Date: Fri, 31 Jul 2020 14:10:24 -0700
Reply-To: David McNeely <davmcneely40@GMAIL.COM>
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: David McNeely <davmcneely40@GMAIL.COM>
Subject: Re: Project Farm Battery Test
In-Reply-To: <BN6PR22MB0115FEE72725FCD813CC9A58BA4E0@BN6PR22MB0115.namprd22.prod.outlook.com>
Content-Type: text/plain; charset="UTF-8"
Thank you David. I see that I edited poorly. Where I wrote "standard
deviation," I should have written "standard error," once more. mcneely
On Fri, Jul 31, 2020 at 2:00 PM David Boan <dboan@outlook.com> wrote:
> What a refreshing response after seeing such a lack of critical thinking
> on much of social media. Thank you.
> ------------------------------
> *From:* Vanagon Mailing List <vanagon@gerry.vanagon.com> on behalf of
> David McNeely <davmcneely40@GMAIL.COM>
> *Sent:* Friday, July 31, 2020 4:43 PM
> *To:* vanagon@GERRY.VANAGON.COM <vanagon@GERRY.VANAGON.COM>
> *Subject:* Re: Project Farm Battery Test
>
> Well ......... . Anyone who knows anything about experimental design knows
> that this "experiment" is deeply flawed. Every time you measure something,
> the value you get is an estimate. It takes a series of measurements of
> each of a sample of items to give a mean value for items of that type. One
> measurement is not a fair estimate. Typically, in an experiment to compare
> two different sorts of things, one wants a mean value from measuring some
> 20 individuals of each sort. Then one can calculate a mean and standard
> error. Roughly, a large overlap in standard error would then suggest that
> the two sorts are no different from each other. No or a small overlap in
> standard deviation would suggest that the two sorts of things are
> different. It's a bit more complicated than that, but that's the idea.
> Getting two different values from measuring one individual of each sort
> like was done in this "experiment" might be within experimental error.
> Next time the experiment is done, the values might be reversed between the
> two sorts. Experimental design theory generally suggests that it takes 20
> individual measurements, one measurement of each of 20 individuals, to give
> a fair estimate of the true mean value.
>
> Bottom line: Based on this "experiment," I do not know that one of the
> battery brands is better than another. In fact, if one Wal-Mart battery
> were compared with one "identical" other Wal-Mart battery, the same
> difference between the two as between the Wal-Mart and the Autozone battery
> might show up.
>
> You can take all this for whatever you deem it to be worth, just as I will
> take the "experiment" in the video for what I deem it to be worth.
>
> mcneely
>
> On Fri, Jul 31, 2020 at 10:56 AM Jack R <jack007@comcast.net> wrote:
>
> > Very educational.
> >
> > https://www.youtube.com/watch?v=h7rTcBanpMk
> >
> >
> >
> > Interesting independent test of batteries
> >
>
|