top of page
  • rmacculloch

Can Polls Drive Election Results under MMP? Short answer: Yes, they can

The Centre for Mathematical Social Sciences at Auckland University held a conference last year that I attended. Part of the discussion centered on what is the best voting system for a country to have, a field called "social choice" which lies at the junction of economics and mathematics. NZ of course moved away from the UK-style First-Past-the-Post system to a German-style Mixed Member Proportional Representation (MMP) system many years ago.

Our MMP is so complicated, with concepts like "overhangs", that none of us pretend to understand many of its implications. For example, whilst we think of polls as being quite innocent, merely reflecting who people are intending to vote for, Arkadii Slinko, a Maths Professor, noted that under MMP the polls can affect the outcome of the election itself. How does that work? Well, if a poll comes out just before an election saying that 4.9% of people are intending to vote for, say, NZ First, then some folks may react to the poll and decide to change who they support and vote for NZ First instead just to get it over the 5.0% line, enabling the party to win a bunch of seats in parliament.

If you're interested in more of the nuances of this issue, then the research topic is whether or not the proportional representation system is manipulable. The answer depends on the type of voter. If she's interested in maximizing the number of seats for her preferred party, then she must vote sincerely. However, if she wants to maximize the chances of her favorite party to form a government, then sometimes it is worth voting insincerely for a potential coalition partner increasing an index of power in the post election government formation game.

Which just goes to show how complex MMP is - a system whereby polls can drive who you intend to vote for, not the other way around.



bottom of page