Nedarim 82a: Does Proximity Affect Girsology? And Did Rav Yehuda cite Rav or Shmuel?

From my translation of Rif on Nedarim 82a-b:

{Nedarim 82a}

אמר שמואל משמיה דלוי כל הנדרים בעל מפר לאשתו חוץ מן הנאתי על פלוני שאינו מפר
אבל הנאת פלוני עלי מפר

{Nedarim 82b}

א"ר יהודה אמר שמואל נדרה משתי ככרות מאחת מתענה ומאחת אינה מתענה מתוך שמפר למתענה מפר לאינה מתענ' ורב אסי א"ר יוחנן מפר למתענה ואינו מפר לשאינה מתענה

Shmuel said in the name of Levi: All vows, a husband annuls for his wife, except for "my benefit upon Ploni," which he cannot annul. But "the benefit of Ploni {should be forbidden} upon me," he can annul.

Rav Yehuda cited Shmuel {our gemara: Rav}: If she vowed off two loaves, one of which involves self-denial and one which does not involve self-denial {e.g. if one was of fine flour and one was of course flour}, since he can annul in respect of that which causes self-denial, he can also annul in respect of the one which does not cause self-denial.
And Rav Assi cited Rabbi Yochanan: He may annul the one causing self-denial, but may not annul the one which does not cause self-denial.

Rav Yehuda often cites Rav and often cites Shmuel, and this is a pretty frequent girsological variant -- whether he cites Rav or Shmuel. Sometimes this makes a practical halachic difference -- in terms of whether we rule like him over a different Amora, or whether we draw consistent opinions within Rav or Shmuel.

This is an interesting instance, though. Besides substituting Shmuel for Rav, Rif also trims a section of gemara which is not lehalacha, immediately before. This brings the earlier statement, which was Levi citing Shmuel, into a much nearer proximity. In turn, this might influence a sofer's error in substituting Shmuel for Rav, assuming Rav Yehuda citing Rav was the original text. In our gemara, there is a moderate amount of material separating the two statements, such that such a slip of the quill would be less likely.

I wonder if we can establish this as a consistent occurrence, using more than just anecdotal evidence. If so, it can serve as a useful tool in identifying girsological variants, both within a single text which has a proximate matching name, and where we have different texts with greater distance between statements. Perhaps someone has already done such a study, manually. If I had the time, I might design the following experiment. (Because I take an interest in computational girsology.) Take a corpus of text from Bavli, and a corpus of parallel text drawn from Rif. Align the two texts. Find each name substitution (=girsological variant) and within each text, measure the distance to the nearest occurrence of that name, and see if there is a tendency to echo the more proximate name. (Of course, not only names influence other names. I have seen things akin to the word שמא influencing the introduction of שמואל.)

a