Tagged: spreadsheets

Special votes are increasingly turning towards Labour, away from National

So the final election results are out. The results weren’t too surprising… They produced the second of the two likely scenarios I outlined in my last blog:

  1. Golriz Ghahraman (Green) and Angie Warren-Clark (Labour) are in. Nicola Willis and Maureen Pugh (both National) are out.
    • Winston remains king/queenmaker. A Nat-NZF government would have 65/120 seats. A Lab-NZF-Green government would have 63/120.

I’m pleased to have Ghahraman and Warren-Clark in there, as they both seem like good people. They also make the Greens’ caucus 75% women and Labour’s caucus 45.65% women (but meanwhile National’s caucus is lowered to 30.36% women):

Screen Shot 2017-10-07 at 7.55.20 PM.png

Here’s how everyone did in the end (for similar numbers for previous elections, see my previous blog):

Screen Shot 2017-10-07 at 6.58.14 PM

In 2020 we could have even more seats swinging from National to Labour after special votes are counted, if the following four trends continue:

  1. Labour did their best ever on special votes: 18.91% better than on preliminary votes. Their previous best was last election, where they did 14.16% better.
  2. National did their worst ever on special votes: 21.12% worse than on preliminary votes. Their previous worst was last election, where they did 16.86% worse.
  3. There were more special votes than ever, even more than the Electoral Commission predicted: 422,094 or 16.29% of votes.
  4. This was the worst election ever for minor parties. National and Labour together won a massive 81.34% of party votes: the highest since party votes have existed:

Screen Shot 2017-10-07 at 7.21.49 PM

Meanwhile, TOP did an impressive 63.18% better on specials than on preliminary votes. Only a few parties have done better than this on specials:

  • Internet Mana in 2014 (97.58% better),
  • Mana in 2011 (77.34% better),
  • Māori in 2005 (68.76% better),
  • Green in 2002 (85.52% better).

This brought TOP up to 2.4% of total votes. This was towards the upper end of what we could have expected based on the polls, but it isn’t anything like the massive polling error that would have been required for them to crack 5%.

One fact from the final results that could give some hope to left-leaning people: The combined National-ACT vote has dropped below 45% for the first time since 2005.

Anyway, now that the final results are in, Winston claims he’ll finalise a coalition agreement within the next five days. Will he do it? Who will he go with? How productively will he work with them? Only time will tell… I can’t predict Winston Peters’ actions with spreadsheets.

Advertisements

Realistic possibilities for Parliament post-specials (all two of them)

You, like me, may be wondering what will change in the parties’ seat numbers when the final vote count (including the special votes1) is released on Saturday. Well, I’ve crunched the numbers, and…

Short answer: There are only two likely scenarios for how the preliminary seat allocations in Parliament will change after special votes are counted:

  1. Golriz Ghahraman (Green) is in. Nicola Willis (National) is out.
    • Winston remains king/queenmaker. A Nat-NZF government would have 66/120 seats. A Lab-NZF-Green government would have 62/120.
  2. Golriz Ghahraman (Green) and Angie Warren-Clark (Labour) are in. Nicola Willis and Maureen Pugh (both National) are out.
    • Winston remains king/queenmaker. A Nat-NZF government would have 65/120 seats. A Lab-NZF-Green government would have 63/120.

Warning: by reading any further you are entering into the spreadsheet dystopia that is Caleb Morgan Day’s mind.

Long answer: On the basis of past election results since 19992, there are only two likely scenarios for how the preliminary seat allocations in Parliament will change after special votes are counted:

  1. Golriz Ghahraman (Green) is in. Nicola Willis (National) is out.
    • This is what will happen if the special vote results are like what they’ve been most MMP elections in NZ.
    • On this scenario, A Nat-NZF government would have 66/120 seats. A Lab-NZF-Green government would have 62/120.
  2. Golriz Ghahraman (Green) and Angie Warren-Clark (Labour) are in. Nicola Willis and Maureen Pugh (both National) are out.
    • This is what will happen if the special votes are like what they were like last election.
    • On this scenario, a Nat-NZF government would have 65/120 seats. A Lab-NZF-Green government would have 63/120.

