Compare to the conditional logistic regression models with fixed and mixed effects for longitudinal data

  • انتصار عريبي فدعم
  • يوسف خليل الخفاجي
Keywords: طريقة الإمكان الأعظم، الانحدار اللوجستي الشرطي، البيانات الطولية، نماذج التأثيرات المختلطة، معيار شبه الإمكان في ظل نموذج الاستقلال (QIC)، معيار اكايكي التجريبي (EAIC)، التلوث البيئي، التحليل العنقودي

Abstract

Mixed-effects conditional logistic regression is evidently more effective in the study of qualitative differences in longitudinal pollution data as well as their implications on heterogeneous subgroups. This study seeks that conditional logistic regression is a robust evaluation method for environmental studies, thru the analysis of environment pollution as a function of oil production and environmental factors. Consequently, it has been established theoretically that the primary objective of model selection in this research is to identify the candidate model that is optimal for the conditional design. The candidate model should achieve generalizability, goodness-of-fit, parsimony and establish equilibrium between bias and variability. In the practical sphere it is however more realistic to capture the most significant parameters of the research design through the best fitted candidate model for this research. Simulation studies demonstrate that the mixed-effects conditional logistic regression is more accurate for pollution studies, with fixed-effects conditional logistic regression models potentially generating flawed conclusions.  This is because mixed-effects conditional logistic regression provides detailed insights on clusters that were largely overlooked by fixed-effects conditional logistic regression.

Published
2018-08-01
How to Cite
فدعما., & الخفاجيي. (2018). Compare to the conditional logistic regression models with fixed and mixed effects for longitudinal data. Journal of Economics and Administrative Sciences, 23(98), 406. https://doi.org/10.33095/jeas.v23i98.290
Section
Statistical Researches

Most read articles by the same author(s)