Multiple statistical comparisons (many tests of association) on the results of a study increase the likelihood of one or more associations (e.g. between a treatment and an outcome) having a low p-value (probability of having occurred by chance) when, in fact, the association did occur by chance.
Multiple statistical comparisons increase the likelihood of attributing a difference to a treatment effect when chance is the more likely explanation. This mistake is sometimes referred to as a Type 1 error.
If you feel that this definition hasn't helped you to understand the term, click on our monkey to let us know.