Glossary » internal validity

internal validity

the degree to which the design of a study allows you to accurately attribute an observation to a specific cause rather than alternative causes. An undesirable alternative cause for your result is known as a “confound”.

An example is to compare the amount of time it takes in using 2 alternative interface designs. You would like to claim that if one design is faster than another, then the time difference is due solely to differences in the interfaces. An example of a confound is when different groups of people are tested for each design–if one group is more skilled than another, that would be an alternative explanation for the observed time difference.

A “control variable” is a potential confound that has been controlled for so that it can reasonably be ruled out as a possible explanation. For instance, to avoid the above confound, subjects can be a control variable by assigning them randomly to each test condition.