Parliament’s next top model: simulating elections
Either National or Labour would have won a clear majority, while the Greens, with the third largest party vote in 2008,
would have few or no seats in parliament if recent elections were run under some of the voting systems in this year's
referendum.
These scenarios, and others, can be tested using an election simulator created by researchers at The University of
Auckland's Centre for Mathematical Social Science to help voters understand the consequences of their referendum vote.
"The simulator can be used by anyone who plans to take part in the voting system referendum," says Dr Mark Wilson from
the Department of Computer Science, who created the simulator with colleague Dr Geoff Pritchard from the Department of
Statistics.
"A lot of descriptive information about the upcoming referendum has been made available to voters, explaining, for
instance, that coalition governments are more or less likely under particular voting systems. But as scientists
interested in collective decision making, we wanted to know more precisely what the voting systems would mean in terms
of seats in parliament and we think that voters should have this information too."
"Some of the voting systems, such as Single Transferrable Vote (STV), are quite complicated and it's essentially
impossible for people to figure out for themselves what parliament would look like from a particular distribution of the
party vote.”
"The simulator allows anyone, from researchers to members of the public, to test their hypotheses. For example, it has
been argued that the Supplementary Member (SM) system is very similar to First Past the Post (FPP), and with the
simulator you can certainly see that it usually delivers a comparable number of seats in parliament."
The simulator calculates the distribution of seats in parliament based on the party vote. Users can either enter their
own estimation of the level of support for each party or use pre-programmed results from each of the elections since MMP
was introduced in 1996. The number of electorate seats won by minor parties can also be adjusted.
Some voting systems require voters to rank candidates in order of preference. Since this information was not collected
in past elections, the simulator uses data from the 2008 New Zealand Election Study to deduce how voters are likely to
have ranked the parties.
The simulator also allows alteration of the five percent threshold that, under the current voting system, parties must
reach to enter parliament unless they win an electorate seat. The threshold is one aspect of MMP that will be reviewed
by the Electoral Commission in 2012 if the system is retained in this year’s referendum.
The simulator can be accessed at http://cmss.auckland.ac.nz/2011-referendum-simulator. All of the source code and the
assumptions made by the research team can be viewed online.
The Centre for Mathematical Social Science at The University of Auckland is an interdisciplinary research group with
expertise in mathematics, statistics, computer science, economics, and philosophy. It is interested in collective
decision making by self-interested agents, ranging from members of human society to multi-agent computer systems. Its
work has applications for politics, economics and artificial intelligence.
Notes
Information about the 2011 referendum on the voting system can be found on the Electoral Commission website at
www.referendum.org.nz
The New Zealand Election Study, by researchers from The University of Auckland and University of Exeter has analysed
political behaviour over seven successive New Zealand elections. For more information see http://nzes.org