In their article For want of a nail, Howard Wainer and Richard Fienberg of the National Board of Medical Examiners explore the issue of test length through the lens of reliability – a standardised measure for the stability and consistency of test scores. The longer the test, the more reliable it is. But reliability gains become ever more marginal as test lengths increase. Can we, and should we, reduce the length of exams and free up some time for individuals to devote to other beneficial pursuits?
Elsewhere this issue you can read How machines learned to think statistically, our special report on artificial intelligence and the contribution statistics has made to advances in the field. It's a timely piece, coming hot on the heels of the Future of Life Institute's open letter warning of the potential dangers of unchecked research into AI. The special report includes a Q&A with Zoubin Ghahramani, the Cambridge professor who's trying to build an AI for data science, called The Automatic Statistician.
The full list of contents is here, with links through to individual articles. All articles are free to read to mark the start of our 12th year in print. And, in case you missed the news, we recently opened up the Significance archive to make volumes 1-10 free to read also. Enjoy!