🚀 go-pugleaf

"Jim Ley" <jim@jibbering.com> wrote in message
news:3ca72566.11210124@news.cis.dfn.de...
> On Sat, 30 Mar 2002 15:10:08 -0500, "Jim Davis"
> <newsmonkey@vboston.com> wrote:
>
> >Looking at the Superstats page we can infer (although we can't see
directly
> >as they don't, in my opinion, make very good use of the data collected)
the
> >following for the site being looked at:
>
> Well as they're javascript isn't very good. relying on browser
> detection, even the statistics they do gather have to be regarded as
> dubious - the sample is obviously biased towards those browsers they
> do detect properly.  The examples page is also on a very limited
> sample size 80,000 chosen by visitors to a page is not statistically
> significant for anything other than the page in question...

I agree.  The more I examine the site to more I see it as oppurtunity lost.
Most of the data is ambiguous and pretty misleading.

> >So now let's attempt to related this to the Superstats page.  Well you
can't
> >get a client area at 800x600 of 800 or greater unless you make the window
> >larger than the available viewing space.  However about 60% of browser
> >widths are 800 or greater!
>
> Of course you can, you can have a window of at least 10,000 by 10,000
> on any windows OS I've seen, I don't know what the limit is, but
> screen size certainly isn't it.

Shot that one off before finishing the sentence, did you?  ;^)  Note I say
(just a few lines above): "unless you make the window larger than the
available viewing space."

From a usability perspective however the point is moot.  The abilty to
enlarge a window bigger than the screen res IS there, but the number of
people that do it out of 10,000 can be counted on a leper's bad hand.

> >The numbers just don't add up.
>
> Interesting analysis, but as you notetotally flawed due to the data
> available.

Sadly so.  It's annoying however.  I'd rather show less, better data (which
I think I do with my res stats) than lots and lots and lots of broken data.

This leads to "binder blindness" (my term): the old school-boy idea that
anything in a neat plastic binder is inheirently more "correct" and
deserving of higher marks.  As long as the reports look good (and these do
look pretty good - all professional and all) then the data will be seen as
more relevant.

Edward Tufte book's cover this topic in some detail (more from the much more
valid point of data correlation however).

I just finished a project revamping some web site metrics reports.  The
reports had been generated in 4-year old Perl code: they hadn't been correct
in 3 1/2 years.  But they "looked right" and so were used for all sorts of
business decisions.  Even after knowing this however they've rejected my
proposal to set aside budget money to periodically test the stats for
continuing accuracy.  Some people never learn.

> >So, I guess what I'm (still) saying is that screen resolution data isn't
> >irrelevant.
>
> So how is it relevant, and what do you do with it once you get it?

Screen resolution alone is relevant in some ways.  I made some statements
further up the thread:

"Resolution is an inidication of monitor sizes (it's NOT monitor sizes) and
can thus be used as an indication of usage/buying trends.  Certain
resolutions are so device dependent (for example 320x240) that they can be
used to indicate the audience using that device (in this case PocketPC style
handheld devices).

Very (VERY) roughly resolution data can be used to make inferences on
technical level of the audience.  A group using predominantly very large
resolutions can, at this point at least, be considered more technical than
groups made up of lower (but not lower than 640x480 which would indicate
handheld users) resolution users."

Now the stats that I present are more relevant as they also correlate the
data with the reported color depth.  This, I think, is the key usage for
that report.  I also think that the resolution thumbnails are very useful
even if only to represent very clearly the differences in sizes (I'm
actually pretty proud of this).

I think that too many designers "abstract" smaller resolutions.  They look
at their paint program's ruler but rarely look at their actual sites in
those resolutions.  I hope that the thumbnails will drive the difference
home to those more visually-minded.

In the end tho' it's still only one small part of the puzzle.  Even if
(when) I do display more complete metrics on browser width, etc it's still
only those browsers that both can and do report their stats being counted.
And no matter how well I (or anybody) track things it's still only those
people that choose to visit my site.

My stats are just another data point (and, depending on the questions you're
asking, possibly not a very good one).  Just because some people may misuse
them is no reason not to present them for those that won't.

I would LOVE somebody like Yahoo to take this job on seriously and do some
good reports on what they can gather.  That's the kind of information we
need to make decent generalizations: (good) information from massively
popular general interest sites.

Jim Davis
--
President, Depressed Press of Boston:  http://www.DepressedPress.com/
Webmaster, Virtually Boston:  http://www.vboston.com/
Webmaster, First Night Boston:  http://www.firstnight.org/
Senior Consultant, Metlife eCommerce IT:  http://www.metlife.com/