The spreadsheets to back this up:

Preliminary, final, and special vote %s for the four main surviving parties (plus “other”), 1999-2017:

The important figure is “% diff”. This is how much better or worse a party does on special votes than it did on the preliminary count: ie its relative special vote performance.

specialvotes1

What would happen if the special votes act like five “average” scenarios:3

The five “averages” are:

  • mean of past special vote performance (1999-2014),
  • median of past special vote performance (1999-2014),
  • each party having the same special vote performance as what they did in the election where they got the closest to their 2017 preliminary result (different elections for different parties),
  • each party having the same special vote performance as what they did in the 2005 election, which was arguably the most similar to the 2017 election, and
  • each party having the same special vote performance as what they did last election (2014).

specialvotes2

What would happen if each party did as good as it’s ever done before on special votes, and its rivals did as badly as they’ve ever done before:

Note that each of these still only produces the two basic scenarios of seat numbers.3

specialvotes3

There are a few other technically-possible scenarios, but all would be surprising.

First, the spreadsheets!

What it would take for each of these scenarios to come true:

Figures in bold involve a party doing better or worse than it’s ever done before. I’ve tried to approximate the most likely way for each scenario to actually happen.

specialvotes4

Then, the text summary that doesn’t really deserve the term “summary”! In order of likeliness (to my mind), here are the other technically possible scenarios:

  1. Labour gain not one but two seats from National (Green also still gain their seat from National).
    • This is somewhat unlikely. This would require National to do 25% worse on specials than on normal votes. The worst they’ve ever done is 16.86% worse, so this would be a surprise, though National are trending downwards in their special vote performance. It would also require Labour to do 25% better on specials than on normal votes, and the best they’ve ever done is 14.16% better. This is more likely to happen, as Labour are trending upwards in their special vote performance, and they’ve probably taken some Green voters this time.
    • On this scenario, A Nat-NZF government would have 64/120 seats. A Lab-NZF-Green government would have 64/120.
  2. The status quo: nothing changes from preliminary results to final results.
    • This is unlikely. It would require the Greens to do only 15% better on specials than on normal votes. The worst they’ve ever done is 38.2% worse (in 2011). However, they didn’t do that much better (43.37%) in 2005, which was arguably the most similar election to this one. It’s also possible that the Greens’ usual special vote effect will be largely swallowed up by Labour and TOP this time.
    • On this scenario, A Nat-NZF government would have 67/120 seats. A Lab-NZF-Green government would have 61/120.
  3. National retain all their seats, and the Greens gain their one from Labour instead of from National.
    • This is very unlikely. It would require National to do equally well on specials as on normal votes, and the best they’ve ever done is 2.23% worse, in 2002. Each election since then they’ve done at least 10% worse. This scenario also requires Labour to do as badly on special votes as they did in 1999, ie doing 4.87% worse on specials than on normal votes. Each year since then, Labour have done better on specials than on normal votes, and this has been trending upwards.
    • On this scenario, A Nat-NZF government would have 67/120 seats. A Lab-NZF-Green government would have 61/120.
  4. New Zealand First gain one seat from National (Green also still gain their seat from National).
    • This is very unlikely. It would require New Zealand First to do equally well on specials as on normal votes, and the best they’ve ever done is 1.71% worse, in 1999. Every other election, they’ve done at least 15% worse, and they’re trending downwards in their special vote performance. Finally, to balance out the numbers, it requires “Other” to do 70% better on special votes than on normal votes. We can’t rule this out because TOP are an unknown quantity, but TOP are less than half of the “Other” votes, so to push the overall “Other” figure up by this much, they’d have to do even better than the Greens’ amazing 2002 special vote performance.
    • On this scenario, A Nat-NZF government would have 66/120 seats. A Lab-NZF-Green government would have 63/120.
  5. The Greens gain two seats from National (Labour don’t gain any).
    • This is very unlikely. It would require the Greens to do 110% better on specials than on normal votes. Only once have they come remotely close to this (2002, when they did 85.52% better). All other elections they’ve done 38-55% better. This scenario also requires Labour to do as badly on special votes as they did in 1999. As discussed under #3, this is unlikely. Finally, to balance out the numbers, it requires “Other” to do 65% better on special votes than on normal votes. As discussed under #4, this would be amazing.
    • On this scenario, A Nat-NZF government would have 65/120 seats. A Lab-NZF-Green government would have 63/120.
  6. New Zealand First gain one seat from Labour (Green also still gain their seat from National).
    • This is extremely unlikely. It would require New Zealand First to do even better on specials than on normal votes, and as discussed under #4, even doing equally well is unlikely. It would require Labour to do even worse than they did in 1999, and as discussed under #3, it’s unlikely they’ll even do equally badly. Finally, to balance out the numbers, it requires the highest TOP/Other vote yet. As discussed under #4, this would be amazing.
    • On this scenario, A Nat-NZF government would have 65/120 seats. A Lab-NZF-Green government would have 63/120.

