## Projection of special votes for 2017

One nerdy post-election exercise I find fun is to project (not “predict”) the result including special votes, based on the 2017 preliminary results and the impact of the 2014 special votes. The most significant assumption I make is that the relationship between special votes and the preliminary count in 2014, is a good predictor for the same relationship in 2017. Now, this isn’t a great assumption: in 2014, the Greens actually did even better than projected, and National worse. But it’ll probably provide some guidance.

A full description of the method is in the post I wrote for the 2014 election.

This election, special votes comprise 15.4% of the total votes cast, compared to 12.6% in 2014. This fact is accounted for in the projection, so the impact looks slightly larger than in 2014.

Here’s what I got:

Party | Preliminary results | Projected final results | Gain/loss | ||||

Votes | % | Seats | Vote | % | Seats | ||

National | 998,813 | 46.03% | 58 | 1,150,396 | 44.85% | 56 | -1.18% |

Labour | 776,556 | 35.79% | 45 | 938,382 | 36.59% | 46 | +0.80% |

NZ First | 162,988 | 7.51% | 9 | 187,707 | 7.32% | 9 | -0.19% |

Greens | 126,995 | 5.85% | 7 | 162,564 | 6.34% | 8 | +0.49% |

Act | 10,959 | 0.51% | 1 | 13,044 | 0.51% | 1 | +0.00% |

TOP | 48,018 | 2.21% | 0 | 56,783 | 2.21% | 0 | +0.00% |

Māori | 23,456 | 1.08% | 0 | 28,698 | 1.12% | 0 | +0.04% |

So under this projection, Labour and the Greens would each pick up a seat at National’s expense. This wouldn’t really affect the underlying “balance of power”, which would still rest with NZ First. But it would put the left (Labour/Greens) and right (National/Act) blocs closer together, having 54 seats against 57 (currently 52 vs 59).

It turns out that this projection’s not especially sensitive this year. Golriz Ghahraman, number 8 from the Greens, would pick up the 114th seat, leaving plenty of room to spare. Number 9, Mojo Mathers, wouldn’t be due until the 129th seat, miles away. As for Labour, Angie Warren-Clark gets the 120th and final seat, Labour’s 46th quotient of 10,312, a bit ahead of National’s 57th quotient, which is 10,181; Labour have about 2 percentage points on special votes to spare. For Labour to get Helen White in, their 47th quotient (currently the 122nd seat) would have to leap-frog National’s 46th (currently 119th). This would require them to do a full 9 percentage points better in special votes than ordinary votes—a tall order.

So small shifts in special voter behaviour relative to 2014 won’t really affect this outcome, though of course large shifts would.

Now, in 2014 I said that it would be “close”, then the Greens picked up so many special votes that their 14th quotient leap-frogged into the 119th position. So maybe a “large shift” isn’t out of the question. Then again, the Greens would need to win about 12% of the special votes to get Ms Mathers into Parliament, more than *double* their percentage of votes in the preliminary count.

Graeme Edgeler published his version over at Legal Beagle a few hours ago. Our methods are the same in principle but different in some subtle details, so our projected vote counts are marginally different, but not in a way that impacts how many seats each party would get.