This week, I get to offer a sneak peak on my work as a researcher. My colleagues and I have put out a working paper (PDF link), describing our early efforts to produce a detailed measure of educational acquisition in Denmark from the 19th century onwards. Our longer-term ambition is to use this (and other) data to explore the role of education — and in particular, changes in the types of higher education — in Denmark’s industrialisation.
The data we describe in this paper are old records of university admission exams, and students’ exam results during their studies at the University of Copenhagen and the polytechnical institute that is nowadays the Technical University of Denmark. The records offer a rich level of detail on enrolments across disciplines and on student performance. Whereas education is commonly modelled as years of schooling or highest level of education attained, we expect to produce more precise measures of what students learned — broadly speaking, to capture “quality” rather than merely “quantity”.
The major challenge in this task is converting old paper records into a usable database. Our colleagues at the University of Southern Denmark are applying machine learning techniques: training a computer to read the various tables from the published material, extract names and grades, and convert the information to data. It’s technically impressive work, and the results to date are very promising.
This working paper is just the beginning. Our overall project is considering data from across Scandinavia: in addition to Denmark (as discussed in this paper), we’re also compiling equivalent data for Norway (which until 1814 was united with Denmark under the Danish crown) and will later tackle Sweden. Combining this data with census data and other records offers the potential to deepen considerably our understanding of the contribution of education to economic development.