Making sense of the polls

Making sense of the polls

Labor gains slightly as both leaders’ ratings slump

This week we had three polls from Ipsos, Newspoll and Essential. Newspoll is not the old Newspoll, which was conducted by landline telephone interviews. The new Newspoll is conducted by a mixture of robopolling and online panel methods, and the first new Newspoll has a sample of 1630, where the old Newspolls usually had samples of about 1150. The last Ipsos and Newspoll were both taken three weeks ago. Here are this week’s three polls.

polls mid July.

Newspoll probably came close to a 53-47 Labor lead, since Labor’s Two Party Preferred (2PP) share, when calculated from the published primaries, was 52.6%. The Coalition probably benefited from rounding in this Newspoll.

Ipsos replaced Nielsen as Fairfax’s pollster in November 2014, and is now the only live interview pollster remaining (it calls mobiles as well as landlines). Until May this year, Ipsos had shown a clear lean to the Coalition, but its last two polls have been more in line with the overall trend, and it now appears that Ipsos’ lean has disappeared.

Despite, or possibly because of, the recent focus on national security and the ABC, Abbott’s ratings in both Ipsos and Newspoll have slumped. In Ipsos, Abbott’s approval rating was down four points to 36%, and his disapproval up five points to 59%, for a net approval of -23, down from -14; Abbott’s net approval peaked at -8 after the budget. In Newspoll, Abbott’s satisfied rating is down one point to 33%, and his dissatisfied rating up four points to 60%, for a net approval of -27, down from -22. This is Abbott’s worst performance in a Newspoll since late March, and well below his post-budget peak net approval of -13.

Up to June, Ipsos had given Shorten mild approval ratings, but no more. Shorten’s ratings crashed from -6 to -20 net approval in Ipsos. In Newspoll, Shorten’s ratings became only slightly worse, going from -26 to -28, but both polls now have Shorten’s ratings at record lows. Shorten’s recent dive has probably been caused by the Trade Union Royal Commission, but the previous Newspoll was probably too harsh on both Shorten and Labor. Accordingly, his ratings in this poll did not change much.

The Greens’ support varied across this week’s polls; they received 16% in Ipsos (up 2), 13% in Newspoll (down 1) and only 10% in Essential (down 1).

Kevin Bonham’s poll aggregate is now at 52.5% 2PP to Labor, a 0.2% gain for Labor since last week and 0.7% better for Labor than in the post-budget polls.

The Poll Bludger’s BludgerTrack is at 52.2% 2PP to Labor, a 0.2% gain for Labor. Primary votes are 40.0% for the Coalition, 35.7% for Labor and 13.8% for the Greens, with both major parties shedding about half a point, and the Greens picking up half a point. Labor is at its lowest primary vote since November 2013, and the Coalition at its lowest since March, while the Greens continue to register record primary votes. Graphs show the big slump in both leaders’ ratings. BludgerTrack’s voting intentions did not include this week’s Newspoll; if it had, Labor would be performing slightly better.

Notes on these polls

In Ipsos, the respondent-allocated 2PP was also 53-47 to Labor, the same as the previous election method, and 85% supported constitutional recognition of indigenous Australians.

Questions were also asked on leader attributes, with comparisons made with November 2014. Shorten was down 15 points to 56% on “has confidence of his party”, nine points to 34% on “strong leader”, seven points to 36% on “vision for Australia’s future” and six points to 52% on “competent”. However, Shorten still leads Abbott by 68-34 on being “open to ideas”, by 59-30 on “social policy”, and by 52-45 on “competent”. On the economy, Abbott has only a 47-43 lead, which is weak for an incumbent Coalition PM.

Essential recorded very positive attitudes to women’s sport, but these responses may be affected by pressure to answer in a pro-woman way. Now, 69% think it is likely that same-sex marriage will be legalised in the next few years, and only 20% think it is unlikely; in February, likely led by 56-28.

Asked about the ABC, 22% think the national broadcaster is biased to the Left, 3% think it is biased to the Right, and 36% think it unbiased. The 40% who were undecided probably reflect voters who simply don’t care about ABC bias. Among Coalition voters, 42% think the ABC is biased to the Left, 3% to the Right, and 21% think it unbiased.

Essential asked whether the government was doing a good or poor job on various issues. The government received a net positive rating on only one issue, supporting businesses, but improved on most issues since January. The biggest improvements were in supporting businesses (from -5 to +15), health services (-31 to -19), education (-24 to -13) and the economy (-14 to -5). The government went backwards on relations with other countries (+5 to -2).

WA and SA Newspolls

We have had two final state polls from the old Newspoll series, conducted from April to June. In WA, Labor holds a 52-48 lead, unchanged from January-March. Primary votes are 33% for Labor (down 2), 40% for the Coalition (steady) and 14% for the Greens (steady). Premier Colin Barnett has a net approval of -21, down from -15. Opposition leader Mark McGowan has a net approval of +16, down from +25. The sample size was 870.

There is very little change in SA; Labor holds a 54-46 lead, from primary votes of 36% for Labor, 33% for the Liberals, 10% for the Greens and a hard-to-believe 21% for Others, with all voting intentions steady on January-March. Premier Jay Weatherill’s net approval is steady at +2, and opposition leader Steven Marshall’s net approval is at +2, down from +4. The sample size was 870.

Morgan’s most recent polls in June had the Coalition leading in both states, by 52.5-47.5 in WA, and by 51-49 in SA.

Greek referendum result a huge miss for pollsters

While Australian pollsters have continued to be accurate, it has not been fair sailing for pollsters internationally in recent times. At the US midterm elections, the Israeli election and the UK election, pollsters underestimated the vote of the more conservative party or parties. At the Greek referendum, pollsters missed in the opposite direction, as the left-wing option to vote NO greatly overperformed the polls.

This Wikipedia table has all the Greek referendum polls. The first polls, conducted the weekend before the vote, were accurate, but polls conducted in the final few days were terribly wrong, with some recording YES leads. Even polls taken on election day only had NO leading by 3-4 points, but NO actually won by a crushing 22.6%.