Topic 3.2 Introduction to Planning a Study: distinguish observational studies from experiments, identify explanatory and response variables, and recognize that only an experiment with imposed treatments can support a causal conclusion.

What this topic is asking

The College Board (Topic 3.2) wants you to distinguish an observational study from an experiment, identify the explanatory and response variables, and understand why only an experiment that imposes treatments can support a cause-and-effect conclusion.

Observational study versus experiment

The single test you apply is: did the researchers do something to the subjects, or did they only watch and record? Assigning people to a diet, a drug, or a teaching method is an experiment; recording the diet, drug, or method people already chose is an observational study. Surveys are observational. This one distinction governs the entire scope of what the study can conclude.

Explanatory and response variables

Identifying these correctly orients the whole analysis: the explanatory variable goes on the horizontal axis of a scatterplot, defines the groups being compared, and is the variable a causal claim would be about. In an experiment, the levels of the explanatory variable are the treatments.

Why only experiments prove causation

The reason an observational study cannot establish causation is confounding. Two variables are confounded when their effects on the response cannot be separated. In the coffee-and-blood-pressure example, people who drink a lot of coffee may also be more stressed or sleep less, and those factors raise blood pressure too, so the observed link between coffee and blood pressure might really be a link between stress and blood pressure. Because the researchers did not control who drank coffee, the groups differ in many ways at once, and any of those differences could explain the result. An experiment removes this problem by imposing the treatment and (in Topic 3.5) randomly assigning subjects to treatment groups, which tends to balance every other variable, measured or not, across the groups. Then a difference in the response can be attributed to the treatment. This is why the College Board repeats the mantra: association from an observational study, causation only from a well-designed experiment.

Planning the right study for the question

Choosing a study type is a trade-off the exam expects you to reason about. Experiments give causal answers but are not always possible or ethical: you cannot randomly assign people to smoke for twenty years, so the link between smoking and cancer rests on observational evidence and biological mechanism, not a human experiment. Observational studies are often the only feasible route for questions about harmful exposures or naturally occurring groups, and they are perfectly good for describing associations and generating hypotheses, as long as you do not overclaim causation. So the planning question is not "which study type is better?" but "which type can answer my question, and what will I be allowed to conclude?" If the goal is to establish that a treatment works, an experiment is needed; if the goal is to describe a relationship in a population you cannot manipulate, an observational study is appropriate, with the confounding caveat stated plainly.

Try this

Q1. State the one feature that distinguishes an experiment from an observational study. [1 point]

• Cue. An experiment imposes a treatment on subjects; an observational study only measures or records, without intervening.

Q2. A study links eating breakfast to better grades from survey data. Why can it not conclude breakfast causes better grades? [2 points]

• Cue. It is observational, so a confounding variable (such as a stable home life) could be linked to both breakfast and grades; the link may reflect the confounder, not breakfast.