|Date: ||Mon, 12 May 2003 17:10:00 -0400|
|Reply-To: ||Richard Ristow <email@example.com>|
|Sender: ||"SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>|
|From: ||Richard Ristow <firstname.lastname@example.org>|
|Subject: ||Re: The final and ultimate analysis question|
|Content-Type: ||text/plain; charset="us-ascii"; format=flowed|
Well, let me try a quick one. Others may know better.
At 05:01 PM 5/12/2003 -0400, Chris Fisher wrote:
>Here's what should be the final question...it's a little complicated.
>Generally, the ultimate question for the project is...is there a
>connection between the reception of certain homeless services and
>advancement to self-sufficiency?
>To answer this, I need to see if, for example, a client received one
>or more related services (each a different variable)(e.g., housing,
>life skills, rent loans) and whether the client moved from unstable
>housing (one variable) across the course of being tracked.
What about a survival analysis? Your end-point would be transition to
stable housing; clients that never made it would be 'censored' at the
last follow-up date you have for them; and reception of services would
be independent variables.
(Watch for confounding. Clients who are homeless longer may simply
receive more services, which may produce an apparent perverse effect.)
This is an unusual survival analysis, in that the "good" outcome is
short 'survival', not long; but otherwise, it's pretty typical.
I'll let others comment on the different survival procedures in SPSS,
with which I'm not very familiar.