Social network methods for estimating adult mortality: preliminary evidence from Brazil
*Dennis Feehan, Princeton University
Keywords: Social networks, Adult mortality, Survey methods
Measurements of adult mortality are a vital part of understanding the health and well-being of populations everywhere. Death rates above age 15 are needed to compute life expectancy, to understand socioeconomic and occupational differences in health, and to measure the impact of large health events like the availability of new medical treatments or the spread of HIV/AIDS. In countries that lack high-quality death registration data, adult death rates must be estimated using alternative strategies and in many of the world's poorest countries, no measurements of adult mortality are available at all. In order to expand the array of techniques available for measuring adult mortality, we introduce a new, survey-based estimator for adult death rates based on respondents' reports about members of their social networks. We test our approach using a large survey of about 25,000 households which was conducted in 27 Brazilian cities in 2012. We compare the estimates the new method produces to estimates from the sibling survival method, and also to the vital registration system. We discuss the advantages and disadvantages of the new estimator, and outline directions for future work.