I S K O

[top of entry]

1. Background

ANZSRC was first released in 2008 as the outcome of a collaborative project between the Australian Bureau of Statistics (ABS), Australian Research Council (ARC), the then New Zealand Ministry of Research, Science and Technology (MoRST) — now Ministry of Business, Innovation and Employment (MBIE), and Statistics New Zealand. The classification is one of a suite of joint Australian and New Zealand Standard Classifications used in the analysis and comparison of official statistics in the two countries. The other classifications include industry (ANZSIC), occupation (ANZSCO) and offence (ANZSOC). The collaboration is initiated by the general approach of the statistical agencies to harmonise statistical classifications wherever possible as part of the wider Australian and New Zealand Closer Economic Relations Trade Agreement (ANZCERTA).

ANZSRC is owned by both statistical agencies but the ARC and MBIE act as custodians. The original development leading to the release of the first ANZSRC in 2008 stemmed from the revision of the Australian Standard Research Classification (ASRC), and the decision by the then New Zealand Ministry of Research, Science and Technology (MoRST) that this provided an opportunity to look at producing a standard that could also be used in New Zealand. The ANZSRC classification contains three component parts: Type of Activity (ToA), Fields of Research (FoR) and Socio-Economic Objectives (SEO).

Implementation of the 2008 ANZSRC was delayed by the creation of the new Ministry of Business, Innovation and Employment (MBIE), which absorbed the previous Ministry of Research, Science and Technology (MoRST) plus other New Zealand Government departments. Consequently, no maintenance strategy was ever embedded to maintain the relevancy of the classification for the New Zealand research community. Any plans for an update were also driven by the ABS decision to progress a five yearly review cycle commencing in 2013.

An investigation was undertaken in 2013 to consider whether ANZSRC should be revised, but at that time, there were no strategic drivers in the New Zealand research community that indicated a desire to undertake a review, although the ARC did seek some changes at that time. Subsequently the New Zealand Research, Science and Innovation Domain Plan was developed which highlighted the need for updating the ANZSRC classification to address data gaps in the official statistical system (Ministry of Business, Innovation and Employment et al. 2016)

In 2018 a full analysis of whether a revision of ANZSRC should commence was undertaken by the four lead agencies and key stakeholders that which resulted in the decision to fully review ANZSRC and to release a new classification in June 2020 (ANZSRC 2020).

[top of entry]

2. Research and development standards

Research and development (R&D) plays a pivotal role in driving a country’s economic growth and supporting the sustainable development goals (United Nations 2022). It is not just about measuring the expenditure in the R&D sector or focussing on traditional sciences alone as research and development encompasses many multi-disciplinary domains and sectors used for policy and decision making. But other factors such as the cost of conducting research, shifts in research activities and changes in the composition of institutes or organisation conducting research and development influence the measurement of R&D and the need for standardised approaches (Curtis, McVay and Toynbee 2019).

However, as with all official statistics, there is a need to provide comparability and definition of a variable in a standard way and statistical classifications and statistical standards enable that. Statistical frameworks and classifications have always been necessary to enable grouping and organisation of information and data into meaningful sets based on common criteria, concepts or definitions. This is exemplified in the United Nations Fundamental Principles of Official Statistics which states in Principle 9 that: “The use by statistical agencies in each country of international concepts, classifications and methods promotes the consistency and efficiency of statistical systems at all official levels” (United Nations Fundamental Principles of Official Statistics 2014).

The International Organisation for Standardisation (ISO) endorsed the following definition of an international standard:

International standards are usually documents established by consensus and approved by a recognised body that provides for common and repeated use, rules, guidelines for activities or their results, aimed at the achievement of the optimum degree of order in a given context. (ISO/IEC Guide 2 1996, Section 3.2)

In addition, the Guidelines for the Template for a Generic National Quality Assurance Framework (NQAF) state that:

Standards refer to a comprehensive set of statistical concepts and definitions used to achieve uniform treatment of statistical issues within a survey or across surveys, and across time and space. Standards assist in maximising the effectiveness of statistical outputs and the efficiency of the production process in terms of inter-temporal, national and international comparability and coherence (i.e. the capacity for integration) of the statistics. (United Nations Statistical Division 2012, 14)

Whilst there is no single international standard statistical classification for → classifying research and development (Hjørland and Gnoli 2022), contextual content is included in international statistical classifications such as the Central Product Classification (CPC), International Standard Industrial Classification of all Economic Activities (ISIC) and the International Standard Classification of Education (ISCED). At the regional and national level there are examples such as the Australian and New Zealand Research Classification (ANZSRC 2020) which is discussed in this paper, the → Canadian Research and Development Classification (CRDC) (Legendre 2019), the Dutch National Academic Research and Collaborations Information System (NARCIS) (Scharnhorst and Doorn 2019; National Science Foundation 2022).

