100% agree There's a simple statistical argument behind it: the variance drops as you increase the number of tests.

Let's say your true ability is A and you have one measurement M1. The variance (or, let's put it in another way: the probability that your M1 is waaaay different from your A) goes down if you also get M2, M3, M4 and take an average.

In other words: your measurements become more precise and the stress goes down a lot