Previous Next

With the benefit of hindsight: Valedictory reflections from departmental secretaries, 2004–11

5. The last count—the importance of official statistics to the democratic process1

Dennis Trewin

Dennis Trewin joined the Australian Bureau of Statistics (ABS) in 1966 as a statistics cadet and was Australian Statistician from 2000 to 2006. Prior to that he was Deputy Australian Statistician responsible for economic statistics and a Deputy Government Statistician in New Zealand. Other appointments he held include Australian Electoral Commissioner, Chairman of the World Bank Board on the International Comparison Program, a member of the Committee responsible for preparing the 2006 report on the State of the Environment and Adjunct Professor at Swinburne University of Technology, Melbourne. Dennis Trewin is a past President of the International Statistical Institute, International Association of Survey Statisticians and the Statistical Society of Australia. He is currently Chair of the Policy and Advocacy Committee of the Academy of Social Sciences Australia

In this valedictory address I will reflect on the Australian Bureau of Statistics (ABS) − both past and future − before making a few comments on the APS more generally.

The Australian Bureau of Statistics

I started work in the ABS in December 1965 as a vacation student and then as a statistics cadet before starting work full-time in 1968. At that time, the ABS was actually the Commonwealth Bureau of Census and Statistics and a Branch of Treasury rather than a statutory authority (but largely independent except for budget considerations). The Australian Statistician was referred to as the Commonwealth Statistician. Four out of the last five Statisticians had been Tasmanians as a consequence of their early signing of an Integration Agreement with the Commonwealth government. Core outputs were publications full of statistical tables and few words. Also there was no seasonal adjustment or summary of main features and as a consequence relatively little media reporting of official statistics.

Moreover, social statistics were based on the Census or administrative systems (e.g. births, deaths, migration cards). Household surveys had just started. These were quarterly labour force surveys. The first social supplementary survey was conducted in 1968 − a survey of chronic illness in NSW. Official labour force statistics were based on registered unemployed and were very unreliable. Monthly labour force surveys were not introduced until 1978, and Statistics cadets were our main source of graduates; it was difficult to otherwise persuade graduates to come to Canberra.

But there were more things in common than different. Our core purpose has not changed: we remain the provider of official statistics for the government and the community. There has always been a strong emphasis on the quality and integrity of outputs, and professionalism − our core values have not changed. Similarly, confidentiality was always regarded as paramount. The Census and Statistics Act is still recognisable from that period, indeed from when it was first created in 1905. We have had the support of governments throughout my time at the ABS.

ABS staff numbers over this period have been up and down: 10 staff in 1905 not including those working in the state statistical offices; 2500 in 1968; 3500 in 1975 to 2900 now. But productivity has improved substantially. Although staff numbers have declined since the mid 1970s, outputs have increased substantially, including the number of statistical collections. Technology and methodology developments have been key contributors. Close alignment between the technology and methodology areas and the business areas has been a key factor and one which I will return to shortly.

During my time as Australian Statistician, I have tried to:

  • maintain the quality and integrity of statistical outputs, particularly core statistics like population, national accounts and prices, and extend our outputs to new areas of policy interest. Such new areas have included environment (water), social policy and innovation;
  • improve productivity (as required by the Certified Agreement) by redesigning business processes, utilising our technology/methodology capabilities;
  • improve user engagement by listening, explaining and acting instead of just listening (we were always fine at listening but not at the follow-up);
  • in recognition of the increasing demand for detailed statistics, increase access to data and more generally improve usefulness of data for analytical purposes, particularly by improving access to microdata. Free data on the web has been a positive and well received step;
  • make the ABS website the core way of accessing our outputs, including regular publications;
  • be media friendly and accessible, not just myself, but specialist staff throughout the ABS (most Australians probably find out about official statistics through the media);
  • keep compliance cost to a reasonable level (the number of complaints has declined substantially). Access to taxation data has been of great assistance in reducing the compliance cost on business. Compliance cost on all business, particularly small business, has reduced by more than 40 per cent since the Small Business Deregulation Task Force reported in 1996.

