Logistic Regression: A Self-learning Text, Third Edition (Statistics in the Health Sciences)

Working Correlation Matrix

Col1 Col2 Col3 Col4 Col5 Col6 Col7 Col8 Col9

Row1 1.0000 0.5254 0.2760 0.1450 0.0762 0.0400 0.0210 0.0110 0.0058
Row2 0.5254 1.0000 0.5254 0.2760 0.1450 0.0762 0.0400 0.0210 0.0110
Row3 0.2760 0.5254 1.0000 0.5254 0.2760 0.1450 0.0762 0.0400 0.0210
Row4 0.1450 0.2760 0.5254 1.0000 0.5254 0.2760 0.1450 0.0762 0.0400
Row5 0.0762 0.1450 0.2760 0.5254 1.0000 0.5254 0.2760 0.1450 0.0762
Row6 0.0400 0.0762 0.1450 0.2760 0.5254 1.0000 0.5254 0.2760 0.1450
Row7 0.0210 0.0400 0.0762 0.1450 0.2760 0.5254 1.0000 0.5254 0.2760
Row8 0.0110 0.0210 0.0400 0.0762 0.1450 0.2760 0.5254 1.0000 0.5254
Row9 0.0058 0.0110 0.0210 0.0400 0.0762 0.1450 0.2760 0.5254 1.0000

Analysis of GEE Parameter Estimates

Empirical Standard Error Estimates

Parameter Estimate Standard Error 95% Confidence Limits Z Pr>|Z|

Intercept 1.3978 1.1960 3.7418 0.9463 1.17 0.2425
BIRTHWGT 0.0005 0.0003 0.0011 0.0001 1.61 0.1080
GENDER 0.0024 0.5546 1.0846 1.0894 0.00 0.9965
DIARRHEA 0.2214 0.8558 1.4559 1.8988 0.26 0.7958

Contrast Estimate Results

Label Estimate

Standard

Error

95% Confidence

Limits

Chi-

Square Pr>ChiSq

log odds ratio
(DIARRHEA 1 vs 0)

0.2214 0.8558 1.4559 1.8988 0.07 0.7958

Exp(log odds ratio
(DIARRHEA 1 vs 0))

1.2479 1.0679 0.2332 6.6779

Contrast Results for GEE Analysis

Contrast DF Chi-Square Pr>ChiSq Type
Score Test BIRTHWGT and DIARRHEA 2 1.93 0.3819 Score

The output includes a table containing “Analysis of Initial Parameter Estimates.” The
initial parameter estimates are the estimates obtained from running a standard
logistic regression assuming an independent correlation structure. The parameter
estimation for the standard logistic regression is used as a numerical starting point
for obtaining GEE parameter estimates.

Tables for GEE model information, the working correlation matrix, and GEE
parameter estimates follow the initial parameter estimates inthe output. Here,
the working correlation matrix is a 9
9 matrix with an AR1 correlation struc-
ture. The table containing the GEE parameter estimates includes the empirical
standard errors. Model-based standard errors could also have been requested

SAS 625