Why Medical Experiments May Use a Placebo
In Control Groups and Comparison Groups, you learned that a control group provides a baseline for comparing outcomes under different conditions. In a medical experiment, that baseline may be a placebo: a treatment designed to resemble the treatment being studied but without its active ingredient for the condition.
A placebo can help researchers ask whether an active treatment produces a different outcome from receiving a similar-looking treatment without that active ingredient. This comparison matters because people may feel or report a change after starting treatment for reasons besides the treatment’s specific medical action. Their expectations, the attention they receive, and the natural course of an illness can all be relevant.
For example, a placebo pill might look and taste like the study medication and be taken on the same schedule, but lack that medication’s active ingredient. A placebo is not necessarily literally “nothing”: it may contain inactive ingredients needed to make the pill, and participants still have an experience of taking a treatment and interacting with study staff.
The placebo effect refers to a change in a person’s symptoms or other response associated with expectations or the treatment context, rather than the active ingredient being tested. Expectations can influence how someone experiences or reports a symptom. They may also be associated with changes in some bodily responses. For instance, a person who expects pain relief may experience pain differently. This does not mean that belief alone reliably cures a disease or replaces effective medical care.
Placebo Response and Placebo Effect Are Not the Same Claim
Suppose participants who receive a placebo report less discomfort by the end of a study than they reported at the beginning. That is an observed change among the placebo-group participants. It is often called a placebo response. By itself, however, the change does not establish that the placebo or the participants’ beliefs caused the improvement.
Symptoms can change over time for other reasons. A minor illness may improve naturally. Participants may have good days and bad days, or may report symptoms differently at different times. Attention from study staff and the routine of taking part in a study can also affect what participants notice or report. These possibilities are reasons to avoid treating every improvement in a placebo group as proof of a placebo effect.
A randomized experiment can compare the group assigned to the active treatment with the group assigned to a placebo. If the groups are treated similarly apart from the active ingredient, a difference in their responses can help researchers judge whether the active treatment provides a benefit beyond the placebo-treatment context. As in Why Random Assignment Is the Key to Causation, random assignment helps make the groups comparable on average; it does not guarantee that the particular groups will be identical.
When the outcome is a participant’s report of pain, nausea, or another subjective symptom, expectations may influence the experience or the report. A placebo comparison can be especially useful for understanding that kind of outcome. For an outcome that is less subjective, such as a laboratory measurement, researchers still need a suitable comparison, but should not assume that expectations affect every response in the same way.
What a Placebo Comparison Can Show
A placebo-controlled experiment compares the active treatment with the experience of receiving a similar treatment without its active ingredient. If the active-treatment group has better outcomes, that difference is evidence about the active treatment’s benefit relative to the placebo condition in that experiment. It does not mean that every participant improved, or that the placebo group had no response.
The comparison is most informative when the groups’ experiences are similar in relevant ways: for example, they follow similar schedules and have similar contact with study staff, and the response is measured using the same method. If participants know which treatment they received, their expectations could differ between groups and affect their experiences or reports. Researchers may use design features to address that concern; blinding and double-blind experiments are the subject of the next tutorial.
Specify the condition, the active treatment, and the response researchers will measure.
Make the placebo resemble the active treatment in relevant ways while leaving out the active ingredient being tested.
Random assignment supports a fair comparison by helping make the groups comparable on average.
Describe what the active-treatment and placebo groups received, what response was measured, and what the comparison can address.
Worked Example: A Placebo Control for a New Pain Medicine
Worked Example: Comparing Two Pill Conditions
A fictional research team wants to study whether a new medicine reduces short-term muscle soreness after a minor procedure. The team enrolls 80 adults who have agreed to participate. By chance, 40 are assigned to receive the new medicine and 40 to receive a placebo pill that looks and is taken like the study medicine. The team records each participant’s soreness rating using the same scale at planned times.
State: The explanatory variable is the assigned pill condition, with two levels: the new medicine and the placebo. The response variable is each participant’s soreness rating. The question is whether participants assigned to the new medicine have different soreness outcomes from participants assigned to the placebo.
Plan: Randomly assign the participants to the two conditions. Use the same schedule, symptom-rating scale, and study contact for both groups. The placebo is a relevant control because it lets the team compare the medicine with a similar treatment experience that lacks the medicine’s active ingredient. The team should also consider whether the study is medically appropriate and ensure participants give informed consent.
Do: Collect the soreness ratings as planned and compare the responses from the two assigned groups. If the new-medicine group reports less soreness, that difference is relevant to whether the active medicine provides a benefit beyond the placebo condition. A change in the placebo group’s ratings could reflect expectations or the treatment context, but could also reflect natural recovery or other influences.
