By North Essay Help Editorial Team · Reviewed by a subject specialist (master’s or PhD) · Editorial policy · · Updated
Confounding vs Extraneous Variables in Psychology: The Difference, With Examples
The short answer: an extraneous variable is any variable other than the independent variable that could affect the dependent variable. A confounding variable is an extraneous variable that also changes along with the independent variable, so it offers a second explanation for your result. Every confound is an extraneous variable, but most extraneous variables are not confounds.
Every study has variables you did not plan to study. Some add noise. Some quietly ruin your conclusion. Knowing which is which is the difference between a design that works and one a marker can pull apart.
This guide shows how to spot a confound in a real design and how to control it.
The difference in one table
| Extraneous variable | Confounding variable | |
|---|---|---|
| What it is | Any variable other than the independent variable that could affect the dependent variable | An extraneous variable that changes along with the independent variable |
| Effect on your study | Adds noise or random error | Creates a rival explanation for your result |
| Can you separate its effect from the IV? | Yes, usually | No, not without further control |
| Example | Participants' mood on the day | Group A tested in the morning and Group B in the afternoon |
The rule to remember: every confounding variable is an extraneous variable, but not every extraneous variable is a confound. An extraneous variable becomes a confound when it differs systematically between your groups.
Spot the flaw: a worked example
The study below is invented for illustration. A school wants to know whether a new reading app improves reading scores. Ms. A's morning class uses the app for a term. Mr. B's afternoon class does not. At the end of term, Ms. A's class scores higher.
Step 1: List everything that differs between the groups besides the app.
- Teacher (Ms. A or Mr. B)
- Time of day (morning or afternoon)
- Students' reading level at the start
- Class composition and size
Step 2: Decide which are systematic. All four differ between the groups in a consistent way, so each is a confound. If the students in each class had been picked at random, reading level would be an extraneous variable, not a confound.
Step 3: Choose a control for each.
| Confound | Control |
|---|---|
| Teacher | Same teacher for both conditions, or several teachers with both conditions in each class |
| Time of day | Hold constant, or balance both conditions across morning and afternoon |
| Prior reading level | Pretest all students and compare gain scores, or use the pretest as a covariate |
| Class composition | Randomly assign students to conditions within each class |
Step 4: State the improved design. "Students in each class are randomly assigned to the app or to regular reading time. A reading pretest is given at the start. The same teacher runs both groups. At the end of term, gain scores are compared." Now a difference in scores is much harder to explain without the app.
Quiz: spot the confound
Eight short studies. For each one, decide whether the variable named is a confound (it changes along with the independent variable) or an extraneous variable that is not a confound. Tap an answer to see the explanation.
Question 1 of 8
A researcher tests whether caffeine speeds up reaction time. The caffeine group is tested at 9 am and the placebo group at 4 pm.
Variable: time of day.
Show the answer
Confound. Time of day changes along with the independent variable, so a difference in reaction time could come from caffeine or from the time of testing. Test both groups at the same time, or balance both conditions across morning and afternoon.
Question 2 of 8
Students choose whether to join a four-week meditation program or not, and the meditation group reports lower stress afterwards.
Variable: how motivated the students were to reduce stress.
Show the answer
Confound. Students who chose meditation may already have been more motivated to manage stress. Self-selection makes motivation vary with the condition. Random assignment would spread motivation across both groups.
Question 3 of 8
Participants are randomly assigned to two conditions. The room temperature varies from day to day, and both conditions are run on every testing day.
Variable: room temperature.
Show the answer
Not a confound. Temperature varies, but not systematically with the condition, because both conditions are run on every day. It is an extraneous variable that adds noise rather than a second explanation.
Question 4 of 8
In a memory study, the music condition learns word list A and the silence condition learns word list B. List A has shorter, more familiar words.
Variable: word list difficulty.
Show the answer
Confound. Each condition gets a different list, so list difficulty changes along with the condition. Use the same list for both conditions, or counterbalance the lists.
Question 5 of 8
A town finds that months with higher ice cream sales also have more drownings.
Variable: hot weather.
Show the answer
Confound. This is a third variable in correlational data. Hot weather increases both ice cream sales and swimming, so the link between the two does not mean one causes the other.
Question 6 of 8
The researcher who knows the hypothesis runs every session in the treatment group and gives more encouragement, while an assistant runs the control group.
