For each statistical test undertaken with p=0.05, there is a 5% chance of incorrectly rejecting the null hypothesis, assuming it is true. As the number of tests increase, the error rate is compounded and an adjustment must be made to prevent bias. You might, for example, look up Bonferoni as an example of one way to make such a correction. Though this is often too stringent it is useful to see the principle. A more sophisticated approach to pairwise comparisons can be found in package multcomp.

For each statistical test undertaken with p=0.05, there is a 5% chance of incorrectly rejecting the null hypothesis, assuming it is true. As the number of tests increase, the error rate is compounded and an adjustment must be made to prevent bias. You might, for example, look up Bonferroni as an example of one way to make such a correction. Though the Bonferroni adjustment is often vviewed as too stringent (conservative) it is pedagogically useful to see the principle. A more sophisticated approach to pairwise comparisons can be found in package multcomp.