The Power of Silence: Using Sentiment Text Analysis to Examine Twitter Responses to Sexual Harassment Accounts

Keywords: sexual harassment, apology, denial, text analysis

Abstract: This research explores how alleged sexual harassers are perceived when using apologies, denials, or reticence in responding to real sexual harassment allegations. Understanding which responses elicit greater negative responses has important conflict management implications (Sitkin & Bies, 1993). We build on prior research showing conflicting findings about the effectiveness of apologies and denials(Dunn & Cody, 2000; Ferrin, Kim, Cooper & Dirks, 2007), and add to the literature investigating reticence. Using a state of the art neural network approach to sentiment analysis, we analyzed 214,000 Twitter posts about 315 high-profile accusations of sexual misconduct. We found that reticence elicits less negative sentiment than apologies and denials. We also found that denials elicit less negative sentiment than apologies in responding to sexual harassment allegations. We discuss future opportunities for experimental research of this phenomenon and the implications this research may have for the #MeToo movement.