Improved understanding of geographic variation and inequity in health status, wealth, and access to resources within countries is increasingly being recognized as central to meeting development goals. Development and health indicators assessed at national scales conceal important inequities, with the rural poor often least well represented. High-resolution data on key social and health indicators are fundamental for targeting limited resources, especially where development funding has recently come under increased pressure. Globally, around 80% of countries regularly produce sex-disaggregated statistics at a national or provincial scale on a range of key indicators. However, their utility in supporting local policy making, intervention targetting, monitoring, and national resource prioritisation aiming to improve gender equity, is often poor.
Geolocated cluster data from the Demographic and Health Surveys (DHS) on rates of literacy, stunting and use of modern contraception methods in a range of low-income countries were used in this project to produce high-resolution spatial gender disaggregated maps using predictive modelling techniques. Bayesian geostatistical and machine learning methods were used to take advantage of the fact that many of the key social and health indicators are correlated with environmental and social factors that can be mapped in locations where surveys have not been performed.
This study demonstrates that geolocated household survey data, integrated into novel and robust statistical models and validation techniques, can be used to create high-resolution maps of a wide range of gender-disaggregated indicators in low-income settings, with appropriate confidence intervals. The results reported here show that high-resolution gender-disaggregated maps can be developed for a range of countries and indicators, heralding a paradigm shift in our understanding of gender inequities, and in the availability of gender disaggregated data at a resolution that can support local policy making and monitoring.


Excerpt (Disclaimer)

Conclusions. The rising international focus on inequalities requires a detailed, strong evidence base with an explicit quantification of uncertainties. Some of the maps produced in the present study have sufficiently low uncertainty to be summarized to a level of administrative unit that is relevant for policy and decision makers for planning and allocation of resources. In particular, the maps of female literacy in Nigeria and Kenya, use of modern contraception methods in Nigeria or male and female stunting in Nigeria have levels of certainty that make them highly applicable for planning purposes. These maps can support the development of policy to ensure women’s equal access to education and economic resources, and to promote equal opportunities.
The work we have performed shows the value of combining data from geolocated household surveys with spatial covariates using advanced modelling architectures. It allows for the quantification of the distribution of many different indicators in resource-poor settings at the gender-disaggregated level, and estimation of the associated level of uncertainty. With geolocated household surveys being undertaken regularly, the potential exists for continuous updating and monitoring of these indicators at a continental or global scale. The presented work needs to be considered as a preliminary study to test the strength and limits of the proposed approach. Creating or selecting a superior set of covariates with a higher correlation with modelled indicators will lead to improved estimates.
We strongly recommend scaling these analyses to develop regional estimates of a larger number of key indicators in support of local policymaking and prioritization of national and international resource allocation.