By JR on Tuesday, June 23, 2015
"New Matilda's" idea of survey research
I am probably one of the very few who take any notice of Matilda but they are so determined to come to their foreordained conclusions that I find them amusing.
I find the latest effusion particularly amusing because I spent 20 years as a university survey researcher -- resulting in many academic publications. So I am quite sure how the sleight of hand below works.
What they do is to compare a properly conducted survey with responses gathered over the internet. And they found a great divergence of results. But such comparisons always do diverge. People contacted over the internet are not representative. They differ from the mean in being better educated, more socially isolated, more Leftist and in various other ways. As far as generalizing to any known population is concerned they are invalid and useless. It takes Matilda to hang their hat on such a survey
A survey distributed on social media has recorded dramatically different results to those released by the government-funded group, finding stronger support for constitution reform among white Australia than black. Amy McQuire reports.
Only 25 per cent of Aboriginal and Torres Strait Islander people support the “grassroots” Recognise group, and a majority would vote no in a referendum if it delivered only symbolic recognition, according to a new poll that pours cold water on the Recognise’s claim that nine out of ten blackfellas support their campaign.
In May, the government-funded Recognise group released research suggesting nine in ten Aboriginal and Torres Strait Islander people supported constitutional recognition.
The polling was conducted by Polity Research, which surveyed 750 Aboriginal and Torres Strait Islander people, and 2700 non-Indigenous voters, according to Recognise.
Recognise said the “research confirms continuing support for recognition from the vast majority of Aboriginal and Torres Strait Islander people”.
The group released its media statement without detail of the question that was posed to interviewees, or where it had retrieved its sample of Aboriginal and Torres Strait Islander voters.
But the claim of overwhelming Indigenous support is one that was controversial among many who believe the issue is being met with a much more diverse sample of opinions in Aboriginal communities.
While conservative opposition to constitutional reform has been highly publicised, the concerns from Aboriginal and Torres Strait Islander people has been largely overshadowed by the Recognise campaign’s claims of almost unanimous support.
In February, the National Congress of Australia’s First Peoples acknowledged that there was a growing opposition to constitutional reform from Aboriginal and Torres Strait Islander communities, with co-chair Les Malezer saying “we are already receiving messages from our people that they are determined to vote against any referendum”.
This opposition is confirmed by an online survey conducted by Luke Pearson at IndigenousX, a social media platform across Twitter and Facebook, who released the results last night.
The survey drew responses from 827 Aboriginal and Torres Strait Islander people across the country, with the majority of responses coming from New South Wales, Queensland and Victoria.
The poll, conducted on Survey Monkey, was widely shared across Twitter and Facebook, the latter of which has high rates of Aboriginal and Torres Strait Islander users.
The survey acknowledges that, like the Recognise polling, the sample size was “still quite small”.
“Questions regarding the distribution of data collected by Recognise should be similarly asked and caution must always be exercised when claiming that a survey is ‘representative’,” the group says.
But the IndigenousX poll produces radically different results to those promoted by the Recognise campaign.
It found that rather than Recognise’s stated 87 per cent Indigenous support, only 25 per cent of IndigenousX respondents supported the group. 58 per cent were in opposition, while a further 17 per cent were still unsure.