Date: Tue, 25 Oct 2005 13:35:06 -0400
Reply-To: dhssresearcher@NETSCAPE.NET
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Frank Schiffel <dhssresearcher@NETSCAPE.NET>
Subject: Re: SUDAAN proc rlogist check
Content-Type: text/plain; charset=iso-8859-1
First off, why bother?
There have been studies that the uber rich die of the same old common complaints that other lesser mortals do. Stroke, heart conditions, pneumonia, etc etc etc. Any frequency count of death data will tell you that. Same with top ten admission types in a large database. pneumonia, heart conditions are up there.
Anyway. Stroke is irrelevant to income. Where's the medical connection? Lack of medical care? No primary care? No affordable ER care? No treatment of the standard stroke causes, i.e. diabetes, heart conditions, the usual drill.
Income isn't the underlying condition.
Its one of those questions that may be statistically significant, but not medically significant. Might get you a pub in a journal if that's your point.
regards
frank
ps. I'm one of those who has read way too many crappy proposals to use restricted data.
Adam <dprkexchange@HOTMAIL.COM> wrote:
>Hi everyone,
>
>I'm fairly sure I'm reading this SUDAAN output (and input) correctly,
>but if someone has a minute to double check to make sure I'm not making
>any obvious mistakes, it would be greatly appreciated.
>
>The program is a simple logistic regression using SUDAAN 9.0's proc
>rlogist, one dependent variable, one independent variable, and a set of
>raked weight variables from the jackknife design (all data from
>California Health Interview Survey 2003).
>
>Purpose of program: to model the odds of reported stroke among those
>reporting $20,000 annual income or less compared to those earning more
>than $20,000 yearly in the survey.
>
>My understanding: According to the output and my understanding of the
>input, the odds of reported stroke of 1.4 among those with annual
>household incomes equal to or less than $20,000 compared to those with
>incomes greater than $20,000.
>
>Thanks in advance for your help.
>Adam
>
>Here's "the goods" (the program and output):
>
>data temp1;
>set chis.pufa1;
>where ac6 > 0;
>if ac6 = 1 then stroke = 2; /**stroke yes**/
>if ac6 = 2 then stroke = 1; /**stroke no**/
>if ak22_p =< 20000 then lowincome = 2; /**low income yes**/
>if ak22_p > 20000 then lowincome = 1; /** low income no**/
>if (ab22 = 1 or ab22 = 3) then diabetes = 2; /**diabetes or
>pre-diabetes yes**/
>if ab22 = 2 then diabetes = 1; /**diabetes no**/
>run;
>
>proc rlogist data=temp1 filetype=sas design = jackknife;
>weight rakedw0;
>jackwgts rakedw1--rakedw80 / adjjack=1;
>recode stroke = ( 2 ) lowincome = ( 2 ) doct_yr = ( 2 );
>model stroke = lowincome; /* srsex srage_p doct_yr ovrwt ab22 ab29
>ab34;*/
>/*class lowincome;*/
>run;
>
> The SAS System 11:38
>Tuesday, October 25, 2005 896
>
>
>
> S U D A A N
> Software for the Statistical Analysis of Correlated Data
> Copyright Research Triangle Institute February
>2005
> Release 9.0.1
>
>
>Number of zero responses : 7919
>Number of non-zero responses : 749
>
>Independence parameters have converged in 5 iterations
>
>Number of observations read : 8668 Weighted count:
>3770275
>Observations used in the analysis : 8668 Weighted count: 3770275
>Denominator degrees of freedom : 80
>
>
>Maximum number of estimable parameters for the model is 2
>
>Sample and Population Counts for Response Variable STROKE
> 0: Sample Count 7919 Population Count 3437417
> 1: Sample Count 749 Population Count 332858
>
>R-Square for dependent variable STROKE (Cox & Snell, 1989): 0.002172
>
>-2 * Normalized Log-Likelihood with Intercepts Only : 5175.69
>-2 * Normalized Log-Likelihood Full Model : 5156.84
>Approximate Chi-Square (-2 * Log-L Ratio) : 18.85
>Degrees of Freedom : 1
>
>Note: The approximate Chi-Square is not adjusted for clustering.
> Refer to hypothesis test table for adjusted test.
>
>
>
>Date: 10-25-2005 Research Triangle Institute
> Page : 1
>Time: 12:41:45 The LOGISTIC Procedure
> Table : 1
>
>Variance Estimation Method: Replicate Weight Jackknife
>Working Correlations: Independent
>Link Function: Logit
>Response variable STROKE: STROKE
>by: Independent Variables and Effects.
>
>------------------------------------------------------------------------------------------------
>Independent
> P-value
> Variables and Beta Lower 95% Upper 95%
> T-Test
> Effects Coeff. SE Beta Limit Beta Limit
>Beta T-Test B=0 B=0
>------------------------------------------------------------------------------------------------
>Intercept -2.47 0.06 -2.59
>-2.34 -39.57 0.0000
>LOWINCOME 0.33 0.10 0.14 0.53
> 3.35 0.0012
>------------------------------------------------------------------------------------------------
>
>
>Date: 10-25-2005 Research Triangle Institute
> Page : 2
>Time: 12:41:45 The LOGISTIC Procedure
> Table : 1
>
>Variance Estimation Method: Replicate Weight Jackknife
>Working Correlations: Independent
>Link Function: Logit
>Response variable STROKE: STROKE
>by: Contrast.
>
>-------------------------------------------------------
>
>Contrast Degrees
> of P-value
> Freedom Wald F Wald F
>-------------------------------------------------------
>OVERALL MODEL 2 1014.99 0.0000
>MODEL MINUS
> INTERCEPT 1 11.22 0.0012
>INTERCEPT 1 1565.75 0.0000
>LOWINCOME 1 11.22 0.0012
>-------------------------------------------------------
>
>
>
>
>Date: 10-25-2005 Research Triangle Institute
> Page : 3
>Time: 12:41:45 The LOGISTIC Procedure
> Table : 1
>
>Variance Estimation Method: Replicate Weight Jackknife
>Working Correlations: Independent
>Link Function: Logit
>Response variable STROKE: STROKE
>by: Independent Variables and Effects.
>
>-----------------------------------------------------------
>Independent
> Variables and Lower 95% Upper 95%
> Effects Odds Ratio Limit OR Limit OR
>-----------------------------------------------------------
>Intercept 0.08 0.07 0.10
>LOWINCOME 1.40 1.15 1.71
>-----------------------------------------------------------
>
__________________________________________________________________
Switch to Netscape Internet Service.
As low as $9.95 a month -- Sign up today at http://isp.netscape.com/register
Netscape. Just the Net You Need.
New! Netscape Toolbar for Internet Explorer
Search from anywhere on the Web and block those annoying pop-ups.
Download now at http://channels.netscape.com/ns/search/install.jsp
| 
