Editors dedicated a recent issue of Science, Technology, and Human Values to the topic of ‘ghost variables.’ Articles focused on this topic through the lens of race. The brief idea is that certain features don’t appear directly within the scientific data. However, they ‘haunt’ the data, existing as invisible traces. I think we can put this idea to good use.
Let’s look at a topic I wrote about awhile back: standardized testing. Could there be ghost variables at work? Standard disclaimer: as some of you know, I work in the testing industry. My opinion – as always – is my own and only my own.
Ghost Variables
I’ve been involved with the field of science studies for about a decade. The field focuses on many kinds of social implications of science. Applied to race, the thought is that race formed the intellectual basis for many ordinary scientific concepts. Much of modern science does not explicitly concern race. But many earlier ideas – from eugenics to early models of evolution – carried regressive ideas about race.
Some science studies scholars argue that modern science eliminated these racist ideas only at the surface. They don’t advocate for eugenics or use explicitly racist ideas. However, they carry with them the baggage of these racist ideas. What they do, therefore, still counts as racialized science. And so, the thought is that race haunts much of what happens in science today.
How does it do this? Race might appear in many ways in modern science via ghost variables. A scientific study might not discuss race, but it might discuss hair. As many readers know, U.S. society still racializes hair. Two authors in the ghost variables issue discuss, for another example, the racialization of hair and hirsutism in the context of Polycystic Ovary Syndrome. It’s not about race, at least not explicitly. But U.S. society so thoroughly entwines these things with race that, ultimately, it is about race.
Standardized Testing
So, what does any of this have to do with testing? For one, testing is a science. Or, at least, it incorporates scientific data and process. Companies use social science and quantitative methods in the process of building tests and reporting accurate scores. I discussed some of these issues in my previous post on the topic. And so, it’s possible the work contains ghost variables.
But, more to the point, I discussed Ibram X. Kendi‘s objections to standardized testing in that post. Kendi cited the racist origins of the industry, particularly sympathies with the eugenics movement. In response, I pointed out that Kendi’s argument commits a rather straightforward genetic fallacy. That is to say, Kendi cites a racist history, but he doesn’t give us any reason to believe the history persists to the present day. Indeed, testing companies go to great lengths to make sure it doesn’t persist.
However, could Kendi appeal to ghost variables to save the argument? Could he argue that race and racism persist in the industry through ghost variables?
Maybe. The persistence of race in this tacit form would explain why the presence of race isn’t obvious in the data. But I think defenders of that line of reasoning have their work cut out for them. It’s one thing to point to the possibility. But it’s quite another to specify exactly what these variables are and how they haunt the data. However, done well, defenders of Kendi’s line of reasoning could use these ideas to defend themselves from the charge of a genetic fallacy.
