ANOVA

ELEMENTS

treatment - level of factor
factor - variable of interest
experimental units - object of interest

ASSUMPTIONS

• Variance of DV are all the same for each popu.

• All DV are normally distributed

COMPLETELY RANDOMIZED DESIGN

1. between treatment - MSTR = SST/K-1

2. within treatment - MSE = MSE/n(T)-k

• F = MSTR/MSE

  • If mean of k populations do no equate to F, MSTR overestimates the variance

  • Reject null
    

HYPOTHESES

Null - all popu means are equal
Alternative - not all

• IF null is rejected THEN there is sufficient evidence not all are the same
  

RANDOMIZED BLOCK DESIGN

• experimental units - object of interest

• If the experimental units are heterogeneous, blocking can be used to form homogeneous groups, resulting in a randomized block design

FACTORIAL DESIGN

• draw conclusions about more than
  one variable or factor

• experimental conditions include
  all possible combinations of the factors.