Nick Xenophon could be South Australia's next premier, while Turnbull loses his 25th successive Newspoll

The next South Australian election will be held in three months, on March 17, 2018. A South Australian Newspoll, conducted October to December from a sample of 800, had primary votes of 32% for Nick Xenophon’s SA-BEST party, 29% Liberal, 27% Labor, and 6% Greens.

On the better premier measure, Nick Xenophon had 46% of the vote, followed by incumbent Jay Weatherill on 22% and Opposition Leader Steven Marshall on 19%.

Xenophon’s strong performance is partly explained by the dire ratings of both Weatherill and Marshall. Weatherill had 53% dissatisfied, 34% satisfied, for a net approval of minus 19. Marshall had 50% dissatisfied, 27% satisfied, for a net approval of minus 23.

The previous South Australian Newspoll was taken in late 2015. A Galaxy poll taken for the Australian Bankers’ Association in early October gave the Liberals 31% of the primary vote, SA-BEST 30%, and Labor 26%. The better premier measure in that poll had 41% Xenophon, with Weatherill and Marshall both on 21%.

If the primary votes in Newspoll were replicated at the March 2018 election, SA-BEST would probably win a clear majority of lower house seats. Both major parties’ supporters dislike the other major party, so most Labor voters will preference SA-BEST ahead of the Liberals, and vice versa.

There are still three months until the election, and SA-BEST will be attacked ferociously in the coming weeks. However, the disdain for both major parties, and Xenophon’s popularity, gives SA-BEST a real opportunity to end the major party duopoly in South Australia.

The total vote for all “others” in Newspoll is 6%. At the 2014 state election, Family First won 6%. This poll does not suggest the Australian Conservatives, formed by Cory Bernardi, are surging.

Turnbull loses his 25th successive Newspoll, 53-47

This week’s Newspoll, conducted December 14-17 from a sample of 1,670, gave federal Labor a 53-47 lead, unchanged from last fortnight. Primary votes were 37% Labor (steady), 36% Coalition (steady), 10% Greens (steady), and 7% One Nation (down one).

This is Malcolm Turnbull’s 25th successive Newspoll loss. Tony Abbott lost 30 in a row before he was dumped.

Turnbull’s ratings were unchanged at 57% dissatisfied, 32% satisfied, for a net approval of minus 25. Bill Shorten’s net approval fell three points to minus 24. Turnbull led by 41-34 as better prime minister (39-33 last fortnight).

The 7% for One Nation is its lowest support in Newspoll since December 2016, before One Nation was included in the party readout. On the overall vote for left- and right-wing parties, the left leads by 47-43 in this Newspoll (47-44 last fortnight). This is the first change in the overall left/right balance since October.

The passage of same-sex marriage legislation through parliament and the media furore over Sam Dastyari do not appear to have improved the Coalition’s position. Most voters realise that the large “yes” vote in the plebiscite forced the Coalition to act. It is likely that only partisans are interested in Dastyari.

Newspoll (paywalled) asked who was better at handling the economy, national security, asylum seekers, cost of living, and tax cuts. Turnbull had more than 20-point leads over Shorten on the first three issues, and a 40-33 lead on tax cuts. Shorten led by 43-41 on cost of living.

These questions are biased in favour of Turnbull, as they appeal to the Coalition’s perceived strength on the economy, national security and asylum seekers. There were no questions regarding issues like health, education and climate change, where Labor is perceived to be better than the Coalition.

Incumbent prime ministers tend to outperform their party, so Labor would probably have obtained more favourable results had Newspoll asked Coalition vs Labor, not Turnbull vs Shorten.

Essential 53-47 to Labor

In this week’s Essential, the Coalition gained two points since last fortnight, reducing Labor’s lead to 53-47. Primary votes were 38% Labor, 37% Coalition, 9% Greens and 7% One Nation. Essential uses a two-week sample of about 1,800, with additional questions based on one week’s sample.

All proposed reforms of political donations were very popular, with the exception of banning donations and making all political party spending taxpayer-funded (50-30 opposed).

Respondents were asked whether the last 12 months had been good or bad for various items. The economy had a net +11 rating, large companies a net +22, your workplace a net +34, and you and your family a net +27. The planet had a net -22 rating and Australian politics a net -36.

Respondents were also asked about their expectations for 2018, though Essential apparently thought the next year is 2017.

By 54-29, voters disapproved of the proposed A$50 billion in company tax cuts to medium and large businesses (50-30 in October). By 47-8, voters thought personal income tax cuts were more important than business tax cuts, with 33% for both being equally important.

37% thought interference in Australian politics by foreign countries is a major problem, 36% a minor problem, and 12% did not think it was a problem at all.

Bennelong preference flows

With virtually all votes counted in Saturday’s Bennelong byelection, Liberal John Alexander defeated Labor’s Kristina Keneally by 54.9-45.1, a 4.9-point swing to Labor since the 2016 election. Primary votes were 45.1% Alexander, 35.8% Keneally, 6.7% Greens, 4.3% Australian Conservatives, and 3.1% Christian Democrats.

The informal rate of 8.1% is far too high, and indicates savings provisions should be introduced so that votes can still be counted even if voters do not number every square. There is confusion in New South Wales because state elections use optional preferential voting.

I have information about preference flows from one booth in Bennelong from a Labor scrutineer. At this booth, Keneally won 88% of Greens preferences, but Alexander won 85% of Australian Conservatives’ preferences and 77% of Christian Democrat preferences.

Preferences of other candidates split evenly between Keneally and Alexander. This booth may not be representative of the whole electorate, but these preference flows seem reasonable.

While Newspoll and Galaxy understated Alexander by four-to-five points, these polls were of one seat. Australian pollsters have been bad with seat polling, but very good with national and state polls. The error in Bennelong does not affect the trustworthiness of Newspoll’s national polls.