Improving Diversity in Engineering: A Data-Driven Approach to Support Resource Mobilization and Participation in Hashtag Activism Campaigns

Abstract

A critical barrier facing engineering is inclusiveness of women in the profession. In recent years, engineering diversity advocates have taken to social media platforms to raise awareness of the issue and redress this problem. A recurring challenge for their initiatives though is attracting and mobilizing participants efficiently. For a successful mobilization campaign, organizers need real-time information about their users and also need to understand what messaging works to attract and mobilize them. We hypothesize that participants in any given campaign related to engineering diversity will also be interested in other campaigns related to that issue. Furthermore, since the primary signal for a social media campaign is a hashtag, by using clustering patterns of various co-occurring hashtags along with relevant topics and relatable sentiments, we can better understand participation and also mobilize users for the target campaign. To empirically examine our hypothesis, we study two diversity hashtag activism campaigns on Twitter (#ILookLikeAnEngineer and #WomenInEngineering) using a real-time predictive analytics framework. We design and evaluate the framework with a set of novel features that uses retweetability as an indicator of participation. Our result analysis for topical features found that monetary gain and advertisement-oriented content were less likely to be propagated in the campaigns whereas messaging aligned directly with the issue at hand such as breaking stereotypes in engineering was deemed more retweetable and engaging. In terms of sentiments, informal tone in the messages were considered desirable whereas short-form messaging were not very popular in either movements. These analytical insights can inform activists in effective resource mobilization through message content design, in order to expand the reach of an activism campaign. Our work shows how data-driven techniques can assist in increasing the participation of women in engineering education and the workforce.