Date: Sat, 22 May 2004 00:21:31 -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: LOOP - END LOOP help (i think)
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 10:41 AM 5/21/2004, alex lantsberg wrote:
>I have a 7400 case dataset representing a workforce with a current
>tenure (in years) as well as other attributes. [...] I would like to
>model how the composition of that workforce changes after a certain
>date (between old employees and new employees) over the next six years
>based on an imputed probability of separation/survival constant (S)
>based on full/part time designation.
>My basic strategy is to have SPSS creat a set of seven variables (t)
>for each year of the six years i'm looking out into the future (y),
>with the probability (p) that someone occuping position (yt) is new -
>essentially the equation
>p_y_t =p_[y-1]_[t-1]*S where p_y[1...6]_t = 1-S
>I am trying to tell SPSS to do that for me using the LOOP-END LOOP command.
First: I don't think LOOP/END LOOP will help you. LOOP loops *within a
single SPSS case (or record) only*; it doesn't do calculations that
extend across several cases.
Second: Your problem isn't quite clear. If I read you correctly, you
want to calculate the likelihood, from your data, that a full-time
employee will still be with the firm after 1, 2, ... 6 years, and the
same quantities separately for part-time employees. That sounds like a
total of 12 new quantities for the whole data, 6 for full-time and 6
for part-time employees; not really "seven [new] variables".
Third: I don't know what logic you want to use for this -- the actual
logic; skip the SPSS syntax, for now. It may not be possible. If you
have a set of employees at one point in time, and know how long each
has been with the firm, there's no information about how many joined in
the previous six years (say), and how long they stayed before leaving.
And you don't know final tenure of the present staff, either; only that
it's at least as long as their current tenure.
If you have data on new hires over time, and how long they served
before leaving, you can do something with this, at least if you're
willing to assume that the future will be like the past; but with what
you say you have, I don't see how.