A Zero-Truncated Generalized Poisson Approach to Under-Dispersed Count Data: Fertility Patterns in Bangladesh, 2022

Background The number of children ever born (CEB) to women of reproductive age is a key factor of fertility and a critical factor in shaping population dynamics, particularly in developing countries like Bangladesh. It reflects cumulative reproductive behavior and is influenced by a range of demographic, socioeconomic, and cultural factors, making it a vital metric for understanding fertility trends. Based on data from Bangladesh Demographic and Health Survey 2022 (BDHS), this study focuses on women who have given birth to at least one child during the reproductive age. Methods To fit an appropriate model to the zero-truncated count and under-dispersed data effectively, both Zero-Truncated Poisson (ZTP) and Zero-Truncated Generalized Poisson (ZTGP) regression models were used. The ZTGP model provided the best fit and outperformed ZTP by addressing the under-dispersion of CEB, as resulted in narrower confidence intervals for the Incidence Rate Ratios (IRRs) and lower AIC and BIC values. Results Findings of this study indicate that maternal age, education, religion, place of residence, wealth index, and fertility preferences significantly influence the number of children. Specifically, higher education and economic status were associated with lower fertility, while rural residence and Muslim affiliation correlated with higher fertility rates. However, paternal occupation showed no significant impact on the number of children ever born in both ZTP and ZTGP models. Conclusion The Zero-Truncated Generalized Poisson (ZTGP) model effectively addressed the under-dispersion in the fertility data, yielding more accurate estimates of children ever born. The findings reveal significant socio-demographic determinants of fertility, reflecting the complex interplay of social, cultural, and economic factors and underscoring the importance of targeted, evidence-based family planning policies in Bangladesh.