Harmonized INSEE socio-demographic IRIS-level data and IRIS conversion file (2010-2020)

The smallest level of aggregation for sociodemographic data made publicly available by the French INSEE is the IRIS. Data is downloadable on INSEE websites but decomposed by type of variable and by year. I thus combined all datasets into a single homogenous dataset for practicality. It includes on the period 2010-2020 :

• Population structure (age, gender...) data
• Economic occupation CSP data
• Available income data
• Family data

Since a sizeable amount of IRIS change each year, timewise comparisons are limited. I therefore created a conversion file. Each IRIS is expressed as a % combination of previous IRIS. This allows to track IRIS merge, IRIS split and border changes. Measurement errors linked to border overlap were detected when the overlap of 2 IRIS was less than 1 percent. The IRIS overlap without the measurement error is the variable _ajuste (pardon my French). This allows to express the IRIS of a year as the wieghted sum of the IRIS of any other given year.

You will also find the codes I used to create each of the files included in this project on my github. The annotation may be lacking, it is currently being improved.