“Natural” Hazards": New CSSS and Sociology professor Sasha Johfre brings research on unquestioned biases to UW
Why do people think that just because something is “natural,” it is somehow better than things that are not natural?
When people talk about “age,” do they mean chronological age, or how old someone appears or acts, or where someone is in their career?
And when social scientists present explanatory models involving different social categories, why is the “reference category” to which others are compared so often white and/or male? Does it have to be?
These are some of the questions that Sasha Shen Johfre worked on while a graduate student in Sociology at Stanford. “The types of questions I’m interested in are all about the ways we know things. I like to interrogate things we take for granted, investigate how are they come to be, and think about their impact on different kinds of people,” says Johfre, who will join the UW faculty this summer as an Assistant Professor in Sociology and the Center for Statistics and the Social Sciences (CSSS).
“You think it’s just the way the world is, but actually these are things that people are creating all the time.”
While some unexamined ideas about social life are innocuous, others can have serious consequences, Johfre says.
A good example is what Johfre calls the “nature card,” or the conventional wisdom that, all else being equal, things that are “natural” are healthier or otherwise preferable to things that seem in some way “unnatural.” While the belief that there is a clear boundary between natural and unnatural things -- and that natural things are better --may inspire some to avoid overprocessed foods or limit use of garden pesticides, Johfre argues that it can also produce social harms, as when women are expected to take care of children and households because that is their “natural” role, or when queer people are persecuted because their ways of being are considered “unnatural.” Johfre hopes that by pointing out that the “natural is better” idea is a cultural logic that people use when making various kinds of decisions, social scientists will begin to study how this belief in the superiority of naturalness could be used to promote social inequalities.
Another example of a taken for granted idea that Johfre has unpacked is the idea that age is simply the number of years one has lived. This view overlooks many of the ways in which people assess age in each other: personal appearance; health status; life experience; position on a career ladder; and role within a family or organization are all other dimensions of the concept of “age” that people pay attention to. And these social dimensions of age matter just as much as chronological age: If someone is socially classified as “old” they may be accorded more--or less--status, resulting in more respect in some communities, and fewer opportunities in other contexts.
Even how social scientists set up their statistical models can involve unexamined biases. For example, a multivariate regression using data from the sinking of the Titanic might aim to explain how gender, class of ticket, age, and parental status affected chances of surviving the disaster.
While such a framework might seem unproblematic, common models using categorical variables require one value to stand as the “reference category,” or the baseline to which all the others are compared. With the Titanic data, for example, the reference category for class of ticket might be set as a first-class ticket holder; for gender it might be ‘male.’ What this means, then, is that interpreting the effects must be done relative to this reference group (first class, male): compared to those in first class, having a second-class ticket meant a reduction of survival probability by a certain amount, and compared to men, females’ survival probability was that much higher, and so on.
In a recent paper, Johfre reviewed publications in two prominent social science journals, and found that three-quarters of studies with race or gender used a dominant group (white, male) as the reference category. Johfre argues that the use of dominant social groups as reference categories in social science research reinforces the idea that these social categories represent the “normal” humans, and that other categories are somehow deviant forms.
Not content to merely identify the biases that lurk in common assumptions, Johfre also works to improve social science methods by developing better analytical tools.
For example, Johfre has developed a typology that measures age in terms of generation, physical markers, responsibilities and more, so that social scientists can capture the multidimensional nature of age.
As for the problem of the reference category in social science models, Johfre highlights several ways that researchers can remove built-in biases toward some social groups as the “norm” while also improving the legibility of their results. One option is to calculate the overall average of the outcome variable and use that as a baseline. For the Titanic data, that would mean using the average probability of survival for all passengers as the baseline and evaluating the effects of class, gender, and age on survival probabilities compared to that.
Another approach involves identifying the category with the lowest overall probability, which means that all compared effects appear in positive numbers. For the Titanic data, that would mean setting the male holder of a second-class ticket as the baseline, as that was the demographic with the lowest survival rate. Every other variable, such as femaleness, or other ticket class, would only improve the probability of survival from there, making the model very easy to quickly grasp.
Johfre is looking forward to continuing her work on these issues when she arrives at the University of Washington this summer.
“I’m excited to say more about how we can improve the specific methods we are using, and what implications that has for how we know things and communicate and learn about and engage with the social,” she said. “I’m also looking forward to collaborating on that. Having people from different backgrounds with different questions and how to think about them means there are exciting opportunities to innovate.”