As a computer scientist (a guy who writes programs, really), I have very little use for probability & statistics in my line of work, which is to say that I’m not what they call a natural scientist, a scientist who studies the natural world. My use of probability is more or less nil, and statistics & instrumentation only comes into play when I’m working on performance and scalability problems – and then it’s nothing more than rudimentary use of the services of that field. I know the term p-hacking has come to the forefront in science, a term indicating scientists are manipulating the data they collected in order to find something significant to say about their latest study, but it and its relationship to statistical significance are not explicit parts of my life.
So I’m fascinated and bemused to see this interview Retraction Watch published with Professor Nicole Lazar (principally; Ron Wasserstein and Allen Schirm contributed to the answers, as co-editors of a publication on the subject) of the University of Georgia on the obsolescence of the phrase statistical significance:
So the [American Statistical Association] wants to say goodbye to “statistically significant.” Why, and why now?
In the past few years there has been a growing recognition in the scientific and statistical communities that the standard ways of performing inference are not serving us well. This manifests itself in, for instance, the perceived crisis in science (of reproducibility, of credibility); increased publicity surrounding bad practices such as p-hacking (manipulating the data until statistical significance can be achieved); and perverse incentives especially in the academy that encourage “sexy” headline-grabbing results that may not have much substance in the long run. None of this is necessarily new, and indeed there are conversations in the statistics (and other) literature going back decades calling to abandon the language of statistical significance. The tone now is different, perhaps because of the more pervasive sense that what we’ve always done isn’t working, and so the time seemed opportune to renew the call.
Dr. Lazar helped edit and publish an issue of The American Statistician devoted to this subject, but, sad to say, I shan’t try to read it because of my abysmal ignorance of the subject. I’ll be fascinated to observe, however, if this recommendation takes hold in the world of science, and how it’ll change how we do science.
