Date:         Wed, 7 May 2008 10:29:57 -0500
Reply-To:     William Elliott <elliott.willie@SBCGLOBAL.NET>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         William Elliott <elliott.willie@SBCGLOBAL.NET>
Subject:      Re: Longitudinal data question
In-Reply-To:  <10060895.1210172653624.JavaMail.root@mswamui-thinleaf.atl.sa.earthlink.net>
Content-Type: text/plain; charset="us-ascii"

I have all variables with the exception of savings for college and college
enrollment (because the children were not old enough to enroll in college in
2002) for both 2002 and 2005. It is survey data with weights (PSID for those
who are familiar).

1) Is savings a "yes/no" variable, or do you have the amount saved?
        Yes it is yes/no

2) Is enrollment in college a "yes/no" variable, or do you have type of
college (e.g., public, private, 2 year, 4 year, community)?
        I have it both as a yes/no and as a 2yr/4yr

3) What controls have you got?
Examples of controls:

/*Parent controls*/
Head's education
Marital status
Parent engagement
/*Child controls*/
Race
Gender
/*Academic control*/
Special education
/*Psychological controls*/
Children's self-concept
Children's math efficacy
Children's college expectations
Parent's college expectations
/*Economic controls*/
Household income
Household wealth
Children savings for college

4) What's your sample size?
Approximately 400




-----Original Message-----
From: Peter Flom [mailto:peterflomconsulting@mindspring.com]
Sent: Wednesday, May 07, 2008 10:04 AM
To: William Elliott; SAS-L@LISTSERV.UGA.EDU
Subject: Re: Longitudinal data question

William Elliott <elliott.willie@SBCGLOBAL.NET> wrote

>Hello, I was just wondering what a good method would be for analyzing
>the following question:
>
>Are children with savings for college more likely to enroll in college
>than children without savings for college?
>
>When the savings variable and controls are from 2002 and the dependent
>variable, enrollment in college, is from 2005 (same children)?


Unless you have longitudinal data, it's not really a longitudinal problem.
That would mean you had multiple years of data on at least one variable.
From what you write, it appears that you have one year of data on each
variable.

Some questions:
1) Is savings a "yes/no" variable, or do you have the amount saved?
2) Is enrollment in college a "yes/no" variable, or do you have type of
college (e.g., public, private, 2 year, 4 year, community)?
3) What controls have you got?
4) What's your sample size?


Peter

Peter L. Flom, PhD
Statistical Consultant
www DOT peterflom DOT com