D-Roc - the "normalization" of data trending is very complex. I can speak to certain scenarios, but not being a tribologist, some of my examples may be contorted, so bear with me.
As a hypothesis, let's say you ran 35k miles of the same oil (brand "A") with samples set every 5k miles (7 total runnings). There is reasonable data to start developing historical normality. You would notice that wear metals and insolubles are actually dynamic in that they never are one absolute value; rather, they tend to stay in an acceptable range, moving both up and down. Trends, on the other hand, are when the values show continued movement in a signular direction, and infer that some characteristic is no longer in bounds. That movement can be either desirable or undesirable.
Then, one decides to experiment and change brands (to brand "X"). Now after another 5k miles we pull a sample, and get somewhat disturbing wear metal results. By disturbing, I don't mean grossly high or low, I just mean not "normal" by statistical definition. Further, if you don't get what you wanted (expected), then you would presume it to be bad. Here is where panic overtakes reason, and one would then switch back to brand "A". Now we sample again in 5k miles, and the results from this next UOA, while perhaps not completely in "range", would seem more "normal" and comfortable. But the problem is one didn't allow for "normality" to establish itself with the new brand.
In this one example, the engine/oil combination that was initially established, set up long term wear patterns and chemical reactions had stabilized. Then the chemicals were switched, and reactions changed. But "normality" was not established before yet another change occured. When returning back to brand "A" there is a high potential for what is known as "incidental return concurrence", where the desirable reactions had not been fully overcome by the "intruding" chemical, are easily established in return. In short, the typical effects from chemicals in "A" had not been fully erased by "X", and therefore the move back to "A" came as less of a shock. When brand "X" came to town, it didn't have enough time to erase brand "A", and when "A" returned, it was easy to establish "normality".
The thing to remember about normality is that it only exists in the context of location. It's kind of like the old saying "When in Rome, do as the Romans do." I am a quality process engineer at work; I see this type of stuff nearly every day. Even I sometimes get caught up in the "snapshot" rather than the whole albumn of information.
And if this weren't harrowing enough, don't forget to consider that each and every engine is just a little different, just as each batch of oil blend is a little different. That's why ranges and trends are so necessary to the proper understanding of UOA's. You need to establish long term ranges AND trends, to see the true health of the engine relative to the oil brand selected.