Posted by Riccardo Pentenero (194.116.8.5) on October 07, 1999 at 09:49:03:
In Reply to: I cannot agree with your math posted by Ueli on October 07, 1999 at 09:11:45:
The objective is to have a general average, the group average are OK
-my english DOES NOT permit me to continue for a long time on these arguments, i risk to sink....:-) -
But If i would like to have a general average, or if i want a comparison between groups i have to use the data of the two groups weighted by the fraction of net access of this age group.
With a hyperbole, If i have 100 "young" with 20 successes and 1 "old" with 1 success, i can say the % is 20.8 on the SAMPLE i have, a 20 % on young and (mmmmh i know that the sample is not too valid....) a 100% on old.
But in the "universe", where i have not a parity of presence within the age groups i cannot say that is the same, but i have to increase the weight of the group with less global accesses on the WEB.
The "old" sample have to increase the % much more than the .8%, i have to imagine there are many "old" not connected.
I know that is an approximation, but in my opinion is more true than the other (yes, as the politicians :-) )