Legal Beagle by Graeme Edgeler


Election '20: The Special Votes

The 2020 General Election has a preliminary result. For reasons I am unable to really explain, we will not have even a preliminary result for the end of life choice and cannabis legalisation referendums for some weeks (I dropped the ball on that one when the referendum legislation was before select committee, focusing on other concerns with the bill).

The preliminary result is mostly meaningless from a legal standpoint, but we want to know what happened before we go to bed on election night, so we get a rough and ready count of the ordinary votes: those cast by people whose name appears on the printed electoral roll who voted at a voting place designated for their electorate (including during advance voting).

The official count is still to happen. They’re more careful with that, and there's a lot of cross-checking - starting with a check to make sure that no-one voted more than once. The official count also includes the counting of special votes. Special votes are:

  • votes cast overseas;
  • votes cast by the telephone dictation servce, including this year, by people in managed isolation;
  • votes cast on polling day (or in advance) by people voting at a voting place not desginated to serve their electorate;
  • votes cast by people who enrolled after the printed electoral roll was closed (including during advance voting or on election day itself);
  • takeaway votes;
  • votes cast by people on the unpublished electoral roll; and
  • votes cast by people who are not on the electoral roll, who thought they were enrolled.

    There are some differences this year - and managed isolation is only one. The biggest change is that people can be enrolled on election day. Last election around 20,000 special votes were cast on election day by people who were not enrolled to vote. In 2017, those votes were not counted - you could enrol during advance voting, but not on election day itself. That has changed: in 2020 you could do this on election day itself.

    This election there are a lot of special votes. The most ever. The Electoral Commission estimate is that there are 480,000 special votes, which is around 17% of the total. The estimate is always wrong. The Commission know approximately how many special votes were cast in New Zealand. Postal votes don't have to have arrived yet, and add to the total. The Electoral Commission's election night estimate of special votes in 2017 was that there would be 384,072. In the end there were 446,287 (of which 417,980 were counted).

    We do not know what electorates special votes are intended for - the Electoral Commission know where they were cast, but over the coming days, local returning officers will be sending the votes to the returning officers of other electorates, where they will be checked, and if valid, opened and added to the official count.

    Historically, the voting patterns of those who cast special votes differ from those who cast ordinary votes. Special votes in recent elections have tended to favour left-aligned parties. It is probably fair to assume that this general direction of special votes this will continue at this election. But with so many more special votes, and new reasons why some were cast, whether it is fair to assume that the size of the effect will be similar is less clear.

    That said, we don’t have anything better to go one, so using the same rudimentary method I use each election (assuming the variance in special votes is the same size as it was at the preceeding election), along with the Electoral Commission’s estimate of the number of special votes at this election, I estimate the following final result:





    Vote share


    Vote share

















    Green Party





    New Zealand First





    New Conservative










    The Māori Party





    In 2017, National did 20.7% worse in the special votes than they did in the ordinary vote. While Labour did 19.5% better and the Greens did 44.2% better. The Māori Party did 57% better.

    Even with the largest ever special vote, there isn't a lot of change in the estimate: National is down one, and Labour is up one. Labour is close to a second, however. If there were 500,000 special votes, and not 480,000, the estimate would predict a Labour gain of two and a National loss of two. And if there were 525,000 special votes, the estimate would change so that National would be down two, but they'd go one each to Labour and the Greens. An extra seat for the Māori Party is exceedingly unlikely.

    Of course, as the number of specials votes has increased, this will mean the make-up of the special vote cohort has changed, and as the reasons people are casting special votes have changed, there is an extra note of caution, and there is election day enrolment as well.

    We don't have terribly long to wait for the official count, but that is now mostly of interest to the MPs who are on the cusp, as the shape of the new Parliament is known: Labour has a mjaority on its own. More interesting for the rest of us will be the results of the two referendums.


    Some more explanation of how close parties are, and whether some electorates might change

    I've answered some additional questions on twitter, which provide some further context for my conclusion above. I am happy with my conclusion above. It worked last time, and I am not remotely enough of a Nate Silver to advance a serious model, this is rough and ready, just slightly better than people's reckons (which to be fair, reach approximately the same conclusion). I should also note that two election ago, I underestimated the special vote swing against National.

    Assumption: Party Special Vote Swing - the Green Party

    These are some other ways of looking at the numbers: in 2017, the Green Party did 44% better on special votes than ordinary votes. If the number of special votes is as estimated by the Electoral Commission, this would not be enough for them to take a seat off National. But in 2014, the Greens did 53% better on the Special votes. If that's what happened this year, it would be enough.

    Assumption: Number of Special Votes

    The Electoral Commission has estimated that there are 480,000 special votes. I have used this number in my calculation. On election night 2017, the Commission estimated there were 384,072 specialvotes, but in the end there were 446,287, of which 417,980 had valid party votes. I do not know how many more there will be this election, but if there were 500,000, and the swing was the same as last time, National would lose two seats. If there were 525,000, and the swing was the same, one of those seats would be to the Greens.

    Assumption: Party Special Votes Swing - The Māori Party

    The Māori Party tends to do best out of the Parliamentary parties in the special votes. In 2017, the Māori did ~57% better on special votes than ordinary votes. They would need to substantially better than that to get an extra seat this time. If there are 480,000 special votes, they'd need to do 99.5% better on special votes. This isn't impossible (legalise cannabis came close by doing ~97% better on special votes over ordinary votes in 2017). And if there were 520,000 special votes, ~93% better would be enough.

    Estimate of Effect of Special Votes on Whangārei

    Whangārei was the closest result on election night, with National MP Shane Reti ahead of Labour's Emily Henderson by 164 votes. 

    Shane Reti won 47.3% of the ordinary votes in Whangārei in 2017. His Labour opponent, in third on election night had 18.6%. Reti got 38.2% of the special vote, and his Labour opponent 24.2%, enough to move ahead of Shane Jones into second place. I am not making as estimate, but with over 4300 special votes cast in Whangārei at the last election, there is cause for Reti to be worried, and the seat could definitely swith to Labour's Emily Henderson.

    Estimate of Special Votes on Waiariki candidate votes

    Waiariki was won on the preliminary count by the Māori Party's Rawiri Waititi, with a lead of 426 votes over Labour's Tamati Coffee. It is the only seat that the the Māori Party has won this election. In 2017 Coffee won 53.6% of ordinary votes, and did slightly better on special votes with 54.3%. The Māori Party candidate, Te Ururoa Flavell was the inverse (there were only two candidates in 2017), doing slightly worse on special votes. 426 is fair margin however, and a switch would be a suprise.

    Estimate of Special Votes on Auckland Central candidate votes

    Auckland Central was won on the preliminary count by the Green Party's Chlöe  Swarbrick, with a lead of 492 votes. In 2017, Labour's Helen White won 39.8% of the ordinary votes, and 40% of the special votes. The Green Party's candidate at that election, Denise Roche, won 8.8% of the ordinary vote, and 12.4% of the Special vote. If anything, the gap will increase here, and Swarbrick seems safe.

    8 responses to this post

    Post your response…

    This topic is closed.