Date: Fri, 2 Oct 2009 09:48:36 -0400
Reply-To: Peter Flom <peterflomconsulting@mindspring.com>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Peter Flom <peterflomconsulting@MINDSPRING.COM>
Subject: Best way to rearrange dyadic data
Content-Type: text/plain; charset=UTF-8
Good morning all
It's been a while since I've asked a question :-)
Before posting this one, I searched SAS-L and found similar questions (including some by me!) but I also found advice that each case is a little different.... so...
I have a data set with data on heterosexual couples. Currently, each line is an indvidual, and there are ~1000 variables. Each line includes
COUPLE - which is the couple ID
ID - which is the individual's ID
SEX - which is the individual's sex
many many other variables, e.g. marital status. So, it looks something like this
---------------------------------------
DATA HAVE
Couple ID Sex A11
10260 10261 Male Single, Never Married
10260 10262 Female Widowed
10380 10381 Male Married
10380 10382 Female Married
10280 10281 Male Married
10280 10282 Female Married
etc for c. 300 couples and 1000 variables
--------------------------------------------
for some parts of this analysis (it's going to be a long project - many months, anyway, with lots of analysis) this is the best order of data. But some of the initial work is checking whether the two members of a dyad agree.
for this, I *think* the best arrangement is
Couple A11M A11F
10260 Single, Never Married Widowed
10380 Married Married
(that's A with two ones)
Then I could just do
PROC FREQ;
table a11M*a11F;
run;
etc.
The nice thing is that every dyad has one male and one female, and the male always has a 1 suffix and the female a 2.
The hard thing is that there are hundreds of variables, some numeric, some character, many with a lot of missing data. And the variables have been named rather capriciously (e.g. the MARITAL variable is A11, other names are such things as D6_M). The variables have good labels, and have formats attached.
Any hints appreciated as always.
Note that I have to be out from 10 to about 2 or 3, Eastern time, so, if I don't reply, that's why.
Thanks as always
Peter
Peter L. Flom, PhD
Statistical Consultant
Website: www DOT peterflomconsulting DOT com
Writing; http://www.associatedcontent.com/user/582880/peter_flom.html
Twitter: @peterflom