In the New Zealand context, there are also the primary funding standards that sit with the Royal Society’s Marsden Fund [1] and the Performance Based Research Fund (PBRF) [2].

However, the starting point for classifying research and development is the Organisation for Economic Co-operation and Development’s (OECD) Frascati Manual (OECD 2015), along with the Oslo Manual for Innovation (OECD 2018). These, particularly the Frascati Manual, provide the base definitions for the concepts of research, and their associated types of activities, as well as a Fields of Research classification.

Multiple but similar conceptual frameworks are not always the most helpful starting point when striving for a consistent approach to classifying research and development. For statistical purposes, it is important to use one standard with mutually exclusive categories, but it is accepted that a single, all-encompassing standard will not satisfy everyone or have the coverage that everyone wants. The elements of statistical significance and user demand along with historic practice are the main inputs to any classification and as long as the standard can meet the majority of needs of the majority of users then that is the benchmark to achieve. Coupled with a regular maintenance program and a strong implementation plan and supported by the appropriate tools for coding and managing time-series, the use of classifications best practice principles is usually sufficient to provide a consistent and integrated framework.

[top of entry]

3. Why classify research and development?

It is important to ensure there are standard concepts, definitions and classifications used across a statistical system to allow integration, comparability, and interoperability of research data. The standards group and organise information in a meaningful and systematic way, in exhaustive and structured groups that are usually defined according to an agreed set of criteria. It is, in some respects, about providing a simplification of the real world and for collecting, organising and analysing data from statistical and administrative sources.

Fundamentally though the need to classify research is, firstly, from a government perspective, about understanding the contribution of research, science, and innovation to economic, social, health and environmental objectives. This combined with the social return on the government’s investment in research and development, and for determining how the government may support or invest in research and development leads to the need for a consistent framework of concepts and definitions and a reporting mechanism for which a classification is an ideal tool.

A second requirement or need to classify research is for understanding the different fields that a classification measures, and in explaining the categories that are within scope of a particular subject field. Understanding → science or engineering requires a fundamental explanation of the scope of those particular fields, it is not just about categorising into fields like electrical, mechanical, or structural engineering, or biological, clinical, or physical science. There is an inherent need to encapsulate the sociology of engineering or science to distinguish between the application of methods in research and the quantitative aspects of that research. This poses a great challenge for building a statistical classification.

The effect of the two preceding paragraphs leads to the fundamental question of what the purpose of the classification or standard is. On one hand, it is needed for the production of official statistics, particularly within a research and development survey, and for identifying data gaps, whereas on the other hand it is for providing a funding and reporting framework for research organisations and academia to assist with progressing the evolution of research and for developing policy.

The challenges in building a statistical classification are that they are often used in a way or context not initially intended by their developers. It is not always an intuitive consideration to balance the differing user and data system needs and interests when the primary purpose is for the production of official statistics. For example, an occupation classification may be created to simply provide a count (usually from a labour force survey or population census) of the number of persons employed in a particular category whether this is a statistician, a farmer, a manager, or a teacher etc. However, that same framework may be used to underpin policy in skill migration or assist in identifying career pathways through linking of occupation data to skills requirements or job vacancies. Combined with this is not letting the “loudest voice, deepest pockets” syndrome impact what content is represented in a statistical classification.

Whilst the national statistical office may own and drive a development or revision of a classification, its goal is really to support government policy and decision-making needs. A statistical classification cannot be all things to all people. As Millerand and Bowker (2009, 165) wrote: “This is why it does not make sense to see standards simply as things out there in the world that have a predetermined set of attributes. In information systems, standards are in constant flux — they have to migrate between communities and across platforms”.

As the scope of science, technology and innovation policy has widened over time it is useful to note that the innovation system is no longer about the research and science system, it encompasses skills, tax, procurement, enterprise policy and framework conditions (Ministry of Business, Innovation and Employment et al. 2016, 9). All of which has an impact on how a standard classification is developed.

