Making sense of the polls

Making sense of the polls

Labor takes lead in Newspoll as state income tax plan bombs

This week’s Newspoll is the first since Turnbull became PM to have Labor in front, by a 51-49 margin, a 2% gain for Labor since last fortnight. The catalyst for this change was probably the state income tax plan that was proposed Wednesday last week. Newspoll found that 58% opposed this plan with just 19% in favour; even Coalition voters were opposed by 48-27.

While Newspoll had the Coalition falling behind Labor, Morgan had them rebounding back to a clear lead after being just behind last fortnight. Here is this week’s poll table.

polls early Apr.

Some of the difference between Morgan and Newspoll may be because Morgan used a two-weekend sample, and the polling taken on the 26-27 March came before the state income tax proposal. Morgan’s last result also looks likely to have been a pro-Labor outlier. While the Coalition primary vote in the three tabulated polls is similar at about 42%, Labor’s primary vote in Morgan is only 31%, compared with about 36% in the other two polls.

After a long period when Morgan’s respondent allocated preferences had given the same Two Party Preferred (2PP) as the previous election method, this week’s Morgan had the Coalition on 52.5% 2PP respondent allocated, 1% better for them than the previous election method. This highlights a potential source of poll error at this year’s election: most polls use the previous election method to derive a 2PP estimate from primary votes. If the Coalition does better on actual preferences, these polls’ 2PP estimates will be a little too favourable to Labor, even if their primary votes are correct.

Newspoll had Turnbull’s satisfied rating down one point to 38%, and his dissatisfied rating up four points to 48%, for a net approval of -10. Since the first Newspoll this year in late January, Turnbull has lost 32 points of net approval. Shorten’s net approval was -21, up three net points, and he has gained 14 points on net approval since January. The gap between the leaders’ net ratings has closed to 11 points, from 57 points in January.

In my opinion, launching a state income tax plan three months before an election was a major blunder by Turnbull, and I agree with Peter Brent that Turnbull is lucky that the Premiers rejected it. Taking this policy to the election would have given Labor much ammunition for a scare campaign.

Since the state income tax plan has now been abandoned, the damage to the Coalition’s prospects at the next election should be limited. However, this assumes that Turnbull and the Coalition make no further blunders.

Kevin Bonham’s poll aggregate is now at 50.7% 2PP to the Coalition, a gain for Labor of 0.7% since last week. The Poll Bludger’s BludgerTrack is at 50.8% 2PP to the Coalition, a gain for Labor of 0.8% since last week. Primary votes in BludgerTrack are 42.5% for the Coalition, 34.1% for Labor and 11.2% for the Greens. Since last week, the Coalition has lost 1.0% on primary votes. Graphs show the huge dive in Turnbull’s net approval rating.

Notes on these polls

In Newspoll, 35% preferred Turnbull as Liberal leader, with 22% for Julie Bishop, 14% for Abbott and 8% for Scott Morrison; among Coalition supporters, it was Turnbull 48%, Abbott 20%, Bishop 16%. By 57-31, voters thought the Liberals were right to vote out Abbott, compared with a 62-27 margin in October.

In this week’s Essential, there was a surprising 34-34 tie on the state income tax plan, but note the “it would mean Federal income tax rates would be reduced” wording. By 64-24, voters disapproved of tax exempt status for religious organisations. 32% thought the economy was good and 27% poor; in January poor led by 31-28. Scott Morrison was trusted most to handle the economy by 26%, with 23% for Chris Bowen and 51% undecided; this is a 4% rise for Bowen since January.

In last week’s Essential, 44% would approve of calling a double dissolution if the ABCC bill is blocked by the Senate, and 23% would disapprove. Only 14% thought the main reason for a double dissolution was the ABCC bill, with 30% opting for “will have a better chance if the election is early” and 25% for “getting rid of the Independents in the Senate”. Income tax cuts were preferred to company tax cuts by a 62-16 margin.

The Coalition was thought best to handle the economy overall by a 35-22 margin, but Labor did better on other economic issues, such as being best for working people, and people like you. Labor’s positive attributes have declined a little since November, while the Coalition’s have worsened more than Labor’s. A comparison between the two parties shows there are big differences in perceptions of being too close to the big corporate interests (Coalition by 27), leadership (Coalition by 11) and working people’s interests (Labor by 19).

