Date:         Thu, 12 Mar 2009 12:41:35 +0100
Reply-To:     Victor Tarrago <vtarrago@acziona.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Victor Tarrago <vtarrago@acziona.com>
Subject:      Deviation
Content-Type: multipart/alternative;

Dear Colleagues,



Could you please help me with the following problem?



I have two independent measures of the same proportion and, as I am worried
that they might contradict each other, want to measure how good they fit
with each other (or how strong they deviate from each other).

We are talking about data that describe the proportion of incidence of 15
different events (technical problems).



The first measure is a probability measure (percentage) for each of the 15
events - coded into 15 variables [0;100]. The sum of these 15 variables =
100.

The second measure is an absolute value (# of occurrence) for 3 of the
events (the most frequent ones, therefore which 3 of the 15 variables is
quantified varies from case to case) - coded into 3 variables [1;700].

In case it matters, I am working with a file of 416 cases.



How can I measure the fit/deviation of these two measures?

Thank you very much in advance for your suggestions!



Best wishes,



Ulrich Schubert

Product Development

C/Llull 321 Edifici CINC

08019 Barcelona

Tel.  + 34 93.55.14.426

Fax. + 34 93.55.14.427




[text/html]