The need to standardise administrative and statistical data across the national data system and reduce the number of stand-alone frameworks and reporting mechanisms also assists in improving the wider data availability. Classifications make sense out of chaos and the competing needs of research funding versus understanding investment in research benefit from having a standard classification.

[top of entry]

4. Drivers for a standard research and development classification in Australia and New Zealand

One of the challenges for statistical agencies is the need to ensure the data reflects the contemporary reality of the economy and society. The traditional approach to this is through a cyclical review process, often on a five- or ten-year cycle, whereby a revision project is undertaken to update the classification. Factors that influence this process are changes in legislation or policy needs, changes in international reporting requirements (and/or alignment with new international standards), the need to identify and report on new and emerging sectors or fields of research and development, and user demand for classification categories. In considering these drivers, it is important to also consider the costs and administrative burden for stakeholders to implement any changes and the impact on time-series.

A major issue for the Australian and New Zealand research community was the lack of visibility or recognition of indigenous research for Aboriginal and Torres Strait Islanders, Māori, and Pacific Peoples.

From a New Zealand perspective, statistical classifications need to reflect a te Ao Māori (Māori world) view and acknowledge our founding document Te Tiriti o Waitangi (the Treaty of Waitangi). Data on research and development has to be for and about Māori, and in the New Zealand context there is also a need to reflect that same requirement for Pacific Peoples.

ANZSRC when first published in 2008 endeavoured to address research and development categories for “Aboriginal and Torres Strait Islanders”, “Māori” and “Pacific Peoples” as alternative groupings for understanding Indigenous research. The 2018 ANZSRC revision sought feedback as to whether these alternative groupings were still sufficient and appropriate for users to classify Indigenous research.

The consultation highlighted that the classification lacked visibility and recognition of Indigenous research, and that the range of Indigenous research had advanced since the classification had been first published. This was affecting the ability to report and analyse data for government, universities and other users. An additional aspect was the Eurocentric/Western ideals of what constituted research and development affecting which categories might be represented and described in the revised classification. There was, as with the interdisciplinary and multidisciplinary research categories, the issue of mutual exclusivity.

Stakeholders were given three options to consider which included:

• Option 1: Promotion of existing fields (6-digit) to groups (4-digit), additional codes at both levels and a list of alternative grouping of these new codes
• Option 2: Creation of a new division for “Indigenous Research” with cross referencing to other divisions
• Option 3: Creation of new division for “Indigenous Research”.

The outcome was the creation of a new division in ANZSRC for “Indigenous Studies” and a definition was developed that reflected that the new division applied to the subject of the research, not the identity of the researcher. Additional definitions, exclusions and explanatory material was added to provide clarity for users. The change has had a broad support from Indigenous and non-Indigenous stakeholders in both countries.

In addition to this is the ability to clearly and/or easily classify interdisciplinary and multidisciplinary research. In ANZSRC interdisciplinary research is treated as research that integrates tools and concepts from multiple and often disparate fields of research into a single research activity. Multidisciplinary research is treated as involving researchers from different disciplinary backgrounds working together to solve research problems.

For ANZSRC to strictly adhere to classification best practice, particularly the principle of mutual exclusivity, would not enable users to accommodate approaches to, and need for interdisciplinary and multidisciplinary research categories. Both types of research are cross-cutting and the ability to address this in traditional statistical classifications is not possible as much of the configuration of a classification hinges on the constraint of an A4 hardcopy page. Therefore, there can only be one category label, there is only so much available space for definitions, and codes have to be sequential and follow a strict parent-child relationship. Fluidity in content is more easily obtainable through conceptual frameworks, or metadata modelling approaches within cloud-based IT systems and that may be a consideration in any future review of ANZSRC.

Consequently, stakeholder feedback was sought on how ANZSRC could be revised to better classify interdisciplinary and multidisciplinary research. The consultation reiterated that there was no viable solution that could be applied to the classification to resolve or avoid this issue. At best providing general guidance on how to treat both types of research in relation to the classification categories was about the only resolution. In most instances, allowing users to assign multiple codes to research data, and/or apportion research across multiple codes, allows capturing of interdisciplinary and multidisciplinary research and that was needed. The alternative would be to overload the structure with multiple and similar classification categories.

