Date: Fri, 12 Jun 2009 14:26:50 -0700
Reply-To: thumzup <thumzup@GMAIL.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: thumzup <thumzup@GMAIL.COM>
Organization: http://groups.google.com
Subject: =?windows-1252?Q?Re=3A_SAS_=96_multiple_regression_analysis_help=21?=
Content-Type: text/plain; charset=ISO-8859-1
Thanks for the suggestions!
I actually tried proc genmod. Let me attach the code and the output
at the bottom -- can someone help me if I am on the right track?
Unlike multiple regression using (proc reg) which provides an overall
F-test and R-square for the model used, proc genmod only gives out an
output of z-test for each independent variable. How should I know
whether or not the model I used is reliable or significant? How can I
interpret the output??? -- sorry for my statistical ignorance...
Any help or suggestions would be greatly appreciated!
_______________________________________________________________
Here is the code:
proc genmod data=test;
class subj;
model DV1 = IV1 IV2 IV3 IV4 IV5 IV6;
repeated subject=subj / type = ind;
run;
Analysis Of GEE Parameter Estimates
Empirical Standard Error Estimates
Standard 95% Confidence
Parameter Estimate Error Limits
Z Pr > |Z|
Intercept 53.3774 4.6946 44.1761 62.5787
11.37 <.0001
IV1 7.5207 2.8542 1.9266 13.1149
2.63 0.0084
IV2 23.6036 4.2157 15.3410 31.8663
5.60 <.0001
IV3 0.2465 0.0692 0.1108 0.3822
3.56 0.0004
IV4 -0.0192 0.0797 -0.1754 0.1369
-0.24 0.8092
IV5 -0.4015 0.0486 -0.4968 -0.3062
-8.26 <.0001
IV6 -0.1035 0.0876 -0.2753 0.0682
-1.18 0.2373
On Jun 12, 12:40 pm, jeremy.mi...@GMAIL.COM (Jeremy Miles) wrote:
> You should consider multilevel models (with proc mixed) or GEE (with
> proc genmod).
>
> Jeremy
>
> 2009/6/12 thumzup <thum...@gmail.com>:
>
>
>
>
