Surveys and focus groups have been the gold standard when it comes to determining whether a public service announcement (PSA) persuades people to change behavior. People would be asked to view an ad and rate it for what researchers call ‘perceived effectiveness.’ In other words, participants would rate the ad for how likely they would be to remember it or whether it grabbed their attention.

Enter Neuromarketing

“But these questions are subjective,” noted USF College of Public Health professor Dr. Claudia Parvanta, a social marketing expert. “The respondents judge how they feel and report what they might do. The perceived effectiveness scale was ‘as good as it got’ until neuromarketing technologies came on the scene, mostly in the private sector, about 20 years ago.” 

Neuromarketing measures physiological and neural signals, such as eye movement and facial expressions. While it’s been used extensively in the private sector, the technology’s prohibitive cost has put it out of reach when it comes to determining the effectiveness of public health messaging.

Until now.

Parvanta, along with a team of researchers, including Dr. Rob Hammond from USF’s Muma College of Business and Dr. Kim Walker from the Zimmerman School of Advertising and Mass Communications, set out to study whether facial expression analysis—which measures human emotions through facial expressions—can predict the perceived effectiveness (PE) of Tobacco Free Florida’s (TFF) anti-tobacco PSAs. The research was funded by TFF.

Their research, “Face Value: Remote facial expression analysis adds predictive power to perceived effectiveness for selecting anti-tobacco PSAs,” was published in July in the Journal of Health Communication.

How Effective Is Facial Expression Analysis?

In the study, 302 tobacco users watched three PSAs while a webcam recorded metrics for attention, such as head position, and facial action units (e.g., eye and eyebrow movement, facial muscle movement, etc.). 

“What we found was that a specific combination of facial movements (e.g., lip suck, brow furrow, brow raise) shown more and longer COMBINED with higher attention (percent of time looking directly at the screen during the video) COMBINED with a higher PE rating—could predict 82% of the time who would fall into the ready-to-quit group (defined as those who visited the TFF website) or the not ready group (those who didn’t visit the site),” Parvanta said. “No measure alone did as good a job at predicting a participant’s behavior of going to the TFF website.”

And while the study looked only at anti-tobacco PSAs, Parvanta noted it can be extrapolated to other public health messaging. “If you wanted to set up a program—it could be for quitting smoking, but also using PrEP (pre-exposure prophylaxis), WIC, health insurance, anything—you could put up an ad for it and by measuring facial expressions and PE, determine who is most likely to come to your program. You could then create advertising that appeals to this audience and have a better idea of who needs a different approach.” 

How Facial Technology Works

The facial analysis technology used in the study was developed by the company iMotions. It employs artificial intelligence to interpret and code “facial action units” (for example, eye and eyebrow movement, facial muscle movement, etc.). The codes are fed into a computer algorithm that was trained on millions of faces from all over the world.

“While a respondent looks at a specific video ad, the software provides ongoing probabilities that we would see specific expressions exactly when they occur (i.e., which point in the video), and for how long they endure in microseconds,” Parvanta explained. “The other important metric that the software provides is ‘attention,’ which shows whether the respondent is facing the computer screen or looking away.”

Bottom Line

According to Parvanta, all the study participants liked or disliked the same videos, but if they were ready to quit, they rated the videos more highly. “Higher ratings say more about the viewer’s state of mind than the video’s inherent persuasiveness,” she said. “When it comes to messaging, the independent measure of ready-to-quit/not ready prevents us from ‘preaching to the choir.’ As far as improving anti-tobacco ads, we hope the FDOH signs us up to do more of their pretesting work.”

Story by Donna Campisano for USF College of Public Health