Slight movement to Labor as Shorten's ratings improve

The polls taken after the budget had some movement to the Coalition. This week’s polls have that movement reversing. While the overall vote is showing about a 50-50 tie, this is not enough for Labor to win; they probably need at least 51% Two Party Preferred (2PP) to win. The main story this week is the big improvement in Bill Shorten’s ratings.

During the campaign, Ipsos will be published on Saturday rather than Monday, and its fieldwork will be Tuesday to Thursday rather than Thursday to Saturday. The last issues of Newspoll, Ipsos and ReachTEL were all done about a fortnight ago. Here is this week’s poll table.

Once again, ReachTEL used respondent allocated preferences for its 2PP estimate. This showed a 50-50 tie, unchanged on last fortnight, but it disguised the clear primary vote move to Labor. I am continuing to use previous election methods for the 2PP estimates in the table.

Ipsos also showed a primary vote gain for Labor, and this poll would normally be 50-50 on the primary figures, but rounding was good for the Coalition. The respondent allocated 2PP was a 50-50 tie.

Last fortnight’s Newspoll had no movement to the Coalition when other polls did move, so it is not surprising that this Newspoll had no movement to Labor. It is the fourth successive Newspoll with Labor ahead 51-49.

There was also a ReachTEL seat poll for Macarthur, held by the Liberals by 3.3% after redistributions. Labor was ahead 51-49, matching the result from a Galaxy robopoll last week.

Kevin Bonham’s poll aggregate is now at a 50.0-50.0 tie, and the Poll Bludger’s BludgerTrack is at 50.1% 2PP to Labor. These aggregates have changed little in the last five weeks. Primary votes in BludgerTrack are 41.8% for the Coalition, 35.4% for Labor, 10.9% for the Greens and 3.3% for the Nick Xenophon Team.

Leaders' ratings

In Newspoll, 38% were satisfied with Turnbull’s performance (steady), and 50% were dissatisfied (up 1), for a net approval of -12. Shorten’s net approval was up seven points, to also be at -12. This is Shorten’s best net approval since last May, and he has gained 26 points since his nadir in December.

In November, Turnbull led Shorten by 69 points in net approval ratings; now they are tied. Kevin Bonham says this is the second biggest movement to an opposition leader in Newspoll history. Shorten’s recovery of 26 net approval points is the third best.

In Ipsos, Turnbull’s approval was 48% (steady), and his disapproval was 38% (down 2), for a net approval of +10. Ipsos' ratings for Turnbull have been much better than other polls. Shorten’s net rating increased five points to -6, his best rating from Ipsos in almost a year. By 72-25, voters preferred more spending on schools to tax cuts for business.

In ReachTEL, Turnbull’s (total good) minus (total poor) rating was -7, unchanged on last fortnight. However, Turnbull’s “very good” rating is now only 8%, down 2 points on last fortnight, and below Shorten’s “very good” rating of 9% for the first time. Shorten’s (total good) minus (total poor) rating was -11, up 9 points, and his best ReachTEL rating since January 2014, when Shorten was experiencing a honeymoon.

A well-known saying about elections is that governments lose them, oppositions do not win them. However, oppositions are capable of losing winnable elections by selecting unacceptable leaders or having clearly flawed major policies.

I think the main effect of Shorten’s increased ratings is to make him an acceptable leader. If voters decide they no longer want the Coalition government, they will not be as hesitant about going to Labor.

After falling steeply between February and April, Turnbull’s ratings have stabilised in the last month. Turnbull’s current Newspoll net approval of -12 is still better than any of Abbott’s ratings since September 2014. The PM’s rating is much more important to voting intentions than the opposition leader’s rating. So while it is good for Shorten to have better ratings, a further slump in Turnbull’s ratings would be more detrimental to the government.

Shorten has gained in better PM ratings, with Ipsos giving Turnbull a 47-30 lead (51-29 last fortnight), ReachTEL 56-44 (previously 58-42) and Newspoll 46-31 (previously 49-27).

Is the “Think will win” question relevant?

Ipsos found that 57% (up 4) thought the Coalition most likely to win, and only 20% (down 4) thought Labor would win. However, Newspoll was very different, with the Coalition ahead by only 44-33 on this question, down from 55-25 in March.

In my opinion, voters are more likely to think the Coalition will win because most media pundits are saying that the Coalition will win. I do not agree with those who assert that undecided voters will vote for the party they think will win. The “think will win” statistic is more meaningful in voluntary voting countries, where depressed turnout can result if people think the election is a foregone conclusion.

In Australia, the undecided are far more likely to vote for the party they believe is better on the economy, health, education, etc, than be bandwagon jumpers. In fact, the 2015 Queensland election and the 1999 Victorian election are two contrary examples, with Labor winning both elections when very few thought they would. In both these elections, some people probably voted Labor as a protest, thinking that Labor had no chance of actually winning.

If people think one party will easily win, but the polls are close, there is a danger that the other major party will benefit from protest voters, and surprisingly win. That danger also applies to betting market expectations. Betting markets have not performed well at many state elections, and I disagree with those who talk about the wisdom of the betting markets.