26.38 Power analysis

26.38.1 a priori power analysis

  • In an a priori power analysis, the type I error rate is chosen (e.g., 0.05), the power is chosen (e.g., 80%), and then the sample size needed to achieve the desired power is calculated
    • One must have an estimate of the effect size, ie the expected difference between the null and alternative hypotheses
  • The experiment is executed with that number of samples
  • You may need to do a small number of tests to estimate the difference and then re-compute the required number of specimens