To understand what a confounding variable is, let’s go over what an experiment is in brief.
An experiment, in ridiculously broad terms, seeks to understand the relationship between two variables. Let’s take two variables A and B, now the researcher wants to see if AcausesB. For instance, Bandura wanted to see if seeing a Bobo doll kicked or punched (including tool use) would cause children to mimic the behaviour.
Experiments are designed to see whether a causal link exists between two variables, such as observed violence and demonstrated violence as in Bandura’s famous experiment.
However, the researchers have to be on the lookout for what’s called confounding variables. What does confounding actually mean?
If you look up the term in Google and check the second meaning you’ll run into this definition: “mix up (something) with something else”.
In essence our confounding variable might interfere with our results by changing the relationship between the two variables under consideration.
In other words if we run an experiment and see AcausingB, there may be if researchers are not careful, a third variable C which is actually responsible for causing C or A.
Let us go back to Bandura and try and think of a possible confounding variable: One that might come to mind is how one accounted for individual variance and demonstrations of violence. But Bandura had already accounted for this variation by using matched pairs and avoided all the aggressive children ending up in one or another of the experimental groups. But if Bandura had not done this, then the results may have been less useful.
Thus, a confounding variable is one that appears in an experiment and may mess with the results by altering the relationship between two variables or by acting on variables directly.