D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

Re: [LUG] Flash 9 beta

 

William Fidell wrote:
> 
> 2) it is used by ~98% of internet browsers.

Whilst Flash is pretty widespread, I very much doubt the figures put out
by Macromedia.

First they claim 97.3% (+/- 5%) for version 5 or better, which is down a
little in what they call "mature markets" (Europe/Japan/US)

Remember 98% of Internet browsers is not 98% of visitors, the bots are
important too!

Sites have radically different browser patterns (I think one of the
sites Matt used has a majority of users using MacOS, where as most
people see <1%  MacOS), so even if it is true for 98% of all sites they
recruited for the survey from, it is probably not true for your site.

To do this scientifically one couldn't use web recruited candidates, as
some of those may have been recruited from sites using flash
extensively, and thus bias the results (especially as they were using
the results to compare different technologies - I bet java.sun.com has a
lot of users with Java installed, at least when they leave), unless you
can demonstrate there is zero bias in the sites used to recruit.

I suspect the data was massaged to please Adobe. They talk about
corrections for demographics, which assumes your sample survey isn't
representative to start with. Will people who click "Yes" I want to
complete a survey, be more likely than average to click "Yes" when asked
if they want to install a plugin?

Simpler methodology is to create a flash plugin that retrieves an object
via HTTP, and you can compare the retrievals of page with plugin to
retrievals of the object retrieved by the plugin. Put this on widely
accessed page, and you have the number for that page. This number covers
all types of HTTP users agents retrieving your page, which is probably
more useful anyway.

-- 
The Mailing List for the Devon & Cornwall LUG
http://mailman.dclug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html