Editor of The Poll Bludger

Newspoll breakdowns broken down

Newspoll offers a deep dive into its recent polling data, offering unprecedented detail on voting intention by income, education, language and religion, along with more familiar breakdowns by state.

The Australian has published a set of geographic and demographic breakdowns compiled from multiple recent Newspoll results, once a regular quarterly feature of the pollster notable for its results at state level, but now greatly expanded as more elaborate methods are adopted in response to last year’s pollster failure. Where in the such breakdowns were limited to geography, gender and age, there are now also education (no tertiary, technical and university), household income, English or non-English speaker at home, religion (only Christian and no religion are provided, but they presumably have a small sample result for other religions).

Compared with a national result of 50-50, the state breakdowns show level pegging in New South Wales (1.8% swing to Labor), 55-45 to Labor in Victoria (1.9% to Labor), 56-44 to the Coalition in Queensland (2.4% to Labor), 55-45 ditto in Western Australia (0.6% swing to Labor, and 53-47 to Labor in South Australia (2.3%). These suggest statistically indistinguishable swings to Labor of 1.8% in New South Wales, 1.9% in Victoria, 2.4% in Queensland, 0.6% in Western Australia and 2.3% in South Australia. The primary votes are notably strong for the Greens in Queensland, up nearly three points from the election to 13%, and One Nation in Western Australia, who are on 9% after never having done better than 7% in the last term.

The age breakdowns are notable for the 62-38 lead to Labor among the 18-34 cohort, a differential quite a lot greater than that recorded by Newspoll in the previous term, which ranged from 4% to 8% compared with the present 12%. The gender gap — 52-48 to the Coalition among men and the reverse among women — is at levels not seen since the Tony Abbott prime ministership, whether due to genuine churn in voting intention or (more likely I think) a change in the pollster’s house effect.

Analysis of the education breakdowns is made easy by the fact that two-party is 50-50 for all three cohorts, with even the primary vote breakdowns recording little variation, other than university graduates being somewhat more disposed to the Greens and allergic to One Nation. As the table below illustrates, there are notable differences between these numbers and comparable findings for the Australian National University’s post-election Australian Election Study survey, which recorded a strong leftward lean among the university-educated compared with those without qualifications and, especially, those with non-tertiary qualifications.

For income, Newspoll reflects the Australian Election Study in finding the low-to-middle income cohort being Labor’s strongest, with a relative weakness among the low-income cohort presumably reflecting their lack of support in rural and regional areas. However, the distinctions are less marked in Newspoll, which credits the Coalition with 46% of the primary vote among the top household income cohort (in this case kicking in at $150,000) compared with 51% in the Australian Election Study, with Labor respectively at 34% and 32%. Differences were predictably pronounced according to language (51-49 to the Coalition among those speaking English only, 57-43 to Labor among those speaking a different language at home) and religion (58-42 to the Coalition among Christians, the reverse among the non-religious).

The results are combined from the last four Newspoll surveys, collectively conducted between March 11 and May 16, from a sample of 6032, with state sample sizes ranging from 472 (suggesting a 4.5% margin of error on the South Australian result) to 1905 (suggesting 2.2% in New South Wales.

Essential Research: coronavirus and attitudes to China

A major souring in Australians’ attitudes to China but little change on coronavirus (at least since last week), according to the latest Essential poll.

Another week, another Essential Research coronavirus poll — this one focusing on attitudes to China, which have notably soured. As related by The Guardian, respondents were asked if they had a favourable or unfavourable view of China’s influences on Australian life, which produced a net rating of minus 30% on trade, compared with plus 1% last August, and a net rating of minus 40% for Chinese business operating in Australia, down from minus 21%. There were also scores of minus 26% for defence, minus 36% for politics and minus 9% for culture. Conversely, the United States scored net positive scores, albeit that these were quite a lot bigger for defence (plus 29%), business (plus 15%) and trade (plus 14%) than politics (plus 2%) and culture (plus 7%).

Asked which relationship would be more beneficial to strengthen, 42% favoured the US and 18% China, compared with 38% and 28% last August. Respondents had two bob each way on trade in that 53% thought Australia “needs to do all it can to avoid a trade war with China”, with 17% opposed, but 48% felt Australia should impose retaliatory tariffs, with 22% opposed. The poll found “more than half” believe China’s trade sanctions against Australia were motivated by the government’s call for an investigation into the origins of COVID-19.

The poll continued its weekly suite of questions on coronavirus, recording no change on the government’s handling of the crisis, which was rated positively by 73% and negatively by 11%. Levels of concern little changed on last week (79% either very or quite concerned, down one, and 21% either not at all or not that concerned, up one). A divide appears to be opening on restrictions, with higher responses for both lifting them as soon as possible (up five to 14%) and holding off (up two to 27%). The poll was conducted Thursday to Sunday from a sample of 1087; a full report should be published later today.

Note that below this post is a dedicated thread for the Eden-Monaro by-election, which you are encouraged to use if you have something specific to say on that subject.

UPDATE: Full report here.

Eden-Monaro by-election: July 4

A date set for the Eden-Monaro by-election, as the Liberals confirm their candidate and a GetUp! poll suggests local concern about climate changes remains at a high in the wake of the summer bushfires.

Eden-Monaro will be celebrating American independence this year with a July 4 by-election, as announced yesterday by the Speaker of the House of Representatives, Tony Smith. The timeline for the by-election is as follows:

Issue of writ: Thursday, May 28
Close of rolls: Thursday, June 4
Close of nominations: Tuesday, June 9
Declaration of nominations: Wednesday, June 10
Date of polling: Saturday, July 4

Clearly there will be nothing radically different about the way the polling is conducted on account of coronavirus, as Smith noted that the Australian Electoral Commission unusually favoured a date at the start of school holidays, as it would allow ample time for sanitising to be conducted at schools used as polling booths.

