What the Trump victory tells Data folks? #trumpwon
While most surveys & polls “suggested” a landslide victory for Clinton, Trump crosses the finish line with, well, to the amazement and even frustration of many, some ease.
This really set me thinking how the poll predictions went so horribly wrong!
I don’t doubt for one moment the mathematical models or the competencies of the data analysts. In fact, I do believe that poll analysts would have some amazing models built from years of election research and experience that would most likely predict, to a great deal of accuracy, what the election would turn out to be.
As a person who is obsessed with data, this story reminds me of a fundamental principle of data analysis or for that matter life; that of removing biases! I wonder what the predictions would have looked like if the sampling for the poll surveys had been done with lesser biases. In fact, one of the local English channels covering the US elections reported a few people arguably from the Clinton campaign coming out of the White House exclaiming “where are these people who are voting coming from?”
Isn’t it as simple as the equation,
Even great models fitted to samples with bias = Biased predictions = Biased decisions
Now, to think about it, I realized biases are everywhere and define the way we manage our personal and professional lives. As simple as it is and I’m sure this is no eye opener for anyone, but the question is, how open are we to question our own assumptions and how often do we actually see ourselves doing it?
One of my three commandments I follow in my professional life reads: “Challenge the Status Quo. Question everything, but assumptions & not ideas.” For me, the need for this commandment just gets reinforced and I pledge to myself to follow this more vigorously.