Abigail Z. Jacobs and Hanna Wallach observe that computational systems often use unobservable theoretical constructs (e.g. “creditworthiness”, “risk to society”) that cannot be measured directly. These unobservable constructs are operationalized (as latent variables) by defining a measurement model to infer their value from an observable property. However, this practice may lead to a mismatch between the unobservable theoretical construct and its operationalization, which can exacerbate social injustices. Jacobs and Wallach suggest that measurement modeling techniques can be applied to understand construct validity and reliability and to assess fairness in computational systems.