Anaes SAQs · Measurement & monitoring physics

Hypothesis testing, p-values and confidence intervals

Primary — the null (H0, no difference) assumed true until rejected; the alternative (H1, a difference exists); Type I error (alpha, false positive, conventionally 0.05); Type II error (beta, false negative, conventionally 0.20); power equals 1 minus beta (conventionally 0.80); p-value is the probability of the data given the null; 95 percent CI is the range containing the true parameter in 95 percent of repeats and is not significant if it crosses 1.0 (ratio) or 0 (difference); NNT equals 1 divided by the absolute risk reduction.