All these issues pose a challenge against the primary classification principle of mutual exclusivity and whether the research can be put to one single classification category or not. However, the overarching decision in determining a way forward for Indigenous research, and interdisciplinary and multidisciplinary research was to try and be consistent. Providing better guidance and definitions rather than structural change and an increase in categories in the classification was seen as the best approach going forward. A future revision may attempt to revisit the methodology and strategy used.

[top of entry]

5. Methodology

The methodology used for developing the ANZSRC classification followed the statistical classification development and revision model utilised by both the Australian Bureau of Statistics and Statistics New Zealand. Whilst there is not one fully agreed model that both agencies use, the processes are very closely aligned.

The focus for any classification development is based around establishing the need for the classification either as a new classification or as a revision of an existing classification. This is then followed by a process of clarifying the scope of the work in terms of:

• identifying the statistical unit to be measured,
• identifying what data sources are available to inform what is included in the classification,
• clarifying concepts, principles, and criteria for building the classification, the type of structure for the classification, and if it is a revision, the scale of that revision.

The latter may or may not include a review or change to the underlying concepts and conceptual base of the classification, but usually encompasses a process of adding or deleting categories and/or definitions within the classification structure.

The process for additions and deletions, or actual creation of the structure in the first instance is based off issues established in the need process supplemented by stakeholder or user raised issues along with an analysis of any survey or administrative data. For many statistical classifications a numerical threshold may be used as criteria for whether a category is included in the structure, for example, 100 persons reporting a category in a statistical survey (this can be compared to the → literary warrant threshold in libraries, see Barité 2018). However, over time the threshold criteria requirement has diminished as data users want more and more information and content is included unless there is a good reason to exclude based on the scope and purpose of the classification, or because of statistical insignificance. Data from Higher Education R&D Surveys, or even R&D funding applications may be used as a guide but fundamentally it is about the ability of the National Statistical Office to reasonably collect and output the information and articulate user requirements.

A top-down, bottom-up iterative process is then employed to build the structure with input from subject matter experts or data users until such point as an agreed structure and content is achieved. This approach helps refine the structure by looking at the known data and seeing whether the existing categories at the lowest level cover the content of the data. If not, new categories are developed and then evaluated against the scope of the level above to ensure that the new category belongs in that higher-level grouping. This is then done at the next level back to the top. Fundamentally it is about identifying all the possible categories at the detailed level, then identifying some broad groupings, putting the detailed categories into those broad groupings and establishing how many levels for the classification structure are required. Then refining this through revisiting the scope of the groups and categories, their relationships to one another and applying any known data to see if the outcomes are logical and sensible.

In creating a new classification, where one has not previously existed, consideration is given to the availability of any international standards and whether or not these can be adopted as is or adopted through a process of adaptation to the local context. For ANZSRC in 2008 the initial process stemmed from the revision of the Australian Standard Research Classification (ASRC) with adjustments made to incorporate New Zealand content and enable a joint standard to come into existence. Direct adoption of the Frascati Manual content, and even adaptation of Frascati was not considered a viable or cost-effective exercise both in 2008 and in the 2020 revision.

At all times the processes, adhere to the Best Practice Guidelines for Developing International Statistical Classifications developed by the United Nations Expert Group on International Statistical Classifications (Hancock 2013) and the associated Approval Process for an International Statistical Classification (Hancock 2017). The two documents represent the approved procedures for classification development that national statistical agencies should follow.

All of this constitutes a form of domain analysis when defining the content and structure of ANZSRC, albeit as a sub-conscious by-product within the development process rather than a formal exercise. The ANZSRC (2020) revision had a previous but dated version to refer back to, and to build on. However, with the scale of change proposed from the consultation process, there was limited benefit for much of the existing classification content, particularly at the six-digit level, to be carried over.

From a system perspective having the ANZSRC 2008 stored in a classification repository did make it easier to build and reuse existing content. However, with technology changes and the overarching drivers to move from hardcopy to a variety of formats such as XML, PDF, MS Word, MS Excel, and SDMX etc., the storage of the new ANZSRC and its mappings were configured appropriately for those formats within the system. That said, the conceptual model and componentry that now sits behind ANZSRC enables a simpler entity and facet management going forward such that an updating can be of a dynamic nature rather than through the traditional cyclical review process. Also, the future vision is to link related concepts and categories together so that the user of ANZSRC data can produce richer and more detailed stories about that. For example, using neural networking principles enables research to link to industry, to occupation, to products, to qualifications and so on.

