[R] Using R for modelling Australian Senate Election in NSW?

Philip Rhoades phil at pricom.com.au
Wed Oct 1 02:32:57 CEST 2014


People,

I am setting up an Australian Science and Technology party with Human 
Health and Longevity as it's defining, first platform plank:

   http://lestp.org

and intend to run Senate candidates at the next Federal Election (due 
late 2016).  Senators are elected from each State and Territory using a 
Proportional Representation (PR) method but it is complex and odd things 
happen with direction of preferences when the smallest groups or 
individuals are progressively eliminated from the count when producing 
the final quotas.  I have a little experience with R but it seems like 
it might be useful for modelling this situation?  A quota for a 
half-senate election (six senators to be elected in each state) is one 
seventh of the total number of votes + one vote.

My current thoughts are these:

- each Senate vote corresponds to a vector with say 80 numbers on it 
(corresponding to 80 candidates - some grouped into parties, some as 
individuals)

- the order of the numbers from 1-80 could be random on the ballot but 
in practice, there will be many identical ballots corresponding to the 
voting preference recommendations of the major parties

- the groups of votes with enough first preference votes (number "1"s) 
that constitute a "quota" will have their first candidate declared 
elected and their quota subtracted from the party's total votes - this 
process continues until there are no full quotas left to allocate

- the next stage is the elimination process - the party / individual 
with the lowest number of votes is eliminated and their second 
preference votes (their number "2"s) will be added to the count 
corresponding to the that individual - this process continues until 
another person has enough votes for a quota

- the previous exercise is repeated until all quotas have been allocated

- I can generate the approximate simulation vectors from previous data 
OK but am not sure how how to proceed with the "elimination" process 
coding etc.

Suggestions (so I don't waste too much time re-inventing wheels etc) 
would be much appreciated!

Regards,

Phil.

-- 
Philip Rhoades

GPO Box 3411
Sydney NSW	2001
Australia
E-mail:  phil at pricom.com.au



More information about the R-help mailing list