Exploring Characteristics of Effective Apologies in Online Harassment Settings
Presenter
Doctoral Student at the School of Information
Time and location
North Quad 4330, Thursday (1:00-2:00) pm
Abstract
The prevalence of harassment in online spaces has led to increasingly negative experiences for several members of these communities. While the harm done is irrevocable, receiving an apology can help the victims move past the experience and continue to be members of the online community. Prior work shows that apologies serve this restorative purpose, but what makes an apology work in this online context remains unclear. In this project, our goal is to understand which characteristics make for an effective apology for cases of online harassment. We vary different components of an apology (acknowledgement of responsibility, explanation for the situation, a promise of non-recurrence), timing, visibility, and tone of empathy as potential factors. Ideally, these would be potential factors in an apology from the offender. However, offenders do not always apologize and cannot be coerced to do so. As a result, we also explore whether apologies in these settings can be mediated by AI and how that impacts their effectiveness. Our setup relies on hypothetical scenarios based on different types of online harassment—we first establish the credibility of using these scenarios, followed by a study of apology components used individually or in combination, and finally conduct a survey to study the effect of different factors on the efficacy of an apology.