Quote:
He is not stating that he advocates less data per se. But that he actually goes through from experience different methods than most to get accurate consistent data.
Hard to understand what you're saying here, but it sounds to me like you're assuming that no one else has learned proper laboratory procedures. I believe I had it well nailed down by the time I completed my first two undergraduate Chemistry labs 31 years ago. Three science and engineering degrees and 25+ years engineering experience later, I think I got it down to where I can smell a bad procedure from 1000 yards away, while blind folded.
Quote:
On another thread he was mentioning how you could get pre-cleaned bottles for sample or just use a double rinse method. There is more to it than just sticking a bottle in the stream. Did you clean your sample thief etc etc.
See the above.
Quote:
Now with that and a lab he trust for consistency he got results based on his experience that verified what he already suspected. Could he have had a bad sample sure. Did he? Probably not. It just backed up data from years of experience and hands on practical application testing.
Good point on lab consistency. But unfortunately George has yet to give us any details HIS PROCEDURES and the lab he used. Further he has yet to give us any details on the procedures his lab used to generate the data, even though he has been repeatedly asked.
Quote:
Now with good samples I am sure he'd have no problem comparing notes and seeing it proven again. However the first, second, or third result that did not agree, he is cautioning that, that would not mean the filters were sub-standard or even that there is a problem at all. That is probably his major concern.
Again, you're assuming George's data is somehow automatically more valid than anyone else's. If you want to BELIEVE that, that's fine. But that isn't how science works. Science is based on repeatable experimental results than can be verified. That's why I've been advocating more than 1 or 2 results. I'd like to see at least 5 test runs with all the data more or less pointing in the same direction before I'd be willing to say we're seeing a significant difference.
If the accumulated data is inconsistent, then maybe we need to look at why. Maybe it's the competence of the labs, or the PC procedure, or even different sampling procedures. But you don't start out automatically assuming Goerge's data is correct and that if everyone elses's doesn't agree, they all must be wrong.
And finally, which I'll be repeating this for the umpteenth time, there's a distinction between a "significance difference" and the "size of the difference" and how that correlates with various cost/benefit models.