7 Questions with Democratic Pollster Celinda Lake

by Elena Veatch

Democratic pollster Celinda Lake

Pollster Celinda Lake answers our 7 questions.

Celinda Lake, a Democratic pollster and president of Lake Research Partners, is one of the top political pollsters and strategists of our time. Celinda Lake weighs in on what happened in 2016 and how Democrats can be thoughtful about polling and strategy in 2018 and beyond.

1. How do you explain the challenges around polling the 2016 presidential election?

There were certainly swings at the end, and I think the turnout models were off. The biggest critique I have around 2016 polling (this serves as a self-critique as well) is that had you asked many pollsters on Election Day, they would’ve said Hillary Clinton was going to win albeit by a narrow margin. In many ways, we suffered more from groupthink than from errors in the polling itself. A lot of pollsters on both sides stopped polling the election too early. We thought we knew the outcome. The lesson is that if you really want to know what the numbers will be, you have to poll through Election Day.

2. There was a lot of discussion around the concept of the “hidden Trump voter” this past cycle. How can any Democratic pollster work to account for similar phenomena in the future?

Pollsters are divided on this, but I’m a believer in the hidden Trump voter phenomenon. Two things jumped out during the cycle that can help us moving forward:

  1. Trump was performing better in online polls than in phone polls. A lot of people didn’t want to tell an interviewer that they were voting for Trump.
  2. Firms take different approaches to the order in which questions appear in a survey. There's a balance to strike between getting people to say who they’ll support on the ballot upfront and not intimidate people by asking that question too soon in a survey. As a firm, we’ve found that the drop-off effect is worse than the intimidation effect, so we’ve moved the ballot question up earlier in the questionnaire. This way, we get that vote preference even if people decline to answer the rest of our survey questions. After every poll, we analyze the responses from those who terminated the survey early and report back to our clients on the partisan scores of these respondents (they’re usually more conservative).

It’s also important to keep in mind the long-term trend of the Wilder/Bradley Effect, or the discrepancies between polling and election outcomes that result from the social desirability bias among survey respondents. People often don’t want to demonstrate racial or other biases when asked about their candidate preferences in a survey. Research has shown that white Democrats who thought they were talking to black interviewers in 2008 were more for Obama than were those who thought they were talking to white interviewers. Similar research has shown that women tend to be less pro-choice in surveys when they’re interviewed by other women – they feel as if they can express ambivalence about abortion to other women, but they don’t to do the same with a man who might think he then deserves a say over regulating women’s bodies. A combination of online and phone surveys can control for this phenomenon, though it can be expensive.

3. What can Democrats say to Trump supporters who had voted Democratic in the past to try to win them back in future elections?

The Democratic brand is weak enough among Trump defectors that it will be hard to win them back. There are a couple of things we can do:

  1. Get labor union members back to the Democratic side. When a union tells its members that Trump has not followed through on his promises regarding prevailing wage and overtime, they’ll listen – members like and trust their unions.
  2. The Republican tax bill is a real opportunity to turn one of Trump’s anchors of support against him. Instead of draining the swamp, he’s letting the swamp get kickbacks.
  3. Trump’s style is starting to wear thin among women and millennials in particular. His constant Tweeting, a penchant for chaos, and lack of empathy turn people off, whether he’s responding to the disaster in Puerto Rico or escalating his bellicose approach to North Korea. We don’t necessarily have a way to take this on yet, but it’s important to recognize responses to this behavior.

Our task in 2018 is not to beat Trump, though; it’s to defeat Republicans in Congress. In some cases, it will be better to go after those Republican members directly (taking advantage of their own weak brand) than to tie them to Trump. In looking to persuade Independent voters (rather than to mobilize Democrats), it can be more effective to accuse these Republican members of pushing an agenda on their own. Attacking Trump can muddy the issues, as there will always be Independents who still like Trump, but who might vote against their Republican member of Congress. Many wary voters are skeptical that anyone is like Trump, but they do respond negatively to anyone voting with him 95 percent of the time.

