Interviewed Women
Word network from sexual violencecomments showing keywords and themes
used in the comments
The malignant influence of social media, as a ubiquitous and powerful medium of communication in countries all over the world, has brought about a growing era of challenges such as hate speech, online violence, fake news, and other forms of harmful content perpetuated by online bullies a.k.a. byte bullies.
From the physical world to the digital realm, gender-based discrimination persists and thrives through malignant online practices. Kenya is no exception, as online harassment targets both prominent women and everyday users. These harmful behaviors are becoming all too common in our online spaces, perpetuated by social, economic, cultural, and political structures that echo through the medium of digital networks.
In order to further understand these malignant practices, Pollicy conducted a comprehensive study that involved tracking and analyzing the online activities of 268 women and men candidates on Twitter and Facebook during the 2022 Kenya general elections specifically, during the campaign and election period. The methodologies encompassed a range of techniques, including lexicon-building focus group discussions, data scraping of publicly available profiles, qualitative data analysis, and the development of a Machine Learning model capable of identifying and categorizing instances of online violence and hate speech in both English and Swahili languages.
RESEARCH QUESTIONS USED TO GUIDE TOWARDS ACHIEVING THE RESEARCH OBJECTIVE
- How do women politicians in Kenya use social media platforms for campaigning during the election period?
- How does the use of social media platforms differ amongst men and women candidates?
- How does online violence against women candidates differ from that experienced by men in regards to factors such as the form of violence experienced and frequency across platforms?
- What evidence of OVAW-P exists on social media platforms and how does it manifest?
- What is the association between OVAW-P and factors such as their age, frequency of social media use, and electoral results?
FINDINGS FROM THE STUDY REVEALED THE FOLLOWING
- The use of social media platforms for engaging with voters and constituents by women politicians remains low in Kenya. 93 percent of the accounts belonging to men candidates were used at least once during 2022 on Facebook compared to 80 percent of the women politicians' accounts. On Twitter, 49 percent of accounts belonging to women had less than 5 tweets a month during the campaign period.
- OVAWP was more prevalent among women candidates than men, especially on Twitter where 2 out of 5 women's Twitter accounts monitored experienced sexual harassment.
- The attacks against Kenyan women politicians were often focused on personal and sexual aspects rather than policies or qualifications.
- The cultural and traditional barriers that limit women's participation in politics contributed to online violence, and factors such as age and party affiliation also made women more vulnerable to harassment and abuse.
How did women politicians in Kenya use social media platforms for campaigning during the general elections?
Facebook usage was generally higher across the men candidates with 93.8 percent of the accounts
belonging to the men candidates being used at least once a week compared to the women candidates at
80 percent.
On Twitter, 51 percent of the women candidates used their account at least once a week
compared to their male counterparts at 56 percent.
Is there evidence of OVAW-P? If yes, what are the manifestations of OVAW-P?
- OVAW-P was more prevalent among women candidates on Facebook with at least one in two of the women's accounts monitored (55.7 percent) experiencing some form of OVAWP compared to 35.4 percent among the men candidates.
- Additionally, each of the forms of violence observed was more prevalent towards those women candidates, especially sexual violence which was predominantly experienced by women candidates.
-
On Twitter, online violence disproportionately affected women, with significantly higher percentages of
women experiencing sexual harassment, hate speech, and disinformation compared to men.
Does OVAW-P correlate with other factors such as age, election result, and frequency of social media use?
According to the data from Facebook, sexual violence increased with an increase in age.
While the two forms of violence: insult and hate speech and trolling were more associated
with younger women age groups. Online violence was predominantly associated with the losing
candidates, and 72.7 percent of the women who lost the election experienced disinformation and
60 percent of the women who lost the election experienced insults and hate speech, and 75% of the
women who lost the election experienced sexual harassment and 77.4 percent of the women who lost the
election experienced trolling.
In terms of social media frequency of use, higher usage of social media platforms resulted in
higher levels of online violence.
Percentage of women candidates on Twitter experiencing a specific form of online violence grouped by their age
Percentage of women candidates on Facebook experiencing a specific form of online violence grouped by their age
Percentage of women candidates experiencing a specific form of OVAWP grouped by their usage frequency on Facebook
Evaluating the Nature of Discourse Observed in the Twitter Replies to Women Candidates
1 / 2
Word network from sexual violencecomments showing keywords and themes
used in the comments
2 / 2
Word network showing the most frequentlyused words in replies to women candidates
on Twitter
Note: As part of this project, a text classification model was used to classify tweets/posts in various categories such as: Trolling, Sexual violence, Insults, etc. This model can be tested out here!
