2968.0 - Information Paper: Parliamentary Inquiry into the Treatment of Census Forms: Second Submission from the Australian Bureau of Statistics , 1998  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 15/04/1998  First Issue
   Page tools: Print Print Page Print all pages in this productPrint All



This submission has been prepared following the completion of public hearings by the Standing Committee in its inquiry into the treatment of census forms. It focuses on four key issues raised in the submissions to the inquiry and in the public evidence:

(i) There is strong affirmation from key users that high quality census data is important;

(ii) There is strong evidence that the retention of names and addresses will adversely affect census data quality;

(iii) There is no evidence of substantial public good from retaining census forms; and

(iv) The compromises proposed are not viable.


Without exception, submissions to the Committee confirmed that the census is the single most important statistical collection in the country.

The Australian Electoral Commission confirmed the importance of reliable State/ Territory population estimates for the purpose of determining the number of members from each State/Territory in the House of Representatives. The AEC reminded the Committee of the possible Constitutional consequences of unreliable estimates. This was supported by the Australian Joint Roll Council.

The Commonwealth Grants Commission explained how it relies on accurate State/Territory population estimates and high quality census data to determine the distribution of about $21 billion of Commonwealth Grants each year to the States/Territories. The importance of high quality information to support this process was confirmed by the Queensland Premier (s. 224) and South Australia Premier (s. 87) and the ACT Chief Minister (s. 102).

Other government agencies that gave evidence of their need for high quality census data included Treasury, the Department of Social Security, the Department of Employment, Education, Training and Youth Affairs, the NSW Cabinet Office, the NSW Treasury, the NSW Department of Health, the NSW Department of Training and Education Coordination, the NSW Department of Urban Affairs and Planning, the WA Ministry for Planning, the Vic Department of the Premier and Cabinet and the NT Department of Housing and Local Government.

The views presented to the Committee on the importance of high quality census data are consistent with the views that these and many other users have consistently expressed to the ABS over time, particularly when there is a suggestion that the frequency or content of the census might be reduced (eg the submissions presented to the Inter Departmental Committee inquiry in 1993, chaired by the Department of Prime Minister and Cabinet).

The suggestion in the public hearings that agencies had somehow been coerced into providing support for high quality census data is ludicrous and offensive.


(a) There is widespread public concern about privacy and the census

In the public hearings the Committee seemed to take the view that the ABS was relying solely on the surveys conducted by AGB-McNair to support its contention that there is widespread public concern about privacy and the census, and particularly relating to the storage of names and addresses. While the AGB-McNair surveys are convincing (see later) the ABS did provide significant other evidence in its submission (eg see paras 2, 3, 5, 9, 19, 22, 23 of Section 4), and at the public hearing on 25 September 1997.

The ABS evidence was added to at the public hearings. A number of government agencies with experience in collecting data from the community gave evidence that assurances of privacy and confidentiality are essential to ensure a high level of response to data collection. These included the Department of Employment, Education, Training and Youth Affairs (LCA 524), the New South Wales Registry of Births, Deaths and Marriages (S724; LCA 224) and the NSW Department of Training and Education Coordination (LCA213).

(b) The AGB-McNair Surveys

From the outset of the public hearings some Committee members seemed intent on undermining the credibility of the surveys undertaken by AGB-McNair. In the final analysis the first and the second survey results represent the only quantitative evidence presented to the Committee on public attitudes to census form retention. The second survey not only confirmed that the strong public attitudes to census form retention remained 12 months after the census, but demonstrated that neither the lead in statements to the first survey nor the ordering of the statements biased the results. It must be accepted that a very large proportion of the population opposes the retention of census forms.

ABS notes the evidence from other witnesses supporting AGB-McNair work:

  • "I have been with an organisation that has commissioned research from AGB-McNair in the past. AGB-McNair is faultless." (Mr Cavalier, Chairman of the Advisory Council on Australian Archives - LCA 366);
  • " ... within the limits of what is possible in professional market research, the A.C. Neilson-McNair survey is a proper and professionally conducted survey" (Professor Sless - LCA 465).

Committee members seemed to have given great weight to other evidence given by Professor Sless, particularly his response to a question from Mr Mutch.

