
Election pollsters got it wrong, again. As of last Wednesday (Oct. 28), an ABC News/Washington Post poll showed Joe Biden enjoying a massive 17-point lead on Donald Trump in Wisconsin and a comfortable nine-point cushion in Michigan. FiveThirtyEight had Biden ahead by 8.6 percentage points in Wisconsin and 8.9 points in Michigan two days later. Yet the Democratic presidential nominee ended up winning those two states by less than two points combined. “That’s not a miss, that’s a huge miss and it makes polling very difficult to rely on,” said Adam Grossman (CEO, Block Six Analytics).
Like politics, the sports industry is reliant on polling, survey and focus group data. Whether the focus is sponsorship, events or ticket sales, there is a need to understand the audience. But if polling biases are for an event as significant as a presidential election are so pronounced and widespread, it’s fair to question the accuracy of the data that sports properties are getting. Grossman says that in order to gain more accurate insight into how fans will ultimately act, pro teams and leagues should be leaning on social media listening tools which track organic posts and online activity of fans on social media platforms.
Our Take: The three core problems found within political polling are also present within sports surveys and focus groups. The first issue pollsters run into is selection bias. It’s not easy to find an unbiased, reflective sample of the population willing to participate in a survey. In Q4 2019, the Pew Research Center reported that telephone survey response rates had dropped from 36% in 1997 to just 6%. While the results are weighted to more accurately reflect the overall population, “the fewer the number of people [surveyed], the greater the importance of weighting becomes and the more likely there is to be error,” Grossman said.
Unlike many of the political polls, the majority of sports surveys are conducted online or at a stadium or arena instead of over the phone. But that doesn’t mean the results will not have similar bias issues. “You are still asking people to respond to questions and response rates are going to be low,” Grossman said. So, it’s not as if the pool of respondents is going to be drastically bigger. While it would be reasonable to assume those polled at the venue may be more representative of “sports fans,” remember that Adam Silver has said that 99% of NBA fans will never go to a game. By definition, those spending money to attend are more avid than the average fan.
Even if sports pollsters managed to find participants at the stadium who are representative of the greater population, there would still likely be significant response bias in their survey results. “Often times, people will respond in an interview the way they think they should respond—not in the way that they [intend to] act,” Grossman explained. That’s because people try and avoid negative interactions. They’re more likely to give the response they believe the interviewer is looking for than their actual feelings—particularly on polarizing issues (like the presidency).
Social media listening is one way for sports rights owners to derive insights on their fans without incurring the biases native to polling. “Organic posts are typically a much better reflection of what [the fan] actually thinks or feels. The closer [a pollster] can get to what people organically think, the closer they will get to [fans’] actual actions. And because you can use machine learning, artificial intelligence and natural-language processing to crunch data quickly, [the pollster] can look at a much larger sample size over an extended period of time,” Grossman explained. That’s not to say there aren’t biases in social media listening, but the technology can significantly help address issues with polling.
It’s fair to wonder why the country continues to rely on polling, survey and focus group data when social media listening is an option. Grossman suggested it’s simply a combination of inertia and social listening technology being relatively new (think: machine learning became commercialized at a larger scale in 2015). “This is how polling and surveys have always been done.” And while perhaps not necessarily a great reflection of what people will do, surveys and polls can provide value in their ability to accurately measure peoples’ biases. “The third bias that comes up in polling is confirmation bias,” Grossman said. “People confirm what they already believe or think they believe. If [a team] knows what [the fans] believe, that’s helpful from a marketing perspective—even if it’s not 100% reflective of what they’re actually going to do.”
Since the sports world came to a grinding halt, there has been a lot of time, effort and money spent on trying to understand if and when fans will feel comfortable returning to the venue. For example, a recent Morning Consult poll indicated that just 55% of “regular sports attendees” are comfortable attending an outdoor sporting event at 50% capacity. Grossman believes that in reality, the figure is likely much higher than that: “There’s a clear right answer to the question “would you return to the sports venue?” in terms of the health and safety response versus what people’s actual actions likely would be.” The response bias is likely significant.
The B6A founder added that the presidential election results seemingly support that notion. “Trump and Republican House and Senate candidates all outperformed [expectations],” he said. “That’s an indication that people want the economy open, that they want to be out doing things [including a return to the stadium].” Of course, getting the virus under control is the surest way to ensure that happens Grossman said.