What’s in a name?
On the one hand, names are just the words that we use to refer to others. As Juliet famously said of Romeo, a rose by any other name would smell just as sweet. On the other hand, names also communicate who belongs and, conversely, who is seen as foreign. For example, to counter being viewed as foreign, many Asian immigrants adopt an Anglicized first name: Americans are more likely to welcome Alex than Xian into their workplace. Names also drive other forms of bias. Experiments in which researchers send out resumes with randomly selected names, find that applicants with distinctively Black names (such as Lakisha and Darnell) are less likely to receive interview offers than those with typically White names (like Emily or Greg) – even with identical qualifications.
Do the same biases afflict judgments in the courtroom? The decision of whether or how long to imprison a person can significantly curtail their freedom and life trajectory. To help judges make data-driven decisions and not rely on their instincts, courts use many different tools, including artificial intelligence, to predict recidivism and otherwise inform decisions. When judges make these decisions, do they punish the crime or the name? In other words, do racial stereotypes attached to names lead to harsher sentences for Black men with more distinctively Black names?
To answer this question, we examined judges’ sentencing decisions of Black male inmates in Florida. A panel of online participants rated how stereotypically Black or White each defendant’s first name was. We created difference scores representing how distinctively Black the name is, and correlated these ratings with the severity of judges’ sentences, given equal crimes and past criminal records —key factors that are used to determine sentence length.
The results are sobering. For two defendants with the same criminal history and facing equally severe charges, judges issued a harsher sentence for the defendant with the more stereotypically Black (and less stereotypically White) name. Indeed, a man with the first name of Jamal would be imprisoned an additional year (409 days) compared to one named James. These results are all the more disturbing given their context. Judges in Florida are required to use standardized criminal sentencing score sheets designed to mitigate bias in sentencing.
[O]ur findings suggest that, instead of humanizing Black men or encouraging people in the courtroom to see them as individuals, names can reinforce racial biases.
This first-name bias isn’t limited to judges, however. Our findings held up in a follow-up online experiment, in which American participants were asked to sentence a series of Black defendants. We held everything constant but the defendants’ names, which we randomly set to a more or less stereotypically Black name. The results from this controlled experiment mirrored what we found in the courtroom: the more stereotypically Black the name, the more punitive the assigned sentence.
Why might this be the case? One reason is that people make assumptions about defendants based on the racial stereotypical associations with their names. Indeed, when asked “how stereotypically Black” a person’s name was, and the degree to which the name called to mind a violent criminal, people gave almost identical ratings – a near perfect correlation. This reveals that stereotypical Black names in the criminal justice context call to mind racialized stereotypes.
Our findings are particularly consequential because they point to a bias blind spot in decision-making that has severe consequences for individuals’ freedom. Many times, we expect names to set us apart from others. Parents give their children names that they hope will confer dignity and identity. Yet, our findings suggest that, instead of humanizing Black men or encouraging people in the courtroom to see them as individuals, names can reinforce racial biases.
This blind spot has the potential to become more pervasive, impacting more lives, as courts increasingly supplement judges’ sentencing decisions with AI technologies. While these AI tools were introduced to make sentencing more fair and less prone to human biases, they have perpetuated and in some cases amplified biases that judges exhibit. For example, AI models that use the length of previous sentence(s) to determine recidivism (e.g., COMPAS), would assign disproportionately higher recidivism scores to Black men with distinctively Black names, leading to disproportionately longer sentences again.
First-name biases are pernicious because we tend to think of names as identifying individuals. The blind spot operates in the gap between this intuition and the social assumptions tied to names. As a result, we may feel as if we are judging individuals when we are actually applying group stereotypes. To fill in this gap, we need to increase awareness of the potential for names to drive race-based discrimination. This will be particularly critical for the people developing AI models and the judges who use them to guide their sentencing decisions. Otherwise, we will perpetuate the same racial discrimination, just by another name.
Photo via Getty Images
