"Small Sample Adjustments for Wald-type Tests using Sandwich Estimators"

by Michael P. Fay and Barry I. Graubard

All Simulations were done in Splus and written by Michael Fay.

We performed simulations on 4 types of data

1. Balanced Normal Model
2. Poisson GEE
3. Logistic Independence Estimating Equations derived from GUIDE data
4. Conditional Logistic Regression
5. Cox Regression

Balanced Normal Model

This simulation/program produced the results in Section 3.1 of the paper.

• Results
• Program

Poisson GEE

This simulation produced the results in Table 1 of the paper.

• Results:  Columns 1 and 2 (without last row)  and  Columns 3 and 4 (without last row)  and last row
• Programs:  Main Program, Columns 1 and 2  and  Main Program, Columns 3 and 4 and Program for Last Row
• Called functions: Version 1.5.1.2 of YAGS (see  VJ Carey's website ), and my modifications to it  (myags , myags.pmat.fit.mancl , LZ.exchalp all modifications are  under the  GNU General Public License)

Associated rows of Table 3, Program and Results

Logistic Independence Estimating Equations derived from GUIDE data

This simulation produced the results in Table 2 of the paper.

• Results(except last row) and Results for Last row
• Main Program and Program for Last Row
• Called functions: Version 1.5.1.2 of YAGS (see  VJ Carey's website ), and my modifications to it  (myags , myags.pmat.fit.mancl , LZ.exchalp,  all modifications are under the GNU General Public License)

Conditional Logistic Regression

This simulation produced the results in Table 4 of the paper.

• Results
• Main Program
• Called functions: clogist.calc,  clogist

  Cox Regression

This simulation produced the results in Table 5 of the paper.

• Results
• Main Program
• Called function: mcoxph

Return to SAWS homepage