I just found this not too old op-ed gem by David Brooks for the NY Times, “What Data Can’t Do”. The best quote is this:
‘Raw Data’ Is an Oxymoron’ One of the points was that data is never raw; it’s always structured according to somebody’s predispositions and values. The end result looks disinterested, but in reality, there are value choices all the way through, from construction to interpretation.
I get giddy about data. My friends do too. That said, no matter how excited we get about what we can do with it and no matter how the strength of our good intentions, it is just plain difficult to fully analyze and account for your own biases. It’s hard to step outside and say to yourself:
- Hmm, where am I structuring towards the initial solution I had in my head? Yes, the data speaks to me, but life is not a movie and I’m not John Nash and I have neither schizophrenia or a beautiful mind.
- Not all correlations are red herrings.
- Just because it’s ridiculously obvious means that it’s wrong (i.e., sometimes it is that simple and don’t doubt yourself).
- Even the social can be quantified (i.e., the dating charts, anyone, OKCupid?)
- But the quantification of the social is fundamentally suspect no matter how elegant the final value.
- “Edges form definitions” – therefore, the structuring of a problem question already operates in the sphere of limiting context; just because you can now get an answer doesn’t mean it can be perfectly extrapolated to the larger issue at hand
How do you tackle your biases and your challenges? I’m curious to hear any tips and frames that can adjust for the “adjustments” we make to interact with data.