Pollster Ben Lazarus on State Legislative Races, COVID-19, and the 2020 Election.
Pollster Ben Lazarus is currently the Director of Research Solutions at TargetSmart. He is a specialist in state legislative races and works as a principal pollster and analytics consultant for the New Jersey Democratic Assembly Campaign Committee, the Maryland Senate Democratic Caucus, and the Indiana House Democratic Caucus. We asked him 7 questions on polling in state legislative races.
1. What has changed in the world of political polling since you started working in the field?
The essence of what we did over a decade ago when I started doing this work in generally unchanged. We’re attempting to, as accurately as we can, tap into the truth of public opinion in targeted pieces of political geography and use that truth to run strategic and authentic campaigns that win. What has changed is how we go about measuring that truth. When I started polling in 2008, we reached roughly 1-in-4 telephone respondents on cell phones, the other 3-in-4 on landlines, and almost no one online. Today, our cell phone blends represent upwards of 60-70% of the telephone interviews we complete, and we are mixing multi-faceted online recruitment techniques into polls for races at every level of the ballot. In this same period of time we’ve seen response rates to our polls steadily decline. These factors require us to always be tapping into a constantly evolving set of tools for reaching different types of people in different ways to measure how they feel about politics and elections. Finally, the use of predictive analytics in Democratic campaigns has taken off since I first started working in this field, and the way those tools integrate and compliment public opinion research has changed dramatically as well.
2. How should state legislative campaigns determine if they should conduct political polling?
Typically, I recommend any campaign, be it state legislative races or any other level of the ballot, poll when they have strategic use of it, and when the significant investment in polling makes sense relative to the campaign’s overall budget and expected spend for things that can be guided by polling, like paid communication and field. For example, a state legislative benchmark poll typically costs between $15,000 and $20,000. If a state legislative campaign’s entire budget is $60,000, it would be hard for me to recommend spending upwards of 25% of that budget on a poll, even if that last $45,000 is spent more wisely due to the information gleaned in the poll. I think research accounting for somewhere in the realm of 10% or less of the campaign’s overall spend (after accounting for things like overhead and payroll) is a good rule of thumb to use to determine if the budget for a state legislative campaign can and should support polling. Of course, there are circumstances in which a state legislative campaign can defray the cost of polling through partnerships with state party caucus operations or think about trying to leverage polling to help jumpstart fundraising. Every campaign is different, and usually a brief conversation with a pollster about what your goals are for polling will help to determine if it’s the right fit for your race.
3. How can campaigns effectively use polling to inform overall campaign decisions?
Polling generally informs decisions on two axes – message content and message timing/sequencing. This information can and does materialize in lots of different ways. What should the campaign’s broad narrative be? How can we most effectively define a contrast between the candidates in a way that matters to the voters who will decide the election? What should or shouldn’t the tone of the campaign be? Which issues or debates are most salient for voters? How do candidate and community specifics interact with the broader trends in the state or national political environment? Polling can help to shed light on all of these questions, and align the strategic answers to them with resource and communication constraints every campaign faces including the overall budget for paid advertising, the mediums through which the campaign will advertise, and of course, the limited time horizon for communication with which every campaign must grapple.
4. How is polling in state legislative races different from polling in presidential and state-wide elections?
It’s really all about the scale and budget of the campaign, rather than the technical level of the ballot in which a campaign is taking place. Statewide and (some) Presidential campaigns tend to spend and communicate significantly more than state legislative campaigns, so the scale of the associated research program tends to increase as well. That materializes in the form of larger sample polls and more frequent polling, particularly in the context of broadcast television advertising. Broadcast is very expensive, but also tends to be the faster “mover of numbers”, so when a statewide or Presidential campaign goes “up” with its television advertising, it will frequently use polling as an accountability mechanism for making sure the big spend on TV is working, and help make strategic decisions about moving resources to new tracks of creative, or refining the buy to ensure it’s reaching the right people. All that said, some state legislative campaigns are incredibly well funded and take place in affordable broadcast markets where a similar approach to polling makes sense.
5. Could COVID-19 create false trends that make polling numbers less accurate?
I think it’s unlikely (spoken like a true pollster!). The fact is polling measures public opinion at a distinct point in time. COVID-19 and the general social upheaval we are seeing around the murder of George Floyd and the protest movement for Black lives has created a tremendous amount of volatility in public opinion. So, what was accurate two weeks ago may not be accurate today. I think understanding this is an inherent limitation of polling speaks to polling more frequently this year to make sure we’re not missing trends as they are moving beneath our feet!
6. What are some challenges that occur regularly in polling state legislative races that might be exacerbated because of the ongoing pandemic?
One of the main challenges associated with all polling is declining response rates over time, and as state legislative districts tend to be relatively small, the combination of limited numbers to call and limited response rates yields smaller sample surveys, which are subject to more variance and statistical error. State legislative districts are also more challenging for administering multi-modal online polling. This is because voter file matched survey panels tend to have limited reach into individual state legislative districts, and the other ways to prompt online survey taking (through SMS and/or email invitation) tend to yield even lower response rates than telephone calls. This problem is not unique to the age of COVID. In fact, we’re actually observing observably better response rates in the age of COVID than we saw pre-pandemic, though we think it’s still too early to assess if this is a stable trend.
7. How powerful do you think polls are in the outcome of state legislative races, especially in the COVID-19 era when candidates do not have as much in-person interaction with voters?
This is a great question, as the core tool any legislative candidate tends to have in their campaign toolbox is the sweat equity that comes with in-person, grassroots door-knocking. And since the pandemic is limiting the safety of this core tactic, campaigns lose out not only on a great way to persuade and mobilize voters, but also a key information gathering mechanism out in the community. I’d be wary of attempting to replace such a vital qualitative pulse on the community with polling, as polling tends to be a tool less suited to picking up on key community-specific nuances as told through the language voters are speaking on a day-to-day basis. I’m encouraging my legislative clients to find innovative ways to move the conversation from the door to anywhere else that is safe, like the telephone, or through texting, or through online engagement using the candidate’s personal and/or campaign social media accounts. This way, engagement is happening, AND the campaign is able to pick up vital intelligence in the field, without making big expenditures on extra polling or qualitative research.
Bonus: What are state legislative races that you'll be keeping an eye on during the 2020 election?
Fun question! I’m really interested to see how gains Democrats made in swing suburban districts in 2018 hold up in 2020. Like will these races we won by a few dozen votes remain competitive in 2020, or will the trajectory we started to observe in 2018 continue? A few examples of these types of contests in states where I work include re-elects for Representatives Jen O’Mara (HD 165, Delaware County, PA), Mike Zabel (HD 163, Delaware County, PA), Wendy Ullman (HD 143, Bucks County, PA), Karin Derry (HD 39, Polk County, IA), and Heather Matson (HD 38, Polk County, IA). I also hope folks watch Missouri’s 15th state Senate district in the St. Louis suburbs where Rep. Deb Lavender is taking on the author of Missouri’s abortion ban law Andrew Koenig in one of the state’s top targeted races.
Ben Lazarus Thanks for answering our questions. Check out TargetSmart's Insights about COVID-19 and the 2020 election. If you have you any questions or comments, drop them below!