There has been substantial publicity surrounding poor statistical practice in published work, casting doubt on how many published results that appear to be statistically significant are in fact true or reproducible. This is sometimes called the statistical crisis. It is treated as a recent phenomenon, focused predominantly on traditional hypothesis testing, but that largely ignores the fact that potential problems in traditional statistics have been well known for decades and are usually connected to poor use or interpretation, not intrinsic weaknesses in the statistical techniques themselves. Furthermore, the same problems are equally possible with alternatives such as Bayesian statistics. There is a positive side to this recent publicity in that it has focused many fields of science on improving statistical practice. However, the danger is that researchers either abandon statistical testing where it is in fact appropriate, or adopt alternatives whose weaknesses they understand even less.
Defined on page 98
Used on pages 98, 99, 100, 142