What Comes Next: Multi-Channel Polling That Meets Voters Where They Are
By: Global Strategy Group Research Team
“What’s the best methodology for polling these days?”
Our polling team at GSG gets asked this question all the time. “Are phone polls really accurate anymore?” This topic has been of particular focus since Trump’s election but it did not start with that event. Questions about the viability of the “gold standard” methodology began with the decline of landlines and the rise of mobile phones — and have only been amplified as the internet has become Americans’ preferred medium for communication.
Telephone polling is still a strong and accurate approach in many situations. We recommend it to clients every day. However, response rates are declining and cost is skyrocketing.
The industry has already begun transitioning to more online methodologies, and we only see that transition accelerating. But here’s the challenge: most online surveys are conducted using non-probability samples. Instead of every American having a chance of being polled, only respondents pre-recruited to opt-in panels can take the surveys. The problem here is that extensive research has shown that many of these panels — particularly when respondents are not matched to real voters on a voter file — can have significant biases that are difficult to adjust for since the survey results cannot be calibrated to the voter file.
What’s our answer? Multi-Channel Surveys using Validated Voter Technology.
We believe there is no one-size-fits-all solution for polling. Gone are the days when telephone polls could solve every problem, though they aren’t dead yet either. Meanwhile, there is still no single, magic online solution that is a perfect replacement. That’s why at GSG we are embracing what we call a Multi-Channel Approach to Polling. This strategy is headlined by our new approach to online surveys developed and tested via extensive experimentation over the past year.
Here’s how we developed our approach:
Over the final month of the 2018 midterm campaign, we performed a large-scale in-house experiment. In seven different statewide and congressional races across the country, we conducted surveys using several different experimental online methodologies in parallel to traditional phone tracking being conducted in those same races. We continued this experimentation in 2019 — from the Chicago mayoral election to the Louisiana Governor’s race among others. This led us to develop a new approach that reaches voters where they are: on their smart phones via both traditional calls and text messaging, online at home, as well as on landlines for older Americans who still have them. This approach results in accurate measurements of voter preferences and it provides all the benefits of data that is matched to the voter file — all at a lower cost than telephone-only surveys.
How does it work?
The backbone of this approach is a combination of two or more of the following:
· Polling via text using a random, scientific sample of cell phone numbers drawn from the voter file
· Polling via the best online panel vendors, matched back to the voter file
· Traditional telephone polling, with voters reached via both landline and cell dialing
We may also supplement these methods with other new techniques as they become available to us, as long as they meet our methodological standards, pass our tests for accuracy, and allow for matching to the voter file.
In all channels, only validated voters are surveyed. This provides all the benefits of data that is matched to the voter file, including the ability to use demographic information and campaign modeling scores to ensure we have a representative sample — typical of all the best professional campaign polling, but very few online polls. We also use benchmarks attained from probability samples (phone and/or text) to calibrate the non-probability sample (panel), reducing potential bias even further.
The combination of the various channels that we recommend for each project is optimized to match the characteristics of the population we are looking to survey and the budget of the client. In some cases, we recommend an online — only approach that combines text and matched panel samples. In some cases, we recommend phone and text. And in some cases, the best approach is to use all three.
GSG was among the first major polling firms to incorporate probability text samples as part of an accurate, multi-channel approach — an approach that we have now used successfully in dozens of surveys this year.
There is no doubt that our industry faces challenges. But we see this as the beginning, not the end, of the discussion about moving beyond phone-centric political polling. We’re proud to be on the cutting edge. In fact, when Politico sent a reporter to this year’s American Association of Public Opinion Research conference to learn what industry leaders were saying about the future of polling, they found our presentations some of the “most interesting” — including one on the experimentation that we outlined above.
We are committed to providing the most accurate data and strongest possible analyses to our clients. We will continue to lead the industry with that standard as we continue to experiment with new methodologies.
Contact: Nick Gourevitch, Partner and Head of Research, Global Strategy Group