According to an online test developed by Harvard psychologists, I have a moderate preference toward white people.
I—a liberal white male who deeply cares about racial equity—was quicker to assign positive words (e.g., joy, love, peace) to faces of white people and negative words (e.g., awful, failure, hurt) to faces of African Americans. I feel shock, shame, and disgust. The fact that 27 percent of the 732,881 people who took the test had the same result, while another 27 percent had a “strong preference” toward white people, provides some consolation but not much.
Welcome to the world of implicit bias research—an emerging field that is illuminating how real and pervasive discrimination is. The gist of implicit bias is this: we have opinions that lie beneath our consciousness, beyond our control, and that influence our decision-making process. We (inaccurately) ascribe traits to an individual on the basis of the social group we associate them with. Our decisions are often biased because they are based on these implicit associations, not evidence.
Implicit bias is a relatively new field of study, with many seminal works published in the 1990s. Project Implicit—founded in 1998 by scientists at the University of Washington, the University of Virginia, and Harvard—serves as a clearinghouse for implicit bias research and allows visitors to take various Implicit Association Tests (IATs). In addition to the race version described above, the tests can identify biases according to a variety of characteristics including: weight, ethnicity, skin tone, gender, sexuality, disability, religion, and gender. They work by assessing subconscious responses—measuring, for example, how much longer it takes you to sort positive and negative words by racial categories. An explanatory video is here.
Implicit bias research is interesting, and the tests can certainly spur conversation and controversy among liberal-minded folks, but does it have real implications for public health? Absolutely.
One place where implicit bias can directly affect health is in the doctor’s office. As described in a recent review article in the Journal of General Internal Medicine, physicians are not immune to implicit bias, and it can affect their medical decisions. For example, studies have found that physicians with preferences towards whites, as assessed through these tests, are significantly less likely to prescribe pain medication to African Americans than to whites.
Implicit bias can also affect the health of minority groups (racial, ethnic, sexual, religious, or otherwise), separate from the decision-making processes. A robust body of research has demonstrated that perceived discrimination (i.e., the feeling of being discriminated against) activates stress response systems and is associated with poor physical and mental health outcomes. Perceived discrimination is a contributor to racial disparities in health.
If implicit bias exists beneath our conscious thought, and seemingly beyond our control, can anything be done to prevent it, or at least dull its effects? I think so.
One idea, as proposed by UCLA Law Professor Jerry Kang, is to address a major source of implicit bias—media policy. As Kang describes, the Federal Communication Commission’s 2003 Media Ownership Order redefined “public interest” in a way that encouraged companies to broadcast local news. This aspect of the policy had the consequence of increasing the proportion of news coverage devoted to crime, which in urban areas ends up highlighting racial minorities and thus adding to bias against them. As Kang describes, “as we consume local news, we download a sort of Trojan Horse virus that increases our implicit bias.”
To fight this, you could write letters to the FCC and to local television news directors. You could also become aware of your own implicit biases, a major step toward not letting them affect your attitudes and decisions. For me, taking the test was enlightening and helps me to work toward aligning my decisions with the evidence and my personal values—or else the Trojan Horse will run amuck.