Discussion of k=2...
The model:
Thus we have added an error term, denoted e,
representing experimental error.
If we hold the values of factors A and B constant, and repeat
the experiment r times, we obtain a set of values for
response y. Statistical analysis of this set of values will
yields:
- what fraction of variation in the set of (2^k)r
experiments is due to experimental error
- an estimate of the interval containing the true mean
of y, or confidence interval