It would seem that you are familiar with Textual Criticism.
If so, then you should understand that just because there are more copies of one rendering; does not mean that the Majority Text is the correct rendering.
If you already understand this; I'll explain it for those who don't.
We don't have the Autograph Manuscripts (original manuscripts) of any of the books of the Bible. What we have are copies of copies; but let's pretend that we did have an Autograph Manuscript; and someone copied it. However, when he hand copied it; he made a mistake. He then went down to the copy shop, and had them run off 1000 more copies of his flawed copy. Now we have 1001 flawed copies of the Autograph Manuscript.
Should we then dismiss the Autograph Manuscript, by the Author himself, as being in error?
This is just one of many factors that scholars of Textual Criticism weigh, in discerning which of the thousands and thousands of variants that we find in the many manuscripts available to us today, to most accurately restore the Autograph Manuscripts.
The debates over these many factors probably won't be fully settled until Messiah comes back to set us all straight.
I agree with everything you wrote. Yes, counting words and calculating percentages does not offer a definitive proof. But some facts are better than no facts, and even a few facts can guide us in our analysis.
Take the facts regarding 'ei' in Matthew 19:9.
Fact 1: There are zero examples of the presence of that word in any
Uncial Greek New Testament Manuscripts.
Fact 2: The first example appeared in the 12th century Majuscule MSS.
Fact 3: In the wild, the idiom 'ei me' is only translated as 'except' 1% of the time.
In your example above, a person makes one copy with a mistake and makes thousands of copies of the error. Well, in the case of 'ei' there are no examples before the 12th century, so that means, until someone comes up with one example, we must conclude that 'ei' is a mistake.
Now, in the other posts, it is suggested that we don't need 'ei'; that 'me' can be translated as 'except' all by itself. Ok, where is the proof of this. Where are the facts? The UBS5 and NA28 give us lists about what the various manuscripts say. Where are the examples of 'me' being translated as 'except'? How can I know that this is the best translation? In other works, such as Jones (2005)[cited in another post above] it is only translated that way 1% of the time? Such a low percentage does not inspire confidence that that is the right and true translation.
And what about the other competing views: that 'me' can be 'inclusive' as well as 'exclusive'? What if 'not [even] for fornication' is a legitimate interpretation? - seeing that 'even' is not in the Greek language, therefore it is 'possible'.
How about the facts concerning the last phrase of Matthew 19:9:
Fact A: that phrase is present in 93% of all manuscripts
Fact B: that phrase is present in 89% of the uncials
Fact C: that phrase is in the Complutensin Polyglot and the Textus Receptus
Fact D: that phrase is absent in the NA28, and has never been included even from the first edition.
Fact E: that phrase is present in 2/3 of 4th century uncials
Fact F: that phrase is present in 2/3 of 5th century uncials
Fact G: that phrase is present in 95% of post-5th century uncials
With these facts, it is a weaker case to argue than the case with 'ei', because of the two examples: one 4th century and one 5th century MSS that did not contain that phrase.
At least, facts help to expose the problem, and it appears from the facts, that the NA28 has some explaining to do as to why they left out the last phrase and it appears like we need more facts to help us puzzle out these ongoing problems.