Estimating Bias in Educational Inequalities in Mortality: A Simulation Study

Ana C. Gómez-Ugarte, Max Planck Institute for Demographic Research (MPIDR)
Ugofilippo Basellini, Max Planck Institute for demographic Research
Carlo Giovanni Camarda, Institut National d'Études Démographiques (INED)
Fanny Janssen, Netherlands Interdisciplinary Demographic Institute (NIDI) and University of Groningen
Emilio Zagheni, Max Planck Institute for Demographic Research (MPIDR)

In many countries, unlinked cross-sectional data is the only available data to study educational inequalities in mortality. However, such data is subject to three data-quality issues: under-coverage, age misreporting and education misreporting. Many studies have looked into the first two issues, whereas education misreporting has been less explored. The first goal of this study is to provide a comprehensive evaluation of how estimates of educational inequalities in mortality can be affected by this source of error. For this, we rely on simulation scenarios with varying direction and magnitude of education misreporting, and on different measures of educational inequalities. We find that education overstatement downward biases educational inequalities in mortality based on life expectancy and upward biases those based on life span variation measures, while education understatement upward biases inequality. The second goal is to quantify the degree of education misreporting in observed data, and to introduce an adjustment procedure to correct educational misreporting. We adjust the distance between education-specific log-mortality rates such that its pattern follows the theoretical and observed patterns in previous studies. This, in turn, allows us to adjust the education-specific log-mortality rates and derive a new estimate of educational inequalities in mortality.

Keywords: Civil Registration and Vital Statistics, Mortality and Longevity, Inequality, Disadvantage and Discrimination

See extended abstract.