Textor thesis fails empirical test

The 2016 federal election kills off Liberal Pollster Mark Textor’s theory of how the Liberal Party would win under Malcolm Turnbull.

Quoted in an article in The Australian he agreed that the ascent of the “centrist” Malcolm Turnbull would not result in a loss of conservative votes saying:

“The qualitative evidence is they don’t matter,’’ Mr Textor said. “The sum of a more centrist approach outweighs any alleged marginal loss of so-called base voters.’’

Now that the results are in we can test that thesis quantitatively, using election results, as well as qualitatively, using the results from our online qualitative polling.

The first part of the thesis is that the Coalition will win more votes in the centre.

After distribution of preferences the Coalition will have beaten the Labor Party by 50.37% to 49.63%, a two-party preferred swing of 3.12%. On first preferences the Coalition lost 3.37% and the ALP gained 1.32%.

Where did the rest of the first preference swing from the Coalition go?

The Greens grew by 1.26% leaving just an additional 0.79% to go to all the other minor parties, including the new whales of the minor party pool, the Nick Xenophon Team and Pauline Hanson’s One Nation.

These are the net figures. It is not as clean as this looks. While the ALP gained 1.32% net, they will also have lost some to the Coalition, and vice-versa. But in so far as the Greens are on the left and Labor and Liberal meet in the centre, Textor’s expectations were wrong. The major centre-left parties have advanced and the major centre-right parties retreated.

However, it is possible that some of the centre may have decamped to the Greens, as a protest party, or one of the other minors and might have come back via preferences. Preferences from the other minors are also a test of the second part of the thesis, which is that there will be a loss of right wing voters which is less than the gain from the move into the centre.

The electoral commission has not done a full distribution of preferences yet, so it is impossible to exactly know the answer to that, but a simple model demonstrates this is unlikely to be the case. In the last election Greens voters allocated 83.03% of their preferences to the ALP. If they did the same this election, then Labor won 49.71% of preferences from the rest of the minor parties. This is up from the 46.69% they won at the previous election.

So not only was there a loss on the left, but there was a loss on the right.

And it only gets worse if the Turnbull Libs got a smaller percentage of Greens preferences this election than last. If Greens preferences only flowed 78.84% to Labor, as they did in 2010, then Labor won 52.99% of the non-Greens third party vote.

Of course it is possible that the thesis was initially correct, but defeated by actions of other players, including Malcolm Turnbull himself. We will never know that, because it would require a number of parallel universes in which to run multiple simulations of the election.

However, it is unlikely because in the universe we do have, and looking at Australian, UK and US politics over at least the last 40 years, Centre Left and Centre Right parties have tended to win secure majorities when they won a significant slice of the blue-collar conservative vote. This was certainly true for Ronald Reagan, Margaret Thatcher, Malcolm Fraser, John Howard, Tony Blair and Bill Clinton.

In this universe we looked at the appeal of Malcolm Turnbull to centre left voters at the end of last year when he was hugely popular with the centre to left. In our qualitative polling panel he had the support of 31% of ALP and 21% of Greens voters.

We determined at the time that he was on probation, and was popular because these voters thought that he was similar to them in views on climate change, gay marriage and refugees and would also be more in tune with them on the economy. They also found him more likeable than either Abbott or Shorten and they expected him to change the Coalition’s policies.

By the time of the election the only thing that remained was the likability factor giving him a slight edge over Shorten, but with him and Shorten each having a net negative personal approval rating.

Many commentators describe the non-Greens minor party voters as being fundamentally driven by fear for their economic circumstances. In as much as this is the case then the Coalition pitch on the economy should have worked in Turnbull’s favour.

But we find these voters voice concern over cultural issues much more than economic ones (although the economy is definitely an issue). These cultural issues are to do with gay marriage, climate change, Islamic terrorism and refugees. They have a “team Australia” approach, rather than the “camping” approach of the cosmopolitans who tend to populate the inner city centre-to-left political spectrum.

Turnbull appeared unable or unwilling to address these concerns, so these voters felt alienated. Contra the Textor thesis they also have somewhere to go. As demonstrated above, somewhere around 50% of them are happy to preference Labor before the Liberals. It is only a short jump from this to giving Labor a first preference.

Three issues in the election campaign were nominated as being vote changers. They were Medicare/Mediscare, Superannuation, and the Economy. From this it appears that the LNP probably won the Mediscare debate but lost the economy debate, even though they were ranked better than Labor to deal with the economy. They also lost the Superannuation debate, although it didn’t make much difference because not many cared about it.

15% of respondents mentioned Medicare, 3.6% Superannuation, and 36% the economy of economics. We then compared how these respondents reported their first preference vote last election versus whether they cast their preference to favour Coalition or Labor this election.

On Mediscare we found that there was a relative 1.88% swing towards the Coalition from last time Labor voters, but a relative 1.33% swing to the ALP from last time Liberal voters. The relative effect with Greens voters was a relative 3 percent greater swing to the ALP. Let’s call that a dead heat.

Superannuation was an own-goal, with a relative 9.43% swing against the Liberals in this group but the percentage of respondents who were interested in super was very small.

The economy was a surprise with a relative swing to the Coalition of 0.35% from last time Labor voters, but a migration away from the Liberals of 3.29% over and above the general swing away. However virtually all Greens voters who came across to the Coalition mentioned the economy.

These figures demonstrate how small the movements in voter preference are that determine governments. They also show that none of the issues run this election grabbed the public.

In the end it is little wonder that Malcolm Turnbull just fell over the line. Neither major party gave voters much to vote for, and in so far as the Textor thesis has any validity, it is that the “centrist” Turnbull was just a fraction more personally acceptable than Bill Shorten.

But this is not a strong basis for winning an election. The next election will be won by the party that manages to reap more than its fair share of the non-Greens minor party voters. They are up for grabs for Liberal or Labor.