Mr Mutch: "It has been said that those surveys (ie by AGB-McNair) indicate significant opposition to the retention of census forms. Is that drawing a long bow?
Professor Sless: Yes, it is ......". (LCA 468)

Professor Sless has strong views that lead him to the conclusion that approaches to communication research other than his own cannot provide accurate information. All his evidence to the Committee reflects this 'anti market research position'. At times his evidence suggests that he does not understand the research that was undertaken by AGB-McNair, or that he had not studied it closely. For example:
  • In his response to Mr Mutch Professor Sless talked about ".....the critical question. If you look at the survey question - and I am thinking here of the first one ......... It seems to me that that is not a question but an argument or a statement". He fails to appreciate that the survey had no questions, only statements. As ABS evidence to the Committee pointed out, attitudinal market research is not about asking questions at all, it is about making statements to elicit either strong approval or strong disapproval. It is perhaps worth noting that Mr Mutch corrected him on this point.
  • Referring to a statement in the survey that 'census forms should be destroyed to protect people's privacy', Professor Sless stated that "I would be hard pressed to answer in the negative to that" LCA 469 implying that everyone would agree with that statement. In fact, as the second AGB-McNair survey shows, 30% of people responded in the negative.
  • From a number of references to "the first question" he seems not to appreciate that the order of the statements in AGB-McNair's second survey had been rotated. The statement that 'census forms should be destroyed to protect people's privacy', which he and others suggested was a leading statement, was only presented as the first statement to 1 in 4 respondents in the second survey.
  • In implying that the 'questions' put were 'leading questions' Professor Sless says that "even on the second survey there is an implied correct answer, if you like....". It should be noted that of the four central statements included in the second survey, three were positive for the retention of forms and only one was negative, the one he quoted. Nevertheless, the survey results show that respondents did not accept the "implied correct answer" but consistently opposed the retention of census forms for any reason.

Whilst agreeing with Professor Sless' view that it is not easy to measure "attitudes", his assertions that "attitudes don't predict behaviour" and hence "changed attitudes do not necessarily result in changed behaviour" border on the extreme. All information campaigns, including those run by private businesses and political parties, are aimed ultimately at changing behaviour. The many millions of dollars spent each year on market research, polling, and advertising campaigns suggest that many people do not agree with Professor Sless. Indeed, if the Committee were to accept his advice that the results of these surveys should be disregarded then the Committee would have to believe that all market research results should also be disregarded. Of course, acceptance of Professor Sless' message that "running campaigns .... doesn't necessarily result in predictable changes in attitude", would rule out the suggestion that a slick ABS public relations campaign prior to the next census will be sufficient to sway community attitudes and behaviours!

(c) The ABS position

The findings of the two AGB-McNair surveys are consistent with the other quantitative and qualitative evidence available to the ABS. This evidence strongly supports the ABS contention that the retention of names and addresses will adversely affect census data quality. With 89% of the population agreeing that 'census forms should be destroyed to protect people's privacy and confidentiality', and between 34% and 45% saying they 'would be less likely to complete a census form, if forms were kept', a non-response level of 10% or more in the census is possible. Such a level of non-response, quite likely of uneven impact across different segments of the community, would immeasurably damage the quality of census data. In turn many important uses of census data would be affected.

(iii) There is no evidence of substantial public good from retaining census forms.

Those in favour of retaining census records for family, general history or social research did not present convincing evidence of substantial public benefit from their work through the use of census records. Indeed, most did not address this issue, focusing instead on perceived personal benefit. Few articulated the actual uses that might be made of census records in 2076 or 3001, other than to say it could help compile family histories or conduct social research. In fact some researchers told the Committee that it is not possible to say whether the records would in fact be utilised.

The Committee heard evidence that there are abundant alternative sources that are already available for much of this work. Census records would merely provide another source that might, possibly, be more convenient to access and more complete. For example, the evidence from Mr Trujillo, an expert practitioner from the Genealogical Society of Utah:

Acting Chair: "If Australian census records were retained on, for instance, microfiche and were kept confidential for 100 years and were then released to the public, would they contain information which would not otherwise be available to researchers through any other source?"

Mr Trujillo: "No. There would be duplication under births, deaths and marriages and other records, but they would have a unique value to the genealogist. Every little bit of information, such as the street and suburb that somebody lived in and who else lived in the house, is of utmost importance to genealogists and family historians." (LCA459)

It was apparent that some of the social research anticipated would not need identifiable census data but could be undertaken from the already available aggregate census data relating to small areas or communities of interest. Many social research projects of high contemporary value have been conducted and widely reported in the media using aggregate census data (eg Professor Bob Gregory's work on income distribution). It is not clear from the evidence what additional value such research would have if identified census records were to be available.

It was argued that destroying records is wrong because "we never know what question society may wish to ask or seek answers to in the future" (LCA284). That is tantamount to an argument for retaining all records forever without the onus of establishing their value. Of course, unless the law were changed, the Privacy Act would not allow the use of records for any purpose not explained at the time their collection.

Some geneticists proposed (LCA 254) that the census forms be retained to assist with genetic research. In its initial submission the ABS raised some practical problems which make this difficult to implement. But ABS believes that the general public would strongly oppose the development and maintenance of a social register of all persons in Australia, based on the census (see the reaction to the Australia Card proposal).

The Australian Institute of Health and Welfare (LCA 337), proposed that the census forms be retained by the ABS, that ABS immediately commence matching the records against various administrative files (eg birth and death records) and then ABS make available those records, in a way that they could not be identified, to support epidemiological research. Evidence by New South Wales Health (LCA 206) outlined some of the failings of census records for such purposes, but ABS can accept that there could be some value to such epidemiological research. However, there would be significant privacy and ethical issues to be addressed, including likely community concerns about data matching across data bases, and data being available (albeit in unidentified form) immediately for researchers. While undoubtedly there would be some support for such work it is doubtful whether there would be a high degree of public support.

