From: t-tests, non-parametric tests, and large studies—a paradox of statistical practice?
Statistical procedure
1978–1979
1989
2004–2005
t-tests∗
44%
39%
26%
Non-parametric tests‡
11%
21%
27%