Re: [Summary-Talk] average view time

[Bill_Royds@xxxxxxxxx]

Thanks for your response. You seem to create such a lot of detail with 
low 
resources that I hoped you could also have a magic approach to median.

 Perhaps one good way to look at page view time would be percentiles in 
time ranges (<30 secs,  30sec - 1 Min,1-2Min, 2-5 min, 5+). The two hump 
clumping would seem to indicate that many pages are either not what 
visitor wanted (very short time) or are very important to visitor (what 
we 
want) or very difficult to understand (what we don't want). A range would 
help find these pages.


Bill Royds
Knowledge, Information and Technology Services
Department of Canadian Heritage
Tel: (819) 994-0507





Jason Linhart <jason@summary.net>
Sent by: owner-summary-talk@lists.summary.net
07/01/2004 10:30
Please respond to summary-talk

 
        To:     <summary-talk@lists.summary.net>
        cc: 
        Fax to: 
        Subject:        Re: [Summary-Talk] average view time


On 1/7/04 9:53 AM Bill_Royds@pch.gc.ca (Bill_Royds@pch.gc.ca) wrote:

>Is there an ability get the median view time statistic, which is a better 

>measure than average for skewed distributions?
>Median is time for which half samples have less, half more.

Calculating a median is significantly more complex, in both memory and 
CPU usage, so Summary does not attempt to calculate them. Casual testing 
on simple web sites shows that the average view time is often very close 
to the median view time in the common case.

However there are often interesting things going on at the highest levels 
of detail. View times often clump up near zero and near another number 
related to the complexity of the page. The average and the median 
generally fall between these two clumps. Some pages show even more 
complex behavior, with clumping at more than two values. Tracking this 
level of detail for every page in a site is complex, and it seemed to me 
that few people would be interested in it, or interested enough in it to 
justify the large multiple on the memory usage that calculating it would 
entail.

Jason

-----------------
Jason@Summary.Net
-----------------
Dr. Seuss books . . . can be read and enjoyed on several levels. For
example, 'One Fish Two Fish, Red Fish Blue Fish' can be deconstructed
as a searing indictment of the narrow-minded binary counting system.
  -- Peter van der Linden, Expert C Programming, Deep C Secrets
-------------
Go to <http://summary.net/list.html> to update subscription info.