==================================================================
The gateway between this list and the sci.stat.edu newsgroup will
be disabled on June 9.  This list will be discontinued on June 21.
Subscribe to the new list EDSTAT-L at Penn State using the web
interface at http://lists.psu.edu/archives/edstat-l.html.
==================================================================
.
Hello,

I am facing a challenge that somebody else on the list might already
have solved and so I decided to post.

I would like to estimate the lenght of time a given user spent on the
last web page viewed during an interaction with a given web site. That
is, while I can quite easily compute the approximate lenght of viewing
any page before the last by simply substracting the datetime stamp of
the current web page from the following one, I do not know when a user
left the web site.

I thought I use the following procedure:

=> Given the X is the last page of the specific interaction between user
A and web site B, I take the average page X viewing time for all users
of  web site B
=> I compute the average page viewing time for all users, that is how
long on average any page of the web site is viewed
=> I compute the average page viewing time for user A, provided he/she
has viewed at least Y pages, that is how long on average any page of the
web site is viewed by user A
=> I compute the "speed factor" for user A as the ratio between
user-specific and overall web site average page viewing time (the ratio
of the prior two measures)
=> I compute the average viewing time for user A for that session
=> I compute the estimated viewing time for user A on page X as the
average viewing time for user A for that session times the speed factor

I am just wondering how reliable can be a procedure as this. I suspect
reliability to be weak when user specific data are scarse, as well as
when page X happens to be very often a web site exit page and little
information about page X viewing time exists. I am also quite convinced
I should use some sort of weighting to correct for high variability and
low reliability of computed averages, but I am not sure which to use and
how to use it.

Can you provide any suggestion/references?

Thank you very much in advance,
Luca

Mr. Luca Meyer                          
Consumer research advisor: http://www.lucameyer.com/en/ 
Italian Online Research Mailing List:
http://it.groups.yahoo.com/group/ior
Tel: +390122854456 - Fax: +390122854837 - Mobile: + 393355217628

- One world, one human race -     


Reply via email to