The way users access statistics has changed dramatically over the last six years. The following table is indicative of this trend.

Table 1: Demand for different forms of statistical output





Pages viewed



Products downloaded



Telephone inquiry service



Publications − revenue



Confidentialised Unit record files − applications processed



As the table shows, website access has increased fivefold. Products downloaded are now 70 times greater than what they were six years ago [the year 2000]. Not surprisingly, the demand on our printed publications has reduced substantially (although phone inquiries have increased). This is a good outcome. We prefer as many users as possible to be able to ‘self-help’.

Although use of Confidentialised Unit Record Files (CURFs) have not increased that much over this period, you didn't have to go back much further to find their use was almost negligible. The increase of CURFs increased considerably as a result of a special agreement with the Australian Vice Chancellor's Committee. Statistical leadership across all providers of statistics is another area where I have tried to put more focus but progress has not been as great as hoped. But I think the direction is correct.

I am fortunate that I inherited an organisation that was in very good shape and had a strong reputation. There was no need for major change. We could afford a continuous improvement strategy reacting to new demands, the changes in our environment and new opportunities that arose.

There are four key challenges for the ABS for the future that I would like to mention:

  • improving the national statistical system;
  • making greater use of administrative and transactional data bases;
  • improving the user experience;
  • enhancing the international comparability of statistics.

I will discuss each in turn. The first challenge is improving the National Statistical System (NSS). Clearly the ABS has some technical skills in statistics that might be of use to others. And increasingly our information management skills are being recognised as being particularly strong.

Our involvement in NSS work recognises that the ABS is not the only provider of statistics. There are advantages to both producers and users of statistics if we can work in a more coherent way. Government agencies increasingly need to work in a 'connected' way. This will only happen if they are prepared to share information, including statistical information. It is important that this information be coherent − for this we need to be using the same concepts and definitions to the extent possible. This requires leadership on standards and classifications, a role which the ABS is well suited to play. It is also important that the range of statistics be of good quality − not perfect but based on sound statistical methods. Again the ABS has a constructive role to play.

Key NSS activities might include: developing and promoting standard classifications and coding tools; developing and promulgating statistical frameworks, standards and definitions for general use; providing access to statistical infrastructure such as our business register; promoting good practice (e.g. through training); maintaining networks of statistical practitioners; and possibly over-recruiting statisticians recognising that some might move to other agencies. Another thing we are doing is developing statistical centres of expertise that have knowledge of all relevant statistics not just those of the ABS (e.g. environment statistics).

From our perspective, the NSS will be considered successful if more good quality statistics become publicly available, something that is good for both democracy and effective policy. For example, macro-economic statistics/policy is very dependent on transparent and trusted statistics. This is also true for other fields of statistics. It is good for government, particularly if things are going well − better to have policy debate focussed on facts rather than advocacy or anecdote. And yet, there is reluctance by some to make information publicly available in case it might embarrass the minister (at both Commonwealth and state levels). It will occasionally, but I would argue that in most cases it is better for governments if debate was based on trusted facts rather than the alternatives.

The second challenge is greater use of administrative and transactional data bases for statistics, something that is presently underutilised. These are beneficial because:

  • they can reduce compliance cost (e.g. ABS use of taxation data has been a major reason for the 40 per cent reduction in compliance cost);
  • they can provide more detailed data than from sample surveys because you effectively have a census of the population in scope of the administrative system (e.g. regions or small populations);
  • they often have an in-built longitudinal element which enables you to better understand the dynamics of the population.

Of course, there are privacy issues associated with this issue that have to be carefully managed, but we should not be scared off by them.

Australia is well behind many other countries in its use of administrative data for statistical purposes. The use is most prevalent in Scandinavia with their register systems. But Canada, which has a lot of cultural similarities to Australia, also has a history of record linkage for statistical purposes. It is not controversial.

The ABS could act as custodian of these data bases for statistical purposes (as we do for births and deaths, immigration cards, customs and taxation) or could act as statistical and information management advisers to agencies who prefer to retain custodianship. In many cases we are finding access to administrative data more difficult than in the past. There are a variety of reasons such as privacy, statistical support not being core business and, in some cases, concern about information becoming public.

