The final count of the votes, including special votes, has saved us from having to revisit  our ideas about majority governments under MMP. Oh - and I (sort of) told you so.

As I said here on the eve of polling day (and intimated again here the day after), the 2014 election wasn't over until Saturday, October 4*:

We could concievably be going to see a largish number of special votes remaining to be counted after the election night results are announced. In 2011, there were 263,469 of these that had to be counted after the election - well over 10 percent of the final total. It is possible we'll see even more this time around. And these special votes don't even start getting counted until 10 days after election day, meaning that we don't see the results of the count for at least 2 weeks. 

All of which means it may not be over even when it looks to be over. Because if past special vote trends hold true for this election (no guarantee that they will, but), special votes tend to favour parties on "the Left" (i.e. the Greens and Labour). Which, in a tight-run result may prove crucial to the final outcome. Furthermore, in some important electorates - Te Tai Tokerau, especially - the counting of specials may swing the result one way or the other. So just bear that in mind.

I was, of course, completely wrong to even speculate that the contest might be "a tight-run result". Even after the counting of specials (and official recount of ordinary votes), National remains towering above the nation's political terrain like the statue of Ozymandias in his glory days. While it may now have lost its outright majority in the House, thus allowing us political pundit nerds to continue to pontificate that is "near impossible" for one party to ever achieve this feat, I'm betting that there isn't a single item of business that it will no longer be able to progress into law as it wants. Its three-way choice for a majority gives it almost unlimited room for manoeuvre.

Nevertheless, it ain't a majority party no more - in fact, it is now less popular than it was in 2011. Which was not the result that you would have predicted on September 21, based upon the way previous special votes panned out (as outlined here and here). In addition to demonstrating the problems associated with assuming that instances of which we have had no experience resemble those of which we have had experience, the change also seems to have caught John Key a bit by surprise:

It's just odd actually more than anything, because it's a very, very complex mathematical formula, but the advice that we had was that it was pretty unlikely we would lose one [seat] because we got the 119th seat and not the 120th, but effectively what happened was the special's [votes] were so massively skewed against us.

It just seems odd - we got about 48 per cent of the vote across the country.

I'm pretty sure Key isn't using the term "odd" in the sense that "I think someone's fudged the numbers and it needs investigated." The great thing about New Zealand's electoral process is that it works so damn well! Seriously - aside from a few deluded loons who tried to start a petition asking the chair of the Electoral Commission to recount the votes because the result didn't "feel right", everyone accepts that the numbers that the Electoral Commission gives to us are a true and correct reflection of how the country voted. That's a remarkable and precious advantage to enjoy.

Instead, Key is reflecting surprise that special votes (and the final recount, too, of course) moved the parties' relative vote share quite as much as they did. Because a drop of over 1% for National and corresponding jump of 0.5% for Labour and the Greens is a pretty big shift from the election night's figures. What, then, might account for it happening?

I don't know for sure, but that's no reason not to speculate wildly. So here goes.

First of all, there were quite a lot more special votes available to influence the final result. In 2011, there were 263,469 special votes (or 11.8% of the total) cast. In 2014, there were 330,985 special votes (or 13.5% of the total) cast. 

Second, there was some reason to think that the make up of special votes might be different this election compared to the last one (as I already speculated here). One difference may be in overseas votes. In 2011, only 20,333 special votes were cast by overseas voters - less than 1% of the total vote. Given the number of ex-pat Kiwis living overseas and legally eligible to vote, that's a pretty small figure. It's also a lot lower than the estimated 32,000 overseas votes cast in 2008.

I suspect we'll see a bounce back in overseas votes in 2014, which (according to the received wisdom of New Zealand psephologists) favours the parties of the left as young Kiwis on their OE in the UK or the like turn in their votes for Labour or the Greens. That received wisdom may be complete hogwash, of course ... but there it is.

Furthermore, and here I really am starting to speculate without any very firm basis, I think that the big numbers of advance votes in 2014 may have had something of a delayed effect on the final results. On election night itself, everyone was on tenterhooks as to what it meant that some 700,000-odd votes had been cast prior to election day. In the end, it actually didn't seem to be all that relevant to the final outcome. The parties' share of the vote stayed pretty much flat from the start of the night (when the results of those advance votes were fed into the count) to the end (when the last of the election day votes from the big polling booths were counted).

But two things weren't being seen in those election night results . First, some of the advance votes cast before election day and counted on election night may have been votes that previously were special votes. So, for instance, a travelling salesperson who in the past may have cast a special vote outside of their electoral district may this time around have cast an advance vote before polling day. If enough of a certain type of voter changed their behaviour in this way, it will change the make-up of special votes from past elections.

Second, the election night results did not reflect any success that the left parties may have had in mobilising voters who weren't on the printed electoral roll to enrol and cast a vote. Any such votes (whether cast before election day or on it) had to be "specials". And the fact that this was a "turn out the vote" area into which the left parties put much greater efforts in 2014 means that some sort of effect should be seen ... as it indeed transpired, with Labour, the Greens and Internet/Mana all doing significantly better in their share of special votes than they did on the ordinary votes count on election night.

Sure, this effect wasn't anything like enough to change the overall outcome - and there's a good argument to be made that it was in the end a poor move because of the opportunity costs involved in generating those few extra special votes. But it may go some way to helping explain why 2014 wasn't like 2011. Or so I'm speculating wildly, anyway.

[Update: For anyone interested in the Math behind this (along with some analysis that largely mirrors my own, I recommend this excellent post by Chuan-Zheng Lee.]

 ----------------------------------------------------

* Insurance against inevitable electoral law nerd response: candidates and parties have until Wednesday 8 October 2014 to lodge applications for a judicial recount. If none are forthcoming (and it seems unlikely any will be made this time around**), then the return of the writ will occur on Thursday 9 October 2014. Furthermore, individuals wanting to challenge the result in an electorate contest, or party secretaries wanting to challenge the allocation of list seats under the party vote, have until Monday 3 November 2014 to lodge an electoral petition in the High Court.

And then the 2014 election really will be over!

[**Update 2: And wouldn't you know it - Hone Harawira has gone and made a fool of me. Cheers for that!]