Heterogeneity within demographic categories means group variables as predictors can disadvantage atypical group members and underrepresented groups¶

Claim · All claims
Evidence · 1 study (2 entries) · theoretical r3 · q1
Currency · newest study 2023 · not yet checked against its sources · not reviewed by a person · recheck by 2026: tool and model accuracy changes quickly

Subclaims¶

q1 i? If there is significant heterogeneity within a group, using the group variable will disadvantage atypical members, biasing towards a group mean. → Baker 2023 q1 i? Including one group's demographic variable can eliminate a shared mediating variable, worsening prediction for other underrepresented groups not represented in the data. → Baker 2023 (2)

Evidence¶

Baker 2023¶

Baker, R. S., Esbenshade, L., Vitale, J., & Karumbaiah, S. (2023). Using Demographic Data as Predictor Variables: a Questionable Choice. Journal of Educational Data Mining, 15(2). https://jedm.educationaldatamining.org/index.php/JEDM/article/view/619 doi:10.35542/osf.io/y4wvj

q1 · i? · theoretical · r3

Theoretical argument (type e) in section 4.3 on categorization limits: "If there is significant heterogeneity within a group, then the use of the group variable wi ll disadvantage atypical members"; the article adds that heavily demographic models may predict typical members better but atypical members worse, biasing towards a group mean.

"If there is significant heterogeneity within a group, then the use of the group variable wi ll disadvantage atypical members."

Baker 2023 (2)¶

Baker, R. S., Esbenshade, L., Vitale, J., & Karumbaiah, S. (2023). Using Demographic Data as Predictor Variables: a Questionable Choice. Journal of Educational Data Mining, 15(2). https://jedm.educationaldatamining.org/index.php/JEDM/article/view/619. doi:10.35542/osf.io/y4wvj

q1 · i? · theoretical · r3

The article's worked example: if only African American can be included as a variable, "including that variable may eliminate disciplinary in fractions from the model", worsening prediction for Native (North) Americans, Native Canadians/First Nations, and Native South Americans; Baker & Hawn (2021) are cited that Native Americans are often insufficiently represented in US education training sets.

"then including that variable may eliminate disciplinary in fractions from the model, reducing model effectiveness for Native (North) Americans, Native Canadians/First Nations individuals, and Native South Americans while only mildly improving model performance for African American students."

Discussion¶

Related Claims¶

• The argument that excluding demographic predictors amounts to color-blind racism is contested: inclusion can also produce racist decision-making — related
• Using demographic variables as predictors risks reinforcing biases embedded in training labels, including self-fulfilling prophecies — related
• Evidence is mixed on whether including demographic variables as predictors improves model performance in educational prediction — related
• Including demographic variables as predictors can reduce actionability by de-emphasizing correlated actionable risk factors and by discounting intervention effects — related
• Learning conditions worsen slightly but significantly over a school year absent intentional efforts to improve them — related
• Model accuracy of vision-based proctoring systems can vary across demographics, lighting conditions, and hardware setups, potentially disadvantaging some students — related
• Anonymity in computer-mediated collaboration can produce a stronger group identity than non-anonymous face-to-face situations — related