Introduction to Analysis of Variance

• Logic of F ratio

  • Variance among subjects within the various treatment groups provides an estimate of random or chance variation among subjects treated alike.
  • Sampling error
    • Subjects within S(A)
  • Variance of Means about Grand Mean, provides an estimate of Variance due to treatment + random Variation
    • A

• If treatment had no effect this will be just another measure of random variation (sampling error)
• If treatment had an effect, this will be larger than sampling error alone.

• F ratio is A/S(A)
  • If we have no treatment effect, ratio would be 1

  Previous slide

  Next slide

  Back to first slide

  View graphic version