The problem is that these tests are taken by humans, who, for better or for worse, are subjective in the way they see themselves. Combine this with unclear delineations, such as T/J and F/P confusion, and I find many of these results to be silly.
For example: Sometimes people confuse terms like "feeling" with "random" and "logic" with "order". So someone who scores ISFJ may actually be an ISTP like their real life actions suggest.
Or they mark the test the way they would like to be, or how they see themselves relative to their ideal self. Like when my wife marks that she is "disorganized", because she is not as organized as she would like to be, however, objectively, she is ten times more organized than me, especially in her own head (which is what matters since she's an N).
Also, many people have misgivings and prejudices about some of the functions as inferior, especially S, F, and P. Sensing in particular, since I would imagine that many people would look at the numbers S 75%/ N 25% and quickly make the leap that S = "Stupid" and N = "Intelligent", and that in order to seem smart, they need to be an N (which is, to be fair, how it works in real life, since college-educated doctors, scientists, and computer programmers make more money than bricklayers.)
Or that feelings are generally looked down upon as being irrelevant and invalid, with thinking being "smart" and "logical". Or that unclean people who are constantly late are subject to ridicule as being "flaky", while our constant, structured, and punctual brethren get a promotion.
Or take the word "methodical", which has the dual meaning of being both "ordered" (J) and "logical" (T). Confusing...
The real issue is that this is an objective test, with subjective people taking it.