This package is no longer maintained. I recommend using the dwnominate
package for R instead of this gem. It should be more convenient in most cases.
The 'nominate' gem, to run W- and DW-NOMINATE from Ruby
W-NOMINATE uses rollcall votes to map legislators along a political spectrum. In American state and federal elections, it typically finds the traditional left-right spectrum. DW-NOMINATE uses legislators who have served in multiple legislative sessions to compare sessions and track changes through time.
Installation: sudo gem install nominate
- Linux
- R
In Ubuntu:sudo apt-get install r-base r-base-dev
- GFortran (only for DW-NOMINATE)
In Ubuntu:sudo apt-get install gfortran
Let's say you have a file of voting data in the format
name1|party1|rollcall1 info|vote1
name2|party2|rollcall1 info|vote2
...
name1|party1|rollcall2 info|vote1
name2|party2|rollcall2 info|vote2
...
which is similar to what you can download from the New Hampshire state legislature. And these are collected in file '1999NHrollcalls.csv' and so forth.
The following code will read the rollcall data into Wnominate objects, and add the sessions to a Dwnominate object.
chamber.dwnominate
starts the nominate process, first calling wnominate
on each Wnominate object, then using the results to write the input files for the DW-NOMINATE program, and then running DW-NOMINATE. All results, including output graphs from the R wnominate package, will be written to a folder called 'nominate'.
An alternative is to call wnominate(prefix)
on Wnominate objects, which will run only the W-NOMINATE program, adding the specified prefix to each output file.
require 'nominate' chamber = Dwnominate.new # votes other than Yes or No are counted as missing ('M') vote_key = Hash.new('M') vote_key['Yes'] = 'Y' vote_key['No'] = 'N' # Wnominate object only accepts 'Y', 'N', and 'M' votes ['1999', '2001', '2003', '2005', '2007'].each do |year| file = year + 'NHrollcalls.csv' lines = IO.readlines(file) session = Wnominate.new # default party is 'unknown' parties = Hash.new('unknown') old_issue = 0 # default vote is 'M' for 'missing' rollcall = Hash.new('M') lines.each_with_index do |line, i| data = line.split('|') name = data[0] party = data[1] issue = data[2] vote = vote_key[data[3].chomp] # every time we get to a new rollcall, add the hash with the old rollcall's votes to the # Wnominate object, and start a new rollcall hash. if issue != old_issue session.add_rollcall(rollcall) unless i == 0 rollcall = Hash.new('M') old_issue = issue end rollcall[name] = vote parties[name] = party if not parties.has_key?(name) # get the last rollcall when you reach the end of the file session.add_rollcall(rollcall) if i == lines.length - 1 end # add a hash of legislator names and parties if you want party info to be included session.parties = parties chamber.add_session(session) end chamber.dwnominate(prefix = 'dw_')
- Scaling Roll Call Votes with W-NOMINATE in R
- R wnominate package documentation
- DW-NOMINATE at voteview.com (with an explanation of the output format)
- Spatial Models of Parliamentary Voting, by Keith Poole, an in-depth explanation of the programs
- Ideology and Congress, by Keith Poole and Howard Rosenthal