Inspired by the recent thread about the minor agreements, and kind of enjoying a brief foray back into the synoptic problem, I looked up some of these minor agreements of Matthew against Luke, and the following one struck my eye as a sort of example of the complications that this entire issue tends to bring up.
Q skeptic Mark Goodacre takes a look at Matthew 22.27 = Mark 12.22b = Luke 20.32:
There is no denying that Matthew is comfortable with this usage of ὕστερον (Matthew 4.2; 21.29, 32, 37; 22.27; 25.11; 26.60). That Luke, who never uses it elsewhere, should change Mark's ἔσχατον (which he also never uses elsewhere as an adverb, and which appears in the rest of the NT as an adverb only in 1 Corinthians 15.8) to ὕστερον independently of Matthew does seem unlikely. If he did not like Mark's ἔσχατον, he could have gone with εἶτα ("then," "next," Luke 8.12), ἔπειτα ("then," "next," Luke 16.7), τότε ("then," "at that time," multiple instances throughout Luke-Acts), δέ ("and," "but," often used to get to the next event in a story), or even with something like μετὰ ταῦτα ("after these things," "after this," about 5-6 times each in Luke and Acts).
The manuscripts are of no help in relieving us of assuming that Matthew and Luke are less than independent. Matthew 22.27 has no variants that I can find which would do a thing to explain this agreement. Luke 20.32 has the feminine adjective ὕστερα instead of the adverb ὕστερον in Washingtonianus, but even the adjective would constitute an agreement with Matthew against Mark. Mark 12.22, on the other hand, evinces a lot of textual variation, including the feminine adjective ἐσχάτη instead of the adverb ἔσχατον in Alexandrinus; none of the variants that I can find, however, have the ὕστερον which would explain Luke 20.32.
So case closed, right? Luke must have drawn upon Matthew at this point in his narrative.
Except... really? I perused the SQE at this pericope (#281) and decided to create a composite image of the synopsis:
I moved through the columns fairly quickly, underlining significant agreements between any two of the gospels using a simple color code. Matthew is blue, Mark is red, and Luke is green. If you find blue underlining in Mark or Luke, then that means that Matthew is agreeing with Mark or Luke for those words or phrases. If you find red underlining in Matthew or Luke, then that means that Mark is agreeing with Matthew or Luke for those words or phrases. And, if you find green underlining in Matthew or Mark, then that means that Luke is agreeing with Matthew or Mark for those words or phrases. I did not select every single agreement, since some of them are meaningless or easily explained on the grounds of well established writing patterns. But you can go through and see for yourself whether I was fair or not to all possible solutions to the synoptic problem. (I did not mark triple agreements, of which there are plenty, because triple agreements do not favor any solution over another.)
It is very noticeable that Matthew and Mark agree against Luke a number of times, and it is equally noticeable that Luke and Mark agree with Matthew a number of times. Both sets of agreements include both words and phrases. It seems indisputable that there is a literary connection between Matthew and Mark and also a literary connection between Mark and Luke. But, as I try to find Matthew and Luke agreeing against Mark in any significant way, all I seem to come up with is (A) the prefixed πρός right in the first verse of the pericope and (B) the ὕστερον agreement highlighted by Goodacre. The mutual addition of πρός to the main verb, in a context of Sadducees coming to or toward Jesus, seems to me to be innocuous; it could easily be coincidental: a smaller coincidence than two editors independently changing "went into" to "entered." The mutual change of ἔσχατον to ὕστερον, however, is more serious, as Goodacre surmises.
And yet what does that particular point of agreement do for us? On Goodacre's preferred Farrer theory, it seems to imply that Luke copied his entire pericope from Mark with the exception of that single word, for which he relied upon Matthew instead of upon his own wordsmanship. Faced with an adverb in Mark which he never uses elsewhere, he decided to change it out for a different adverb which he never uses elsewhere, simply because Matthew, whose text exercised no recognizable influence on him at any other point of the pericope, has done so. Why would he rather not have just gone with something he is more accustomed to? Those available options I have listed above (εἶτα, ἔπειτα, τότε, δὲ, μετὰ ταῦτα), as well as more that I have not listed, show both (A) how striking the agreement between Matthew and Luke is and (B) how strange it is that Luke would opt for the word which created this agreement instead of any of a number of more likely choices.
If Matthew had offered a bit of handy information absent from Mark, or perhaps an alternate and simultaneously more attractive version of information present in Mark, it would be quite understandable that Luke should select Matthew over Mark; and this sort of procedure is one very viable option for how things might have happened in the so called Mark-Q agreements. But here? For this? I do not buy it. At least, the solution does not satisfy. It does not sound like what Luke really did.
So do I opt for a textual solution? Maybe Luke, already copying Mark anyway, copied Mark's ἔσχατον and then a later scribe harmonized it with the more popular Matthew's ὕστερον. That kind of harmonization (changing either Mark or Luke to match Matthew) happened all the time. But this particular instance of such harmonization? Completely unattested. So I would be just assuming that the harmonization happened in between the autograph and the archetype of Luke. Certainly possible, but it is also possible that Luke used Matthew just for this single word, or maybe kind of had Matthew's lingo running through his head while he was deciding how to alter Mark here... or something. And I feel like I do not have enough information to tell which of the possibilities is more or less likely.
Or hey, does Marcion help? Well, Luke 20.32 is apparently unattested for Marcion. Maybe Marcion had some other (but necessarily bland or noncontroversial) transition into the resurrection question besides the death of the woman. Or maybe Marcion had exactly what Luke has, or even what Matthew has, but the verse was not important enough for Tertullian's or Epiphanius' purposes to write about. Not attested means not attested (as opposed to attested as absent or attested as present); therefore, Marcion is of no actual help for this one.
So I wind up in this case at a spot I know very, very well: not sure.
YMMV. Let me know where I have gone wrong.