I’ve often wondered why we still use the QWERTY keyboard, which dates from 1868. It is terribly inefficient. More of the common letters are on the left hand, and valuable central real estate is wasted on less-common consonants like F, J, V and B. The right pinky is bizarrely wasted on a semicolon, of all things. There is a story that the inventor chose this layout in order to slow down typists so that the keys would not get jammed, but that was actually just a joke that sounded plausible enough to be true. All in all the layout atrocious, even if as a lefty it is nice to have at least one thing in the world that works slightly in my favor.
I thought for sure the generation that grew up typing on phones and tablets would come up with something better than the 150 year old approach. While there were some scattered attempts at this, none caught on. And so here I am in 2019, typing this message on a device that would be partially recognizable to my great-great-grandparents.
So it is with statistics. Many of the peculiar, non-intuitive aspects of the field are embedded in decisions that were made in the first half of the 20th century. Just as the development of probability theory was largely tied to gambling, the development of statistical theory was largely tied to industrial manufacturing. I recently discarded my undergraduate statistics textbook because the binding was coming apart, but it was full of examples like this:
A sample of 1,000 3-inch bolts found that 6 of them were less than 2.95 inches in length and thus defective. What is the probability that at least 1% of the entire production line of bolts are defective?
If you are interested in manufacturing bolts, then you start framing every problem as if it was a special case of manufacturing bolts. Most of public health does work this way. The subjects are people, who tend not to behave like items on assembly lines. The statistics still work, to a point, but it is like typing on a QWERTY keyboard, inefficient and nonintuitive. At times you have to imagine the existence of time travel and parallel universes. As much as possible, in my teaching I try to steer clear of these conundrums, using Harvey Motulksy’s Intuitive Statistics as my well-thumbed guide.
Pumphandle LLC 