When discussing the differences between parametric and non-parametric tests, it is often mentioned that non-parametric tests have lower statistical power. I would appreciate any examples of formal literature (journal articles) which support this assertion. Thanks.