Salary: Per Ministry of Finance contract staff salary scale

Terms of employment: Contract

Duration of contract: One year with possibility of extension

Place of work: Ethiopian Public Health Institute, Addis Ababa

Quantity: One

Background

The Ethiopian Public Health Institute (EPHI) is the national public health institute mandated to lead, conduct and coordinate the prevention and control of public health emergencies; strengthen the capacity of health laboratories and provide referral and reference testing services; and undertake research on priority public health and nutrition problems, programmes and strategies. EPHI is also mandated to establish and manage the national health data management and analytics center by gathering health and health-related data from multiple sources. The National Data Management and Analytics Center for Health (NDMC) manages national health data and applies advanced analytics, statistical modelling, machine learning, visualization and evidence synthesis to generate timely, policy-relevant evidence for decision-making at national and subnational levels.

With support from the Gates Foundation, EPHI is implementing the project “Strengthening Modeling Capacity to Inform Ethiopian Health Policies,” which supports the development and application of advanced analytical and modelling methods for priority public health questions. The project focuses on producing policy-relevant evidence, establishing reproducible analytical workflows, strengthening the skills of EPHI staff and improving collaboration with the Ministry of Health and public health programmes. The position will contribute to these objectives through its assigned technical responsibilities and capacity-strengthening activities.

Purpose of the Position

The purpose of the position is to lead and apply advanced Bayesian spatial modelling and small area estimation to priority public health and policy questions. The expert will develop and validate reliable subnational estimates, quantify and communicate uncertainty, produce decision-relevant maps and analytical outputs, and strengthen EPHI capacity to undertake and sustain spatial and small area analyses.

Roles and Responsibilities

3.1 Problem Formulation and Data Preparation

1. Work with EPHI, Ministry of Health and programme teams to define the policy use, target indicator, population, geography, time period, estimand, acceptable uncertainty and decision threshold for each assigned analysis.

2. Assess whether direct estimation, model-assisted estimation, small area estimation or a Bayesian spatial or spatiotemporal model is appropriate for the available data and intended decision.

3. Compile and harmonize survey, census, routine health information, surveillance, programme, demographic and geospatial covariates while maintaining traceable provenance and metadata.

4. Address administrative-boundary changes, spatial misalignment, incomplete geocoding, displaced survey-cluster coordinates, missing data and measurement error in a manner appropriate to the data-generating process.

5. Produce design-based direct estimates and standard errors, where applicable, as a benchmark and diagnose sparsity, instability and effective sample-size limitations before fitting model-based estimates.

3.2 Advanced Bayesian Spatial Modelling and Small Area Estimation

6. Develop fit-for-purpose area-level or unit-level small area models and Bayesian hierarchical spatial or spatiotemporal models that account for sampling variability, spatial dependence, temporal structure and relevant covariates.

7. Apply defensible model classes such as Fay–Herriot-type models, generalized linear mixed models, conditional autoregressive models, Besag–York–Mollié (BYM) models, reparameterized Besag–York–Mollié (BYM2) models, Gaussian processes, stochastic partial differential equation (SPDE) approaches and nearest-neighbour Gaussian processes (NNGPs) for large spatial datasets.

8. Fit and compare models using integrated nested Laplace approximation (INLA), Markov chain Monte Carlo (MCMC), or Hamiltonian Monte Carlo (HMC) with the No-U-Turn Sampler (NUTS), as appropriate to the model structure and computational requirements.

9. Account for complex survey design, weights, stratification and clustering in estimation or benchmarking, and document the implications of the chosen modelling approach.

10. Specify priors transparently and evaluate prior sensitivity, identifiability, convergence, spatial confounding, over-smoothing and model complexity.

11. Integrate multiple data sources only where the assumptions linking them are clear; propagate relevant uncertainty rather than treating model inputs as error-free.

12. Develop spatial and spatiotemporal projections or scenario analyses when required, clearly separating interpolation, prediction and extrapolation.

3.3 Validation, Interpretation and Responsible Dissemination

13. Evaluate models using posterior predictive checks, convergence diagnostics, residual spatial autocorrelation, out-of-sample or spatial cross-validation, sensitivity analysis and comparison with credible external information.

14. Compare model-based estimates with direct estimates and report point estimates together with credible intervals, exceedance probabilities or other uncertainty summaries appropriate to the decision.

15. Assess calibration and coverage and avoid presenting fine geographic detail when data and validation do not support it.

16. Prepare maps, estimate tables, technical reports, policy briefs, presentations and manuscripts that explain methods, assumptions, uncertainty, limitations and appropriate use in clear language.

17. Apply disclosure-control and privacy safeguards, particularly for sensitive outcomes, small populations and geographically displaced or restricted data.

3.4 Reproducibility and Institutional Capacity

18. Develop version-controlled, documented and testable R or Python workflows for data preparation, model fitting, diagnostics, estimation, mapping and reporting.

19. Create reusable functions, templates and methodological guidance for future small area and spatial analyses at EPHI.

20. Mentor statisticians, epidemiologists and modellers through hands-on analyses and deliver structured training on Bayesian spatial modelling, survey estimation, validation and uncertainty communication.

21. Collaborate with multidisciplinary teams and provide constructive peer review of spatial analyses and maps produced within the project.

22. Contribute to project work planning, progress reporting, partner coordination and other technically relevant assignments agreed with the supervisor.

Reporting

• Reports to: Project Principal Investigator (PI) / Director of the National Data Management and Analytics Center (NDMC) / Head of the Data Analytics, Modelling and Visualization Division.
• The position will be based at EPHI/NDMC in Addis Ababa.

• Doctor of Philosophy (PhD) in statistics, biostatistics, survey methodology, demography, spatial epidemiology, data science or a closely related quantitative field, with at least two years of relevant experience; or
• Master’s degree in one of the above fields, with at least six years of relevant experience.
• Demonstrated applied experience in Bayesian hierarchical modelling, spatial or spatiotemporal statistics, disease mapping or small area estimation.

Desired Skills and Experience

• Strong knowledge of advanced statistical inference, survey sampling, complex survey analysis, generalized linear and mixed models, spatial statistics, Bayesian computation and uncertainty quantification.
• Advanced programming skills in R and experience implementing Bayesian spatial and small area models using established software such as INLA or Stan; Python experience is an advantage.
• Proficiency in managing, analysing and mapping geospatial data using R and established geographic information system software.
• Experience with area-level and unit-level small area estimation, Bayesian disease mapping, spatial smoothing or spatiotemporal modelling.
• Experience analysing Demographic and Health Surveys (DHS), Multiple Indicator Cluster Surveys (MICS), census, EPHI national surveys, routine health information or disease-surveillance data is an advantage.
• Experience integrating population, environmental, climate, accessibility or other spatial covariates is an advantage.
• Experience with model diagnostics, simulation studies, sensitivity analysis, spatial cross-validation and reproducible research.
• Experience developing and delivering training and mentoring analysts in advanced statistical or spatial methods.
• At least six public health-related articles published in reputable peer-reviewed journals, including at least three as first author.
• Strong analytical, problem-solving, scientific-writing and communication skills.
• Ability to work independently and collaboratively and to manage multiple tasks within agreed timelines.

Contract Duration and Remuneration

• The initial contract will be for one year, with possible extension based on satisfactory performance, project requirements and availability of funds.
• Salary will be determined in accordance with the applicable Ethiopian Public Health Institute and Ministry of Finance contract staff salary scale, taking into account the candidate’s qualifications and experience.