4. How much of a campaign’s energy and decision making should be focused on what polls are telling them?

Polling is a very important piece of the puzzle, but it has to be integrated with the full range of information sources and tools at our disposal. Big data and polling should be integrated, and 2016 didn’t show the right mix in my opinion. We need more comprehensive feedback from the field. Field organizers should be brought into the rooms where campaigns make strategic decisions more frequently, as they’re the ones who are constantly in contact with voters. If field organizers say something is happening in the communities they’re working in, it’s worth listening.

Another important piece is analyzing social media to catch the nuances of voters’ positions on issues. Qualitative research is important and was underutilized at the end of the campaign for Clinton. We used to do focus groups five nights a week until the end, where we could better pick up on the “yes, but…” answers. Most campaigns don’t do this anymore toward the end, which can be a big mistake if they aren’t paying attention to information sources other than tracking polls.

5. What do you see as the biggest barrier to entry for women looking to run for office?

One important barrier is money. Women are raising the same amount of money as men, but they don’t have the same number of big donors as men. They tend to raise money in smaller increments, raise it later in the campaign, and suffer from greater psychological barriers around the ask than do men. When we surveyed women in state legislatures with aspirations for higher office, the number one thing they said was that they needed more donors from their personal networks. Members of female candidates’ kitchen cabinets tend to be composed of women with great community relations, but less money than is the case for most male candidates’ inner circle members.

Another barrier for women is the balance they have to strike between likeability and qualification. Voters will support a man they don’t like if they think he’s qualified, but they won’t support a qualified woman who they don’t like. It’s a critical double standard we saw play out dramatically in 2016. Voters who disliked both Trump and Clinton voted in the double digits for Trump, even though they saw Clinton as more qualified.  

6. What can Democrats do to win more in 2018 and beyond?

Democrats need to recruit good candidates and raise money, which we’re doing. Things that we aren’t doing enough of:

  1. Forming a cohesive economic message. We still don’t have one – the Republicans are ahead of us on the economy. We’ve never won a presidential election when we haven’t been ahead on the economy, and we’re not going to start now. People support issues that Democrats tout, like equal pay and raising the minimum wage, but these policies don’t add up to a major economic plan to get the economy going for everyone.
  2. Developing and funding better field organizations. We need to get into communities now to talk to voters about their concerns and priorities. We can’t show up in communities two weeks out from an election and expect to convince voters to turn out and be on our side. Working America has a great model for this that’s worth following – they have long, textured conversations with people early.
  3. Stop debating persuasion vs. turnout. This is a dangerous debate Democrats have that Republicans don’t have. We have to persuade voters in order to get them to turn out; there can’t be one without the other.
  4. Don’t confuse message and issue. People in our party tend to confuse message and issue. Issues, policies, acronyms, and program descriptions do not constitute a message. A message starts with values and emotion. People respond to aspirations, but right now, Democrats are much better at expressing the existence of problems than being aspirational. Trump was rated more optimistic than Clinton during the 2016 election, even if it appeared otherwise to Democrats. We have to be better about finding positive messages. Obama did this with his message of hope and change in 2008, and Bill Clinton also did in 1992 (“There is nothing wrong with America that cannot be cured by what is right with America”). People want to move forward in this country, and we have to give them the vision to move forward that they can get behind.

7. Have you read any interesting books lately that you think shed light on the American electorate?

  • Dog Whistle Politics: How Coded Racial Appeals Have Reinvented Racism and Wrecked the Middle Class (Ian Haney López)
  • Strangers in Their Own Land: Anger and Mourning on the American Right (Arlie Russell Hochschild)
  • “Durably Reducing Transphobia: A Field Experiment on Door-to-Door Canvassing” (David Broockman and Joshua Kalla)

Thank you to Democratic pollster Celinda Lake for answering our questions. Get in touch with Celinda Lake: visit www.lakeresearch.com, or reach out to Celinda Lake directly at clake@lakeresearch.com.