[thelist] Screen Resloution Detection.

Paola Kathuria paola at limitless.co.uk
Mon May 21 10:33:57 CDT 2001


aardvark wrote:
> we seem to still
> be having trouble getting people to recognize that screen resolution does
> not equate to *window* size...

I suspect it's because screen resolution stats are very easy to
come by, for instance from StatMarket and MyComputer.

Even if knowing screen resolution was useful, StatMarket's own
browser stats doesn't describe all, or even half, the visitors
to the sites that they monitor.

I just sampled a few stats pages now and found that the screen
resolution stats only describe 42% of the visitors they claim these
sites have.  That means, if StatMarket publish stats that 70%, say,
people have screen resolutions of 800x600 it's actually 42% of 70%
which is 29%.  They have *no* screen resolution stats on the
remaining 58% visitors.

If you're interested, here's how I arrived at the 42% :

The number of "daily unique" visitors (V) are shown alongside
the number of visitors that screen resolution stats (S) were
collected for.

1) CenterFind  Apr '01  V: 2,092,673  S:       214  0.01% [1]
   CenterFind  Dec '00  V: 2,472,188  S: 1,897,239   78%
2) imsisoft    Apr '01  V:    75,026  S:    31,183   42%
   imsisoft    Dec '00  V:    64,727  S:    29,015   45%
3) microwho    Apr '01  V:     2,760  S:       442   16%
   microwho    Dec '00  V:     4,528  S:       430    9%
4) InetUK      Apr '01  V: 4,915      S:     2,518   51%

Totals                  V: 4,716,817  S: 1,961,041   42%

Stats sources [2]:
1) http://vwr1.hitbox.com/HitboxUI?acct=WQ5011120DEF&p=1
2) http://vwr1.hitbox.com/HitboxUI?acct=WQ500824D2FE94EN0&p=1
3) http://vwr1.hitbox.com/HitboxUI?acct=WQ590901I5RN57EN0&p=1
4) http://vwr1.hitbox.com/HitboxUI?acct=WQ5102141LZB62EN0&p=1

This is compounded by the fact that HitBox (the s/w StatMarket
use to collect stats) can't accurately count *all* visitors to
a web site.  It relies on JS, cookies and images being viewed,
and the HitBox s/w is on sites that are biased to certain
browsers, etc, anyway.

The percentage of visitors used for the other stats seems to be
also less than 100% in StatMarket, even for browser stats.

StatMarket's stats are promoted as something web developers can
use to make informed decisions.  Although the site says "accurate
internet stats", I believe that their terms are more believable:
"WEBSIDESTORY MAKES NO WARRANTY THAT [...] iii) THE RESULTS THAT
MAY BE OBTAINED FROM THE USE OF THE SERVICE WILL BE ACCURATE OR
RELIABLE, [...] AND (V) ANY ERRORS IN THE SOFTWARE WILL BE
CORRECTED." (from http://www.hitbox.com/cgi-bin/page.cgi?disclaimer)

Pah.


Paola, who can also rant about the MyComputer's stats


[1] the 214 is possibly a result of a bug - which is why I 
included Dec 2000 figures.  Although missing out that month
produces a new average of 75% (the percentage of total visitors
that there were screen resolution stats for), since StatMarket
aggregates *all* stats from all HitBox sites, it is fairer to
include the buggy figure (since they will be including it too).

[2] For "Daily unique visitors" - go to Traffic -> Daily Uniques,
click the Year tab and read off the Daily Uniques for the
month.  For the number of visitors that screen resolution stats
were collected, go to System -> Screen Resolutions, click the
Month tab, select a month/year, read off the bold Total visitors.




More information about the thelist mailing list