A computer program suitable for fitting linear models when the dependent variable is dichotomous, polichotomous or...

Area of Science:

• Biostatistics
• Computational Statistics
• Statistical Software

Background:

• Regression analysis is crucial for understanding relationships between variables.
• Existing software may lack flexibility for diverse dependent variable types.
• Need for robust computational tools in statistical modeling.

Purpose of the Study:

• To develop a versatile computer program for regression analysis.
• To accommodate dichotomous, polychotomous, censored survival, and quantitative dependent variables.
• To implement advanced statistical models and estimation procedures.

Main Methods:

• Utilizes Cox's linear logistic models for dichotomous and polychotomous data.
• Employs Feigl and Zelen models for censored survival data.
• Implements maximum likelihood estimation via Van der Voort and Dorpema algorithm.
• Supports user-defined non-linear functions for quantitative data, including direct Gompertz function fitting.

Main Results:

• The FORTRAN IV program successfully performs regression analyses for multiple dependent variable types.
• The program integrates established statistical methodologies for accurate modeling.
• Demonstrates flexibility in handling both categorical and continuous outcomes.

Conclusions:

• The developed program offers a comprehensive solution for complex regression tasks.
• It enhances statistical analysis capabilities in biostatistics and related fields.
• The software provides a valuable tool for researchers requiring advanced regression techniques.