Exercises 26.17, 26.18 and 26.19 of PSLS 3e
-------------------------------------------

26.17:
------
The correct option is (a). The two other options are pretty far fetched
and probably just serve to make the point that the "two" in two-way
ANOVA has nothing to do with the number of populations. Indeed, it
refers to the number of factors.


26.18:
------
The correct option is (c). In (a) and (b) it is not even obvious what is
meant by "values" of a factor. 


26.19:
------
The intended correct option is (a), but one can definitely make up 
scenarios where the main effects should effectively be ignored (b). If the
interaction is very strong there may no value in looking at the main
effects at all. For example, if the interaction is clearly significant
and both main effects are non-significant (yes, that is possible!), the
interaction needs to be described whereas there is little to say about
the main effects. 

The situation when one should describe the two main effects separately, 
option (c), is when there is *no* interaction.