Using Intersectional Representation & Embodied Identification in Standard Video Game Play to Reduce Societal Biases

Title of the Paper

Using Intersectional Representation & Embodied Identification in Standard Video Game Play to Reduce Societal Biases

Paper Information

• Authors: Marie A. Jarrell, Reza Ghaiumy Anaraky, Bart P. Knijnenburg, Erin M. Ash
• Publication Date: May 2021
• Conference: ACM Conference on Human Factors in Computing Systems
• Field: Human-Computer Interaction (HCI), Societal Bias, Video Games
• Keywords: Video Games, Identity, Racial Bias, Gender Bias, Intersectionality, Representation, Virtual Characters, Embodied Identification

Research Background and Problem

• Issues and Challenges:
  • Existing studies are predominantly limited to fully immersive virtual reality (VR) and games with empathy-driven narratives, focusing on single dimensions of societal bias. Research on intersectional biases in traditional non-immersive video games is scarce.
  • Representation in video game characters has long lacked diversity, with relatively little research on cross-gender and cross-racial studies.
• Significance:
  • The potential societal harm caused by racial and gender biases includes systemic discrimination in areas such as education and employment. Groups with intersectional identities (e.g., Black women) face discrimination and oppression from multiple dimensions.
• Motivation and Related Work:
  • The interactivity of video games can subtly influence players' attitudes toward specific groups. While the impact of empathy and social issue-related content has been demonstrated in VR, its effects in non-VR games remain unclear.
  • Kimberlé Crenshaw's theory of intersectionality underscores the necessity of studying biases across multiple identity dimensions (e.g., race, gender) simultaneously.

Solution

• Research Methodology:
  • Developed a 2D traditional video game featuring characters with diverse racial and gender attributes. Players were randomly assigned to play as characters with different traits (White male, White female, Black male, Black female).
  • After gameplay, participants evaluated a LinkedIn job application. The applicant's photo and name were manipulated based on racial and gender factors (experimental variables).
  • Tested whether experiencing and internalizing virtual characters with intersectional identities could reduce participants' racial biases against Black individuals and gender biases against women, particularly dual biases against Black women.
• Key Techniques and Implementation Steps:
  • Differentiated character design: Ensured that character visual appearances and language settings avoided triggering stereotypes while maintaining design consistency.
  • Evaluation scales: Designed questionnaires based on existing literature to assess the applicant's technical skills, likability, and application quality.
  • Data collection and analysis: Used structural equation modeling (SEM) to study the relationship between biases and character interaction, with subgroup analyses to identify participants with ineffective impacts.

Research Findings

• Specific Results:
  1. Racial and Gender Biases: White male participants rated Black applicants' technical abilities significantly lower than White applicants (racial bias) and rated female applicants lower than male applicants (gender bias). Black female applicants faced dual biases of both gender and race.
  2. Character Internalization and Bias Mitigation:
     • Playing as female characters reduced participants' gender bias against applicants.
     • Playing as Black characters alone did not significantly reduce racial bias, indicating that single-factor interventions are not entirely effective.
  3. Intersectionality Support Model:
     • Participants who highly internalized and identified with Black female virtual characters showed the highest ratings for Black female applicants' technical abilities and likability, significantly reducing biases.
     • This combined effect (intersectionality effect) highlights the unique power of intersectional character identities in mitigating biases.
  4. Reverse Effects of Low Identification:
     • Biases against Black employees were reinforced when participants showed low identification with virtual Black characters.
• Strengths and Innovations:
  • Demonstrated the modifiability of societal biases through embodied identification in non-immersive games without explicit empathy-inducing narratives.
  • Introduced a novel analytical framework for understanding the intersectionality of cross-racial and cross-gender biases.
• Limitations and Future Directions:
  • The simplicity of the game may have limited the demonstration of stronger interactive effects, such as the dynamic changes in biases over prolonged gameplay.
  • Future research should explore additional identity dimensions (e.g., age, weight, religion) and incorporate richer character designs and experience modalities.
  • Improving and optimizing character design to enhance player identification with characters, avoiding potential bias reinforcement caused by low identification.

Conclusion

This study provides a foundation for incorporating racial and gender representation into traditional video game design. It is the first to demonstrate the combined potential of intersectional identity and embodied identification in mitigating biases, offering significant practical implications for both academia and the gaming industry.