Fifty patients suffering from knee pain have volunteered to par- ticipate in an experiment to examine the effects of pain relief medication.

Question

Fifty patients suffering from knee pain have volunteered to participate in an experiment to examine the effects of pain relief medication. Half will be assigned to take 400 mg of the pain reliever every six hours, and the other half to take 200 mg every 3 hours. The patients will report their pain levels on a 1-10 scale at the beginning of the experiment and again after 8 days on the medication.

a) Define the experimental units and the response variable. b) What are the treatments?

c) Describe an appropriate method to assign the volunteers to treatments.

d) An alternate design could include a control group who take a placebo. Describe an advantage of adding a control group.

Answer 1

As always, start with definitions:

1. from standard statistics texts experimental unit : In statistics, a unit is one member of a set of entities being studied. in this case the experimental unit is a single one of the 50 patients in the study
2. The dependent variable is the variable whose variation is being studied and depending on the text and situation can be called "response variable", "regressand", "criterion", "predicted variable", "measured variable", "explained variable", "experimental variable", "responding variable", "outcome variable", "output variable", "target" or "label". So our dependent or 'response' variable here is the difference in relative pain measures made before and after the treatment. (since this is a repeated measures experiment)
3. Treatments. In an experiment, the factor or treatment (the independent variable) is an explanatory variable manipulated by the experimenter. The manipulation is considered the 'treatment. In this case the 'treatment' is the administration of a medication at two different levels and time schedules. Treatment A is 400mg every 6 hours for 8 days. Treatment B is 200mg every 3 hours for 8 days.
4. In simple repeated measures experiments such as this, simple random selection should be used. To make calculations easier the selection, while random, should result in exactly the same number of experimental units being assigned to one of two treatments. A simple yet valid method of doing this would be to sort the list of volunteers in a random order and than assign treatment A to the first half and treatment B to the second half. Since the initial order was random, the selection to treatments will also be random.

Placebo Studies:

In placebo studies a portion of the subjects are given no treatment or a sham treatment (placebo) instead of a treatment without the subject knowing. This attempts to prove that any effects seen in the trials are due to real treatments and not just a random event or the psychological healing effect where people will feel better if they expect to feel better (placebo effect). These 'placebo' groups are often called control groups because they provide a control against external factors, including psychological effects.

Deliberately using placebo testing on human subjects, especially double or single blind tests where the subjects believe they are getting a true treatment carries ethical and moral questions. Is it ethical to withhold treatment where one is available to a person who is sick and who believes they are being treated, even if the illness is not deadly? Is it moral to do even if the patient knows of the chance that they may be randomly assigned to a control group? Is it ethical or moral to do in cases of severe illnesses even when the patient knows they will be assigned to a control group and not treated? These are constant considerations with human trials.