Last week I rode on a shuttle from Denver airport to Boulder sitting next to a PhD candidate in statistics. What a pleasure.
What I was delighted to find, was in the 50+ years since I taught statistics as a TA at San Francisco State College, there have been no significant improvements in statistics.
My shuttle friend described her various doctoral projects and all of them were easily understood by me with my lifetime of work in the field.
There was one particular statistical term that was new in the past couple of decades: bootstrapping. You can look this up in Wikipedia.
Bootstrapping did not come as a surprise to me in any way. One of the perennial problems of statistics has been to fill in the gaps in a typical database in such a way that you do not distort the outcome. Over the years, many techniques have been developed for filling in empty data points.
Bootstrapping is a logical and mathematically defensible method. I have been using many similar approaches to my own data without having the mathematical proof built into bootstrapping. (Aside: the proof uses the Bayes Theorem which I consider a weak theorem.)
My great delight was learning that statistics remains pretty much the way I learned it; the innovations have been pretty much the simple solutions I evolved on my own.
I wonder how many other powerful fields of study have seen so little evolution.