A study of 172 college students at Penn State shows that interactivity, perceived interactivity, and a conversational tone can all impact how engaged users are with a health assessment tool and what impressions they walk away with after using the tool. The study was conducted by S. Shyam Sundar, a professor of communications at Penn State and co-director of the Media Effects Research Laboratory, and Saraswathi Bellur, an assistant professor of communication at University of Connecticut.

"If a communication scenario was to involve strictly task-oriented exchanges (e.g., ATM machine dispensing cash), the display of affiliation behaviors, such as a friendly conversational tone, may not be important," researchers wrote in the paper, published in the journal Human Communication Research. "However, as studies in patient-physician communication have shown, healthcare interactions are different in nature. They not only feature informational messages but also a good deal of psychosocial discourse characterized by factors such as empathy, active listening, and humor. Thus, examining the role of conversational tone can help us translate effective strategies from offline patient-physician communication to online user interactions with a system."

Sundar and Bellur randomized students into low, medium, and high interactivity conditions and conversational and nonconversational conditions. All of the assessments asked users the same questions, but in the higher interactivity conditions more information about their past responses was displayed on the screen, to create the impression that responses were being tailored to the user.

But that design may have had the opposite effect to the one intended, as participants reported finding the low interactivity condition the most interactive. 

"In terms of design considerations, it is likely that the lesser visual display of the interaction (i.e., less “clutter”) in the low-interactivity condition may have led users to form richer perceptions of interactivity," researchers wrote. "As Walther (1992) noted, even very lean forms of CMC are often powerful enough to evoke rich perceptions of communication. With its instant message or texting-like interface, the simple Q&A interaction in the low-interactivity condition possibly appealed to the avid smartphone users who comprised the study sample."

This unexpected finding made it hard for researchers to draw good conclusions about the effect of interactivity on engagement, though the literature review points to a positive correlation. There were clearer effects when it came to the tone. Researchers found that the more conversational tone, which added cues between messages along the lines of "let's move on to the next question", led users to report feeling less susceptible to obesity, diabetes, and heart disease.

"[T]he presence of turn-taking cues seemed to significantly reduce perceptions of relative susceptibility," Bellur and Sundar wrote. "Thus, the presence of even minimal turn-taking cues serves an important 'hand-holding function.' This finding offers practical implications for the design of online health tools, in the form of attentive and empathetic virtual coaches, with risk reduction goals."

Of course, making people feel less susceptible can be a bad thing if the goal of a health communication is to get someone to take action and change their behavior.

"This conversational tone may make them warm and fuzzy, but that's not what you want to do with a health assessment tool," Bellur said in a statement. "If you want people to stand up and take action, this type of friendly turn-taking softens the effect. However, if the goal of the interaction is to promote a sense of comfort among individuals, the same conversational tone strategy could work well, with the online tool acting like a virtual coach and providing reassurance."