web visitors?

[Stanley A. Gorodenski]

The counter increasing when you go from some place in the web site back
to the home page could still remotely be caused by someone else hitting
the web site between the time it took you to go to and from that part of
the web site back to the home page.  You might do this test ten times
and it could still be possible that, by a remote chance, there were 10
valid visitors.  You should conduct your test a few thousand times
(maybe even more than that - the more the better) so that you can make
some kind of probabilistic statement of the web site not accurately
reflecting visitors.  Something like, the probability of getting 10,000
visitors when 10,000 transactions in the web site were made, and when on
average there are only x number of hits per day is .000000000001.  Of
course, that would not prove anything.  There still could have been, by
a very very remote chance, 10,000 valid visitors when you conducted your
test 10,000 times. 

> fine.  If not, there is a problem.  Also, you can do what I did.  Go to the
> site, make a note of the visitor number and leave.  Come back in about one
> minute or two and you will see that the number has gone up by one - your
> second visit.  Now, move around a good bit on site and go back and  forth
> to the home page and see if the number changes most of the time you hit
> home.  If this happens it is counting hits not visits.
> 
>

 
 ------------------------------------------------------------ 

   For subscription and related information about LEPS-L visit:

   http://www.peabody.yale.edu/other/lepsl