Goodness of Fit Statistic for Logistic Regression Models with only Continuous Predictor

Authors

  • Jalal A. Moaiti Department of Statistics, Faculty of Sciences Al-Marj, Benghazi University, Libya Author
  • Ali Meelad Abd Allateef Department of Statistics, Faculty of Sciences Al-Marj, Benghazi University, Libya Author
  • Nawal Mohammed Othman Mustafa Department of Statistics, Faculty of Sciences Al-Marj, Benghazi University, Libya Author

DOI:

https://doi.org/10.65421/jshd.v2i4.389

Keywords:

Logistic regression models, goodness of fit, Fisher Approximation, Variance Stabilization, Simulation Study

Abstract

 Logistic regression models have become widely used statistical models to model binary responses. However, assessing the goodness of fit (GOF) in these models, when we have only one continuous predictor is not clear. The deviance and the Pearson chi square tests are unreliable and don't follow chi square distribution. Furthermore, some researches showed that Hosmer and Lemeshow test is slightly conservative. In this research, the Pearson chi square test and its moments were investigated via a simulation study. The simulation showed that the problem is that the variance of this statistic is not stable and doesn't equal to theoretical variance for chi square test. However, to stabilize the variance Fisher approximation to standard normal was used to solve or relax this problem. The approximated standard normal test had slightly stable variance and good type one error rate, compared with chi square test. 

Downloads

Published

2026-11-16

Issue

Section

Articles

How to Cite

Jalal A. Moaiti, Ali Meelad Abd Allateef, & Nawal Mohammed Othman Mustafa. (2026). Goodness of Fit Statistic for Logistic Regression Models with only Continuous Predictor. Journal of Scientific and Human Dimensions, 2(4), 448-453. https://doi.org/10.65421/jshd.v2i4.389