The only broad surveys covering population issues in the period here considered were those conducted for the Scanlon Foundation in 2009, 2010, 2011, 2012 and 2013. These surveys were based on large samples (national surveys of 2,000 respondents between 2009-2012, 1,200 in 2013), with additional local surveys of 1,800 respondents in 2009, 2,000 in 2012, and 2,500 in 2013) which provided scope for disaggregation of the data to consider attitudes within specific demographics and communities. The Scanlon surveys utilise random sampling and telephone interviewing, with mobile phones added to the national sample in 2013. A rigorous weighting methodology is applied to the data: while the industry standard provides for weighting of two variables (sex and age), the Scanlon surveys are weighted to four variables (sex, age, country of birth, educational attainment).
Polls commissioned by news media – Newspoll for the Australian, Galaxy poll for various media outlets, including News Limited, and the Nielsen poll for The Age and Sydney Morning Herald – are based on telephone interviewing, typically a random sample of 1000-1200 respondents aged 18 and above, with Nielsen polls sampling a larger number, 1400.
The news media also solicit opinion through on-line single question polls, open to all with access to the internet. These polls are not based on a scientific polling methodology and have no value in providing an insight into public opinion; at best, they inform the media source on the views of those of their readers or viewers who are interested in responding to a question of topical interest. This method of polling is open to manipulation by interest groups (for example, supporters of a political party may contact a designated list of activists to request completion of an on-line poll.) To take one example, polls have consistently indicated that the Coalition is preferred over Labor on asylum policy. Yet an on-line Sydney Morning Herald poll in July 2010 indicated majority support (54%) for Labor policy. The paper published the result with its standard disclaimer: 'these polls are not scientific and reflect the opinion only of visitors who have chosen to participate'.
Also in the period considered, a poll conducted by Roy Morgan research (March 2010) asked eight questions related to population issues. It was based on a small sample of '670 men and women' aged 14 and above. A second Morgan poll (July 2010) was administered to 719 men and women electors aged 18 and over.
Other polls with questions related to population issues include the annual survey of opinion on foreign policy conducted for the Lowy Institute, based on a stratified random sample of 1000 conducted by fixed and mobile telephone. The 2010 Lowy Institute poll contained five questions related to immigration issues; in 2011 there was one question on immigration and a series of questions on asylum issues; in 2012 there were three questions on immigration; in 2013 there were five questions on asylum issues.
The Australian Survey of Social Attitudes (AuSSA) 2009, conducted by researchers at the Australian National University, employed a comprehensive questionnaire, including two or more different versions answered by sub-samples. Both versions contain the same substantive question and three follow up questions related to population issues. AuSSA is based on a mail-out, self-administered questionnaire and was completed by over 3000 respondents. Although limited in the range of its questions, AuSSA provides an important insight into Australian public opinion. One problem, however, is that due to financial constraints AuSSA offers to sell space to individuals and organisations for the inclusion of questions, which has the potential to impact on the integrity of the questionnaire.
The Essential Report is a weekly omnibus (general purpose) survey conducted by Essential Research with data provided by Your Source, an Australian social and market research company. Surveys include questions on voting intention and specific questions on issues of topical interest. It utilises a 'self-managed consumer on-line panel' of over 100,000 members aged 18 and above. Invitations to complete the weekly survey are sent to approximately 7000-8000 members, which typically yields 1000+ responses. The survey is live for six days, from Tuesday to Sunday. It provides the most frequent polling available on several questions related to population, immigration and asylum issues, but it suffers from the limitation that it is not based on a random sample and although the achieved respondent profile is weighted the extent of sampling error is a matter of dispute.
Additional indication of public opinion was provided by Vote Compass hosted by the ABC in August-September 2013, during the election campaign. Vote Compass is described as 'primarily and fundamentally an educational tool intended to promote electoral literacy and stimulate public engagement in the policy aspect of election campaigns.' It is further stated that 'vote Compass stimulates voter engagement by offering the electorate the opportunity to engage with and compare the policy platforms of political parties in a clear and accessible manner that addresses points of differentiation between the parties on relevant public policy issues.'
The Vote Compass comprised 30 questions and was completed by 1.4 million respondents. Four of the questions dealt with immigration and asylum.
- 'How many new immigrants should Australia admit?'
- 'There should be more restrictions on 457 visas issued to temporary foreign workers.'
- 'Asylum seekers who arrive by boat should not be allowed to settle in Australia.'
- 'Boats carrying asylum seekers should be turned back.'
On 11 November 2013 the final dataset was released, weighted against the census using eight variables – gender, age, education, enrolment as a student, religion, marital status, industry and state – and yielded a weighted sample size of 573,444.
The data was released with a somewhat confusing message: in one context, it was stated that Vote Compass is not a poll and 'is not a random sample.' But the data was featured on the ABC internet site under the headline 'What Australians think about the big political issues,' with the introductory statement that 'we ...are releasing the full data so you can find out who thinks what about the key political issues facing Australia' and 'view the Vote Compass data by location to see what voters in every electorate think about the big issues.' It was further stated that 'online surveys are inherently prone to selection bias but statisticians have long been able to correct for this (given the availability of certain variables) by drawing on population estimates such as Census micro-data. … We apply sophisticated weighting techniques to the data to control for the selection effects of the sample, thus enabling us to make statistical inferences about the Australian population with a high degree of confidence.'
The challenge is to correct for sample bias produced by the requirement to go on the ABC website and respond to a broad range of questions. A sample obtained by such means, however numerous, is not representative of the total population. There is the basis for correcting for broad range of demographic variables, but there is no simple and agreed method for correcting for attitudinal differences. The specified weighting procedure does not include attitudinal variables.
It is possible to consider reliability by comparing results obtained by Vote Compass and random sample surveying. The wording of three of the four questions on immigration and asylum issues is similar to questions in the Scanlon Foundation surveys. One problem in comparing the results, however, is that Vote Compass results were released without the 'don't know'/ 'refused' categories.
The results obtained by Vote Compass for the proposition 'Boats carrying asylum seekers should be turned back' were 'strongly agree' 30.2%, 'somewhat agree' 17.6%, total in agreement 47.8%. The Scanlon Foundation survey was differently worded, presenting four policy options; it found that the proportion agreeing that boats should be turned back was much smaller, 33%. The different results, however, may be explained by the different response options.
A more direct comparison is provided by the question on attitude to the immigration intake. The Scanlon Foundation survey asked 'What do you think of the number of immigrants accepted into Australia?' and presented three response options: 'too high', 'about right', and 'too low'. Vote Compass asked 'how many new immigrants should Australia admit' and presented five response options: 'many fewer', 'somewhat fewer', 'about the same as now', 'somewhat more', and 'many more'. A consistent finding of the Scanlon Foundation surveys is that while there is majority support for the current level of immigration, only a small minority favour an increase: thus in 2012, 14.4% favoured increase, in 2013, 12.8%; in contrast, a significantly higher 33.8% of Vote Compass respondents favoured an increase ('somewhat more', 'many more'); 30.2% of Vote Compass respondents considered that fewer immigrants should be admitted, compared to 41.9% of respondents to the 2013 Scanlon Foundation survey.
Opinion on the level of immigration
'How many new immigrants should Australia admit?'
Vote Compass, August-Sept 2013
'What do you think of the number of immigrants accepted into Australia?'
Scanlon Foundation survey, June-July 2013
Scanlon Foundation survey, June-July 2012
About the same as now