Statistical power

From WikEM
Jump to: navigation, search


  • null hypothesis: there is no relationship between two measured phenomena
    • often denoted H0 (read “H-nought”, "H-null", or "H-zero")
  • alternative hypothesis: there is a significant relationship between the two measured phenomena
    • often denoted H1

Common Statistical Terms


  • α: probability of type I error
  • β: probability of type II error
  • χ²: ≥ 2 percentages or proportions of categorical outcomes


  • ANOVA: compares means of ≥ 3 groups
  • Confidence interval: range in which real mean expected to fall
  • Negative skew: mean < median < mode
  • Positive skew: mean > median > mode
  • Power: probability of correctly rejecting H0
    • to increase power increase sample size (most important), and/or increase expected effect size, and/or increase precision of measurement
  • Standard deviation: variability from mean
  • Standard error of the mean: variability between sample mean and true mean
  • t test: compares means of 2 groups
  • Type I error: H0 incorrectly rejected
  • Type II error: H0 incorrectly accepted

Also See