The decomposition of demographic measures is a fundamental tool in a demographer’s toolkit. Through decomposition of demographic measures, researchers can gain crucial insights into the importance of different mechanisms in producing demographic change. Most decomposition techniques decompose changes in demographic measures over time or, more generally, differences between demographic rates. However, fewer techniques are available to researchers needing to estimate the contribution of different subpopulations to a demographic measure for the total population. This paper develops such a decomposition technique and discusses one practical application. A full description of this approach is developed in the manuscript "groupwise-decomposition".
- groupwise-decomposition.pdf: describes the decomposition, proves some of its property, and shows an example application. All the content should be considered preliminary at this stage.
- /R: contains R files to replicate the results of the paper.
- /data: contains the data needed to replicate the results of the paper
- /output contains the resulting estimates
- /figures: contains the figures generated by the R code.
The functions carrying out the decomposition are in the 00_functions.R and 00_functions_extended.R files. If you want to use this approach in your own work I would recommend starting there. As shown in the paper, the approach is quite flexible but changes to the code are required.
For any question, you can reach me at eugenio.paglino@helsinki.fi