Oh, and I would also like to say that the change is likely greater than the reported 1% since both Opera and Mozilla browsers are capable of identifying themselves as MSIE.
I used to be a student of statistics. I have seen a lot of prople argue here that WebSideStory's statistic may not be scientific or significant enough. Whether or not it is scientific enough remains to be debated. Yes it does use a sample size, but the sample can be assumed to be reasonably random, if only for its large size.
Now, the significance part. Some have argued 1% is not much. You just can't say that statistically. It depends on what the rangle of 95% confidence level (this 95% has NOTHING to do with MS's browser market share) is. And since the article clearly mentions that IE has been steady at 95.7% since June 2002. Then, the real situation is this: It has been 95.7% for 24 months, and then declined in ONE month to 94.7%. The estimated standard deviation is far less than 0.5% (I would estimate definitely no more than 0.05% by the way it is reported, only one place after decimal point). Therefore, any change of over 0.1% over one month is certainly statistically significant. 1% is 10 times that amount. It is a VERY significant change - more than most people realize.
The problem with looking at 95.7 and 1 is that 1 does look very insignificant. But when total amount of time, and other statistical analytical methods are taken into account, that 1 is more than enough to ring warning bells all over Redmond.
Slashdot Mirror
Oh, and I would also like to say that the change is likely greater than the reported 1% since both Opera and Mozilla browsers are capable of identifying themselves as MSIE.
I used to be a student of statistics. I have seen a lot of prople argue here that WebSideStory's statistic may not be scientific or significant enough. Whether or not it is scientific enough remains to be debated. Yes it does use a sample size, but the sample can be assumed to be reasonably random, if only for its large size.
Now, the significance part. Some have argued 1% is not much. You just can't say that statistically. It depends on what the rangle of 95% confidence level (this 95% has NOTHING to do with MS's browser market share) is. And since the article clearly mentions that IE has been steady at 95.7% since June 2002. Then, the real situation is this: It has been 95.7% for 24 months, and then declined in ONE month to 94.7%. The estimated standard deviation is far less than 0.5% (I would estimate definitely no more than 0.05% by the way it is reported, only one place after decimal point). Therefore, any change of over 0.1% over one month is certainly statistically significant. 1% is 10 times that amount. It is a VERY significant change - more than most people realize.
The problem with looking at 95.7 and 1 is that 1 does look very insignificant. But when total amount of time, and other statistical analytical methods are taken into account, that 1 is more than enough to ring warning bells all over Redmond.