Prev: vuxmdiag
Next: filling in the gaps
From: Espen on 21 Apr 2010 07:53
I have over 700.000 records of pregnancies with variables for
pregnancy length (in weeks) and birth weight. I want to know whether
or not (0/1) the baby weighs less than the population mean (-2
standard deviations) born at the same age (weeks of pregnancy). I
would like to be able to do it in one step, preferably. I've been
looking into proc univariate, but I can't seem to find a solution. Is
there anyone out there with a good idea?
From: Paige Miller on 21 Apr 2010 08:30
On Apr 21, 7:53 am, Espen <espe...(a)gmail.com> wrote:
> I have over 700.000 records of pregnancies with variables for
> pregnancy length (in weeks) and birth weight. I want to know whether
> or not (0/1) the baby weighs less than the population mean (-2
> standard deviations) born at the same age (weeks of pregnancy). I
> would like to be able to do it in one step, preferably. I've been
> looking into proc univariate, but I can't seem to find a solution. Is
> there anyone out there with a good idea?

PROC UNIVARIATE (and almost all PROCs) cannot do comparisons less
than, or greater than, or equal to, etc.

Comparisons must be done in a data step. You store the results of the
comparison for each record, and then you can tabulate them in a SAS
PROC.

--
Paige Miller
paige\dot\miller \at\ kodak\dot\com
 | 
Pages: 1
Prev: vuxmdiag
Next: filling in the gaps