Polity: Who do the “missing million” like?

notices and features
10 years ago
Categories: election 2014, john key, Politics, Steven Joyce
Tags: david farrar, kiwiblog, polity, turnout

When I first saw DPF quote some anonymous correspondent talking about voter turnout, I thought he was having everyone on. But over the weekend John Key was quoting this exact same research as a justification for why National need to Get Out The Vote this time, and so was Steven Joyce on The Nation. It seems this really is the high-water mark for National election analysis. Colour me underwhelmed.

National appear to think most of the tide of new non-voters in 2011 were National supporters, because many of the places where turnout dropped a lot were safe National seats. The problem is that this pattern – big drops in turnout in safe National seats – is consistent with at least two theories:

The new non-voters were complacent National supporters, so seats with more National voters have more complacent ones who don’t bother to vote.
The new non-voters were peeling away from the good ship Labour, whose energies were turned defensively towards protecting support from their core low-income urban base rather than expanding their reach into more challenging electorates. This second pattern is also consistent with the idea of Labour’s vote collapsing into its core areas in 2011, compared to earlier elections.

To better understand which theory might be right, we have to first look more closely at the data. Of course, reading the tea-leaves like this is always a bit hand-wavy, but we box on nevertheless. Take Clutha-Southland, for example, which had lower turnout in 2011 by about 1,900 votes compared to 2008. In 2008, National won 20,235 votes in Clutha-Southland, while Labour won 8,091. In 2011, National won 20,020 votes in Clutha Southland, while Labour won 5,160.

We can see which story is better supported by the evidence here. Labour lost 3,000 votes overall, while National stayed about the same. The Greens’ in Clutha-Southland total went up by around 1,000 during that period, leaving around 1,900 previous Labour supporters unaccounted for. A couple of things might have happened:

They may have stayed home, which is the simplest explanation.
Or they might have switched over to vote National – driving National’s total up, but then had that impact cancelled a pile of National supporters may have not bothered voting – driving their total back down again.

Second, as I pointed out earlier in the year, National’s form of analysis here (inferring individual-level behaviour from electorate-level results) is highly suspect. There are better ways to do this work, but it looks like National doesn’t know what they are.

Third, survey evidence can help us, too. The New Zealand Election Survey is the best resource for this. Of the people who (1) admitted to non-voting in 2011; and (2) remembered who they had votes for in 2008, 92 said Labour, 13 said Greens, 70 said National. On that imperfect measure, you might conclude that the missing million splits about 3:2 to the left.

The 2011 NZES also finds no net switching between Labour and National after 2008. 2.1% of the population reported voting for Labour in 2008 and National in 2011, and 2.1% of the population also reported voting for National in 2008 but Labour in 2011.

I am sure there were some National supporters who chose not to vote in 2011 out of complacency. But I think it is a minority, and that most of the new non-voters (who voted up to and including 2008, and then stopped) are lefties. The survey evidence points that way, and so does the E9 evidence when looked at properly.