Date: Sun, 17 Mar 2013 13:55:02 -0500
Reply-To: mcneely4@COX.NET
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: Dave Mcneely <mcneely4@COX.NET>
Subject: Re: Interesting oil analysis articles
In-Reply-To: <CAG0-GTzPxVwjvX5MarcUyQGUEM7pq2cumkqb15sdYPa3qYFedg@mail.gmail.com>
Content-Type: text/plain; charset=utf-8
Frankly, I can't tell that the "experiment" done to determine whether the magnetic drain plug made any difference or not showed anything one way or another. There are two few data points to have any relevance. Certainly, one vehicle, with the only control being to run the vehicle as usual at times with a non-magnetic drain plug, is meaningless. I know, there were two vehicles, but they were different makes, and almost certainly driven differently.
The experiment that could be done, but not given the resources of a consumer, or even a shop, would be to randomly assign vehicles (new) to an experimental group and a control group. Drive each car in each group through identical test runs with exactly identical stops, starts, speeds, and so on, with the only difference being with a magnetic plug in the experimental cars, a non-magnetic plug in the control cars. In this simple setup, there would need to be 20 cars in each group. The experiment could be made more sophisticated so as to have more groups and fewer cars per group, but more cars overall would be required. Data could be collected at set intervals on wear, or on metal content of oil. Analysis would be by ANOVA. If one measured more than a handful of dependent variables per car, then one would run a greater risk of finding a difference in one or more of the experimental variables when no real difference existed, so the number of variables measured should be kept to a minimum, or the analysis done by a more sophisticated design.
The experiment could be run simultaneously with other test objectives, if great care was done not to create any confounding variables in the experiments. More sophisticated experimental design and statistical analysis would be required in that case also.
The "experiment" described in the newsletter is nothing more than an anecdotal report.
Alternatively, metadata analysis could be done from data collected over a course of several years by consumers or their shops (shops better) to determine wear and or breakdown frequencies, or even metal content of oil.
Gut feeling: After all had been done and said regarding the above, a marginal difference might be found, experimentally and statistically meaningful. It would not be meaningful so far as useful life of the vehicle is concerned. But as a gimmick for selling drain plugs, it works now, and data from the processes I've described could be exploited to sell even more, even if the results were negative or even opposite (non-magnetic plug "better"). Marketers are great at finding ways to make data sound wonderful to consumers.
mcneely
---- Roger Whittaker <rogerwhitt1@GMAIL.COM> wrote:
> Dear OIly Newsletter
> Thanks
>
> On Sat, Mar 16, 2013 at 9:28 AM, Stuart MacMillan <stuartmacm@gmail.com>wrote:
>
> > Blackstone Lab's newsletters are always interesting (to me, at least), and
> > some of you might like reading them as well. The latest one answers the
> > oft
> > asked question "are magnetic drain plugs worth the money?" (No.)
> >
> >
> >
> > Starting in April 2012 they purchased several vintage oils off of eBay and
> > analyzed them. So, if anyone is interested in how oil technology has
> > advanced it's fun reading.
> >
> >
> >
> > http://www.blackstone-labs.com/newsletters.php
> >
> >
> >
> > Stuart
> >
> > (I do an oil analysis every year or two)
> >
>
>
>
> --
> roger whittaker 604.414.6266
>
> ----------------------------------------------------------
> Explore printed work at: http://www.prliving.ca/
> View the growing list of video work at:
> http://www.youtube.com/user/LastonLastof#g/u
> http://vimeo.com/42309497
> http://www.prpeak.com/articles/2010/11/29/multimedia/video/doc4c62e5f80d228504902172.txt
--
David McNeely
| 
