Statistical Method

[Chang,Victor]

Dear all:

While currently working on a report for administrators who have no
statistical background, I am trying to make my project as simple as possible
for better understanding.

I have collected through SPSS and SQL an assessment of victim's profile.  It
consists of the following 5 domains for each individual: living conditions,
financial status, physical/medical status, mental status and social
interaction and support.  For each category, there are 5 responses: severe
problem, problem, no problem, unable to determine and N/A.  Basically, it is
a multiple choice for these 5 domains with only one response being selected
for analysis.

My initial report is to aggregate all of domains by their responses.  The
report looks too much information from the beginning.  My strategy is to
convert each response to score, say, severe problem to 2, problem to 1 and
others to 0.  The sum of the scores will then divided by each month victims'
total.  A higher score will indicate victim's risky factor that resources
may be needed to help them. Thus, a single value instead of 5 responses for
each domain may be easier to communicate.  The report will be broken down by
each month, illustrating the risk trends over the years.  All of the above
procedures will be completed by SPSS.

Does any one have better statistical methods for this work done? Or refer
any research literature for me to review.  Thanks.

Victor
[hidden email]

[Swank, Paul R]

This really does not sound like a good idea to me. First of all you are
adding distinctly non-interval numbers across five domains that may or
may not be related. You could get a 5 by having a problem in all 5
domains or by having severe problems in 2 domains and problems in a 3rd.
Are any of the domains more problematic that others? I would question
the internal consistency of such an index. And I have no idea what you
mean by divided by each month victim's total. Has anyone done the
research which looks at how each domain relates to the likelihood of
victimization? What does that say about the importance of these domains
in predicting victimization?

Paul R. Swank, Ph.D. Professor
Director of Reseach
Children's Learning Institute
University of Texas Health Science Center-Houston

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Chang,Victor
Sent: Friday, June 22, 2007 2:45 PM
To: [hidden email]
Subject: Statistical Method

Dear all:

While currently working on a report for administrators who have no
statistical background, I am trying to make my project as simple as
possible for better understanding.

I have collected through SPSS and SQL an assessment of victim's profile.
It consists of the following 5 domains for each individual: living
conditions, financial status, physical/medical status, mental status and
social interaction and support.  For each category, there are 5
responses: severe problem, problem, no problem, unable to determine and
N/A.  Basically, it is a multiple choice for these 5 domains with only
one response being selected for analysis.

My initial report is to aggregate all of domains by their responses.
The report looks too much information from the beginning.  My strategy
is to convert each response to score, say, severe problem to 2, problem
to 1 and others to 0.  The sum of the scores will then divided by each
month victims'
total.  A higher score will indicate victim's risky factor that
resources may be needed to help them. Thus, a single value instead of 5
responses for each domain may be easier to communicate.  The report will
be broken down by each month, illustrating the risk trends over the
years.  All of the above procedures will be completed by SPSS.

Does any one have better statistical methods for this work done? Or
refer any research literature for me to review.  Thanks.

Victor
[hidden email]