You’ll notice that on both the likely and the unlikely scenarios, NZ First always has the balance of power. (Unless the “teal deal”4 the media are frothing at the mouth over comes to pass. And it won’t.)

By the way, here are some other scenarios which are so impossible that I haven’t even bothered spreadsheeting them:

  1. NZ First lose a seat.
    • Actually I did spreadsheet this one (see the comments section). This is extremely unlikely, because even on their worst precedented special vote performance, NZ First still get almost 7.5% of the effective party vote, and 9 is 7.5% of MPS. They’d have to do a LOT worse than ever before on special votes to be rounded to 6.67% of the party vote for 6.67% of MPs (8 MPs). And by “a LOT worse” I mean unrealistically worse (see the comment for the deets).
  2. TOP cross the 5% threshold.
    • This would require TOP to more than double their party vote from the preliminary count to the final count. I.e. it would require them to get more votes from the 15% of special votes than from the 85% of normal votes.
  3. ACT get a second seat.
    • This would require ACT to more than double their party vote from the preliminary count to the final count. I.e. it would require them to get more votes from the 15% of special votes than from the 85% of normal votes.
  4.  Any other parties make it into Parliament as a result of a change in an electorate vote result.
    • Howie Tamati and then Te Ururoa Flavell (both Māori party) are closest to achieving this, but they’d have to get at least 60-70% of the electorate special votes. Ain’t gonna happen. Marama Fox (Māori party), Hone Harawira (Mana) and Raf Manji (independent) are even further away, no TOP candidates even came close to second place, and Damian Light (United Future)… well, it seems cruel to even link to his results.
  5. Any other electorate seats change hands.
    • I can’t be bothered crunching the numbers, but the closest seat was Ōhāriu and it’s unlikely Brett Hudson (National) would win on specials as the specials tend to favour Labour and disfavour National. There aren’t any seats where the Labour candidate is close enough to have a hope of knocking out the National candidate on specials. Winston Peters isn’t close enough either. And even though Metiria Turei did the best of any Green electoral candidate, she wouldn’t even catch up if she won every single electorate special vote 😦

Don’t say I didn’t warn you about that spreadsheet dystopia.

Footnotes

  1. The Electoral Commission estimates that there will be about 384,072 special votes (about 15% of total votes). I’ve estimated that 90.77% of these will have a valid party vote, as was the case in 2014. (In 2005 it was 92.55%, in 2008 92.78%, and in 2011 92.26%. I couldn’t find the figure for 1999 or 2002. The Commission hopes this number will be higher this time because they think they communicated better about the need to enrol in advance or enrol at the same time if placing an advanced vote. However, I chose the most conservative estimate.) This produces an estimate of 348,603 valid party votes from the special votes. 
  2. I couldn’t find 1996’s preliminary results. 
  3. You may notice that the totals under “% special” and “% final” don’t actually add up to 100%. In practice, on most of these scenarios, at least one party’s special vote performance (e.g. TOP, included as part of “Other”) would have to be higher than what is listed for the scenarios to actually be possible. 
  4. I’m pretty sure I came up with this phrase first. I’m not just a spreadsheet nerd, I’m also the kind of nerd who comes up with the kind of rhyming pun that journalists love!