When I was young, I had trouble with the intuition of statistics and econometrics. One thing that really screwed me up was the error term. That is a pretty big thing to be screwed up by. Managing it is like all of econometrics. Without the error term, all you have is accounting identities and pure theory.
My problem was I thought the world generated errors. * I was not aware that the error referred to us, not it. You can imagine how not knowing that would have been a handicap.
Another thing that bothered me was how we calculated the standard error. Why did we have to add up the squared deviations and then take the root of that? It is not like that is a super-obvious thing to do. Right? Why not just look at the mean absolute error? That is way more intuitive.
Within the normal distribution, the formula that describes the standard error maps directly to the share of the probability found within the standard error. In other words, the distribution is logically prior to the calculation of standard deviation or error. It is not like we start with “the” standard and then go looking for some place to apply it. It is the OPPOSITE. I wish they had told me that.
This came to mind when I saw this excellent tweet from this wonderfully snarky tweeter. He too must work with people who remain every bit as dumb as I was and perhaps still am.
How many times a week do you see guys report a 1 in 400 years event? Hell, things that are unlikely to have happened in the entire history of the universe are fairly common in the minds of innumerate Wall Street analysts. It seems most concentrated in currencies, for some reason. Wow that move in Swissy was 14 sigma.
* Particle physics might bail me out on this, but that is no excuse. We all know what I mean here, not about physics, but about not understanding human ignorance. Sad.