Is ACT liberal?

I gave a talk a couple of weeks back for the ACT Party's Upper South Island conference. I there advanced the argument that I made on the blog back in March: namely, that ACT should guard against positioning itself as a right-wing rump to National and instead should re-emphasize social liberalism.

After many assurances from folks in attendance that ACT was indeed the Liberal party, Rodney Hide gave the counterargument: that ACT would do poorly by emphasizing social liberalism when so many disenchanted National voters are in play. The counterargument isn't a bad one as far as strategy goes. If National has pushed to center under Key, then we might expect Key to welcome a bit of brand differentiation: National for the centrist Blues, ACT for the non-moderates, and a stable coalition. That could be jeopardized if the "keep National honest" crowd voting ACT feared they could instead wind up with a Labour-ACT coalition: especially the rural anti-ETS folks that ACT may well wish to woo.

Of course, ACT's polling numbers, which continue to hover around 1%, suggest that there's not been much exodus from National to ACT; we'll see how things move as ACT's run against the emissions trading scheme continues.

But let's look at some numbers from the 2008 NZES. I ran a couple of (very rudimentary) principal component analyses on economic and social liberalism, pulling out a measure of each respondent's relative economic and social liberalism. I then split the population up into 9 quadrants: economic lefties, centrists, and liberals; social conservatives, centrists, and liberals, using a population distribution cutoff for all: 25th percentiles in either direction.

If we restrict ourselves to the 25% most economically liberal, 51% of those are centrists, 21% are right-wingers, and 28% are liberals. Now, these don't amount to a whole lot in the overall population - 4.8% and 6.3% for right wingers and liberals respectively - but it does suggest that there are more true liberals that could be courted than there are right-wingers.

Let's start by looking at where self-identified ACT voters from the NZES placed themselves, according to the  factor analysis described.

The yellow dots show the positions of ACT voters.  The red lines show twenty-fifth percentiles at the population level: 50 percent of the population lies between the two horizontal lines, with twenty five above the upper line and twenty five below the lower; similarly for the vertical lines.  The upper right quadrant shows the folks in the top twenty-fifth percentile on both economic and social liberalism: true liberals.  The bottom right quadrant shows the folks in the top twenty-fifth percentile on economic liberalism but the bottom twenty-fifth percentile on social liberalism: classic right wingers.  The NZES suffers from small sample problems with respect to ACT voters, population weighting on the other characteristics using the NZES's preferred weighting scheme slightly increases the proportion of ACT voters in the liberal quadrant.  In the weighted sample, 6% of ACT voters fell into the liberal quadrant while 7.6% fell into the right quadrant. So, as far as folks who voted ACT who also answered the 2008 NZES go, ACT isn't really a liberal party.  It attracts few classic left-wingers (top left quadrant) and no authoritarians (bottom left quadrant).  But the folks attracted in '08 were a mix of slightly-more-economically-liberal voters.

What happens if we look at a similar scatter plot for National voters, restricting ourselves to the ones who've answered "no" to the question "Would you never vote for ACT?"  Here's the result:

It looks like there are more right wing National voters who'd consider voting ACT than there are liberals in National who'd consider voting ACT, but the margin isn't huge.  Of those National voters who said they'd consider ACT, using the weighted population sample, 4.4% fall into the liberal quadrant while 6.2% are right-wingers.  But, of course, ACT can draw from more than just National.

So, now let's try it considering the sample of people who don't currently vote ACT but haven't ruled out voting ACT: 68% of the weighted population.

Let's see where those folks sit on the scatter diagram.

If we run the weighted population sample, 7.4% of these potential ACT voters sit in the liberal quadrant; 5.2% sit in the right-wing quadrant.  So there are more potential ACT voters in the liberal space than there are in the right wing space.  If we start limiting the set by restricting it to folks who rank ACT at least a 7 on a 10 point scale, though, the numbers reverse: the non-ACT voters who like ACT the most are right-wingers.  This could mean that there's more chance for ACT to convert these folks over to being ACT voters, or it could just reflect that ACT ran a fairly right-wing campaign in 2008 with David Garrett being fairly prominent.

