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

Practice
Exercises

The following printout summarizes the computer output

from a GEE model run on the Infant Care Study data and

should be used for Exercises 1–4. Recall that the data

contained monthly information for each infant up to 9

months. The logit form of the model can be stated as

follows:

logit PðXÞ¼b 0 þb 1 BIRTHWGTþb 2 GENDER

þb 3 DIARRHEA:

The dichotomous outcome is derived from a weight-for-

heightz-score. The independent variables are BIRTHWGT

(the weight in grams at birth), GENDER (1¼male, 2¼

female), and DIARRHEA (a dichotomous variable indicat-

ing whether the infant had symptoms of diarrhea that

month; coded 1¼yes, 0¼no).

A stationary 4-dependent correlation structure is specified

for this model. Empirical and model-based standard errors

are given for each regression parameter estimate. The

working correlation matrix is also included in the output.

Variable Coefficient

Empirical Std

Err

Model-based

Std Err

INTERCEPT 2.0521 1.2323 0.8747

BIRTHWGT 0.0005 0.0003 0.0002

GENDER 0.5514 0.5472 0.3744

DIARRHEA 0.1636 0.8722 0.2841

Stationary 4-Dependent Working Correlation Matrix

COL1 COL2 COL3 COL4 COL5 COL6 COL7 COL8 COL9

ROW1 1.0000 0.5449 0.4353 0.4722 0.5334 0.0000 0.0000 0.0000 0.0000

ROW2 0.5449 1.0000 0.5449 0.4353 0.4722 0.5334 0.0000 0.0000 0.0000

ROW3 0.4353 0.5449 1.0000 0.5449 0.4353 0.4722 0.5334 0.0000 0.0000

ROW4 0.4722 0.4353 0.5449 1.0000 0.5449 0.4353 0.4722 0.5334 0.0000

ROW5 0.5334 0.4722 0.4353 0.5449 1.0000 0.5449 0.4353 0.4722 0.5334

ROW6 0.0000 0.5334 0.4722 0.4353 0.5449 1.0000 0.5449 0.4353 0.4722

ROW7 0.0000 0.0000 0.5334 0.4722 0.4353 0.5449 1.0000 0.5449 0.4353

ROW8 0.0000 0.0000 0.0000 0.5334 0.4722 0.4353 0.5449 1.0000 0.5449

ROW9 0.0000 0.0000 0.0000 0.0000 0.5334 0.4722 0.4353 0.5449 1.0000

Practice Exercises 559