/riksdag

Bill cosponsorship networks in the Swedish Parliament.

Primary LanguageR

This repository contains code to build cosponsorship networks from bills (and motions) passed in the Swedish Parliament.

The data come from the Riksdag's open data portal.

HOWTO

Replicate by running make.r in R.

The data.r script downloads information on bills and sponsors. Once unzipped to the raw folder, the raw JSON data files will take up to 1.5 GB of disk space -- make sure that you have at least that amount of free space on your hard drive.

The build.r script then assembles the edge lists and plots the networks, with the help of a few routines coded into functions.r. Adjust the plot, gexf and mode parameters to skip the plots or to change the node placement algorithm.

DATA

Bills

The sponsors data are read from the JSON dumps available on the Riksdag's open data portal.

  • uid -- bill unique identifier (int)
  • doc -- document identifier (the first two characters are legislature codes)
  • legislature -- legislature years, imputed from doc
  • date -- bill date (yyyy-mm-dd)
  • authors -- semicolon-separated sponsor unique identifiers
  • n_au -- total number of sponsors

Sponsors

The sponsors data are read from the XML listings available on the Riksdag's open data portal.

  • name -- sponsor name
  • born -- year of birth (num)
  • sex -- gender (F/M), read from the profile
  • party -- political party, abbreviated
  • partyname -- political party, full name
  • constituency -- constituency, stored as the string to its Wikipedia English entry (see note)
  • status -- last parliamentary status
  • mandate -- semicolon-separated mandate years, used to compute the nyears seniority variable
  • job -- occupation
  • url -- profile URL, shortened to unique numeric identifier
  • photo -- photo URL, coded as a numeric dummy (0/1); photo URLs are based on profile URLs
  • uid -- sponsor unique identifier, composed of name and url

Note -- missing constituencies are due to a reform of the constituency system, which merged several constituencies into single ones. There are lots of missing constituencies for the legislatures before 2002, 4% (n = 18 constituencies) missing in 2002-2006, and none afterwards.