radja-ga:
As a summary of our discussion:
I referred you to a press release from OneStat, a web analytics firm.
From my initial comment:
--------------------------------
"In their
most recent tally (Jan 2004), they found the following global
breakdown of browsers in use:
1. Microsoft IE 6.0 68.1%
2. Microsoft IE 5.5 13.8%
3. Microsoft IE 5.0 11.8%
4. Mozilla 1.8%
5. Opera 7.0 0.8%
6. Microsoft IE 4.0 0.7%
7. Safari 0.48%
http://www.onestat.com/html/aboutus_pressbox26.html
Your question asks about users in the US in particular, whereas these
numbers are global. However, as US-based users represent a
significantly large proportion of the total online user population,
these numbers can probably be applied directly. This means that 26.3%
of users (back in January) used IE5.5 or lower.
(The stats regarding non-IE browsers are included in the press
release, that's why I've quoted them above even though you didn't want
them)."
--------------------------------
Upon your clarification that you are looking for information with a
precision of +/- 1-3%, I provided the following analysis:
--------------------------------
"The methodology used in generating the numbers in the OneStat press
release used a limit of 20,000 unique visitors from each of 100
countries, giving a sample margin of error of +/-0.71% when looking at
any one particular country. The overall sample margin of error is
+/-0.071%. However, these statistics are based on the approx. 50,000
sites worldwide that use OneStat's analytics service, which represents
sample margin of error of 0.45%. If you assume that this sample of
50,000 sites is representative of the worldwide distribution of web
servers, then we have a total error range of approx. +/-0.52%.
The challenge is whether you feel that we can make this assumption. If
so, then the numbers listed in the press release are the numbers you
are looking for (remembering that a sample culled from 50,000 websites
has lower margin of error than a sample culled from a much lower
number of websites such as provided in the example links mentioned in
lizzardnub's comment below).
If you do not accept this assumption (that the global distribution of
web servers is representative of the US situation), then it will be
necessary for you to directly contact the large web analytic firms
such as OneStat or Nielsen-NetRatings, to purchase their
country-specific data (which will have higher margins of error due to
smaller sample sizes, but should still be well within your desired
+/-3% error as even a sample culled from 1000 websites will do)."
--------------------------------
If you do wish to pursue this directly with the leading web analytics
firms, here are their contact pages:
OneStat
http://www.onestat.com/html/contactus.html
Neilsen//NetRatings
http://www.nielsen-netratings.com/contact.jsp
I hope that this helps!
aht-ga
Google Answers Researcher |
