Mark Wheldon , United Nations, Department of Economic and Social Affairs
Philipp Ueffing, United Nations Population Division
Aisha Dasgupta, United Nations Population Division
Vladimira Kantorova, United Nations Population Division
In response to several global-level initiatives to promote universal access to family planning, including the FP2020 initiative and the UN Sustainable Development Goals, Bayesian hierarchical models were developed to jointly estimate and project the prevalence of contraceptive use (CP), unmet need for contraception (UMN), and no-need for contraception, among married and unmarried women aged 15–49 (Alkema et al., 2013; Cahill et al., 2018; Wheldon et al., 2018). The quantity of interest was a multidimensional compositional vector. While the statistical model produced a multivariate posterior distribution for the vector, only univariate, marginal indicators were presented in figures and tables in the results. This made interpretation of the relationship between the indicator components across countries more challenging. Here, we show that indicators of family planning can be reported in true multivariate fashion using the ternary-balance scheme of Schöley and Willekens (2017).
Presented in Session 123. Flash Session: Geospatial Analysis in Applied Demography