At this point in our lives, students and researchers may understand, on some level, to be cautious of the information they find. We check our sources to see if their credible, we look at the historical context of the documents handed to us. But there’s still this idea that technology and numbers are, generally, trust worthy.

However, the article by Catherine D’Ignazio and Lauren Klein brings to attention the idea that human factors cannot be removed from the gathering and presentation of that data. Computers may not be capable of internal bias all on their own, but programmers and researchers very much are. Even without lying about the numbers, researcher’s can still often misinterpret the data. In the case of the types of people who are a high-risk for a house loan, the data actually reflects a systemic issue of discrimination and oppression. When a study fails to actually measure what it claims to be measuring, science researchers call it “invalid.”

Impact

By viewing data in a vacuum and ignoring the context in which it was gathered, resarchers allow harmful systemic powers to remain in play. Facts and ideas that support a dominant group reinforce their standing, while allowing minority issues to go unaddressed. Data holds bias. Where was the data gathered? Who gathered it? What was the population like?

This is purpose a flaw with Random Sampling in situations such as the home loan risk assessment. Even with a variety of types of people represented fairly, the systemic issues that Black people face unfairly put them in a High Risk demographic. Even with the assessments, seemingly, attempting to be more about context than personal identity. The data ends up being biased, even if the numbers haven’t technically lied.

Solution to Data Bias

D’Ignazio and Klein claim that researchers and data collectors should bring attention to topics that challenge regimes of power. While neutral language has its time and place, it also works to maintain a harmful status quo of bias and power inequality. Analysts should take a proactive approach to giving voice to minority communities and the damage done to them. When doing research, remaining neutral can be more harmful than helpful. Ironically, one of these neutral words that “work to secure power” is, in fact, “bias.” However, their suggested expansion on the term is “oppression”, which isn’t exactly what this article is addressing. Although it is a negative impact of Data Bias, as seen in the previous example.

What’s more to understand is that the authors are not claiming to replace this language all together. Rather to expand upon how researchers and analysts consider data. They need to look at and acknowledge bias, historical context, and who benefits from an unfaithful interpretation and portrayal of Data.