Linear Statistical Models

Expected Means Squares Further Revealed

A Confession

  • The truth is that I simplified expected mean squares so that we could focus on making correctly formed F-ratios.
  • The fact of the matter is, that the notation for expected means squares changes depending upon whether a variable is a fixed variable or whether it is a random variable.
  • The notational system that was used, so far, in class is correct for random variables but not for fixed variables.
  • Fixed variables use the treatment effect squared divided by degrees of freedom.

    A Completely Randomized Design with Variable A Random

    A Completely Randomized Design with Variable A Fixed

    A Completely Randomized Factorial Design with Variables A & B Random

    A Completely Randomized Factorial Design with Variables A & B Fixed

    Linear Statistical Models Course

    Phil Ender, 27May98