[top of entry]

6. Revising the Australian and New Zealand Standard Research Classification (ANZSRC)

As noted in the background above, ANZSRC was originally published in 2008 after a collaborative project between a number of Australian and New Zealand agencies.

ANZSRC is a framework containing three interrelated classifications for Type of Activity, Fields of Research and Socio-Economic Objectives, and is closely aligned to the OECD Frascati Manual. First released in 2008 it underwent a major revision in 2019/2020 to address gaps in research and development data, recognise new and emerging fields of research and address the low visibility of indigenous research activity in official data.

In New Zealand ANZSRC is used by government, funding agencies, Crown Research Institutes, universities, and independent research organisations. Statistics New Zealand only uses the Type of Activity classification and the Socio-Economic Objective classifications in its Research and Development Survey, and international reporting compliance. The Ministry of Business, Innovation and Employment (MBIE), Health Research Council and Royal Society of New Zealand use the Fields of Research classification to support research funding decisions. In Australia ANZSRC has a similar application across government, university, and other stakeholders. The Australian Bureau of Statistics uses ANZSRC in its R&D data collections whereas the Australian Research Council uses the Fields of Research for researching funding processes, the Excellence in Research for Australia (ERA) and Engagement and Impact (EI) research evaluation exercises, and the Socio-Economic Objectives for reporting purposes.

Type of Activity (ToA)

The Type of Activity classification allows research to be categorised according to the type of research being undertaken. It is a flat classification of four categories: pure basic research, strategic basic research, applied research and experimental development. The definitional text in the 2008 ANZSRC generally aligned with that used in the Frascati Manual with some contextual and colloquial differences for the Australian and New Zealand situation.

ANZSRC (2020) directly embedded the ToA definitions from the Frascati manual but still retained the four categories from the 2008 ToA classification, noting that the definition for strategic basic research is not one contained in the Frascati manual.

Fields of Research (FoR)

The FoR classification is a three-level hierarchic classification consisting of divisions (two-digit) at the top level, groups (four-digit) at the second level and fields of research (six-digit) at the third level. The FoR classifications allows R&D activity to be categorised according to the methodology used in the R&D, rather than the activity of the unit performing the R&D or the purpose of the R&D.

Table 1: ANZSRC Fields of Research divisions
 

30	Agricultural, veterinary and food sciences
31	Biological sciences
32	Biomedical and clinical sciences
33	Built environment and design
34	Chemical sciences
35	Commerce, management, tourism and services
36	Creative arts and writing
37	Earth sciences
38	Economics
39	Education
40	Engineering
41	Environmental sciences
42	Health sciences
43	History, heritage and archaeology
44	Human society
45	Indigenous studies
46	Information and computing sciences
47	Language, communication and culture
48	Law and legal studies
49	Mathematical sciences
50	Philosophy and religious studies
51	Physical sciences
52	Psychology

Socio-Economic Objectives (SEO)

The SEO classification is a hierarchic classification of three levels similar in nature to the FoR classification.

In the 2008 classification an additional level called sector was placed across the top of the classification to provide a level of aggregation for output purposes. This level was removed from the 2020 classification as users indicated that it provided no meaningful purpose and the original decision to use an alpha code for it, when the rest of the code structure was numeric posed system issues for storage and collation.

Table 2: ANZSRC Socio-Economic Objectives divisions
 

10	Animal production and animal primary products
11	Commercial services and tourism
12	Construction
13	Culture and society
14	Defence
15	Economic framework
16	Education and training
17	Energy
18	Environmental management
19	Environmental policy, climate change and natural hazards
20	Health
21	Indigenous
22	Information and communication services
23	Law, politics and community services
24	Manufacturing
25	Mineral resources (excl. Energy resources)
26	Plant production and plant primary products
27	Transport
28	Expanding knowledge

The Socio-Economic Objectives are categorised according to the intended purpose or outcome of the research, rather than the processes or techniques used in order to achieve this objective. The Canadian Research Classification has largely adopted the ANZSRC SEO modified to the Canadian context.

