Code to estimate Bayesian ideal points for French politicians from Twitter data. Forked from pablobarbera's code at twitter_ideology. Edits by briatte and 3wen.

• Read the draft paper and its appendix
• See the HOWTO file for details on how to run the code
• Feel free to suggest improvements or report bugs

• Many Twitter accounts from Elus 2.0, eTerritoire and Regards Citoyens.
• Election results from Ministère de l'Intérieur via Data.gouv.fr.
• French administrative units from Insee via Wikipedia Francophone.
• Geocodes from Google Maps.
• GEOFLA shapefiles from IGN (not included).
• Government composition by Premier ministre via Data.gouv.fr.
• List of French first names by Mike Campbell and Boris New (not included).
• Population age group figures by Irdes (not included).

• This blog post (in French) explains how to plot the data as a network.
• This blog post (in French) shows some intermediary results of the model.
• This paper by pablobarbera explains the nuts and bolts of the model.
• This blog post (in Portuguese) does something similar with Brazilian politicians.
• This paper reviews the relevant literature and does something quite similar with Facebook "likes".
• This blog post by pablobarbera and griverorz discusses partisan polarisation on Twitter.