Any time you perform multiple statistical comparisons, your type I error rate increases at a rate of 1-(1-alpha)^n Where alpha is your p-value threshold and n is the number of tests performed. If you perform 10 tests with a p-value threshold of 0.05 for example, your type I error rate will be roughly 1-(1-0.5)^10 = 0.4. This is much higher than the threshold you set of 0.05.

To counter the quickly increasing error rate as your perform multiple comparisons, many methods have been developed to correct for this multiple comparisons problem, such as Bonferroni, Benjamini-Hochberg, and q-value. That's why for anything with multiple comparisons you should be looking at corrected p-values, and not the raw p-values.