1 October 2018

The Multiple Indicator Cluster Surveys (MICS) measure a range of indicators for various age groups. Many of the MICS indicators have children under the age of 5 years or subsets of this age group as their denominators. For example, the prevalence of wasting is measured for all children under the age of five, while immunization rates are calculated for children age 12 to 23 months, and exclusive breastfeeding is measured for children under 6 months of age.

One of the challenges in developing the sampling strategy for the MICS surveys in some countries is ensuring sufficient cases for sub-groups of children under-five. Sample sizes must be calculated and sampling procedures constructed in such a way that there will be a sufficient number of sample children under the age of 5 years to provide reliable indicators for all children in this age group as well as for specific age subsets of this population.

This issue is particularly salient in countries which have been through the demographic transition, where fertility rates are relatively low and the average household size is small. This phenomenon is found in, but not limited to, many countries in Europe and Central Asia, as well as Latin America and the Caribbean. To resolve the issue of small denominators for child indicators in these surveys, the MICS programme recommends that countries oversample households where there are children under the age of five. This practice began with a few countries during MICS4, continued with more countries in MICS5 and became a standard recommendation (where applicable) in MICS6. 

MICS has developed an oversampling strategy to compensate for low sample sizes of children under-five in low fertility countries. The purpose of this study is to examine how the oversampling strategy has worked in different settings. In this report, we outline the implementation of the oversampling strategy, comparing the results across several countries. The results of this study will be used to develop guidelines for countries planning to use the oversampling approach.


Access the paper