Data and Code

  • Data file containing estimates of constituency opinion on redistribution, Europe, immigration, and same-sex marriage, based on survey data collected between 2010 and 2014 (csv). See also this accompanying codebook and this accompanying technical report.


  • Replication code (zip). Download the zipped archive which contains several files. If you want  to read about how to generate constituency-level opinion estimates, see replication_howto.pdf. If you want to look at the source code used in that document, see replication_howto.R. The remaining files are either data files or config files.


  • Table of estimates of constituency opinion on redistribution, Europe, and immigration for May-June 2014 (pdf). For background information see this booklet or this technical report.
















4 comments on “Data and Code
  1. Rosebud says:

    Very interesting project. Quite coincidentally I am using similar methods to predict the 2015 GE on a seat-by-seat basis. Spatial dynamics and GIS are my area of expertise but I am also using socio-demographics, past results and regional polling data. I should add that I am not a political scientist. I trade political futures markets.

  2. Rosebud says:

    Apologies for not replying sooner; my emails get spooled into a few disparate folders!
    The markets are quite deep on the outrights, for example the Most Seats market or Overall Majority markets. There is less money on what might be seen as minor markets, for example the number of seats won by each party, etc. That’s on Betfair.

    Bookmakers also have a range of prices available for markets, right down to who will win each seat. They can and do close winning accounts though, and can restrict the amount you can get on online. Otherwise it’s a case of placing the bets in the shops, which can be time-consuming. will give you an idea which markets the bookmakers and exchanges provide.

    • admin says:

      Yeah, at least one of us is going to be interested in betting on the seat-by-seat outcomes, where I expect the markets are pretty thin. But I think the overall outcome (Con maj., NOC, Lab maj.) could be worth betting on if people (as last time) persistently overestimate the probability of a single-party majority.

Leave a Reply

Your email address will not be published. Required fields are marked *