_________________________
_________________________
Sampling
and Experimental Design Issues
Work
together with one or two of your classmates to address the following situations. Write
a group response to each question and turn in one completed worksheet for
your group when you are finished. We
will discuss these in class tomorrow.
Sampling
bias
A sampling
method produces bias if it has a general tendency to result in an
overestimate (or underestimate) of a target value in the population of
interest. In each of
the following cases, decide whether there will be bias, and why.
If you suspect bias, say whether an overestimate or underestimate would
result.
1. A
college professor wants to estimate the percentage of students who cheat
on exams. So she takes
a survey of the students in her class, asking those who cheat on exams
to raise their hands.
2. The
goal of a study is to find out the average number of hours per week that
UA students work (for pay) during the academic year. The
study is carried out by obtaining a list of all current UA students, in
alphabetical order, and contacting every 20th person on the
list. (That would be
about 1000 people.)
3. Same
as #2, except they contact every 500th person on the list. (That
would be about 40 people.)
4. The
goal of a study is to estimate the percentage of students who have tattoos. The
study is done by observing every 20th student who walks past
the campus fountain, and observing whether the person has tattoos or not.
5. The
publishers of Money magazine contacted a random sample of 1000 of its subscribers
and asked them who they planned to vote for in the upcoming Presidential
election.
6. A
study is done to see if a majority of the public feels that it is OK to
make copyrighted music recordings available for free download over the
Internet. To do the
study, the Napster webpage is set up with a questionnaire where visitors
to the web page are invited to click yes or no, according to their opinion.
Experimental
Issues
7. A
person wrote a letter to Ann Landers, giving the following remedy for a
headache: stick a banana peel
on your forehead and wait for 30 minutes. Ann
Landers suggested in her column that readers try this and let her know
if it worked for them. Of those
who responded, a substantial percentage said that they tried the banana
peel method and their headaches went away. What
should we conclude about the effectiveness of this remedy? Can
you come up with a better way to conduct a study to determine the effectiveness
of this remedy?
8. In
a study conducted by the Army, soldiers applied antiperspirant on their
feet before going on a ten-mile training run wearing full gear. A
report stated that since only 25% of the soldiers developed blisters on
their feet, the Army felt that the antiperspirant treatment was effective. What
do you think about this conclusion? See
if you can come up with a better way to conduct this study.
9. Taco
Bell has fired their famous Chihuahua. The
company reasons that the ads with the Chihuahua were not effective, because
their sales have remained steady throughout the ad campaign. What
do you think about this conclusion? Can
you come up with a better way for the Taco Bell executives to decide whether
an ad campaign is effective?
10. The
manufacturers of Deep Woods Off want you to do a study to prove
that their product is more effective than Repel at reducing the
number of mosquito bites. You
have 20 volunteers who are willing to get bitten by mosquitoes for the
sake of science (and to collect a few bucks). You
also have been supplied with an enclosed mosquito-infested room. Come
up with an experimental plan. As
part of your plan, you will have to decide how to measure the effectiveness
of each brand, how the repellents will be applied and to whom, and whether
there are other factors that need to be taken into account (and how to
do this).