Methods of geo-spatial sampling

This chapter presents innovative geo-spatial sampling methods for FCV settings, focusing on creating sampling frames without census data, sampling nomadic populations, and using satellite imagery.

It offers practical tools for safer, inclusive data collection in difficult environments.

In this chapter from Data collection in fragile states: Innovations from Africa and beyond, Eckman and Himelein (2020) explore geospatial sampling methods used to overcome challenges in data collection, particularly in conflict-affected and hard-to-reach regions. As traditional sampling techniques face limitations due to high costs and lack of infrastructure, Geographic Information Systems (GIS)-based methods provide innovative solutions. The resource presents three case studies from Somalia, the Democratic Republic of Congo (DRC), and Ethiopia, demonstrating how geospatial tools can improve sampling when conventional census data is outdated or unavailable. These methodologies are crucial for monitoring and evaluation (M&E) practitioners working in fragile and conflict-affected (FCV) settings, offering cost-effective and scalable ways to gather representative data.

Key features

The resource explores several key innovations in geospatial sampling across different challenging environments:

• Creating sampling frames without census data: The chapter details methods used in Somalia and the DRC where outdated or absent census data made traditional sampling impossible. These approaches included using satellite imagery and population estimates (e.g., WorldPop) to generate population densities and create sampling units.
• Sampling pastoralist communities: In Ethiopia’s Afar region, a Random Geographic Cluster Sampling (RGCS) method was developed to include nomadic populations that are typically underrepresented in traditional household surveys.
• Rapid listing in insecure areas: In Mogadishu, Somalia, where full household listings posed security risks, a variety of alternative methods were tested, including rooftop counts from satellite images and a novel random point selection method that considered the spatial distribution of dwellings within enumeration areas.
• Use of satellite imagery: Across all settings, high-resolution satellite data were used to count dwellings, generate population estimates, and delineate Primary Sampling Units (PSUs), making it easier to sample areas without up-to-date population data.

How would you use the resource?

This resource would be useful for practitioners designing surveys in conflict-prone or logistically difficult areas where traditional household surveys are impractical due to insecurity, population displacement, or a lack of reliable census data. The detailed case studies provide practical examples of how satellite imagery, population models, and random geographic sampling can be applied to generate representative data. For instance, the grid-based approach used in rural Somalia can help create manageable sampling units in sparsely populated areas, while the methods tested in the DRC offer valuable insights for urban settings. The resource also highlights the importance of adaptability and innovation in FCV environments, where traditional sampling techniques may not be feasible.

Why are you recommending it?

This resource offers practical, innovative solutions for overcoming the significant challenges associated with data collection in these contexts. The use of geospatial tools and satellite imagery makes it possible to gather reliable data quickly and efficiently, even in areas without census data or where security risks limit access.