PhD Defence: Parity and Prejudice Inquiries into Social Media, Gender Inequality, and Development

This thesis focuses on violence against women and girls, which, in its many forms, is a silent global epidemic and a violation of human rights. While pervasive global gender gaps are shrinking (Bertrand, 2020), addressing violence against women remains an unfinished business. To resolve this systemic crisis, Goal 5 of the Sustainable Development Agenda carries, within its remit of gender equality, an ambition to eliminate all forms of violence against women and girls by 2030 (United Nations Statistics Division, 2015). The main objective of this thesis is to investigate the role of social media in transforming gender inequality and addressing violence against women and girls, two interconnected issues. This thesis comprises four chapters that examine these intersections. 

Chapter 2 studies the extant literature on violence against women and girls and outlines the primary public policy challenges to address this issue through the lens of the ‘Wicked Problems’ framework. The remainder of this thesis focuses on the case of India. 

Chapters 3, 4, and 5 empirically examine the relationships between social media and violence and discrimination against women and girls. Chapter 3 is concerned with the influence of social media technology and examines the impact of Twitter exposure on the problem of son-biased fertility preferences in India. By employing large-scale household survey data and social media data, these findings reflect the voices of over a million women in India and those inhabiting its digital spaces. The results from this chapter show that exposure to social media reduced son preference among women in India, and contributed to an overall decline in discriminatory preferences. The inquiry is extended by studying the effect of the platform on men’s preferences for the sex of the child, which illustrate a similar effect. Moreover, social media not only impacted preferences but also positively influenced outcomes related to these preferences, specifically nutritional outcomes for girls under the age of five. The mechanisms of impact are explored using two approaches. 

First, this research highlights that a majority of narratives related to children on Twitter were either neutral or progressive, indicating the capacity of social media to challenge harmful cultural norms. Second, we focus on a popular Twitter campaign with the goal of demonstrating that daughters are valuable, called #SelfieWithDaughter. Twitter exposure had a striking effect on preferences for the districts in which this campaign was active. This research applies a conjunction of quasi-experimental, computational, and qualitative social science approaches to unearth these findings. Chapter 4 investigates online narratives in more detail with the aim of identifying the footprint of gendered discourse in India. This chapter presents a deep learning approach towards analysing unstructured text data for gender-based discourse in the Indian context. A fine-tuned classifier based on the cross-lingual XLM-RoBERTa model is developed to identify misogyny (regressive) and gender-based empowerment and activism (progressive) messages in georeferenced Tweets. Quantitative analysis of the predictions from the fine-tuned model illustrate the geographical footprint of both these narratives. 

Furthermore, the model’s outputs are harmonised with regional socioeconomic variables to identify their relationships with the production of progressive and regressive Tweets. The findings indicate few significant associations. Lower reported rates of rape in a district and belonging to historically patriarchal North Indian states are associated with more progressive Tweets. Higher rates of urban poverty and, in some cases, scheduled caste populations are positively linked to the production of regressive Tweets. Chapter 5 traces the impact of the Indian #MeToo movement on online narratives of empowerment in the country. Using outputs from the model developed in Chapter 4 and exploiting a natural experiment, this research shows that the #MeToo movement led to an increase in progressive discourse online. However, this impact was neither homogeneous nor enduring. These findings have implications for gender and social inclusion in digital public spheres. They also provide causal evidence on the legacy of fourth wave feminism, in which notable social movements were shaped by the hashtag.