Algorithmic mirrors : an examination of how personalized recommendations can shape self-perceptions and reinforce gender stereotypes
- Megan Rebecca French.
- [Stanford, California] : [Stanford University], 2018.
- Copyright notice
- Physical description
- 1 online resource.
Also available at
At the library
Limited on-site access
Researchers in the Stanford community can request to view these materials in the Special Collections Reading Room. Entry to the Reading Room is by appointment only.
|3781 2018 F||In-library use|
- With the growing prevalence of algorithmic decision-making, scholars have become increasingly concerned about algorithmic bias, or discriminatory differences in algorithmic decisions based one's identity. The present study examines that the extent that algorithmic bias can affect perceptions of the self, as well as the extent that one's understanding of the system moderates this effect. A total of 117 women were randomly assigned to receive a personalized recommendation for a stereotypical "feminine" (i.e., nurse, librarian) or "masculine" career (i.e., lawyer, chief executive) ostensibly based on their Facebook activity. Participants' a priori beliefs about the system's objectivity, personal data collection, and certainty about how the system worked were measured before they received their recommendation. Participant's self-perceptions of masculinity, leadership ability, and self-confidence were measured after they received their recommendation, along with their beliefs about the cause of the recommendation. Women who received a stereotypical feminine career recommendation reported lower masculinity, lower leadership ability, and lower self-confidence than women who received a stereotypical masculine career. Additionally, women who received a masculine career recommendation and believed the recommendation was based on internal characteristics (i.e., internal locus) were more likely to report greater leadership ability than women who received the same recommendation but believed the recommendation was based on external factors (i.e., external locus). The likelihood of making internal locus signal attributions for their recommendation was greater for people who more actively use Facebook and who believed the system was more objective. Together, the dissertation findings suggest that people's self-perceptions can be influenced by algorithmic recommendations, and this effect is magnified by one's understanding of the system.
- Publication date
- Copyright date
- Submitted to the Department of Communication.
- Thesis Ph.D. Stanford University 2018.
Browse related items
Start at call number: