Let's take https://docs.google.com/spreadsheets/d/1kOuDzNdOXpGzSLXq18ux60TX14C5hZuajjzTRQb5PTo/edit#gid=975687386
this test for example.
Click on D tab and sort from Z>to A
You will see Weapon Mod (Mk 8) from 2 different nodes (2 different people testing) and they have the most number of battles that's why I took those into account.
1 person got 0.245 drop rate (233 battles), the other 0.453 (223 battles), the difference is huge.
While "even" this is not the proper testing of this Theory, it is as close as we got from "any" tests and datas provided so far.
(You can also see the same huge fluctuations on other parts, best way to check is to find 2 different battles/people testing the same Gear drop, but on most of other drops the "number of battles" are much more lower, therefore will give a much less realistic result)
PS: Why I linked this Data? It is the combination of results from different people, so the probability of a gear piece being at "5 out of 50" and "45 out of 50" on different battles are far more greater. (Since Data provided from the links Nikoms linked only gives the average chances generally, disregarding the fact that the theory is: "You might be getting less drops when closer to finishing a piece")