What can we as researchers and methodologists do to make sure that our work is not inadvertently contributing to racial injustice? What are things we can do in our work to actively promote racial justice?
I’m asking these questions as a collaborative biostatistician and a white person living in the US. I’m hoping that people from a lot of different professional and personal backgrounds will respond. I will also post some resources and information I have recently become aware of.
I realize these questions are very broad; please feel free to create more specific topics as you see fit.
I will start things off with a recommendation based on a recent report from the Institute for Healing and Justice in Medicine (https://www.instituteforhealingandjustice.org; see “The Report”): Make sure that we are not using medical constructs that rely on the idea of “biological race” in our research.
The IHJM report, whose full title is Toward the Abolition of Biological Race in Medicine: Transforming Clinical Education, Research, and Practice, highlights two areas in which the idea of biological race has caused large problems: kidney disease and lung function. At some point in my career, I learned that the “normal” reference values for eGFR and spirometry depended on race. The IHJM report details the ways in which this information is a) biologically inaccurate and b) rooted in historical racism. The report further argues that using these race-based calculations in clinical practice can lead to Black people receiving inaccurate diagnoses and inadequate medical care. If we use race-based eGFR or spirometry calculations in research, we could be making these disparities even worse.
Thank you so much for starting this conversation, Laurie! One thing that came up in a recent conversation with Ellie Murray is that even our routine statistical methods may be exacerbating access, for example we discussed that common methods such as propensity score trimming may be systemically excluding particular populations, without explicitly stating so in the methods. I suggested using your Visual Pruner as a first step to better investigate and report on who is actually getting excluded when we do this.
Lucy, thank you for bringing that up. And it reminds me that in order to know who we might be “trimming out,” we need to have good ways of asking people to identify themselves. I just started a new thread about that: Best practices for recording race/ethnicity.
Baxter, thank you for posting those here. When I looked at that Twitter thread, another interesting link had been added, so I’ll share it here: How Not To Use Data Like a Racist (event on 24 June 2020).
Consider the sharply formulated theory Dr. Olusanya advances in this short thread:
Constrast that with the vaguer—and scientifically useless!—discourse that typically prevails under an intellectually lazy, biological framing of ‘race’. Probably the best thing to do is to recognize that failing to be actively critical of our constructs is no less sloppy than dichotomizing continuous predictors, calculating post hoc power, or any of the innumerable other bogus statistical practices we all abhor.
That said, this call to formulate more scientifically substantive theories remains anathema to much of biostatistics. Why?