From January 2019 to June 2019, MarvelousAI partnered with Wilson Center Fellow Lucina Di Meco and University of Maryland professor and scholar Sarah Oates to track and identify online narratives surrounding a sample of the Democratic contenders for the 2020 democratic nomination. In their ASPA paper titled “Running While Female” researchers found that female candidates in the 2020 U.S presidential primary election faced significantly more character and identity attacks online than their male counterparts.
The data used in this study focused on Twitter conversations about six leading candidates (three male and three female): Joe Biden, Bernie Sanders, Pete Buttigieg, Elizabeth Warren, Kamala Harris, and Amy Klobuchar. The researchers analyzed conversations surrounding each candidate’s official campaign launch (between January and April 2019, depending on the candidate), and conversations following the first 2019 Democratic debate on June 26th and 27th. They broke down online narratives into five categories: policy, ideology, character, identity, electability.
The team identified profound parallels in how female candidates are treated in online news media and how they were being talked about on social media, specifically Twitter. According to their research, female candidates are frequently marginalized and attacked on character and identity narratives, while their male counterparts are not.
The researchers also found that women running for president received significantly more negative tweets from far-right leaning and non credible sources than did male candidates. After the first democratic debates, these findings became even more prevalent, with one exception. According to the researchers, Elizabeth Warren seemed to rise above the character attacks by the end of the first debate.
Some of the online mentions of Elizabeth Warren were categorized as support narratives. They included playful/ meme-like variations of “Warren has a plan for that.” Examples include:
The researchers used MarvelousAI’s StoryArc, a narrative tracking platform for online news and social media, to collect data from the study. StoryArc is an interactive active-learning loop. It collects and groups similar content. Coders then analyze the automated groups in order to define political narratives. They add labels to narratives in order to refine the quality of automatic identification for future content as well as discover previously unlabeled groupings.
Marvelous AI believes a multi-disciplinary approach augmented by artificial intelligence is key to readily identifying and challenging sexist online narratives and coverage of female candidates. The real-time measurement system set up by Marvelous AI allows candidates to identify and push back against misogynistic comments on social media and take control of their narratives online. When voters and candidates are more aware of the role emotion and manipulation plays in these narratives, they can better understand, embrace, deflect or challenge narratives as they arise in the U.S media ecosystem.