NEW YORK: Elections provide the forum where research and polling get the most exposure, and while flawed 2016 forecasts in the US and UK delivered powerful blows to the credibility of the practice, the main issue seems to be one of "pilot error".

George Terhanian, Chief Research and Analytics Officer at the NPD Group, Inc., based in Port Washington, NY., discussed this topic in "What Survey Researchers Can Learn from the 2016 U.S. Pre-election Polls: How to Fine-Tune Methods and Restore Credibility", a new paper published in the Journal of Advertising Research (JAR).

More specifically, he explores how dire the damage is to polling from its high-profile failure in the electoral race between Donald Trump and Hillary Clinton – and what the consequences might be for marketing researchers.

"As an analogy," Terhanian offers, "consider that when a plane crashes, the National Transportation Safety Board, an independent federal agency responsible for investigating civil aviation accidents, races into action to determine the crash's cause and recommend how to prevent future ones. Such objective, rigorous scrutiny helps restore faith in flying."

From a polling perspective, he insists, "Fortunately, survey research does not seem to be fundamentally awed or broken beyond repair. Pilot error, which can be corrected, is the likely culprit behind the missed calls in the 2016 election.

"It won't be easy, but if survey researchers correct their mistakes, they will do themselves, their industry, advertisers, the public, and democracy a major service, because high-quality survey research continues to be a good way – possibly the best way," he argued, to fulfil legendary pollster George Gallup's aim of "directly approaching the mass of the people and hearing what they have to say."

Continued Terhanian: "That is also why pre-election polling can be hazardous, however – there is nowhere for survey researchers to hide when they make the wrong call, as they did in 2016."

In fact, he also believes that lessons from the last round of elections will help improve pre-election poll accuracy, enhance survey research quality, and increase public confidence in polls.

"What Survey Researchers Can Learn from the 2016 U.S. Pre-election Polls How to Fine-Tune Methods and Restore Credibility" leads off a special "What Do We Know about Word of Mouth?"

Its recommendations include reporting response rates and non-response error, focusing on battleground areas, heightening transparency and increasing representativeness.

Learning from how, when and why advertisers use survey research could be another useful tactic. "These companies do not rely only on survey information. Instead, they regard it as one input in a prediction model," Terhanian said.

Data sourced from Journal of Advertising Research; additional content by WARC staff