Some other extensions are possible with the designs discussed in this

section, Testing for effect can be extended in time because the experimenter

is often interested in long term results. However it is hard to control
for history. To be able to generalize to other populations X has to be
varied. This is juxtaposed to the demand for experimental control. To
find out which aspects are most essential you must either make the control
treatment closer to that of the experimental group or try to reduce the
treatment of the experimental group to the essential elements. Finally

since most meaSures are imperfect, multiple 0"'s are desirable,

Possible Guestions for Discussions

1. How important is external validity? Is repeatability more important?
(pe. 18).

2. If the subjects are in a school setting Campbell and Stanley recommend
using the same setting and teachers(to reduce reactive arrangements
effect). What problems does this create? Is their recommendation a

" valid one? ;