Urban Lifelines

Historical data

The backbone of the historical data consists of manually geocoded full-count census records for a large number of Swedish urban areas between 1880 and 1950.

Individuals are linked across censuses using probabilistic matching methods based on name, age, and place of birth, as there were no unique individual identifiers before the introduction of personal numbers in 1947. Due to the high quality of the underlying data, the proportion of people who can be linked across censuses is higher in the Swedish case compared to what is typical for historical census data, such as in the United States and the United Kingdom.

StockPop

A particular subset of the UrbLong dataset is created for Stockholm, allowing for the annual tracking of all individuals in the city. This subset combines an individual-level, year-by-year transformed and geocoded version of the Roteman register, covering the period from 1878 to 1926, with a digitized version of poll tax registers from 1940 to 1970. The latter dataset is unique in that it also includes information on individuals’ places of work. We are also working on incorporating income data from tax registers into the Roteman archive.

Contemporary registers

GeoSweden

For the period after 1960 we make use of register data from Statistics Sweden contained in the database GeoSweden. Geocoding of individual’s place of residence is done manually for the period prior to 1982. After 1990, the dataset also include information about individuals’ place of work. The data can only be accessed on Statistics Sweden’s external servers.

Linkage between historical data and contemporary registers

To follow individuals from the historical data (-1950) to the contemporary registers (1960-) as special approach has to be applied.