Pollsters just can’t catch a break these days, with recent election prediction misses in Greece, Israel, Scotland and most famously, the UK. It was so bad in the UK that NPR hammered Nate Silver, with Weekend Edition host Scott Simon crowing “Oh so wrong.” Even an experienced hand like David Axelrod got in on the act,via Twitter: “In all my years as a journalist & strategist, I’ve never seen as stark a failure of polling as in U.K. Huge project ahead to unravel that.”
The hand-wringing fits into a larger debate over the value of polling in a Big Data world. Skeptics call public opinion research a relic, created for a world without the technology to observe and analyze the public’s behavior at scale. Why bother asking the public what they are planning to do when data tells us what they are already doing without the risk of response bias?
The skeptics’ view misses a larger point. The value of public opinion research is not just predicting election outcomes or purchasing decisions. The value is far bigger, and it’s tied in a fundamental way to asking questions. By asking questions, polls help us understand how the public thinks. How the public views the world. How the public reacts to certain concepts and messages. The value of public opinion research is the impact it has on strategy — for a company, cause or candidate.
Research shows politicians are not particularly good at gauging public opinion. In our unscientific view, journalists and clients aren’t very good at predicting public attitudes either. They are unduly influenced by anecdote and fringe conversations in social media. A recent example brings this to life.
In 2013 we worked with an education nonprofit having a hard time raising funds and recruiting board members. It was a confounding situation because the group had a track record of success surpassing many competitors. The problem came down to how our client articulated its purpose. The organization described its function as empowering parents, whom they rightly considered a child’s first educators. But opinion research suggested a strategic shift.
The empowerment rhetoric resonated only with libertarians, the group’s most consistent financial supporters. The client was communicating with its base, but its messaging was turning off potential supporters — even those who supported the mission — which explained the fundraising and board recruitment challenges. Our polling showed that if the group wanted wider support, it would need to emphasize themes that resonated more broadly, including fairness and giving kids the opportunities they deserve. In business terms, public opinion research changed the way the organization went to market.
This takes us back to Big Data. Some would argue we should stop asking questions and start analyzing what the public is already saying and doing. We agree. Observing conversations and examining behaviors offers great insight into public opinion. We would go a step further and argue Big Data and behavioral analytics are the most important developments in understanding public perspectives since George Gallup came on the scene. But we would also argue that in the future, the most impactful campaigns — for companies, causes and candidates — will make effective use of both polling and Big Data.
It’s already happening. For instance, when data analytics indicate an issue is gaining momentum in social media and poses a reputation threat to a client, we use public opinion research to discover who is paying attention and explore why it matters to them. We also blend polling and data in an approach that resembles political campaign-style microtargeting. Combining information on public attitudes with a consumer database of demographics and purchasing behaviors, we predict brand favorability, likelihood to recommend, influence and persuadability down to an individual level. This allows us to target the right stakeholders with the right messages to persuade skeptics and activate supporters.
The death of polling has been greatly exaggerated for generations (see Gallup and the 1948 presidential election). No doubt finding a reliable, cost-effective replacement for phone interviews will take time. But technology is already providing high-quality (internet-based) replacements.
In the final analysis, Big Data is not a threat to public opinion research — it’s part of the future. The landscape is shifting quickly, but we have at our disposal the tools to understand the public with more nuance and precision than ever before.
Andrew Sullivan, H+K Strategies and Amber Ott, Research+Data Insights