July 2005 is a huge milestone: Linux just overtook Windows 95... 10793
 |
And your report is 'substantiated' how?
It covers a lot of things besides Linux. For example, on my "atheist" page (which gets hits primarily from people hitting search engines), here're the stats, after removing anything with 'bot' in...
Wait, you tell me an *anecdote* about *one* friend, and then lecture me about 'representative samples'? Admit it, you can't keep a straight face when you're typing this dreck, right?
I claim it's non-representative because the sample size is far too small, and limited to an incredibly narrow geographic area, and the counting methodology is unknown even to *you*.
Now, even if we buttume that your counting is accurate, without knowing the population of the subdivision it's not possible to buttign a confidence value to your study. Let's buttume one member of a family visits the site, and the average family size is 4, so we've got about 1600 people there. Now (again buttuming the count's accurate), you can be 95% certain that between 96.81% and 99.19% of the people *in your subdivision* use MSIE.
But no statistician will allow you to extrapolate this data beyond the subdivision. Consider if I did a poll of ~400 people in a single neighborhood in a particular state and used it to project, say, the proportion of Democrats, Republicans, Libertarians, etc. across the nation. Do you think the results might be suspect? How might the results vary if I picked a neighborhood in Boston, MA or a small town in Arizona? How about a suburb of Orlando, or the neighborhood of South Beach, Miami, both in Florida?
Okay, then, you should have no trouble here: What's the statistical confidence of your results, when extrapolated to the population of the United States (buttume 300e6), in a numerical percentage? Now, extrapolate to the world population (buttume 6e9). An actual number is required. Now, to achieve a 95% confidence level with, say, a +-2% interval, what sample size of the United States population do you need?
Some information on sampling bias:
-- Sincerely,
Ray Ingles (313) 227-2317
The irony is that Bill Gates claims to be making a stable operating system and Linus Torvalds claims to be trying to take over the world. - Drakmere