A program to compute the generalized Mantel-Haenszel chi-square for multiple 2 x J tables

Area of Science:

• Biostatistics
• Statistical Computing
• Epidemiological Methods

Background:

• The Mantel-Haenszel chi-squared test is a standard method for analyzing stratified categorical data.
• Analyzing multiple 2xJ contingency tables requires robust statistical software for accurate inference.
• Existing methods may lack comprehensive output or flexibility for various data types.

Purpose of the Study:

• To present a FORTRAN program for calculating the generalized Mantel-Haenszel chi-squared test statistic.
• To provide essential outputs including expected cell values and stratum-specific odds ratios.
• To offer options for internal checks and corrections for 2x2 tables.

Main Methods:

• Implementation of the generalized Mantel-Haenszel chi-squared test in FORTRAN.
• Inclusion of calculations for expected cell values and individual odds ratios per stratum.
• Development of an internal check for asymptotic assumptions in 2x2 tables.
• Optional incorporation of Yate's correction for continuity in 2x2 table analysis.
• Capability to process both frequency count data and fractional values from life table procedures.

Main Results:

• The program computes the generalized Mantel-Haenszel chi-squared statistic for multiple 2xJ tables.
• Output includes expected cell values and individual odds ratios for each stratum.
• Internal checks validate asymptotic assumptions for 2x2 tables.
• The program accommodates diverse data types, including life table fractional values.

Conclusions:

• The presented FORTRAN program offers a comprehensive tool for analyzing stratified categorical data.
• It enhances the utility of the Mantel-Haenszel test by providing detailed outputs and flexibility.
• The software is suitable for both standard contingency table analysis and life table procedures.