They're all 'tests' trying to explain a complex concept with a multitude of intricate, interwoven factors, into a black and white easy to understand phraze.
It doesn't work fully, and never will, regardless of the test.
Yeu can't dump the entirety of the planet's population into 16 boxes and pretend that we are all clones of 16 people and there is no variation at all.
Obviously, each test is flawed in its' own way, no matter how much, or how little evidence there is for them. They are attempting to quantify things which can't be quantified.
I'm fine with that; I do it myself all the time. It's an easy to understand generic 'close' model that can be reasonably semi-accurate, as long as yeu take it with a grain of salt that it can't measure things perfectly.
A combonation, such as MBTI+Enneagram, tends to be more useful, since they are testing completely different areas of ones' behaviour, and they obviously influence each other, and testing one or the other won't give accurate results. A 7w6 entp is not the same as a 3w4 entp by any means, and even there, there are distinctions among even those subgroups.
The best we can do is classify whot we know, to the best of our ability. Some tests may have further factors that determine more aspects, but none determines all. There is no magical "do this test and we know who yeu are"; the closest we can get is a vague insight through a thick fog.
If yeu understand that, then yeu can appreciate such, and make use of the generic information presented, without taking it too strictly.
MBTI is not some uber-godlike-perfect description of human behaviour. Whot it does do, is try to group people using whot makes "sense for the most part". Everyone is E and everyone is I at different times in their lives; mbti just assumes that the dominant of the two is yeur 'tendancy' to do one or the other more frequently. If yeu take it as general tendancies it's great. Same as most of these other models. Some may be more 'accurate', by testing different factors than others. They're still all inherently flawed however.
I have no perfect clone of myself, no doppleganger. I have some which are vaguely close, which may be very similar in some regards, but no perfect identical twin. I can still be understood, though, in a more generic sense. I think we should at least TRY to understand each other, even if it's flawed or inaccurate to some degree. To give up just because it's impossible to be flawless in the evaluation is foolish.