Inequities Exist but Bias Doesn’t Explain Them

Unlike Mac Donald, most people writing about implicit bias acknowledge the existence of widespread group-based inequities in societies like the United States. Some have claimed, however, that there is little evidence that bias and prejudice—whether implicit or explicit—contribute to those social inequities. For example, Sean Hermanson highlights studies that appear to find “no evidence of racial bias in officer-involved shootings” (2017b). Hermanson also claims that “given base rates for criminality the relationship between race and police shootings is unsurprising” (2018). (A “base rate” is the probability of some event happening, such as having a heart attack, prior to intervention, such as taking medication to lower your cholesterol.) Hermanson implies that the reason police officers shoot more black people than white people is because black people commit more crimes. While

Hermanson (unlike Mac Donald) acknowledges that this fact may be due to race-based inequity, he claims that racial bias is not the cause of this inequity. Police officers’ attitudes toward race, in other words, don’t explain the rates at which they shoot black people.

These claims contribute to Hermanson’s overall conclusion that “I want my money back” with respect to research on implicit bias. His critique is not based on shooting data alone. Like myself, Hermanson is a philosopher, and he is concerned with whether women and members of historically marginalized groups are at a disadvantage within the field. His view is that men are at a comparative disadvantage on the philosophy job market. Based on his own analysis of hiring data, Hermanson argues that there is, for example, “weak evidence” that CVs and resumes are judged differently according to the perceived gender and race of their authors (2017a). “Since 2010,” he writes, “women seem to enjoy significant advantage on the [philosophy job] market. In order to understand demographic variance we need to think more about pre-university influences, not implicit biases” (2018). This point contributes to Hermanson’s broad conclusion that it is a myth that there is widespread evidence that implicit bias contributes to real-world discrimination.

Reply

It may indeed turn out that some of the inequities for which researchers have thought that implicit bias is a partial cause are better explained by other factors. And, it is true that some have “over-hyped” research on implicit bias, suggesting (implicitly or explicitly) that it is the primary driver of discrimination in the contemporary world. But it is, for example, demonstrably false that the relationship between race and police shootings is explained by base rates of race-based criminality alone. An analysis of data from 2015, for example, finds no correlation between rates of violent crimes in 50 major cities in the USA and rates at which police officers killed people in those cities (https://mappingpoliceviolence.org/2015/). A more recent study reaffirms the basic finding; police officers are no more likely to shoot and kill someone who is unarmed in a high-crime area compared with a low-crime area (Nix et al. 2017). Black Americans are, moreover, disproportionate targets of police force, compared with white Americans, even when controlling for whether the person targeted is a violent criminal (Atiba-Goff et al., Center for Policing Equity 2016). Criminologist Justin Nix summarizes his recent study of 990 fatal police shootings this way: “The only thing that was significant in predicting whether someone shot and killed by police was unarmed was whether or not they were black,” continuing, “this just bolsters our confidence that there is some sort of implicit bias going on” (quoted in Lowery 2016).

A much more open question is how much bias (whether implicit or explicit) contributes to police shootings (see Saul (2018) for discussion).

Hermanson cites a much-discussed New York Times op-ed by economist Sendhil Mullainathan which argues that, if bias was a significant factor in contributing to police shooting of black Americans, then there would be a significant difference between the rates at which black Americans are arrested and the rates at which black Americans are killed by police officers. First, however, it is important to note that, despite this point, Mullainathan concedes that “Police killings are a race problem” and that “African-Americans are being killed disproportionately and by a wide margin” (2015). Second, Mullainathan argues that the problem is driven by the fact that black Americans have disproportionate encounters with the police (i.e., base rates for police contact are higher for black Americans than for white Americans) and that the drivers of police encounters are likely not the biases of individual officers, but rather administrative decisions about where to patrol and descriptions of suspects given to the police by citizens. This is a reasonable point, but one which invites further reflection on, and hopefully further research into, the roles bias plays in shaping decision-making at different levels of society. The administrators who craft policies regarding where to patrol may be biased, as may be the people who describe what suspects look like. Indeed, there is a large literature on the latter, indicating that thoughts of crime trigger thoughts of black people (e.g., Kleider-Offutt et al. 2018). Furthermore, differences in the reasons black and white Americans have encounters with the police are crucial. Black drivers, for example, are three times more likely than white drivers to be pulled over for “investigatory” stops, where police officers see something suspicious but no unambiguous violation has taken place (Epp et al. 2014). The phenomenon of being stopped for “driving while black” means that police encounters cannot only be explained in terms of administrative decisions. FBI data relatedly suggest that when initial civilian-police interactions are the least serious, racial disparities in violent outcomes are the greatest (Lind 2015).

Similarly, in the case of academia, there is ample evidence of gender bias against women contributing to harmful outcomes, but there are also many open questions about how much bias contributes to outcomes and in what specific domains bias negatively affects women. In general, research shows that women faculty are asked to do more favors from students than men (El-Alayli et al. 2018), to do more service (Guarino and Borden 2017), are invited less frequently than men to prestigious colloquia (Nittrouer et al. 2018), systematically receive worse teaching evaluations, in both audit studies (e.g., Mengel et al. 2018) and studies in which online instructors teach multiple courses under different gender identities (MacNeil et al. 2014), and more.

In the case of philosophy, Hermanson’s own analysis of the available data over roughly the last decade suggests that women are more likely than men to be hired for tenure-track jobs (Allen-Hermanson 2017). Moreover, Hermanson finds that women are hired for tenure-track jobs in philosophy with approximately half the publications than men who are hired for tenure-track jobs. These findings are consistent with some studies examining gender and hiring in STEM fields too. Williams and Ceci (2015) find that STEiM faculty prefer to hire women over men, at least when both the women and men candidates are exceptionally well-qualified. Hermanson takes these findings to speak “against the hypothesis that sexist attitudes (whether conscious or unconscious) held by philosophers are a major cause of disproportion according to gender” (Allen-Hermanson 2017).

These data are important. More research is needed in reconciling them with conflicting accounts of gender and academic hiring specifically (e.g., Jennings (2016) analyses hiring in philosophy and arrives at different conclusions from Hermanson (Allen-Hermanson 2017); the findings for gender and hiring in STEM fields are similarly mixed (e.g., see Moss-Racusin et al. 2012)). But Hermanson’s conclusion is too strong, even taking his hiring data for granted. His arguments do not rule out, or cast any doubt, on the possibility that philosophers’ sexist attitudes contribute to the decision many women undergraduates make to leave philosophy behind (Thompson et al. 2016), to the gender stereotypes about philosophy that students understand prior to starting college (Baron et al. 2015), to women philosophy professors’ decisions to submit their work to relatively lower-tier journals (Allen-Hermanson 2017), or to the countless experiences of harassment women philosophers have described in public (see https://beingawomaninphilosophy.wordpress.com/). Moreover, another possibility to take seriously, but which Hermanson dismisses, is that women have been faring better specifically in tenure-track hiring in philosophy over the past 10 years because more and more philosophers are taking gender bias seriously. Whether this is the case is, of course, an empirical question, one which scholars will hopefully investigate.