Math 143 C/E, Spring 2001
IPS Reading Discussion Questions
Chapter 3, Section 2 (pp. 237-250)
- What is bias? Give an example of how
an experiment can be biased.
- What is a placebo? How does the use of a
placebo in a blind experiment provide a
control against bias? Describe how an experiment
that is double blind provides further control
against bias.
- What is the big advantage of a controlled
experiment over an observational study? Give
an example of a study for which, despite the
advantage you just mentioned,
an experiment is inappropriate
(perhaps because no experiment would be
practical, or maybe an experiment in this
case would be unethical, etc.).
- Figure 3.3 (p. 245) diagrams the design of
experiment described in Example 3.8. How many
factors distinguish the groups, and what are they?
Describe the treatment (that is, give the level
of each factor) for each group.
- Why is it so important to randomly assign
experimental units to treatment groups? Would
the goals of randomization be realized if
replication were not taken seriously in the
design of the study (that is, say, if all
treatment groups had very few subjetcs)?
- What is block design? If you were studying the effects
of a low-fat diet on the growth of 9-year-old children
in the U. S. , what are some of the blocks you would
pay special attention to as you divided up subjects
into treatment groups? If randomization is supposed
to take care of making the two groups (the low-fat
diet children and the ``normal" diet children) roughly
equal, why might we want to bother with blocking?
- An experimental design that uses ``either two matched
individuals or the same individual to receive each of
two treatments" (Utts, Seeing through Statistics, 2nd
Ed.) is called a matched-pair design.
In which of these two categories does the matched-pairs
example (Example 3.10) on p. 247 belong?