Well it's not a sample size of one as we're relying on the judgement from the tribologist.
It most certainly is a sample size of 1 (one).
In that video, they discuss how the use of the auto stop/start was significantly implemented during this ONE UOA, relative to the others being compared/contrasted. In fact, there are only THREE UOAs shown in the screenshot above, and the "one" being discussed is the most recent, relative to the effects of the s/s function.
In the video, at minute 9:00, quoting LSJ (the acclaimed tribologist ...) "
Why don't you tell everyone what's different about this sample compared to the other two?"
My proof of claim lays directly in this quote ...
IT'S A SAMPLE SIZE OF ONE !!!
The fact that LSJ is a tribiologist means zilch to me; it's clear that he practices bad "science" when he looks at one UOA and concludes a singularly experienced phenomenon is justification to make such a claim as generalized fact.
To any true engineer/scientist, any test process must assure several things are present in any DOE to lend credibility to the conclusions:
- reduction of variables (as best possible); this "test" has WAY TOO MANY
- understanding of gauge R&R accuracy; no mention of this and I SERIOUSLY DOUBT they even know what this is, let alone how to calculate it
- minimum of 30 samples (to understand standard deviation relative to claimed or refuted output effects); as shown above, this is a "test" of one sample
- establish both baseline data and controlled variable averages and trends with enough data to comply with Stdev concept above; not done at all in this "test"
These two people in the video have exhibited nothing but to show their ignorance of how to prove or disprove anything credibly.
Look at this at face value ...
They are claiming that "wear" is reduced by using auto s/s.
What was the Fe wear rates in the two previous samples?
- sample SDS-0730 had a wear rate of 1.72 ppm / 1k miles
- sample AAA-3912 had a wear rate of .81 ppm / 1k miles
- sample AAA-4533 had a wear rate of 9.6 ppm / 1k miles (this was the sample that used auto s/s)
The two samples prior to the "test" had huge variation; the wear rate of the first sample was more than twice that of the second. Yet the "test" sample was very close to the sample before it which supposedly didn't have the auto s/s application as a controlled variable. The "test" sample Fe wear rate (supposedly proving that auto s/s reduces wear) landed in-between the two preceding samples which purportedly didn't use this function. And yet they claim that this proves their theory? OMGosh ....
Also, note that the other common wear metals were also similarly affected; the wear rates showed no real correlation to the auto s/s sample either, relative to previous samples.
LSJ and this Ford tech are saying that auto s/s reduces wear, via the indirect correlation of ONE UOA, by the method of that function (auto s/s) controlling fuel dilution, which in turn controls oxidation, which in turn controls viscosity, which in turn controls wear ....
There are several UNcontrolled variables in that chain of events, and the outcome they claim to prove their hypothesis is blown away by the variation of a factor of two in the preceding two samples, with absolutely no understanding of averages and standard deviation of baseline or variables.
There is no proven correlation between wear and auto s/s in this video data; none whatsoever.
Without correlation, there can be no causation.
JUNK SCIENCE, my good man. This is some of the worst "testing" seen on BITOG in a long time.
Just because LSJ is a tribologist, does not mean he's good at proving anything.
In fact, I'm comfortable stating that if this is the best he can do, he, in fact, is nothing but a YouTube hack who is being monetized by the unwitting fools who believe his "science". If this is all the better LSJ can do, then I'm going to say his "science" right up there with PF videos.
blog.amsoil.com
