Scholarlink

<p>We propose a Bayesian multilevel weighted Poisson regression model to analyse data from the UNICEF 2022–2023 Multiple Indicator Cluster Surveys from Afghanistan to examine the association of individual, familial, societal, and regional factors with the fertility rates. Our proposed methodology probabilistically incorporates survey weights into the model, making the estimates of the regression coefficients more representative to both the population and subpopulation levels. The proposed model demonstrates superior performance over existing models based on the Bayesian Information Criterion (BIC) and Deviance Information Criterion (DIC) and remains insensitive against changing prior distributions with differing amounts of shrinkage. Our findings highlight the significant association of women’s age, age at marriage, education (both of women and household heads), and household wealth with fertility rates. The estimated overall fertility rate was 4.45 children per woman, with notable differences between women without education (4.50, 95% CI: 4.46–4.53) and with education (4.21, 95% CI: 4.15–4.29). Women in female-headed households had significantly lower fertility rates (4.04, 95% CI: 3.96–4.13) than the male-headed households (4.46, 95% CI: 4.43–4.50). Furthermore, women whose husbands had multiple wives had lower fertility rates (3.89, 95% CI: 3.83–3.95) compared to women in monogamous marriages (4.50, 95% CI: 4.47–4.54). Our results provide more interpretable estimates of regional effects on fertility rates, which eventually help to inform tailored and locally sensitive healthcare policies to address the differential of fertility among women in Afghanistan.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A novel Bayesian multilevel weighted poisson regression model to identify socio-demographic factors associated with fertility rates in Afghanistan

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • Hao Xu,
  • Ayisha Cok,
  • Caleb von Maydell,
  • Max Li,
  • Nazrul Islam,
  • Jabed Tomal

摘要

We propose a Bayesian multilevel weighted Poisson regression model to analyse data from the UNICEF 2022–2023 Multiple Indicator Cluster Surveys from Afghanistan to examine the association of individual, familial, societal, and regional factors with the fertility rates. Our proposed methodology probabilistically incorporates survey weights into the model, making the estimates of the regression coefficients more representative to both the population and subpopulation levels. The proposed model demonstrates superior performance over existing models based on the Bayesian Information Criterion (BIC) and Deviance Information Criterion (DIC) and remains insensitive against changing prior distributions with differing amounts of shrinkage. Our findings highlight the significant association of women’s age, age at marriage, education (both of women and household heads), and household wealth with fertility rates. The estimated overall fertility rate was 4.45 children per woman, with notable differences between women without education (4.50, 95% CI: 4.46–4.53) and with education (4.21, 95% CI: 4.15–4.29). Women in female-headed households had significantly lower fertility rates (4.04, 95% CI: 3.96–4.13) than the male-headed households (4.46, 95% CI: 4.43–4.50). Furthermore, women whose husbands had multiple wives had lower fertility rates (3.89, 95% CI: 3.83–3.95) compared to women in monogamous marriages (4.50, 95% CI: 4.47–4.54). Our results provide more interpretable estimates of regional effects on fertility rates, which eventually help to inform tailored and locally sensitive healthcare policies to address the differential of fertility among women in Afghanistan.