Modelling Unwarranted Disparities in Sentencing: Distinguishing between Good and Bad Controls

To minimise confounding bias and facilitate the identification of unwarranted disparities, sentencing researchers have traditionally sought to control for as many legal factors as possible. In this article we challenge such approach. Using causal graphs we show how controlling for commonly used variables in the sentencing literature can introduce bias. Instead, we propose a new modelling framework that clarifies which types of controls are necessary to identify different definitions of sentencing disparities. We apply this framework to the estimation of race disparities in the US federal courts and gender disparities in the England and Wales magistrates’ court. We find that the model uncertainty associated to the choice of controls is substantial for gender disparities and for race disparities affecting Hispanic offenders, rendering estimates of the latter inconclusive. Disparities against black offenders are more consistent, although, they are not strong enough to be seen as definitive evidence of racial discrimination.