The smaller the sample, the more suspect the numbers. Torco was only used by one person, and that could've been a newer car still in the break-in phase, we don't know.
It would be much more relevant to calculate standard deviation per motor-oil, but that wasn't done because the results would look terrible. The small samples are completely understandable since the creator (Resolute?) only had access to UOA's submitted by forum members, and the effort was pretty cool considering the effort involved. Still, the results are not reliable for most brands.
|