26.39 Power analysis

26.39.1 Post-hoc power analysis

  • Post-hoc power analysis has been common in practice (for “insignificant” findings)
  • In post hoc power analysis, the observed samples (the effect size and variance) are examined to “estimate” the “original power”.
    • i.e., “How insufficient was the sample size.”
  • There has been recent debate about usefulness
  • Based on my readings in the math literature, this activity should be discouraged as it doesn’t add value relative to \(p\)-value
    • just be familiar with the concept
  • For a journal article, if asked to do this, recruit me to help rebutt