(iv) The compromises proposed are not viable.

Throughout the public hearings a number of 'compromises' have been canvassed. None are seen as viable due to the privacy concerns they raise, or because they would be difficult to explain and 'sell' to the public, or because they raise significant operational problems, or because of their cost.

Opt in/opt out:

One 'compromise' suggested is to give householders the option on their census form to indicate whether their form can be kept or not.

This proposal would raise significant legal and operational issues:
  • Most significantly, the ABS does not have the authority to collect information other than for statistical purposes. At the very least it would be necessary to amend Section 6 of the Australian Bureau of Statistics Act 1975 to allow this option to proceed.
  • A major operational difficulty would be getting consent from every person in the household. The current census methodology is for one householder to complete the census form on behalf of all household members. This option would require the census form to be redesigned (a time consuming and expensive task), as it would not be acceptable for one person to give consent on behalf of the entire household.
  • Difficulties would arise in getting consent from minors, or on behalf of minors, and for other persons not legally competent to give consent. Would consent need to obtained from minors once they reached adulthood?
  • The Privacy Act would demand that the consent given would have to be informed consent. This would put a substantial additional load and obligation on census operations and the census communication strategy.

Finally, this option will not deliver the data that genealogists et al are seeking. At best it will lead to an incomplete set of census records which would not serve their intended purpose. At worst it could lead to a very incomplete data set, particularly if the proportion giving consent is as low as indicated by the AGB-McNair survey.

A PR campaign will fix everything:

The assertion that a PR campaign will fix everything has very much over-rated the likely effectiveness of such campaigns, especially when a majority of the community has indicated significant concerns about the proposal to retain Census forms. As discussed above, Professor Sless' evidence may also be relevant here.

In its evidence ABS has pointed out that even if a very effective PR campaign could convince the vast majority of the population that there is nothing to fear from the retention of names and addresses, if only 10% of the population are not convinced the quality, and hence value, of the census data would be significantly reduced.

No matter how good an advertising campaign is, a one-off event with universal coverage like the census is very vulnerable to an effective anti-census campaign. It is not good risk management practice to rely on the hypothetical success of a future advertising campaign to totally rid public's concerns about privacy, or the absence of anti-census campaigns, so the Census can continue to deliver quality data for effective decision making. Being a $150m event, there will be no opportunity to run the census again should a judgement that the retention of names and addresses will not adversely affect the census be found to be wrong.

The proposal to assist such a PR campaign by using the term "Centenary gift to nation" is also likely to be ineffective. If the proposition is seen as an unacceptable invasion of privacy, it would still be considered as such by the community, whether it is dressed up as a "gift to nation" or not. No advertising campaign will alter that fact.

Finally, it should be noted that the retention of census forms from just the 2001 Census would not meet the needs of many proponents (eg epidemiologists, social researchers).

Putting genealogists questions on a single page:

It is technically not feasible to put the nine or more topics required by genealogists on a single page - it is likely they would not fit on two pages. However, even if this could be achieved, the sequencing of the questions that is necessary to ensure high quality responses, would be disturbed.

This proposition also ignores the fact that name and address, which are the core items required to be retained, are the two items that arouse the strongest privacy concerns.

Retaining names and addresses electronically and separately from the statistical data:

Again, this ignores the fact that retaining names and addresses is perceived as the main threat to privacy.

Retaining this information electronically substantially, and significantly, increases the capability of matching names and addresses with other information held electronically. This is an area of significant public concern.

Proposals for ABS to retain forms and match census records and other information for AIHW:

The AIHW proposal requires the immediate access and use of the retained census records and the immediate matching of those records with other (sometimes quite sensitive) data bases. It is not clear whether the privacy concerns this would raise could be sufficiently allayed. While it is true that such a longitudinal study has not raised privacy concerns in the UK, their participation in the study is not explained to householders. This would not be acceptable in Australia.


The evidence shows that there is significant public opposition to the retention of census forms. The very important uses of the census should not be put at risk on the basis of an unproved assertion that this opposition does not exist or that an effective PR campaign will allay the public concerns. The very important uses of the census should not be put at risk to satisfy what is essentially a private benefit.

The ABS' professional, frank and fearless advice to the Committee is that there is a very significant risk that a change to the current policy of form destruction will impact adversely on the quality of census data. Nothing put to this Committee demonstrates that the costs to Australia in 2001 and beyond of poor quality census data would be outweighed by the benefits, particularly public benefit, of retaining the forms. The primary purposes of the Census should not be put at risk by the secondary purposes, however laudable they may be, or purport to be.

Australian Bureau of Statistics

April 1998