Variable: experimenter behaviour.
Show the answer
Confound. The person running the session, and how encouraging they are, changes with the condition. Use the same experimenter for both conditions, and keep them blind to the hypothesis where possible.
Question 7 of 8
In a within-subjects study, every participant completes the easy condition first and the hard condition second. Scores improve in the second condition.
Variable: practice from doing the task twice.
Show the answer
Confound. Order is fixed, so practice changes along with the condition. Counterbalance the order, so half the participants do each condition first.
Question 8 of 8
Participants aged 18 to 25 are randomly assigned to two conditions, and the average age is about the same in both groups.
Variable: age.
Show the answer
Not a confound. Age varies between people, but random assignment has balanced it across the groups, so it does not change with the condition. It is an extraneous variable.
Six ways to control extraneous variables
| Method | What it does | Example |
|---|---|---|
| Random assignment | Spreads participant differences across groups by chance | Flipping a coin to place each participant in a condition |
| Holding constant | Keeps a variable the same for everyone | Same room, same instructions, same time of day |
| Counterbalancing | Balances the order of conditions in a within-subjects design | Half of participants do task A then B, half do B then A |
| Matching | Pairs participants on a key variable | Placing participants with similar pretest scores in different groups |
| Blinding | Stops expectations from shaping behaviour or scoring | Researchers scoring recall do not know which group a participant was in |
| Statistical control | Adjusts for a measured variable in the analysis | Including pretest score as a covariate |
Random assignment balances groups on average, but with small groups they can still differ by chance. That is one reason to measure key variables as well as randomizing.
Common confounds in student lab studies
| Confound | How it appears | Fix |
|---|---|---|
| Time of day | One condition run in the morning, the other in the afternoon | Alternate conditions across sessions |
| Experimenter | A different researcher runs each condition | One researcher, or each researcher runs both conditions |
| Order and practice effects | Everyone does the easy task first | Counterbalance task order |
| Demand characteristics | Participants guess the hypothesis and respond accordingly | Use neutral instructions, or a cover story where ethics allow |
| Self-selection | Participants choose their own condition | Assign randomly |
| Different materials | Each condition uses a different word list | Use the same list, or counterbalance lists |
Third variables in correlational research
In a correlational study you do not manipulate anything, so the issue has a different name: the third-variable problem. Suppose students who use study apps score higher in exams. The app may not be the reason. Motivation could drive both app use and higher scores. Motivation is the third variable.
You cannot fix this by design, but you can measure the likely third variables and control for them statistically, and you can avoid causal wording in your report.
How to write about confounds in your report
In the Method, describe the controls you used: "Participants were randomly assigned to conditions. All sessions were run by the same researcher in the same room between 10:00 and 12:00."
In the Discussion, name the confounds you could not remove and say what they might have done to the result: "Participants were tested in groups, so social influence may have affected recall." Our guide to the discussion section shows where this fits.
Frequently asked questions
Is a confounding variable the same as a lurking or third variable?
They are closely related. "Confounding variable" is usually used for experiments and quasi-experiments, "third variable" for correlational studies, and "lurking variable" is an informal term for the same idea.
Can you eliminate all extraneous variables?
No. You aim to control the ones most likely to matter and to spread the rest evenly across groups, typically through random assignment.
Does random assignment remove confounds?
It balances them on average, which makes systematic differences between groups unlikely. With small groups, chance differences can still occur, so check the groups on key variables.
What is an example of a confounding variable?
Testing one condition in the morning and another in the afternoon. Time of day changes along with the independent variable, so you cannot tell which caused the difference.
How is a confound different from a moderator?
A confound offers a rival explanation for your result. A moderator changes the strength or direction of the effect, for example an intervention that works better for first-year students than for fourth-year students.
Related guides and tools
- Independent and dependent variables: 25 psychology examples
- Operational definition examples in psychology
- How to write a hypothesis for a psychology lab report
- How to write the discussion section of a psychology lab report
- Psychology assignment help
- How to write the method section of a psychology lab report
Sources: APA Dictionary of Psychology, entries "confounding variable" and "extraneous variable" (dictionary.apa.org). Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Houghton Mifflin.
Examples in this guide are illustrative. Check your course instructions and rubric before you submit.
----
