“How Would You Like Your Data?”
Would you like your data scrambled, over easy, or sunny side up?
Properly understanding how our data is “cooked” is key in responsible information consumption. People often read or observe data while assuming, as Boyd and Crawford state, that numbers mean objectivity (667). After all, we may think, who can argue with statistics and cold, hard numbers? And what’s more dangerous, subjective data or objective data?
However, some argue that “raw data” is equivalent to saying “jumbo shrimp” — it simply doesn’t make sense. So if numbers don’t equal neutrality and we accept that the concept of raw data is non-existent, then we are forced to choose how we want our data “cooked” (D’Ignazio and Klein 159).
As much as I would love completely objective data, we will never be able to separate naturally subjective people and their opinions from the data they curate.
“Interpretation is at the center of data analysis.” (Boyd and Crawford 668)
This changes our responsibility from finding objective data to understanding how to read past the inherent biases, and a huge part of that is understanding the context. Context means understanding everything that has gone into presenting the data, and it can be as granular as the experimental methods or as front-facing as how the data is presented. For example, even the title of a graph can change the perception and thus play a role in the “cooking” of the data (Boyd and Crawford 165).
It may be overwhelming to consider all the factors to consider when we consume data, but without this context, we risk real danger in succumbing to over-hyped interpretations or spreading dangerously incorrect information, like the misinterpreted abduction trends data in FiveThirtyEight (Boyd and Crawford 150).
Context is what limits our inferences and also what informs our interpretations. Without allowing it to influence the conclusions we draw, consuming our “cooked” data is as dangerous as consuming it raw.
