There have been few attempts at creating data series on levels of educational attainment
of the adult population consistent across time and space by age and sex. They would be
needed to estimate the role played by education and human capital in economic,
technological, environmental models as correctly as possible. In 2007, Lutz et al
developed a methodology to reconstruct (and project) levels of educational attainment
based on the information contained in the base-year source of choice for the most recent
period (Lutz et al. 2007a). The methodology was applied again in the framework of a
new round of population projections published in 2014 online
(www.wittgensteincentre.org/dataexplorer) and in the Oxford University book “World
Population and Human Capital in the Twenty-First Century” edited by Lutz, Butz and
KC. There, the coverage increased to 171 countries and the number of education
categories to six. The back-projection methodology was applied to the updated baseyear
sample in 2010 to arrive at the reconstruction of levels of educational attainment by
age and sex for the period 1970-2005. The purpose of this paper is to compare the
reconstructed datasets to other existing sources of historical data on education, including
the former reconstruction from 2007, collection and other reconstruction exercises.
The validation of the Wittgenstein Centre back-projection model outcomes with
available empirical data source enables the evaluation of our back-projection method for
the establishment of harmonized and consistent time series on the educational
composition of 171 countries in the world. In comparison, the most other available
datasets suffer from severe flaws, hampering any valid trend and regression analysis on
levels of educational attainment.
The back-projection methodology is explained in Section 2 and section 3 describes the
collection of empirical data for the validation of the WIC 2015 dataset and associated
challenges. The validation methodology and results are developed in section 4. Detailed
documentation about the country-specific validation is available from the Appendices.