The third challenge is to improve the user experience. The website is crucial: our strategy is one of continuous improvement with occasional leaps in the improvement of the design of the website. But we will be continually improving usability, increasing content, developing new graphical presentation tools and developing better links to related websites. The establishment of a National Data Network as a single source to all statistics in particular fields (e.g. environment statistics) is another related initiative, as while custodianship will not change but awareness, coherence, access and usability of statistics can improve.

We will also continue to improve access to microdata through remote access facilities (e.g. through the Remote Access Data Laboratory and Table Builder system to access 2001 and 2006 Census data) and by setting up collaborative arrangements with other government agencies that are consistent with our legislation.

Another means of improving the user experience is to develop linked data sets that can be used for research and analysis purposes, particularly understanding the interconnection between different social outcomes. We have started by establishing the first round of a longitudinal link of population census records. It proved to be non-controversial during the 2006 Census.

The fourth challenge is improving international comparability of statistics. Official statistics are being used more extensively for international comparisons, perhaps with the assistance of international organisations like the OECD. Comparing statistics for Australia with those from other countries provide a context for Australian figures. Differences can be illuminating in evaluating the effectiveness of current policy or for assessing alternative policy options. This is one of the main reasons why the ABS is an active contributor internationally, particularly to the development of international statistical standards to support comparability of data.

I am often asked why the ABS is so well regarded, particularly internationally. There is a combination of factors of course but top of the list is core values. Our core values are integrity, professionalism, a focus of service, relevance, trust of data providers and access for all. They are widely accepted and understood. More importantly, they reflect the behaviour of staff at all levels and in all offices. That certainty makes my job much easier.

The Australian Public Service

Now turning to the APS more generally, I wish to make four quick points. First, I want to emphasise the importance of a close alignment between IT providers and the business areas. This needs to work in two directions (whether there is an internal provider of IT or it is outsourced). IT is a great enabler and source for innovation but only if there is alignment with business areas. Furthermore, developments must be treated as business projects rather than technology projects. They should be driven by the desired business outcomes. This may involve redesign of business processes. This is often necessary to get full benefits from the investment. Finally, you should ensure good project management arrangements are in place.

Second, I want to recall the words of Drucker: ‘good leaders don't make many decisions’. This may seem counterintuitive but it implies that good leaders provide the authorities, policies, procedures to enable others to make sound decisions. A question public servants – and perhaps some ministers − should ask themselves is whether they get involved in too many micro decisions. And yet, I realise it is easier for an organisation like the ABS which is not subject to day to day involvement with ministers to set up the authorities, etc to enable effective delegation.

Third, I think we have to assume that new graduates are capable of some difficult work very early in their career. We just can't rely on the recently retired to undertake the more difficult or complex tasks or the APSC will not develop at the rate it should. Our young staff want to be tested − this is the consistent feedback I received from the annual ABS Youth Forum. I think I was fortunate in being thrown in the deep end early in my career. Furthermore, turnover is high among our younger staff so it doesn't make sense to make a high investment in their formal training in their early years. Also, work experience can be the most effective way of learning. That was certainly my experience.

Fourth, I want to discuss how to manage errors – something that is inevitable if only because of volume of work. They also tend to be more public now than what they were even ten years ago. The ABS does make errors but fortunately not too many. The ABS approach is to (a) admit error and subsequently apologise; (b) don't seek blame but try to learn from it particularly and assess whether there are systemic problems; and (c) undertake an independent review if necessary. This generally works well for us but only if we make relatively few errors. Such an approach may have more general applicability.

In conclusion, it has been a great privilege to be Australian Statistician during such an interesting period, and I also greatly appreciated support from government. It has provided us with an adequate budget, but perhaps more importantly it has shown an excellent understanding of the importance of the integrity of official statistics to the democratic process. This is not true for all countries.

1 This speech was delivered in December 2006 at a function organised by the Australian Public Service Commission.

Previous Next