Alternative Sampling Designs for Emergency Settings: A Guide for Survey Planning, Data Collection and Analysis

In emergency settings, rapid but statistically reliable population-based surveys are needed to provide humanitarian organizations and government agencies with essential information on the severity and magnitude of the situation so that appropriate analysis and response planning takes place. The most common method used in emergencies is a two-stage 30x30 cluster survey. This method provides reliable population-level estimates, but is time-and resource-intensive.

The guide provides information on three alternative sampling designs that are proven to be more time-and resource-efficient than the 30x30 cluster survey: the 33x6, the 67x3 and the sequential design. All three designs are hybrid designs, combining aspects of cluster sampling and analysis, with lot quality assurance sampling (LQAS) analysis.

The development, testing and validation of the three alternative sampling designs was completed by FANTA-2 with Catholic Relief Services (CRS), Ohio State University (OSU), Save the Children US (SC/US), and a team of statistical experts at Harvard School of Public Health (HSPH). The guide was funded by the United States Agency for International Development’s Office of Health, Infectious Disease in the Bureau for Global Health, the Office of U.S. Foreign Disaster Assistance in the Bureau for Democracy, Conflict and Humanitarian Assistance, and USAID/Ethiopia.

Download the guide [1.3 mb]