message, in this cycle it can have the power to cut through even against overwhelming force.
 
  1. Turnout and persuasion both mattered, probably in equal proportion. We need to wait until the voterfile comes back with individual turnout models to say with certainty, but our polling and the results indicate that Schachtner won with a combination of turning out more Democrats and persuading independents and even some Republicans. In our benchmark poll, we polled the most likely voters as well as less-likely Democrats. That data showed we had to talk to both at real levels. Even though we were outspent 5-1 or more, we split our resources in half on turnout and persuasion. We were rewarded for that, as Patty won in Democratic strongholds with high turnout but also won or made significant gains in very Republican areas where we could never have made up the difference with just turnout.
 
  1. We need to be more aggressive in thinking about districts to target. The Wisconsin State Senate map was heavily gerrymandered by Republicans, but even still Senate Democrats have more obvious pickup targets in 2018 than SD10 which is why the WI State Senate is now in play after eight years of GOP control. Democrats also have many, many districts that are 0-9 points more Republican than SD10 (Schachtner won by 10 points). Previously we had been advising clients including WI Senate Democrats that a wave was brewing. The obvious target districts were Obama/Trump and Clinton/Romney seats, but we had been advising most clients to look deeper and focus on fielding credible candidates in seats that Clinton and/or Obama had lost by 10 points or less. Based on trends in the last few months, we think this should be expanded even further to districts where one or both lost by 15 or less. We won't win all of those, but now is the time to prepare for a real wave that could open up races for Democrats that would have been impossible in the last decade or more. It also means we should target voters for turnout and persuasion who might not typically be a Democratic midterm voter
 
  1. Polling needed changing, but it isn't broken. Like all/most pollsters, in November 2016 we underestimated Trump and GOP strength in white working-class areas. We polled SD-10 and much of Western/Northern Wisconsin in 2016, and our data was too optimistic. We and many other good pollsters owned those errors in a painful post-election process. The best evidence available is that we all made the following mistakes that were most problematic in heavily white non-college areas:
    • We talked to too many Democrats, partly because our final polling being was right after the Access Hollywood tape release and partly because we were too reliant on 2012 turnout models. We need to be careful about blind reliance on midterm models of which 2010 and 2014 will be the biggest inputs-does anyone really still believe 2018 turnout is going to look the same as those two?
    • We missed types of non-college voters, getting too many in office jobs and not enough in manufacturing, outdoor, and service-industry jobs.
    • We also talked to non-college voters from the wrong places, getting too many in high-education areas and not enough from rural places. In those rural areas, we instead talked to too many college-educated voters.
_____________________________________________________________

We made multiple changes in our approach based on these lessons and are continuing to adapt. At the same time, Schachtner had no chance to win this special election if the exact same partisan voters turned out as in lower-turnout 2010 or 2014. Our polling was optimistic about our chances, and it was backed up by good results across the country in the last year. But we made the conscious choice not to correct too hard and over learn lessons from 2016. We avoided things that would force our polling be too pessimistic, such as weighting the results to Republicans to look like those last midterms.

We believe people are right to question polling in the wake of 2016, and we questioned all our polling ourselves in a systematic review throughout 2017. Pollsters and campaign practitioners can't so dramatically change our approach, though, that we introduce new mistakes bigger than the past.

Looking forward to the midterms,

Brian Stryker
Partner at ALG Research
STAY CONNECTED: