Even though she’s already decided to vote for Donald Trump, Austin resident Diana Welsch was eager to watch last week’s televised Commander-in-Chief forum, which featured the two major party candidates fielding questions about military issues.
But instead of kicking back on her couch at home to watch the forum, Welsch, 59, drove from her home in Northwest Austin to a small grey office building just off Interstate 35.
She put on a pair of headphones, attached a probe to her finger, and prepared to watch the forum while sitting in a narrow cubicle next to dozens of other people. Welsch and the other TV watchers were being closely monitored by researchers in a hidden control room.
Welsch was part of a study conducted at Austin’s MediaScience labs. The eight-year-old company studies consumer response to viewing media, such as a video advertisements, a new TV show, a virtual reality experience or even responses to social media. Facebook is one of the company's clients.
The idea is that people aren't always being honest when you ask them whether a certain advertisement or TV show resonated with them. MediaScience is trying to close the gap between what people say and what people do.
That's why MediaScience studies people's physiological responses, examining things like heart rate and skin conductance, facial responses and eye tracking. The company also does more traditional surveys.
For the Commander-in-Chief study, MediaScience wanted to discover how people really felt about presidential candidates Trump and Hillary Clinton. The firm recruited 28 panelists in Austin and 36 in Chicago, keeping an even balance of supporters for each candidate and undecideds.
What they discovered was a Clinton problem.
“Normally supporters for a candidate are really positive when their candidate is talking," said MediaScience CEO Duane Varan. “What’s fascinating about this data is that Clinton supporters are actually negative when Trump is talking but also negative when Clinton is talking."
That was distinctly different from Trump supporters, who were positive about their candidate and negative towards Clinton.
Varan said MediaScience knows this because of its analysis of facial muscle movements, such as frowns, eyebrow-tightening and other cues that show negative reactions. “The biometrics exposed a deeper sentiment, which I think is problematic for the Clinton campaign right now,” Varan said.
A neuroscience marketing boom
MediaScience is part of the rapidly developing field of neuroscience marketing. This type of marketing research started to gain traction in the early 2000s from companies looking for a more accurate way to tap into consumers' desires by studying the brain or other physical reactions.
The relatively new Neuromarketing Science and Business Association lists nine American companies in this field. Stephen Genco, the co-author of "Neuromarketing for Dummies," said there are between 150 and 250 neuromarketing companies worldwide.
Some companies have focused on using sophisticated and expensive equipment, such as fMRI machines, to study the brain. The idea is that brain activity can offer deeper insights into consumer behavior, showing, for instance, when the brain's pleasure centers are activated.
Other companies, like MediaScience, don't actually examine the brain and look instead at physical responses.
Genco and another industry expert said companies like MediaScience are considered part of the neuroscience marketing field because there is such a strong connection between what the brain does and a person's physical responses.
"The logic behind that is that the control over the facial muscles is involuntary... and cannot be controlled actively by you. And same for the eye movements," said Moran Cerf, an assistant professor of marketing at Northwestern University.
He said even "fake smiles" can be detected with facial coding technology versus a "real smile." And some companies prefer not to study the brain because it's more expensive and perhaps doesn't offer significantly better insight, experts said.
Varan said MediaScience doesn't use techniques such as brain imaging or electroencephalography, or EEG, which can measure electrical activity just below the scalp, commercially because these techniques are still in the "research and development" stage, he said.
"The reality of marketing as a whole is that there isn't a lot of validation in marketing altogether," Cerf said, describing traditional methods of assessing the success of an ad campaign that just revolved around tracking sales. "If I were to put my bet on which field will be the most dominant in the future and will be most validated and better-performing - I'd definitely say neuroscience," he said.
Varan founded MediaScience in 2008, after running a research center out of Murdoch University in Australia called Audience Labs.
He said he picked Austin for the MediaScience headquarters because of its proximity to the University of Texas and its steady supply of doctorate-level researchers, and the diversity of this city, with its “sizeable” Hispanic population and mix of Democrats and Republicans.
“We are a liberal city but it’s not like we don’t have our conservatives,” Varan said. “We have both types in this city.”
He said another factor was the city’s embrace of new technology, important because MediaScience works with clients who want to test emerging technologies, such as virtual reality.
At first, MediaScience was doing work exclusively for Walt Disney Television, running studies for ESPN, ABC, ABC Family and the Disney Channel. After five years, MediaScience re-negotiated its contract with Disney so they could work with other clients. “We still work a lot with ESPN,” said Amy Rask, director of operations for MediaScience. “They have exclusive rights in sports.”
The company has 55,000 people signed up in the Austin area who regularly participate in their panels. They pay $30 an hour, which several panelists said was an important factor in why they chose to participate in the MediaScience studies.
“We’re constantly recruiting new people,” Rask said, and says they wait 60 days before inviting people back.
The company declined to provide details about its finances, but did say it recently opened a second office in downtown Chicago. MediaScience employs 50 people and besides TV clients such as Disney, CBS, Hulu and A&E, it also also works with companies such as Coca-Cola.
One of the company's biggest competitors is media tracking powerhouse Nielsen, which recently created an entire Consumer Neuroscience division.
MediaScience studies often involve testing response to ads, or television shows.
"The participants love it because they get to see something that isn't out yet," Rask said. "Their opinion gets to shape that program." For instance, Rask said they have the ability to tell networks which TV characters they like and don't like.
But MediaScience doesn't just study people watching TV. They also look at how people use social media or virtual reality.
For instance, the company recently published a report on mobile ads, concluding that Snapchat ads commanded more attention and had better emotional resonance than ads displayed on other social networks.
MediaScience's foray into politics is relatively new. Varan said the company completed several studies of the debates between presidential candidates Mitt Romney and Barack Obama in 2012, but did it only for internal research and development purposes.
This year, MediaScience is hoping to partner with a TV network to do studies of each of the presidential debates.
"We think it's a fascinating campaign to conduct research (on)," Rask said. Politics in general is easier to study because "people often feel so strongly about politics," which means heightened physiological reactions.
Genco, the neuromarketing expert, said using biometric data in politics could be useful. "Politics is much more emotion-driven than we pretend it is," he said. "We make our political decisions very much from the gut."
Some TV networks, such as CNN, have used "dial scores," where audience members turn dials to the left or to the right to indicate their interest level or approval. Often, TV networks will display these scores instantaneously on screen.
The problem with dial scores, Varan said, is that people don't want to indicate any negative reactions toward their preferred candidate. But the biometrics data can expose how someone is really feeling, he said, and they can offer this data in "near real time."
Varan said TV networks could find this data valuable because it exposes which topics are salient to their audience.
For instance, Varan said by looking at the data from the Commander-in-Chief forum that when Trump said he had always opposed the Iraq war, it didn't evoke a strong reaction from their panel, even among the Clinton supporters.
But there was heavy media coverage of that comment, with many reporters picking up on it because it contracted his earlier statements. Meanwhile, questions about the ongoing Clinton email saga did generate more interest among panel members, Varan said.
A TV network could offer superior post-coverage analysis with the MediaScience data, Varan said.
And there is the largely unexplored potential of using this data to help political campaigns.
Though MediaScience doesn't engage in brain imaging, a 2010 report on the use of neuro-imaging in marketing suggested that studying brain activity could be valuable in selecting desirable political candidates.
But Varan said his firm is not interested in conducting studies for the presidential campaigns themselves.
"We're truly neutral," he said. "We don't have a side."