Brazil 1900–2010, population 25 years and older
In 1900, only 1.4 % of the Brazilian population older than 25 years had received had received at least lower secondary education. This proportion increased to 8.9% until 1950 and to 51.2% until 2010. By then, about 11.3% of the population aged 25 years or older finished a post-secondary education while 48.8% had primary education or less as their highest educational attainment. Mean years of schooling of the population 25 years or older increased from 0.9 years in 1900 to 2.1 years in 1950 and to 7.0 years in 2010.
In Brazil, gender differences in educational attainment have been only slowly diminishing during the first half of the 20th century. Then the picture reversed and the gender imbalance has turned in favour of women. In 1900, 2.0% of the men aged 25 years and older had at least lower secondary education, compared to 0.8% of the women. By 1950 about 11.9% of the men and 13.2% of the women aged 30 to 34 years had achieved at least lower secondary education while for instance this share was only 4.1% and 2.6% for the population aged 65–69. Universal lower secondary education has not yet been achieved, neither for men nor women. Since 1985, there has been an inverse gender gap in higher education, with larger proportions of women completing post-secondary education than men. In 2010, about 15.8% of women in the 30–34 age group had post-secondary educational attainment, compared to 11.3% of men.
The EDU20C estimates of the population of Brazil by age, sex and education are based on several census datasets dating back to 1900. Brazilian censuses provide information on highest educational attainment in decennial intervals for the time since 1940. Before 1940 data on education or literacy by age and sex are not available. For Brazil, all 6 education categories (aggregated to 4 before 1950) are available in the reconstruction model.
The first historic census data covering information on population by age and sex are complemented with estimates by the Brazilian Institute of Geography and Statistic. For Brazil it was necessary to generate life tables for missing data-points by interpolating/extrapolating life expectancies at birth by sex. The model fits a logistic function to existing life expectancies at birth, given the values of upper and lower asymptotes. Based on these estimated life expectancies we use a function that interpolates the logarithms of the probabilities of dying (nqx) from two life tables to generate a comprehensive set of life tables for the entire reconstruction period. Both R functions are in their methodological core based on the Population Analysis System (PAS) Excel templates E0LGST and INTPLTF/INTPLTM. Furthermore, it was necessary to interpolate the intercensal data-points for population by age and sex using a linear interpolation function.
For Brazil the major source of data on population by age, sex and educational attainment in the 20th century originates from the digitised archive of Brazilian Institute of Geography and Statistic/IBGE as well as from IPUMS. For the EDU20C reconstruction, we also used information on mortality extracted from the life tables published by Arriaga (1968).