The new study inquires whether “implicit bias” is associated with “disparities in breast reconstruction rates, complications, or cost” among women who received mastectomies as a treatment for breast cancer.
Implicit bias refers to supposedly unconscious prejudice that affects attitudes and behaviors toward others. The “science” behind it is tenuous. The test most frequently used to measure implicit bias (including in this new study) is called the Implicit Association Test (IAT). The IAT demonstrates poor levels of reliability, meaning that outcomes tend to be volatile when the same person takes the test multiple times. The IAT also doesn’t appear to predict “biased behavior,” which ought to invite healthy skepticism about whether it measures what it purports to measure.
For woke scholars, however, it is a given that implicit bias is not only real but measurable and profoundly determinative. In this case, the “researchers” task readers to imagine that it plays a role in how doctors treat women with breast cancer and the costs associated with that care.
Precisely why “implicit bias” would lead to differences in cost or the incidence of breast reconstruction is never explained. Surgical reconstruction is a decision made by patients, so their theory compels the belief that patient agency is somehow shaped by the forces of racism.
The researchers assert that average IAT scores across regions are a functional measure of implicit bias for surgeons in that region. To test their hypothesis, they observe the degree to which implicit bias correlates with racial differences in breast reconstruction rates, complications, and costs across these Census regions.
The theory behind the study is flimsy, but the technical execution is worse. Among the myriad problems with technical execution: Featuring US Census regions as the unit of analysis means that their sample size is 9 units. It’s simply too little data to draw any meaningful inference from quantitative analysis.
Using regions as the unit of analysis also means that surgeon “implicit bias” is measured with enormous imprecision. Even if one accepts that the IAT represents a quality measure of implicit bias, the assumption that implicit bias among surgeons varies in the same way that it does among the general population in these regions represents a methodological leap of faith.
Moreover, it turns out that there is almost no variation in IAT score by Census region. Whereas results in IAT score can range from -2 to 2, scores among the 9 regions range from 0.29 to 0.33.
The fact that their study consists of 9 observations and an outcome with almost no variation means that it was all but predetermined that the researchers would not observe statistically significant results, and indeed they do not. Nevertheless, the researchers conclude that “efforts to reduce the observed inequities” – which they did not observe – “should remain a national priority… efforts from individual institutions and national surgical organizations are needed to provide culturally competent, evidence-based care to individuals of all racial and ethnic backgrounds.”
A great deal of postmodern woke “scholarship” features researchers manipulating data to arrive at a preferred outcome or overstating findings. The authors of this new study appear to lack the technical skill needed to do that. Instead, when the analysis does not support their conclusion, they simply provide the same canned policy recommendations that they would have provided if their analysis did support their hypothesis. In other words, JAMA has become so tendentious that its debased standards now apparently allow researchers to perform dog and pony analysis instead of empirical sleight of hand.
Ultimately there is only one lesson to be drawn from the new study: Go woke, go intellectually broke.
Ian Kingsbury is the Director of Research for Do No Harm.
Get up to speed with the threats facing healthcare – and how we’re protecting patients and physicians.
"*" indicates required fields