The College of Administration and Economics at the University of Baghdad discussed, a PhD dissertation in field of Statistics by the student (Fawz Ahmed M.Saleh) and tagged with (Forecasting Population Projections in Iraq Using Bayesian Hierarchical Models) , Under supervision of (Prof. Dr.Qutaiba N. Nayef)
This dissertation focuses on the development and evaluation of an advanced statistical framework for examining future demographic change in Iraq and its governorates. The analysis centers on the total fertility rate and mortality levels, including changes in mortality rates and life expectancy at birth, as the main components shaping population size and age structure. In a broader context, these demographic processes are closely associated with social and economic transformations, such as improvements in educational attainment, advances in health services, and changes in labor market conditions, particularly over the medium and long term.
Given the pronounced spatial heterogeneity and the irregular availability of demographic data characterizing the Iraqi context, the dissertation adopts hierarchical Bayesian models as the primary analytical approach. This framework is complemented by a proposed probabilistic method, while the Maximum Likelihood Estimation (MLE) technique is employed as a benchmark for comparative evaluation. This integrated approach enables a systematic assessment of model performance in capturing temporal fertility patterns and mortality trends, while explicitly accounting for uncertainty and sub-national variation.
The empirical analysis is based on historical fertility and mortality data covering the period 1997–2024 and produces population projections for 2025–2050 under three alternative scenarios (low, medium, and high), representing a range of plausible future trajectories. The results indicate a continued decline in fertility alongside sustained improvements in life expectancy at birth at both the national and governorate levels, although the pace and stage of demographic transition vary considerably across regions. Quantitative evaluation using the Root Mean Square Error (RMSE) demonstrates that hierarchical Bayesian models achieve the highest
estimation accuracy, followed by the proposed method, while the MLE approach shows comparatively weaker performance in most governorates.
Overall, the findings highlight the importance of probabilistic and hierarchical modeling strategies in demographic analysis, particularly in settings characterized by data limitations and spatial heterogeneity. Methodologically, the dissertation contributes a robust comparative framework for fertility and mortality projections, while substantively it provides quantitative evidence to support population planning and the formulation of long-term development policies in Iraq.
