Nate Silver recently published an insightful article on his blog about the state of the polling industry in the U.S. As he rightly points out, longstanding methodologies are at odds with the way consumers behave today. A shrinking number of people have landline telephones and fewer regularly answer them. Silver cites the prohibitive costs and limitations of polling via cell phone as a further challenge to survey response rates. He could have also mentioned that most people won’t answer their mobile phone when called from a number they don’t recognize (*).
Silver then examines why, despite the increasing instability of traditional polling methods, election forecasts continue to perform reasonably well. After extensive analysis, he concludes that accurate predictions are owed to a steady and predictable composition of voters in major elections, which allows pollsters to overcome poor response rates. As with anything authored on Nate Silver’s keyboard, you can disagree at your own peril.
Almost an after-thought in Silver’s piece, however, were the two most important underlying points he made:
First, he notes that the makeup of the U.S. population and the political coalitions it fosters are likely to change. These changes could happen sooner than later, rendering crude demographic models and traditional polling techniques insufficient at best. Silver may have nailed the current state of affairs in polling; he says little about where it is heading or what should be done about it.
Secondly, he touches briefly on why everyone – not just pollsters and candidates – should care about the future of polling. If we cannot accurately measure and report on public opinion, we’re left to guide our policies based on the opinions of “elites” in government, media, and business. Imagine if all we had to go by was the stream of narrow and biased commentary emanating from Twitter every day. Yikes.
If polls are broken, in Silver’s words, “we should be rooting for them to succeed.” Reliable and trusted polling is essential to democracy and free markets.
And it’s this last point that deserves the most attention. There can be no doubt that polling, as it had been understood for generations, is eroding in its effectiveness. It’s up to all of us to find new and better ways to measure public sentiment and share it with our broader community of neighbors and leaders.
The Case for Online Polling – Embracing Trends in Communications Technology
CivicScience is one of many companies that believes the future of public opinion research is a digital one. Silver himself notes that “the average online poll was more accurate than the average telephone poll in the 2012 presidential election.” Declining landline usage is not going to simply reverse. People are not going to begin taking more calls on their mobile phones from unknown callers. Rather, trends in communications and device usage, particularly among young people, suggest that phone calls of any kind will only decrease. The most gifted Devil’s Advocate would struggle to argue otherwise.
Many thought-leaders though, particularly in news journalism, continue to look down their noses at online polling. When the New York Times recently began conducting polls with a web-based firm, the subsequent media coverage was less about the numbers and more about the Times’ radical new partnership. If we named all the reporters and editors who have scoffed at CivicScience’s methodology over the years, refusing to publish our findings, the list could almost qualify as “Big Data.” At the same time, if a university-based phone pollster who was credible 30 years ago releases numbers, most news outlets will publish them without scrutiny. This is puzzling to say the least.
The CivicScience method is certainly not the only web-based polling technique being developed today. Google saw the demise of traditional polling as a big enough problem to invest heavily in its own solution. Companies like YouGov (the NY Times’ partner) are run by very smart, respected, well-meaning scientists. While we don’t love the idea of research based on people who are either forced to share an opinion (in Google’s case) or paid for it (in YouGov’s case), at least they are trying something new.
The future of polling should be based on the voluntary participation of everyone – not just those with the time or inclination to answer lengthy surveys for the chance to win an iPad. We need to reduce the friction in the opinion-sharing process, making it easy for people to weigh in as much or as little as they want, wherever and whenever they want. With enough good data, modern technology can find accurate meaning even in non-probability-based samples.
At the same time, we acknowledge that the CivicScience solution is by no means perfect…yet. We cannot reach people in the U.S. who don’t have Internet access. While our polls are delivered on hundreds of diverse web and mobile sites, there are thousands more we don’t touch. We’ve polled millions of U.S. consumers – yet only a fraction of the full population. We have more work to do.
But make no mistake. A future, in which our polling methodology (or another like it) becomes the standard, is far more likely than a future where telephone polling makes a miraculous comeback. If we can agree that a healthy and reliable polling industry is important to the public good, then we need to agree that our best hope lies in a web-based solution.
Hats off, yet again, to Mr. Silver for shining a bright spotlight on this important cause.
(*82% of US adults say they rarely or never answer calls from numbers they don’t recognize, according to our last research on the subject – N=4,966 US Adults from 3/10-13 2014).