Data for whom? Being mindful of racial disparities

Posted on

As the United States makes progress towards a state of “normality,” the country has set an incredible pace for vaccine production and distribution. But what information is lost in the big picture?

On March 2, 2021, United States President Joe Biden announced that there will be enough COVID-19 vaccines for every American adult by the end of May. Naturally, it will take additional time for those doses to be administered, but it is a cause for hope. As of early March when this blog was written, the US has administered 26.56 doses per 100 people – 4th highest in the world behind Israel, the United Arab Emirates and the United Kingdom. This is commendable.

However, when looking at data in aggregate, it’s easy to overlook the disparity in who is receiving those doses. Across all states, Black and Hispanic Americans are disproportionately overrepresented in number of cases and deaths while underrepresented in vaccinations. As an example, in my state of Washington, non-White Hispanic makes up 5% of vaccinations, compared to 32% of cases and 12% of deaths; the group as a whole makes up 13% of Washington State’s population.

Screenshot of KFF’s comparison of vaccination rates, cases and death rates in the United States.

There are a variety of factors, including technical literacy, proximity to a vaccination center, language, and other barriers. Each racial group faces different challenges, and within each group, there is a diversity of stories that build into the disparity we are seeing at a population level. It’s a complex web that is still not fully mapped out.

It brings up an uncomfortable question. What’s more important, getting as many vaccine doses out as fast as possible, or making sure that vaccine distribution is equitable? The obvious answer is both. Unfortunately, the obvious answer is not what has played out so far in the COVID vaccine rollout. And unfortunately, this racial disparity is not new. How much data do we need to produce that points out disparities before we start using data to make lives better for all?

The last few years, I have had the opportunity to work at an organization focused on community engagement, particularly with immigrant and refugee communities. Coming from an academic background focused on data and statistics, it was quite an adjustment to be put in a setting where quantitative data was limited to non-existent. The reasons behind the data gap could make for another blog post all its own. Nevertheless, it has been humbling to see how policies and services can move forward more equitably after extended conversations with community members, despite the lack of ample sample sizes, tightly managed control groups, or other sampling design best practices. In the initial rise of the COVID-19 pandemic, immigrant and refugee communities immediately felt the technical and language barriers in accessing COVID-19 relief. Even before systems were put in place to address challenges like these, community-based organizations helped marginalized communities navigate these barriers to access COVID-19 resources. Though the backdrop may be different, the barriers are much the same. I am optimistic that similar culturally-relevant, community-centered approaches will become more prevalent (I found this a great read for an academic take on the topic). 

There’s a lesson to be learned here about relying too heavily on quantitative data at the expense of lived experiences. Yes, anecdotal data are unreliable in a vacuum. But listening to frontline communities and those who would be most impacted by a decision can lead to defining relevant data needs earlier, rather than after the decision has already been made.

Still, I would be interested to see how statistics can be used to guide more equitable solutions to the COVID pandemic, rather than simply affirm health disparities that have already been pervasive and studied for decades. Can data science guide the location of new vaccination centers? Or help set operating hours that cater to frontline workers’ schedules? Or direct the routes of mobile vaccination centers? How can models such as this spatiotemporal example (Grauer et al. 2020) be adapted for practical policy?

I hope that as more vaccines are rolled out we are more intentional in addressing the disparities across race and socioeconomic status that are masked by simple percentages.


All posts are written by authors in their personal capacity and in no way represent the view of the organisations, universities, governments, or agencies where they are employed or with which they are associated, or the views of the International Statistical Institute (ISI).

Leave a Reply

Your email address will not be published.