Posted by mbstevens on 07/07/06 22:10

On Fri, 07 Jul 2006 21:27:49 +0000, Chris Tomlinson wrote:

> Seriously though, why does our hit counter on the server report this
> statistic if it is false?

As A.F. explained to you already,
the reason you are getting low numbers
is likely that anyone who has visited with JavaScript turned off will not
return to your site again.

> Why have other programmers, some very respected
> in the Google Maps community, agreed with us that JS is off in less than 1%
> of browsers.

Site an actual study.. Your 'other' 'very respected'
programmers' opinions do not a reasonable statistic make.
I have never seen a study that says that.

>> He might start to correct his deficiency by reading the very short
>> old 1950's classic by Darrel Huff, "How to Lie with Statistics".
>
> Totally bizarre you guys. I'm normally able to think laterally but you've
> defeated me :-S Quoting exactly what our server counter says, and what
> others have agreed with, isn't really a "lie" now is it. Come on.

As I said, your intent does not seem to be to lie. You just don't
understand that a useful statistic takes more than a number.

From Huff's book: "An expressed preference by a 'cross section' of a
magazine's readers for articles on world affairs is no final proof that
they would read the articles if they were published." He might have just
as well have been talking about your "over 80% of a random selection of
people felt it was a good idea in our market research."

That's the kind
of mistake you're making. You keep throwing out number as if they are
actually meaningful and important. What is important is whether the same
80% will actually use your site to buy things. And how did you get
a number that is _exactly_ 80%? It seems likely that your sample size was
small indeed, or that you're rounding. Are you rounding, is it up, or
down? You didn't give us your sample size -- what is it? Is it large
enough to be significant? How were the participants
questioned? How was the question worded?
What kind of _population_ was your 'random sample' taken from?

[Back to original message]