Interpetme: Community-Centered Approach to Address Racial Bias in Social Media Content Interpretation

Background

Understanding social media posts can be incredibly challenging due to their highly contextualized and hyper-local nature. Misinterpretations can lead to severe consequences, such as violence, wrongful arrests by police officers, denial of opportunities by administrators, and the perpetuation of stigmatizing narratives about marginalized communities by journalists. Marginalized communities often endure the most biased and unjust punitive actions stemming from their social media activity, leading to what is termed as "Digital Stop and Frisk." The monitoring and interpretation of social media content exacerbates systemic racial biases, further jeopardizing the lives and livelihoods of these community members. To tackle this systemic issue, we have adopted a community-centered co-creative approach to design and develop InterpretMe (https://www.interpretme.org/), a web tool comprising case-based learning simulations tailored for reporting professionals, including law enforcement officials, educators, and journalists. The aim is to train reporting professionals with the necessary approach and skills to engage in community-centered interpretation processes, identify racial biases, and mitigate wrongful punitive actions and negative digital footprints, particularly affecting Black, Indigenous, and People of Color (BIPOC) communities.

Methodology

This study employs a mixed-methods approach encompassing qualitative, quantitative, and computational techniques. Data collection involved three distinct components. Quantitative data, including demographic profiles and pre- and post-surveys assessing racism, bias, and misinterpretation, were gathered through Qualtrics, with descriptive analyses conducted to evaluate learning outcomes. Qualitative data were obtained through audio-recorded case-based simulation modules, transcribed, and preformed thematic analysis. Sentiment analysis was performed to ascertain the beliefs and emotions surrounding bias and identity during the interpretation process. The triangulation and validation of quantitative, computational, and qualitative analyses were undertaken to derive results for dissemination, adhering to rigorous ethical standards with approval obtained from the Columbia University Institutional Review Board.

Results

The pre- and post-survey findings from the simulation modules confirmed the efficacy of case-based learning simulations in reducing racial bias and misinterpretation among reporting professionals, thereby minimizing harm to vulnerable communities. Sentiment analysis revealed a shift in reporting professionals' emotions from negative (e.g., anger, disgust, fear) to positive (e.g., joy, trust) across the stages of the community-centered interpretation approach. Qualitative results endorse the effectiveness of community-centered case-based simulations in identifying racial bias in social media interpretation.

Conclusion and Implications

Meaningful community engagement is critical in interpreting social media content, bringing cultural context, critical reflection, and meaningful participation to mitigate bias in decision-making processes. Community-centered case-based simulation learning models would serve as innovative learning tool to address identity biases in social media interpretation, thereby reducing wrongful punitive actions and negative digital footprints of BIPOC communities. Researchers utilize a community-centered approach to gain insights into community needs, interests, beliefs, and dynamics to develop tools. Studies have demonstrated the effectiveness of services and interventions rooted in the community-centered approach, with InterpretMe highlighting its relevance in social media interpretation research. These findings underscore the importance of engaging impacted communities as experts in their own culture and lived experiences. Additionally, a multi-stakeholder approach proves effective in community-centered research, it calls for further research in diverse socio-cultural contexts to scale InterpretMe.