So if ACT is chasing fleeing National voters, there may be more of them among the right wingers, but there's absolutely no evidence in the polls as yet of any exodus from National to ACT.  But there are more overall potential voters in the liberal quadrant.  There may be a few strong ACT supporters who don't yet vote ACT in the right wing, but the overall numbers there aren't better than the numbers in the liberal quadrant.

I've also heard rumours of possible courting of the Christian right.  This would be a bad idea.  55% of those rating ACT 6 or better on the "like" scale report they never attend church while only 19% say they attend church at least monthly: pretty similar to the overall population distribution in a pretty secular country.  And I can't think of a better way of chasing away the liberals who still like ACT.

In short, a fairly cursory analysis of NZES data shows ACT voters to be more right wing than liberal but that there's somewhat greater potential for growth of the overall vote among liberals than among right wingers.

For all the protests of being a liberal party, and despite the strong liberal leanings of many in their youth wing, many of their staffers and some of their MPs, it's pretty hard for an outside observer to identify ACT as such. That they attracted a somewhat greater proportion of right wingers than liberals in 2008 says something about their actual policy positioning.  Yes, they mostly supported the medical marijuana legislation.  But by pushing hard on crime and punishment without putting much emphasis on their potentially more liberal social positions on other issues, they're cementing a position at the right wing tail of National.  Which could work out for them if social conservatives aren't much less likely to leave National than are social liberals.  I guess we'll see.

It would be a mistake for ACT to make a lot of noise about keeping options open by courting social liberals when Labour has no hope at all in 2011.  But if ACT is a principled liberal party, they wouldn't have to couch it in strategic terms. Rather, it would just be reminding folks of what their principles are supposed to be: not just in speeches at conventions but also in policy advocacy. Longer term, I still think a liberal positioning would also do them well strategically.  At least the numbers seem to suggest it wouldn't hurt them.  But it's getting late.  ACT supported the civil asset forfeiture legislation that allows police to seize property on "balance of probabilities"; ACT also supported changes to legislation allowing police to collect DNA evidence from folks under arrest rather than just convicted criminals.  And, as the Greens predicted when ACT supported the legislation, we're now hearing rumours that the police are using that power to strong-arm young Maori into giving up DNA samples: give us a sample voluntarily, or we'll arrest you.

It's hard to think of occasions where ACT has stuck its neck out a bit on civil liberties: medical marijuana is the only one that comes to mind.  And the last iteration of the 20-point plan is almost entirely on economic issues; asset forfeiture was something they campaigned on, but on the wrong side.  They also favor New York style "broken windows" policing, though the evidence that such legislation actually reduces crime rates is weak at best.  Liberal?

So, to sum up.  ACT's current voters, as found in the NZES, are more right-wing than liberal.  ACT's potential growth in voters seems stronger on the liberal side than on the right-wing side, but ACT's policy emphasis is more attractive to right-wingers than to liberals. Cementing themselves as the farther-right flavor of National isn't an implausible strategy, but the evidence thus far doesn't suggest it's working, though the ETS campaign may yet yield returns.  But it's not an overall strategy that ought to convince liberals to get out and vote for ACT.  I get the rather strong feeling that, for civil libertarians, it's always going to be "jam tomorrow": that the potential policy gains on economic issues will always be deemed more important than the compromises with National on civil liberties.  And at current margins in New Zealand, civil liberties matter more to me.

More chart porn below, for the folks into that kind of thing.

New Zealand First voters in black; Greens in green (above).  A few confused socially liberal NZ First voters, but otherwise mostly as you'd expect.

A fair few liberal non-voters.  Maori Party voters skew left on economic issues but centrist on social issues.  United Future respondents surprisingly liberal: I was expecting more in the right wing quadrant.  Pretty small sample size on those voters though.

Finally, just how close are National and Labour voters?