Cornell Bowers College of Computing and Information Science
trust graphic


Cornell study shows how trustworthiness is decided in shared economy profiles

What do Airbnb hosts write in their profiles to help potential guests to trust them?

Cornell researchers presented a paper on this question at the 20th Association for Computing Machinery Conference on Computer-Supported Work and Social Computing, Feb. 25 through March 1 in Portland, Oregon. The paper, “Self-disclosure and Perceived Trustworthiness of Airbnb Host Profiles,” has received an honorable mention for best paper at the conference.

Authors Mor Naaman, associate professor of information science at the Jacobs Technion-Cornell Institute at Cornell Tech; Xiao Ma, doctoral student in the field of information science, Cornell Tech; Kenneth Lim Mingjie CS’15; and former Cornell communications and information science professor Jeff Hancock, now at Stanford University, studied the perceptions of trustworthiness in Airbnb host profiles. They used a mixed-methods research approach, combining qualitative analysis, large-scale annotation and online experiment to find out what hosts write about, how trustworthy they seem and whether these perceptions lead to choice of host.

“We are very interested in trust and how it’s formed online, as it will enable the next generation of peer-sharing and shared economy services,” Naaman said. “Airbnb is a great example with a publicly available dataset that allowed us to start examining this topic in depth.”

When researchers asked people to rate Airbnb profiles for trustworthiness, they found the longer the profile text, the more trustworthy it is perceived to be. But length isn’t everything: Not all topics are created equal. The language of hospitality (e.g. “We look forward to hosting you”) is most effective in establishing the perception of trustworthiness rather than listing a life motto as suggested by Airbnb.

In addition, signaling theory predicts that hosts show trustworthiness by disclosing more about their origin, residence, work or study, which are more difficult to fake than interests or beliefs.

“Trust is deeply intertwined with safety,” said Ma, lead author. “Guests want to know if they’ll be safe, treated well and the property is well maintained, etc. We found that profiles which signal hospitality end up being more successful. A show of hospitality is an explicit gesture that is directly relevant to the transaction.”

As part of their research, the team produced the first systematic coding scheme and accompanying dataset for analyzing self-disclosure in online profiles.

“It would be great if Airbnb and peer-sharing communities could formalize or commoditize these findings,” Mingjie said. “Trust is a modality that is a lot of time based on physical appearances. When we are faced with a paragraph of text online, it would be wonderful to have some alternative signals to make good decisions.”

The full paper is available from the Social Technologies Lab’s website, where the researchers also made available all the data used for the study.