Caroline Kelly Blog Post
All too often we hear appeals to data as a counter to any argument that touches upon systemic inequality or involves emotion. In the eyes of many, data has become the “new truth”, it is supposed to be divorced from emotion and be objective. Data has become the manifestation of facts. The selection by Catherin D’Igazio and Laura Klein on “Why Data Science Needs Feminism” explores the biases and entrenched systems that undermine this argument. Data purports to have a pureness, it is simply the collection of quantifiable data points but its current and historical use belies this assertion.
It is often said that what is measured is managed. One of the key arguments made in this selection is the inherent biases that can impact this information. Everything from collection methods to presentation and even intended audience can impact its seemingly unbiased representation. Data can be heavily flawed at two critical points, at the preceding point where the intended use is being conceived, and at the collection point where key factors can be ignored. A misstep at any of these points can radically alter the meaning of data.
Data is all too often presented in an ipso facto manner, and the argument is made that it really can not be. The cited story of Christine Darden, a women data scientist in the early days of NASA is telling of this very clear flaw. When she deigned to ask why there were no women at the upper echelons of leadership the response was a simple: “because we thought they were content where they were”. Any data that would attempt to measure gender representation at such a company would ignore the underlying decisions that impacted it. This simple example demonstrates how data alone cannot be the answer to any question without being placed in context. While this is a particularly telling story, how many companies and organizations never have a Christine Darden to question why and incite change?
The concept of intersectionality is a fascinating one, all the more so because it is a concept that would be difficult to represent with data. Intersectionality thrives in the grey space, the influence of various aspects of a person’s social, racial, and sexual identity are incredibly difficult to quantify. How much categorization would be required? One of the interesting aspects that is suggested is how some minority traits could serve as an advantage when presented against an “otherness” that is more apparent. How would someone classify such a person? Intersectionality is a very real concept, one that suggests an individualism that contradicts raw data collection. And the fact is intersectionality as a concept is not some rare thing but readily present in all people.
The guidelines presented as representative of Data Feminism are a great framework to analyze all data. While the use of the term Data Feminism does play heavily on the binary they purport to question, overall the principles therein are a great guide for anyone using data.
