In the post on Q, I asked whether Matthew or Luke had the “earlier” version of the Lord’s Prayer or the beatitude “blessed are the poor/poor in spirit” or the reference to Wisdom’s “deeds/children.” Michael Goulder (Luke: A New Paradigm) takes the line that Matthew is the creative originator of all the non-Markan material, so Luke’s formulation must always be secondary, while Mark Goodacre (Case Against Q) is open to oral sources that Luke may have drawn on that sometimes could contain pre-Matthew formulations.
Two Source Theorists, on the other hand, believe that sometimes Matthew and other times Luke sticks closer to the wording of their shared source Q. To attempt to reconstruct the possible wording of Q as far as can be done, John Kloppenborg and Robert A. Derrenbacker, “Self-Contradiction in the IQP? A Reply to Michael Goulder” JBL 120 (2001): 57-76 explain the reasoning behind the scholars who put a critical edition of Q:
[A] phrase or word should be treated as secondary (that is, not deriving from Q), (a) when it can be shown by reference to Matthew’s treatment of Mark and by reference to editorial or transitional portions of Matthew that Matthew has a tendency to add the phrase or word, and (b) when Luke has no aversion to the phrase or word. (The same logic applies mutatis mutandis to Lukan phrases and words.) (p. 63)
Additionally, they take from text criticism the preference for the shorter or difficult reading, for a scribe is more likely to expand (elaborate, clarify) than abbreviate and solve than create theological tensions, and use a similar rating system to the United Bible Society on the probability of reconstructions (pp. 59-63).
Stephen Carlson has written a rebuttal to the criterion outlined above in his chapter “Some Problems with the Non-Aversion Principle in Reconstructing the Text of Q” in Marcan Priority Without Q: Explorations in the Farrer Hypothesis (ed. John C. Poirier and Jeffrey Peterson; LNTS; London: T & T Clark, 2015). Check out both links to get a handle on this debate.