ReachTEL Indi poll has big lead for Cathy McGowan

At the 2013 election, Independent Cathy McGowan defeated Liberal frontbencher Sophie Mirabella in Indi by a narrow 50.2-49.8 margin. The Liberals have pre-selected Mirabella as their candidate again. A ReachTEL poll, taken 10 March from a sample of 656 for The Australia Institute, has McGowan leading Mirabella by 37-27 on primary votes, with 11% for the Nationals, 9% for Labor and 5% for the Greens. McGowan won in 2013 despite over a 13 point primary vote deficit, and would certainly win with these primary votes. Kevin Bonham estimates a 58-42 lead for McGowan after preferences from this poll.

Morgan state polls have Labor gaining across the board

Morgan has conducted SMS polls in all states from 25-28 March; comparisons are with late January polls. In NSW, the Coalition leads by 55-45, a 4.5% gain for Labor. In Victoria, Labor leads by 55-45, a 1.5% gain for Labor. In Queensland, Labor leads by 52-48, a 4% gain for Labor. In WA, Labor leads by 52-48, a 6.5% gain for Labor. In SA, there is a 50-50 tie, a 3% gain for Labor; the Nick Xenophon Team is on 24% in SA, a 3.5% gain for them. Sample sizes ranged from 600 in WA to 1100 in NSW.

Morgan’s SMS state polls have been volatile, and not very accurate at the last three state elections. The WA result represents a rebound from a rogueish looking 54.5-45.5 to the Coalition in late January. The trend to Labor in these polls is probably a result of the Federal Coalition’s decreased popularity.

Essential has state voting intentions using samples compiled from January to March. In NSW, the Coalition leads by 56-44. In Victoria, Labor leads by 53-47. In Queensland, Labor leads by 52-48. In WA, the Coalition leads by 52-48. In SA, Labor leads by 54-46.

Cruz easily wins Wisconsin

Ted Cruz crushed Donald Trump in yesterday’s Wisconsin primary, winning by 48-35, with just 14% for John Kasich. As Nate Silver says, Trump’s 35% in Wisconsin was comparable with what he won in neighbouring states, but in Wisconsin Cruz consolidated the anti-Trump vote. Trump has had little vote gains as other candidates have dropped out, and this is now a big problem for him.

Most polls in Wisconsin had Cruz leading, but with Kasich on about 20%. It is clear that Kasich’s vote collapsed as his supporters deserted him for the most viable anti-Trump candidate. This collapse could have implications for Kasich’s competitiveness in future contests.

The delegate split from Wisconsin will be 36 Cruz to 6 Trump, with Trump winning two Congressional Districts to avoid being wiped out. Trump still leads with 759 delegates, to 512 for Cruz and 144 for Kasich. Trump has won 45.8% of allocated delegates so far, to 30.9% for Cruz, and needs 63% of remaining pledged delegates to reach a majority (1,237 delegates) on only pledged delegates.

The next two weekends look grim for Trump with conventions in Colorado and Wyoming to decide 42 delegates; Cruz’s superior organisation has helped him perform very strongly at caucuses and conventions. However, New York on the 19 April is likely to be strong for Trump as he has over 50% in all recent New York polls; this would be likely to give him a virtual sweep of New York’s 95 delegates.

Sanders greatly reduces Clinton’s delegate lead, but he is still way behind

On Saturday 26 March, Bernie Sanders had his best day of the Democratic contest so far, winning Washington State 73-27, Hawaii 70-30 and Alaska 81-19. Washington has a medium population, and its 101 delegates went to Sanders 74-27, while he also gained about ten net delegates in each of Hawaii and Alaska. Yesterday, Sanders won Wisconsin 57-43. As a result, Hillary Clinton’s lead of almost 300 pledged delegates has been reduced to about 210.

However, the three 26 March states were caucuses, and there are only two caucuses remaining on the Democratic calendar, in low population Wyoming and North Dakota. Turnout at caucuses is much lower than at state managed primaries. If one candidate (Sanders here) has passionate supporters, those supporters are more likely to make up a larger proportion of those who turn out at caucuses than at primaries.

Early in the contest, the US elections website FiveThirtyEight had calculated delegate targets based on state demographics that would allow Sanders to win a majority of pledged delegates. This article shows that seven of Sanders’ eight best performances, relative to his targets, have come in caucus states. He has badly missed his targets in primaries, and virtually all remaining delegates will be decided by primaries.

Sanders will probably win the Wyoming caucus easily on Saturday, but there are only 14 delegates there. New York, where Clinton was a Senator for eight years, votes on the 19 April, and has 247 pledged delegates.