Читать книгу Evidence-Based Statistics - Peter M. B. Cahusac - Страница 26

1 1 Taper and Lele (p. 545) emphasis added 'The evidential approach is alone … in having its measure of evidence invariant to intent, belief, and time of hypothesis formulation. The evidence is the evidence. Both belief and error probabilities have been separated from evidence. This is not to say that belief and error probabilities are unimportant in making inferences, but only that belief, error probabilities, and evidence can be most effectively used for inference if they are not conflated' [1].

2 2 The counternull is the value on the other side of the sample mean that is equidistant from the sample mean as the null is from the sample mean. See Section 8.7.

3 3 Often the secondary hypothesis will be the null hypothesis.

4 4 There is an adjustment to Hedges' statistic for small samples, that means multiplying the value by (N − 3)/(N − 2.25), see p. 244 in [7].