Expressing the opinion that gender-based violence is acceptable is not a topic for polite conversation. But sharing this opinion online? It turns out that’s a whole different story.
Over one-third of women in the Middle East and North Africa (MENA) region experience some form of violence (physical, psychological, economic, sexual) in their lifetime. But while many organisations focus on identifying the root causes and solutions for violence, Jihad Zahir, an assistant professor at Cadi Ayyad University in Morocco, measures personal attitudes around the issue by analysing data from social media platforms. Understanding the social norms around gender-based violence is critical for creating effective solutions to eliminate such violence.
“To study gender-based violence, you always need qualitative studies, like focus groups,” explains Zahir. “However, this becomes a limitation for research in this area because these studies are time-consuming and expensive. And also because of something called social desirability bias.”
Social desirability bias is a response bias where survey respondents tend to answer questions in a manner that will be viewed favourably by others. This can result in over-reporting "good" behaviour or under-reporting "bad" (undesirable) behavior. In the case of violence against women, social desirability bias plays out when a person in an open focus group discussion is far less likely to express the opinion that violence can be acceptable.
On the contrary, research shows that people are more honest about their thoughts and opinions on social media platforms. For Zahir, this presented an opportunity to track attitudes around gender-based violence, an area where there is a clear disconnect between opinions that are shared in qualitative research (violence is wrong!) and the actual experiences of women in the MENA region (violence is prevalent!).
Zahir and data science colleagues at Cadi Ayyar University developed Natural Language Processing techniques to analyse attitudes around gender-based violence in the Arab language Twitter and YouTube posts. They found that negative videos — that is, videos that express negative sentiments toward gender equality — were “liked” more than videos with positive messages. They also found that negative sentiments were especially dominant in male-authored comments related to domestic violence and sexual harassment.
The reaction to this research project? “No one found the findings surprising,” Zahir shares. Yet the potential this research has for affecting change is incredibly promising.
Zahir’s research was funded through a grant from Data2X, an organisation housed at the United Nations Foundation that works for more and better gender data to improve the lives of women and girls. In 2017, Data2X issued ten “Big Data for Gender” challenge awards — grants to research projects that sought to use sources of big data — from social media posts, to mobile phone records, to geospatial data — to fill the gender data gap. Zahir was one of the ten grant recipients. The grantees recently shared the results of their research at an event in New York City, Big Data, Big Impact? The Future of Gender-Sensitive Data Systems.
Zahir believes there are several ways her work could positively impact efforts to curb gender-based violence in the region. First, though this data simply confirms what was already known to be true — that women experience violence at high rates — it uses a form of a “big data” which is outside the realm of official statistics. For civil society advocates that want to employ data-driven arguments but are mistrustful of official statistics on gender-based violence, Zahir’s work provides a new body of data to use in their efforts to call for change in policy and programs.
Second, this research has the potential to not just provide a snapshot of attitudes or opinions at a particular time, but to track changes in attitudes in real time. Zahir points out that pulling data from social media platforms is more cost-effective and accounts for a much larger sample of individuals than traditional surveys and focus groups allow. And because this data is constantly generated, it makes it possible to use it to monitor the impact of outside events, like awareness campaigns, economic shocks, and natural disasters, on people’s attitudes and opinions
Finally, while social media may provide the platform for the expression of negative opinions, it also provides a platform for positive opinions, which can be used to influence others. Zahir believes that by using social networks, one could identify connectors and influencers who could then be used to promote values supporting women’s rights.
“Just open Instagram and you see all those influencers advertising products,” Zahir explains. “Imagine what could be possible if, instead, they were used to spread positive messages about women’s rights.”
Sometimes tweets and online comments speak louder than words — and can help paint a more accurate picture of social norms around an issue — in this case, gender-based violence. Zahir’s research presents a unique opportunity to fill data gaps around violence against women and advance solutions that effectively eliminate physical, psychological, economic, and sexual violence in the MENA region and beyond.
Ending violence against women will require commitment from all people, across all platforms, using a variety of data sources. The 16 Days of Activism Against Gender-based Violence, which kicked off on November 25 and lasts through December 10, provides an important 16-day platform for shedding light on this issue. Yet it also reminds us that attention to gender-based violence must go beyond these 16 days; it is an everyday problem that requires everyday attention.
Nina Rabinovitch Blecker is the director of communication for Data2X and Elizabeth Black is a communication associate with Data2X