Homework 1 Solutions (*'d problems)
#1 A dermatologist wants to test which of two products is best for
removing a type of blemish that appears on the face. He instructs the receptionist
that she is to assign each patient diagnosed with this condition with either
Treatment A or Treatment B, and there should be roughly the same number of
people in each group. The dermatologist assigns each patient a severity
score, and each patient is asked to return 4 weeks later. At the followup
visit, the dermatologist (who does not know which treatment the patient has
received) assigns a new severity score. There were 43 patients enrolled
in the study. The results were that, on average, the patients in Treatment
A improved more than those receiving Treatment B.
a) Was this is a controlled experiment or an observational
study, and why?
b) As you might have guessed, with 43 patients enrolled
each treatment group did not have the same number of patients. Is this
a problem? Why or why not?
c) Note there was no placebo. What information
would be obtained by including a third group of patients who reeived a placebo?
Why should or should not this be done? (Assume a placebo is a
completely inactive product.)
d) Suppose the dermatologist has not yet collected
this data, but has proposed this to you as a design for this study. He has
asked whether anything should be changed. What advice would you give him?
a) It was a controlled experiment because the researchers (more precisely
the receptionist) chose who received which treatment. For full credit,
you must both get the correct answer and explain why and your explanation
should use the definition of a controlled experiment (which means it should
refer to the fact that the experimenters were the ones who assigned treatments
-- not the patients or other outside forces.)
b) No, it's not a problem. The groups do not have to have exactly equal
size. We hope that they are balanced with respect to potential confounding
factors, but "balanced" doesn't mean exactly equal.
c) As is, the study allows us to make comparisons between treatments A and
B only. So we can conclude, potentially, that one is better than the
other, or that both are about the same. Including a third group that
received a placebo would allow us to compare the treatments to the "no treatment",
and would let us determine if either treatment was any good at all. In
other words, perhaps neither treatment a or b is worth the expense. To
decide whether this is necessary we'd have to know more about what these
treatments are. Perhaps one of them is a well-known and accepted treatment
that has already been compared to placebo groups before, and so there's no
need to test it again, and we only want to know if the "new" treatment does
better than the "standard". For full credit, your answer should explain
what additional information we could potentially learn by including a placebo.
d) This question is inviting you to go through the checklist of "Things that
Make A Good Study." There's one glaring omission that should be included
in your answer to receive full credit: the assignment was not random.
So the dermatologist could do a better study by randomly assigning
patients to Treatment A or B, rather than let his assistant do it. As
is, the experiment is potentially biased. Suppose the receptionist
doesn't have much faith in Treatment B. She might assign those who
she thinks are not too serious to Treatment B and save Treatment A for those
"who really need it." Then the patients in one group would end up being
more easily cured than those in the other.
p33 #8
As part of a study reported by Kaufmann, 10 subjects with a history of developing
headaches after drinking red wine each consumed 90 mL of a Californian Cabernet
Sauvignon, and each developed a headache. One week later they each consumed
180 mL of the same type of wine, but this time one hour after taking an aspirin-like
drug. Under these conditions only 2 of the 10 subjects developed a
headache.
a) Is this an observational study or an experiment?
Experiment: the researchers controlled who got which
treatment. (The treatments here are "no aspirin" and "aspiring">
Note that each subject received both treatments.)
b) What is the population of interest?
Those who get headaches from drinking red wine.
c) The study is investigating whether taking an aspirin-like drug after consuming
red wine can prevent headaches in those susceptible to the headaches. Frankly,
I myself don't know what more this problem is asking for.
d) Again, go through the checklist. First, there was no randomized
assignment. Of course, in this study everyone got both treatments,
but the order in which they got the treatments could have been randomly assigned.
Perhaps if you drink two glasses of red wine within one
week you don't get a headache on the second time. Ideally, a randomized
mechanism would have some people take aspirin the first time but not the
second, and the others nothing the first time and aspirin the second. There
was no placebo, and so we don't know if the subjects would have not got a
headache even without aspirin. The lack of a placebo also means there was
no blind...perhaps the subjects didn't report a headache because they knew
they weren't supposed to get one, and perhaps the researchers helped mitigate
headaches because they knew this was the intended result. Finally,
it might be nice to try other types of wines.
e) I'd want to do a stronger study, but also try more wines to see if the
treatment works only with certain types of wines.