Conclude in context: The experiment is designed to compare soreness outcomes for adults assigned to the new medicine with outcomes for adults assigned to a similar-looking placebo. Random assignment supports a cause-and-effect comparison of assignment to these conditions for the participants in the experiment, assuming the study is carried out as planned. The design alone does not show that the medicine works; the observed group results are needed to answer that question.
Worked Example: Improvement Does Not Prove a Placebo Effect
Worked Example: A Placebo Group in a Seasonal Allergy Study
A fictional team studies a new allergy tablet during a season when participants’ symptoms may vary from week to week. The team randomly assigns 50 volunteers to the new tablet and 50 to a similar-looking placebo. After several weeks, some participants in both groups report that their symptoms have eased.
State: The treatment of interest is the new allergy tablet, the placebo group is the control group, and the response is participants’ reported allergy symptoms. The study asks whether symptom outcomes differ between those assigned to the tablet and those assigned to the placebo.
Plan: Random assignment makes a comparison of the two groups more informative than looking only at the placebo group. The same symptom-measurement method and study schedule should be used for both groups. Because symptoms can vary naturally, the team should not interpret an improvement in either group by itself as proof of what caused it.
Do: Compare the symptom outcomes for the two groups. The fact that some placebo-group participants improved describes their observed response. It does not identify how much of that change came from expectations, changes in pollen exposure, ordinary symptom variation, or other factors. The comparison with the active-treatment group addresses whether outcomes differed under the two assigned conditions.
Conclude in context: Improvement among participants assigned to the placebo is not, on its own, convincing evidence that belief caused the improvement. A placebo-controlled comparison can help assess whether the active tablet provides additional benefit relative to the placebo condition, but the study’s results and design details are needed to draw that conclusion.
Worked Example: Choosing an Ethical Comparison
Worked Example: Testing an Add-On Treatment
A fictional hospital team wants to investigate whether a new add-on treatment reduces discomfort for patients receiving an established treatment for a painful condition. The established treatment is known to help. The team considers assigning some patients to receive the established treatment plus the new add-on and others to receive the established treatment plus a placebo resembling the add-on.
State: The question is whether the new add-on improves outcomes beyond the established treatment. The control condition is not “no care”: it includes the established treatment plus a placebo for the add-on.
Plan: If the study is ethically reviewed and participants give informed consent, randomly assigning participants to the add-on or placebo add-on can provide a relevant comparison. Both groups continue to receive the established treatment. This design avoids withholding that treatment while allowing the team to focus on the added benefit of the new option.
Do: Measure the same response for both groups under the same schedule and compare their outcomes. If the add-on group has better outcomes, the comparison addresses the new treatment’s added benefit over the established treatment and the placebo context. It does not compare the established treatment with receiving no treatment.
Conclude about the design: A placebo is not automatically an appropriate control in every medical experiment. If using one would deny participants an effective treatment they need, that raises a serious ethical concern. Here, the placebo applies only to the unproven add-on while both groups retain established care, subject to appropriate ethical review and consent.
Common Mistakes and AP Exam Tips
- Calling a placebo “nothing.” Describe it as a treatment designed to resemble the treatment of interest but without its active ingredient for the condition being studied.
- Claiming that all placebo-group improvement is caused by belief. Improvement could also reflect natural change, variation in symptoms, or how responses are measured. A full-credit answer distinguishes the observed response from a claim about its cause.
- Thinking a placebo control proves the active treatment works. A placebo-controlled design creates a comparison; results are needed to determine whether the groups’ outcomes differ.
- Ignoring the treatment context. Expectations and interactions can influence some experiences and reports. Explain how a similar treatment experience in both groups can help make the comparison more informative.
- Assuming placebos are always ethical. State whether participants continue to receive necessary established care, and recognize the need for informed consent and appropriate ethical review.
- Overstating what random assignment guarantees. Say it helps make groups comparable on average. Do not claim that it makes the groups exactly alike or eliminates every difference by chance.
For a strong AP response, name the active treatment, the placebo condition, and the response being compared. Explain that the placebo provides a baseline for the treatment experience without the active ingredient, and avoid saying that improvement in the placebo group proves belief caused it.
Check Your Understanding
Use the treatment conditions and the response being measured to explain each answer.
- What makes a treatment a placebo in a medical experiment?
- Participants assigned to a placebo report less discomfort later in a study. Why does that change alone not prove that belief caused it?
- A study compares a new medication with a similar-looking placebo pill. What question does this comparison help researchers address?
- Why might expectations affect a participant’s report of pain even when the participant receives no active pain medicine?
- A new add-on medicine is being tested while all participants continue to receive an effective standard treatment. Why might a placebo for the add-on be more appropriate than withholding all treatment?