Abstract:
|
NIMH and FDA clinical trials and other mandated studies often limit the sampling frame to people with one clear diagnosis and no comorbidities. This tends to exclude from analysis people who are poorer and nonwhite. In addition, people with serious mental illness are often shielded by their families in various ways, if the family has resources; if not, the first diagnosis usually occurs in the criminal justice system. Here, too, ethnic bias is introduced, mostly unintentionally, into the characterization of the disorder(s). How can statisticians take a more prominent role in recognizing and responding to these biases? (This topic was partially inspired by discussions within the ASA Anti-Racism Task Force.)
|