Split voting results in the NZ referendum

Earlier this week, I reported the polling-place correlations of votes for various options in the New Zealand electoral-system referendum of 2011.

An analysis of the split-voting statistics, as compiled by the Electoral Commission, offers another window on the same questions addressed there. The advantage of these data is that they are based on the Commission’s examination of individual ballot papers, as votes in the two parts of the referendum were cast on a common ballot. ((The same page at the Electoral Commission carries out the split-ballot analysis in each district, although not in each polling place.))

As we already knew, most voters who voted to “Keep” MMP (Part A of the ballot) did not vote at all on Part B, where they could select among several potential replacement systems. In fact, 54.7% of “Keep” voters cast “informal” ballots (meaning blank or invalid).

What is most striking is that of those who cast a vote in Part B, a very large plurality voted for the old First Past the Post (FPTP or FPP) system. The percentage of valid Part B votes for FPTP cast by those who voted in Part A to keep MMP was 40.4%! I wonder how many of these were “insincere” votes, by voters who assumed that FPTP could never defeat MMP in the follow-up referendum that would have happened in 2014 if a majority had voted for change in Part A.

The next most popular choice for “keep” voters was STV, with 24.0%. This, of course, makes sense. Voters who prefer proportional representation ought to rank STV and MMP one after the other as their sincerely preferred choices among the systems on offer.

For third place among “keep” voters who chose any system in Part B, “Supplementary Member” actually edged out “Preferential Voting”, 18.7% to 16.9%. The choice here is an interesting one, as it indicates a preference for any sort of mixed-member system–Supplementary Member is actually a mixed-member majoritarian (MMM) system–over a return to all single-seat districts with ranked-choice ballots. (Preferential Vote = Alternative Vote or Instant Runoff.) In terms of sincere voting, it is not straightforward which of these “should be” preferred by voters whose first choice is MMP.

On the one hand, MMM (SM) retains a tier of nationwide proportional seats, but these would cease to be compensatory (and, under the specific proposal, there would have been many fewer of them). On the other hand, while PV (AV) abolishes all party-list seats, it gives voters for a losing party in a district the potential of determining on preferences which of the big parties wins the seat. Both systems would likely produce majority governments frequently, but those majorities and the campaigns that produced them would have been of a different character.

As for the “change” voters, there are no surprises. Only 3.7% did not cast a valid vote in Part B. Of the remaining voters who did cast a valid vote, just under half (49.8%) voted to go back to FPTP, and 28.0% voted for MMM (SM).

In the earlier correlation analysis of polling place results, I noted the lack of correlation between the vote for Part A and the vote for STV. As we saw above, 24.0% of “Keep MMP” voted for STV, and so did 12.4% of the “Change” voters (for whom STV came in third place, ahead of PV/AV). When the informal votes are included in the denominator, the percentages drop to 10.9% and 11.9%, strikingly close. It is not possible to say for sure from the available data, but maybe STV was a Condorcet winner in the wider electorate, albeit not an especially popular one overall. That would make some sense, as it provides the compromise features of all members being elected in local districts, yet offering a considerable degree of proportionality.

In any case, MMP is now clearly confirmed as the New Zealand electoral system, and so discussion shifts to the formal review of MMP, and how the system might be “improved”.

4 thoughts on “Split voting results in the NZ referendum

  1. I’m more interested in the split vote patterns of the voters who “split” their votes when they cast their electorate vote for a candidate who does not represent the party for which they cast their party vote. At the 2011 election 30.7% of voters split their vote compared to 29.63% in 2008. Is this still higher than in any other MMP jurisdiction?

    Even 16% of National Party voters split their votes, and 18.5% of Labour Party voters. Only in three electorates were tactical votes a reason. In Epsom 60% of National Party voters voted for Banks (local ACT candidate and National’s ally), while 35.5% of Labour voters voted for National’s local man to try to stop Banks and keep ACT out of the House. Similarly, in Ohariu 57% of National Party voters voted for Peter Dunne, United Future leader and National ally. In Te Tai Tokerau 40% of National Party voters voted for the Maori Party local candidate to try to stop Mana leader Hone Harawira, while 25% of Labour Party voters voted for Harawira who had left the Maori Party over its support for the National Party government.

    Other than those three, most of those voters were using their “split ballot” power to prefer a local candidate they liked better than their own party’s candidate, or to punish their party for running an incumbent past his best-before date, or some such motive. NZ voters seem to have taken a serious liking to this feature of MMP, partly to compensate for their inability to rank the list candidates. A candidate who runs well behind his party but wins a list seat may find he is ranked very low three years hence, or even retired. Has anyone analyzed the 2011 split vote patterns yet?

  2. “To have a young candidate from a professional background like Chao-Fu Wu making such a positive impact in such a short space of time is a good thing for the Labour Party and the people of Botany.” Well, perhaps not.

    Labour’s Taiwanese-born candidate was unattractive to 24% of Labour voters and 66% of Green voters. Almost 9% of Labour voters even voted for the local National man to make sure their own candidate lost. So did 34.5% of Green Party voters, who had no local Green candidate. Are so many Kiwi voters unwilling to vote for a Chinese candidate? But he was listed last out of the 70 list candidates; not a strong boost from his party.

    Then there was Carol Devoy-Heena, #68 on the Labour list after running in 2008. In the safe National seat of Bay of Plenty, only 72% of Labour voters voted for her, while almost 11% voted for the National man. Green Party voters, with no local Green, voted only 44% for Carol Devoy-Heena, and 33.5% for the National man. No one seems rude enough to say why she did poorly.

    Sunny Kaushal was the first candidate of Indian origin to contest from an Electorate, representing the Labour Party in the safe National seat of Pakuranga. Only 78% of Labour voters voted for him, while 9% voted for the National man, as did 33% of Green voters, who gave Sunny Kaushal only 48% of their votes (with no local Green on the ballot). He did not make the Labour list.

    I don’t want to continue.

  3. While I’m sure that there is a degree of racism (the nearby suburb of Howick is nick-named ‘Chow-ick’), all three of those examples are new candidates for those electorates.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s