
Chapter 12
438
12.6.1 Null and Alternative Hypotheses in a Factorial Design
A two-way analysis of variance is used to test the hypothesis of a factorial design having two
factors. In light of this, the null and alternative hypotheses for the treatment effect can be stated
as below:
Row effect: H
0
: All the row means are equal.
H
1
: All the row means are not equal.
Column effect: H
0
: All the column means are equal.
H
1
: All the column means are not equal.
Interaction effect: H
0
: Interaction effects are zero.
H
1
: Interaction effect is not zero (present).
12.6.2 Formulas for Calculating SST (Total Sum of Squares) and Mean Squares
in a Factorial ...