Confounding Variable
In short: Confounding variable is an outside factor linked to both the explanatory variable and the response, making it impossible to tell whether the explanatory variable or the outside factor produced the effect.
AP StatisticsStatisticsAP Statistics Unit 1
What is Confounding Variable?
A confounding variable is a third factor knotted together with the thing you are studying. It travels with the explanatory variable and also pushes on the response, so its influence blends into the effect you hoped to isolate.
Suppose students who attend after-school tutoring earn lower grades. Tutoring may not be to blame: students who already struggled are the ones most likely to enroll. Prior struggles are the confounder. Observational studies are especially exposed; assigning treatments by chance in an experiment cuts the tie between the confounder and the treatment.
How to spot it
- An observational study reports that two things go together and the headline hints at cause and effect.
- The groups being compared differ in some other way, such as age, income, health or motivation.
- A surprising result, like more help leading to worse outcomes, suggests who chose the treatment matters.
- The question asks “what other variable could explain this association?”
The method
- State the explanatory variable and the response in the claim.
- Brainstorm a third variable, a possible confounding variable, that plausibly affects the response and differs between the explanatory groups.
- Explain how it connects to both, so that its effect cannot be separated from the one under study.
- Propose a remedy: a randomized experiment, or comparing subjects matched on the confounder.
Confounding Variable: solved example
Problem
A survey of 400 families finds that children whose parents eat dinner with them most nights get higher reading scores. A columnist writes that family dinners raise test scores. Identify a possible confounding variable.
Solution
Explanatory variable: family dinners. Response: reading score. A plausible confounder is parents’ available time and resources, which includes flexible work schedules and education level.
Families with more time and education may both eat together more often and provide more reading support, so the dinners may not be the reason for higher scores. Since the data are observational, the columnist cannot claim cause; only random assignment of dinner habits could isolate the effect.
Try it
Question 1
Cities with more fire trucks sent to a fire have more damage. Name a confounder.
Show the answer
The size of the fire: bigger fires cause more damage and call for more trucks, so trucks do not cause the damage.
Question 2
How does random assignment deal with confounding variables?
Show the answer
It makes the treatment groups alike in every other respect on average, so a confounder is spread evenly and cannot favor one group.
Confounding Variable vs. Correlation vs. Causation
A confounding variable is the specific third factor that can explain a link. The phrase correlation does not imply causation is the general warning. Naming the confounder turns the warning into an actual argument about why the link might be false.
Keep going
Lessons
Worksheets and quizzes
- Experimental Design and Random Assignment – Worksheet
- Experimental Design and Random Assignment – Quiz
- Scatterplots and Correlation – Worksheet
- Scatterplots and Correlation – Quiz
Related words
- Experiment vs. Observational Study
- Correlation vs. Causation
- Sampling Bias
- Correlation Coefficient
- Least-Squares Regression Line
Whole dictionary: A–Z · AP Statistics · Statistics