For the second straight election, political opinion polls got a lot wrong. With a few notable exceptions, the entire industry notably underestimated support for President Donald Trump.
That failure has raised a troubling question for sports marketers and strategists who rely on opinion research as part of the data analytics movement: What if — for whatever reason — polling simply can’t be relied upon anymore to generate a fair read of public sentiment?
It’s a timely concern as the sports world tries to emerge from the pandemic. A majority of fans have told pollsters they want to attend live sporting events again, but ticket sales in these unprecedented times tell a less optimistic story. For example, an Oct. 20 Morning Consult poll said that 67% of sports fans said they’d be comfortable attending an outdoor sporting event at 25% capacity, but less than one-third of 48 NFL games with fans through Week 8 have actually sold out their full allotment of tickets.
So far, analytics experts don’t think it’s a problem. They believe their surveys are less prone to the problems political pollsters face in reaching the right respondents to begin with, and more robust because they can be compared with other, more readily available data than politics permits.
For instance, teams can simply ask every season-ticket holder for feedback, said Russ Scibetti, the New York Giants’ vice president of strategy and business intelligence. That’s much more straightforward than trying to find 800 people randomly who represent the voting patterns of a state.
A team still has to know its audience and adjust, he said. “If younger season-ticket holders don’t respond to outreach as much as older ones or vice versa, then your sample may not actually be representative,” Scibetti said. “They need to compare the participation rate to the audience.”
Polls became commonplace in consumer research not long after George Gallup pioneered the practice in the 1930s. Despite their notable imprecision, poll figures still routinely appear in pitch decks and marketing strategies.
But a key difference in business, said Jessica Gelman, CEO of the sports analytics firm KAGR, is that you have so much more access to other data. Political pollsters have no way of tying an actual cast vote to reported intent, while sports teams can compare actual transaction data to responses.
“What we have found is that often, what people say is not what they do, and we look at both quantitative measures — their behaviors and actions — as well as the qualitative survey information, to understand the truth of their intent,” Gelman said.
Other sports experts believe questions about sports and consumer choices are just less polarizing, and therefore less likely to generate false answers or biased response patterns. But polls should always be seen as an imprecise part of your decision-making information set, and political pollsters and reporters tend to cause their own problems by lending an inappropriate degree of certainty about their findings.
Said Scibetti said: “You should never rely only on one particular form of analysis.”
