• Home
• News and Updates
• 2023
• Understanding Barriers for Women in Politics: Analyzing Language through Quantitative Social Science and Machine Learning

Understanding Barriers for Women in Politics: Analyzing Language through Quantitative Social Science and Machine Learning

August 18, 2023

Sebastián Vallejo Vera is joining the department of Political Science.

Vallejo applies quantitative social science methods to understand the barriers faced by women and marginalized populations as they enter political institutions. He also applies machine learning models, to better understand the type of language used in legislatures and political arenas and how this can create or reinforce barriers.

A native of Ecuador, Vallejo focuses much of his research on politics and institutions in Latin America. Many of these spaces have gendered rules which allow men and women to participate in different capacities. Racial identity also plays an important role in Latin America.

“We have a complex relationship with racial identity. In Latin America, what it means to be indigenous, white, or Mestizo makes these identity complicated and fluid,” he said. “Someone may identify as white in one space, such as an urban space, but as Mestizo in another.”

While he primarily focuses on Latin America, Vallejo said Canada would have some of the same dynamics and barriers preventing women from political success.

“There has only been one female prime minister in Canada, but there would be a similar distribution of (political) talent in men and women,” he said. “If there is only one female prime minister, it means there is a barrier. Someone has purposely created barriers that allow men to access power and women cannot.”

Questions of representation

These barriers are related to the broader topic of representation and consideration of who can represent groups in political and legislative bodies.

“In the long history of these countries, many marginal voices have not been represented. They need to have barriers removed to allow better representation,” he said. He uses many tools to understand these barriers, including applying natural language processing models to political language to track strategic challenges faced by groups, and social changes.

Using transcripts of the legislative record in Ecuador, Vallejo examined how often different speakers were interrupted during speeches in the legislature, including who gets interrupted, and what happened when they were interrupted.

“One barrier women have is participating on the floor. If a man gets interrupted, he can continue talking, and his time to speak isn’t altered. If women are interrupted, they get to talk less, and will not have access to speak again in the future,” he said. This, in turn, informs the informs understanding of strategic behaviour women adopt to avoid interruptions.

He used similar tools to understand public sentiment. In 2019, protests led by Indigenous groups swept across Ecuador. Vallejo was not in the country but was following the events through social media. He saw an increase in racist sentiment and expression through comments about the protest movement. While the sentiment was not necessarily new, the fact it was expressed on Twitter allowed Vallejo to input the text in a processing model, to explore how racism was manifested and expressed, and how it encouraged further racist behaviour.

 

Lowering barriers to entry for Social Science

Vallejo researches political barriers that exist, but he is also interested in lowering the barriers to using different research methods. He is not focused on developing new AI tools, but rather applying those that already exist to societal questions and making these tools available to more researchers.

“Part of my job as a social scientist is to lower the barriers of entry to everyone. At the end of the day, it’s just a tool and everybody should be able to use them,” he said.

If social scientists are reluctant to use new approaches, it leaves “computer scientists to solve social science problems,” Vallejo said. Social Scientists have a comparative advantage in understanding power dynamics in the world, yet most work to identify racist and hate speech in text has been done by computer scientists.

“They (computer scientists) are trying to detect something that has to be understood in a particular way. You need to understand the context of these things before you can apply the models,” he said. “The use of these is to reclaim the part that social scientists do best, that is, understanding the world.”

Vallejo looks forward to working with other faculty members, including David Armstrong, Canada Research Chair in Political Methodology, and the Centre for Computational and Quantitative Social Science.

“I am really excited,” Vallejo said. “I have known some of the faculty members by their work, and I know the type of scholarship department is producing.”