The Liberals conducted their preselection on Saturday, and as expected chose Fiona Kotvojs, a beef farmer who narrowly failed to win the seat when she ran at last year’s federal election. Kotvojs reportedly won a 106-27 victory in the party ballot over Mark Schweikert, a Department of Defence official and deputy mayor of Queanbeyan-Palerang. According to Niki Savva in The Australian ($), “there is speculation, not denied, that the Liberals, with their formidable state and federal apparatuses, are prepared to spend up to $1m to win the seat, money Labor says it doesn’t have”.

The Guardian reports GetUp! commissioned uComms to conduct a poll of 879 respondents in the seat last Wednesday, but it seems they are being reticent with voting intention figures. However, we are told the poll found 59% holding that the government was not doing enough to address climate change, including 48% who strongly agreed, with only 27% disagreeing, 12% strongly. However, the government apparently scored better on its bushfire response, on which respondents were reportedly “evenly divided”.

On failure

A look at efforts to get to the bottom of last year’s federal election pollster failure and the Electoral Commission of Queensland’s recent election night meltdown.

The Association of Market and Social Research Organisations has published a discussion paper for its review into last year’s federal election polling failure. It notes that existing standards set by the Australian Press Council are too lax and readily ignored in any case, and suggests a familiar retinue of suggested new standards including full disclosure of weightings used and detail of how preference flows were determined. While the inquiry’s committee and advisory group are impressively credentialled, it should be noted that most actual pollsters aren’t members of the AMSRO. The recent announcement that YouGov, Essential Research and uComms would establish an Australian Polling Council occurred independently of its process, and is likely to be the more consequential development.

Meanwhile, a parliamentary inquiry has been putting the blowtorch to the Electoral Commission of Queensland over the failure of its results reporting facilities at the local government elections and state by-elections on March 28. Excuses include disruption arising from COVID-19, which extended to “coding resources” being locked down in Wuhan, and the complication of combining elections for two state parliament seats with the statewide council elections. It also appears an American firm contracted to provide a new election management system, Konnech, has found itself bamboozled by what the electoral commissioner described as “the complexity of Queensland electoral law”, which “far exceeded that of any other Konnech customer” (a conclusion it would no doubt have reached in any Australian jurisdiction).

The new results website went belly-up on testing a week out from election day, prompting the ECQ to hurriedly concoct the unfamiliar-looking results website that appeared on the night. Polling booth officials were required to submit results through a shareable spreadsheet application, which threw up formating inconsistencies upon transfer to the ECQ system. The ECQ’s technical staff spent the night dealing with the results website issues, leaving corresponding issues with a horrifyingly complex XML results feed to one side. Consequently, the ABC’s results displays remained stuck on a tiny share of the count all night, and updates remained infrequent beyond election night. It is to be hoped that this will all be sorted out before a state election that will be held on October 31.

Essential Research: coronavirus restrictions and conspiracy theories

A poll suggests a significant proportion of the population believes coronavirus was engineered in a Chinese laboratory, but other conspiracy theories remain consigned to the fringe.

Courtesy of The Guardian, some headline results of another weekly Essential Research poll on coronavirus, the full report of which should be published later today. This includes regular questions on federal and state governments’ handling with the crisis, of which we are only told that respondents remain highly positive, and on easing restrictions, for which we are told only 25% now consider it too soon, which is down two on last time and has been consistently declining over five surveys.

Beyond that, the survey gauged response to a number of what might be described as conspiracy theories concerning the virus. By far the most popular was the notion that the virus “was engineered and released from a Chinese laboratory in Wuhan”, which has received a certain amount of encouragement from the Daily Telegraph but is starkly at odds with the scientific consensus. Agreement and disagreement with this proposition was tied on 39%.

Thirteen per cent subscribed to a theory that Bill Gates was involved in the creation and spread of the virus, with 71% disagreeing; 13% agreed the virus was not dangerous and was being used to force people to get vaccines, with 79% disagreeing; 12% thought the 5G network was being used to spread the virus, with 75% disagreeing; and 20% agreed the number of deaths was being exaggerated, with “more than 70%” disagreeing. The poll also found 77% agreed that the outbreak in China was worse than the official statistics showed.

The poll was conducted Thursday to Sunday from a sample of 1073.

UPDATE: Full report here.

Newspoll: 51-49 to Coalition

The Coalition edges back into the lead in Newspoll, with Labor, the Greens and One Nation all down on the primary vote.

The Australian reports the Coalition has opened a 51-49 lead in the latest Newspoll, after the previous poll three weeks ago recorded a dead heat. The Coalition is up two on the primary vote to 43%, with Labor down one to 35%, the Greens down two to 10% and One Nation down one to 3%. Scott Morrison’s approval rating is down two to 66%, with the disapproval not yet provided; Albanese is down one on approval to 44% and up three on disapproval to 37%. Morrison’s lead as preferred prime minister is all but unchanged at 56-29, compared with 56-28 last time. The poll was conducted Wednesday to Saturday from a sample of 1504. More detail to follow later.

UPDATE: Morrison’s disapproval rating turns out to be up two to 30%. These numbers have been incorporated into the BludgerTrack leadership trends which you can see on the sidebar and in greater detail here. Newspoll has put to respondents the same suite of questions concerning coronavirus in its last three polls, which record soaring confidence in “federal and state governments’ performance” in managing the economic impact (60% satisfied, up 13 points on last time, and 24% dissatisfied, down nine), preparing the health system (up 19 to 78% and down 13 to 15%) and informing Australians about how to protect themselves (up seven to 82% and down seven to 13%).