Riding the NYC subway without something for my brain to gnaw at or without something interesting to freely daydream about (I have an insufferably short attention span) during the hot summer mornings can quickly and painfully become the most unimaginably insipid moments. Lately among other things, Charles Wheelan’s Naked Statistics became a nice distraction from being squeezed like a can of sardines on the 7-line into Manhattan.
I wanted to mention it here because while it’s not a powerfully insightful book (and so I take reservations in writing a full non-blog commentary), if you’ve taken a stats class and are familiar with things like measures of central tendency and regression, the book is still richly chock full of a lot of examples where probability and stats is incredibly powerful but unfortunately grossly underestimated. My particular favorite case in point is the Value at Risk (VaR, not to be confused the with abbreviation for variance, ‘var’) model that had exploded during the ’08 crisis – with many institutions relying heavily on the mathematical sophistication of VaR, very few people actually realized the reality of the underlying probabilities of loss when everything in the markets spirals downward at once. Many other examples abound in Wheelan’s work, from the famous Monty Hall problem (take Stat 110 for Professor Blitzstein’s lecture on the importance of conditioning) to general studies where the setups may not allow us to truly answer the question of interest. For example, in Wheelan’s bran muffin example, understanding the association between bran muffin consumption and reduced incidence of colon cancer involves appreciating the fact that a population that would eat bran muffins is very likely different from a population that would not normally eat bran muffins (the former may be more health conscious in other unconsidered ways that reduce colon cancer risk) and this masks the relationship we are attempting to study.
And if you’re reading this blog, you might want to carefully think about the intuition for wherever critical randomness comes into your own summer work this summer, whether its evaluating the effectiveness of policies or seeing whether your lab experiments have serious enough p-values to take to paper.
… And last but not least, since I’ve all but abandoned my brief stint at photo-blogging, here is a snapshot of what was once my daily subway trek in the grimy underworld of urban travel, the humble backdrop for my pre-market reveries.