[top of entry]

6.1 Challenges

The biggest challenge for statistical agencies is about maintaining relevance and reflecting the contemporary reality of the data about the economy and society within their statistical classifications. In a world where information and data are constantly changing the statistical classifications do not change fast enough. National statistical offices struggle to keep up and consequently have focussed on a cyclical review process for maintaining and updating classifications, whether this be on a five, ten- or fifteen-year cycle. Admittedly part of this has been due to the technology limitations of the past, the need to produce hardcopy and the resourcing required to undertake major reviews. But to a large degree those roadblocks are no longer there i.e. hardcopy has been replaced by electronic media, and technology allows faster creation and dissemination of change. However resourcing and strategic drivers are major factors in determining whether a classification gets reviewed.

Whilst ANZSRC was originally released in 2008 with a proposed five-yearly review cycle borne out of the traditional review cycle of Stats NZ and the ABS, this was with the understanding that the first revision would only be about adding or deleting categories at the lowest level and leaving more significant structural changes for a ten year cycle. However, this scoping aspect impacts on the scale of work to be done at each point. For example, a five-yearly review of issues to create change in the classification can still take six months to one year to complete. A major review on a ten-year cycle can take from 18 months to two or three years to complete. Effectively the classification is still out of step with the contemporary world situation when it is released and the longer between revisions the more issues to grapple with and resolve. It makes the revision cycle somewhat pointless if by the time the work is completed and published it only represents the year in which the revision began.

A fundamental challenge is however posed by virtue of the classification being used to represent the needs of two countries, which whilst loosely aligned, are different in terms of their population size, their economies, and governments. In achieving a consistent approach that benefits all parties a large amount of goodwill and mutual respect is required. The bigger of the two doesn’t have control or say over the final outcome as that leads to an unbalanced classification which only really is suitable for one country, which defeats the objective of a joint standard. Fortunately, the statistical agencies have a long-standing collaboration on classifications which ensures that a commonly agreed outcome is achieved.

[top of entry]

6.2 Scope and principles for the review

The aim of the ANZSRC review was to ensure that the classification reflected current practice and was sufficiently robust to allow for long-term data analysis. Increasingly the content and usefulness of the classification was seen as dated and not reflective of current practice, particularly when it came to classify and analysing higher education research and development activity and funding. It was important to ensure any revision improved coverage, coherence, and consistency of the classification. The scope of the review included all the content and structure of ANZSRC but not a relitigating of the underlying conceptual principles.

Whilst the principles for the review applied international best practice in terms of seeking mutual exclusivity, exhaustiveness, statistical feasibility and time series comparability, these principles were not strictly enforced. This is because the classification has a wider application outside of the usage by the statistical agencies and consequently the revised structure and content more appropriately reflects the external requirements. That said, a robust process and consultation process was undertaken with public participation as well as stakeholder and targeted consultation This included universities and their research offices, discipline peak bodies, users of ANZSRC data and businesses that engage in data services.

[top of entry]

6.3 Consultation process

An initial discussion paper was released in February 2019 which resulted in 237 submissions and over 2,500 issues for consideration. This consultation provided the project team with a range of issues to resolve and consider for the classification structure. All initial issues were analysed and assessed against the guiding scope and principles for the review with the intention to produce a further consultation paper with a draft classification later in 2019.

Issues were incorporated without further analysis where there were no competing suggestions or a strong majority view existed within the submissions, where the change was of low complexity and the issue was consistent with the review principles. Where these conditions were not met the following principles were then applied, in order of greatest to least priority:

1. Views from or support of representative groups of experts
2. Evidence of community of practice
3. Level of impact within the ANZSRC structure
4. Alignment with international practices and standards
5. Evidence of alignment of expertise
6. Bibliographic analysis/volume.

The consultation draft and classification was then published in November 2019 seeking feedback on the proposed changes. This resulted in over 3,500 comments on the Fields of Research (FoR) and Socio-Economic Objective (SEO) classification. No comments were received on the Type of Activity (ToA) classification.

In addition to the primary consultation on ANZSRC, consultation on indigenous research was undertaken given the need to address the lack of visibility on Aboriginal and Torres Strait Islanders (ATSI), Māori and Pacific Peoples research in the classification. Three options were put forward which were:

1. Promotion of existing six-digit codes to four-digit codes, additional codes at both levels and a list of alternative groups for these new codes
2. Creation of a new division for “Indigenous Research” with cross-referencing to other divisions
3. Creation of a new division for “Indigenous Research“.

