Date:         Wed, 23 Dec 2009 10:03:56 -0800
Reply-To:     Ted Clay <tclay@ASHLANDHOME.NET>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Ted Clay <tclay@ASHLANDHOME.NET>
Subject:      Re: how to test the signicanace
Comments: To: Jeff <zhujp98@GMAIL.COM>
In-Reply-To:  <6716d5d0912230654m6c946d93k6beb9aa877c98040@mail.gmail.com>
Content-Type: text/plain; charset="us-ascii"

* I think this might help.  One way to do this is to create a data set with
the 4 cells of a 2-by-2 table, and let PROC FREQ do the test of
significance.
  The key is the "WEIGHT" statement in PROC FREQ.
;
Data ready;
  Group=1; admitted='Yes'; Num=Y1;    output;
  Group=1; admitted='No '; Num=X1-Y1; output;
  Group=2; admitted='Yes'; Num=Y2;    output;
  Group=2; admitted='No '; Num=X2-Y2; output;
Run;

Proc freq data=ready;
 Table group*admitted;
 Weight Num;
Run;

You will get a Chi-square test by default.
A 2-tailed Fisher's exact test would be my choice of test, which requires
adding "/exact".
You might also be interested in relative risk statistics and confidence
intervals, which requires you to pay close attention to which is "row 1" and
"column 1".

TC




-----Original Message-----
From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of Jeff
Sent: Wednesday, December 23, 2009 6:55 AM
To: SAS-L@LISTSERV.UGA.EDU
Subject: how to test the signicanace

Two group of people.  all people in G1 is in a health plan, and these in G2
is not
G1  have  x1 member, and y1 hospital admits. hospital admit rate is
hr1=y1/x1
G2 has x2 member and y1 hospital admits. hospital admit rate is h21=y2/x2

I want to test if there is significant difference between G1 and G2.
What kind test should I use?

Thanks.
JP