Constructs, Tape Measures, and Mercury

This is a Lewinian-field-theory approach to understanding the Graduate Record Examination (GRE) in the context of racism to contribute to the debate about whether graduate schools should remove GRE scores from admissions processes. Woo and colleagues (this issue) review the empirical literature on bias from a psychometric perspective. In this commentary, I challenge the definition of the underlying construct measured by the GRE and offer alternative definitions of what is measured. Next, drawing on an analogy from genome-wide association studies, I discuss how genomic models predicting height that are trained on data from European ancestral populations systematically underpredict the height of West Africans. Our access to data from tape measures, and their correlation with height, provide objective opportunities to audit our prediction. I discuss the implications of this when the criterion variable for validating the GRE is first-year grades. I then probe an analogy used by Woo and colleagues in which they assert that blaming the GRE for disparities in scores across groups is akin to blaming the thermometer for global warming. I describe racism as context for a field-theory approach to thinking about the limitations of this misguided analogy. Finally, I suggest pathways forward.