Due to the unique knowledge domain of Indigenous research in both Australia and New Zealand, stakeholders opted for the creation of a new division for “Indigenous Research”. This outcome makes visible in research the knowledges and methodologies unique to Indigenous peoples and captures research activity that is significantly about or involves them. Whilst this outcome might potentially limit how a researcher categorised their research, the approach follows a multidisciplinary and interdisciplinary process to allow users to assign multiple codes to their research data. This gets around issues of mutual exclusivity. All codes relating to Māori research have labels in te reo Māori and English.

All detail was then finalised across the classification including explanatory detail, group and division definitions and concordance or correspondence tables for time-series mitigation. The classification was then signed-off and released on 30 June 2020.

[top of entry]

6.4 Governance model

The ANZSRC review was guided by a Steering Committee which comprised representatives from the four agencies conducting the review. The Steering Committee provided oversight for the review and was to encourage implementation and manage the project governance. A project team was set up to undertake the work and liaise with stakeholders. An Australian Expert Reference Group and a New Zealand Working Group comprising key external and internal stakeholders were established to provide technical advice, guidance, and feedback on proposed changes. These groups were advisory groups and not a formal part of the eventual sign-off process.

The governance model for the project provided a decision-making mechanism which also was created for any dispute resolution. The Steering Committee provided an initial forum for issue resolution and was to seek at all times to reach consensus on the issues it considered using its terms of reference and the memorandum of understanding established between the two statistical agencies in relation to the ANZSRC review. Where consensus could not be reached after reasonable discussion, the statistical agencies would make the final decision on the review outcomes. Effectively it fell to the project team to arrive at an agreed position on each issue, to work through any disagreements using expert advice and classifications best practice and theory, and report back to the Steering Committee with the agreed outcome. Escalation through the statistical agencies line management was not required.

On advice of the Steering Committee and through project team members in each country working with their line management, sign-off for the final ANZSRC 2020 was advised to, and rested with the Australian Statistician and the New Zealand Government Statistician.

[top of entry]

6.5 Use of ANZSRC

The use of ANZSRC ensures that research and development statistics are useful to governments, educational institutions, international organisations, scientific, professional or business organisations, business enterprises, community groups and private individuals. In New Zealand ANZSRC is used by government, funding agencies, Crown Research Institutes, universities, and independent research organisations. Similarly, in Australia ANZSRC is used by a wide variety of government, university, and other stakeholders. To facilitate the use of ANZSRC users are able to access the three component parts and the relevant correspondences from the statistical agency website: http://aria.stats.govt.nz/aria/.

The dissemination of ANZSRC is enhanced by using this website tool as there are multiple formats available for download, and application programming interfaces (APIs) which allow system to system interaction to extract content relevant to the classification. In addition, users can link to related concepts and other classifications, such as the Frascati Manual or the Canadian Research and Development classification (CRDC). Ultimately the maintenance and refresh of the classification will be done on a more dynamic process thus negating the need for future lengthy classification reviews.

Another usage of ANZSRC is in the Dimensions database [3] whereby the second level (group) is used to cover all areas of academic research for non-granular investigations into broad subject area funding [4]. Hereby ANZSRC exemplify a classification system used for both statistical and bibliographical data.

[top of entry]

6.6 Lessons learnt

Whilst the process for revising ANZSRC followed the traditional classification review process of establishing the need, clarifying the scope of the work, identifying stakeholders and undertaking an iterative top-down, bottom-up structural and content revision, it also utilised a cross-partnering inter-agency approach for the first time. In this instance the statistical agencies did not take the lead and drive the revision. The owners of the classification i.e., the Australian Research Council and the New Zealand Ministry of Business, Innovation and Employment initiated, scoped and led the revision. These statistical agencies provided a process and methodology for undertaking a classification revision, provided best practice expertise, and enabled a level of convergence and consistency with other statistical classifications.

This model worked really well, made good use of resources, knowledge, and in particular allowed the project to succeed via the existing ARC and MBIE stakeholder and user relations in each country. Having the statistical agencies lead and run the project management may well have led to a duplication of effort and greater difficulties in obtaining user buy-in to the revision. The ANZSRC model is now a more preferred approach in New Zealand i.e. divesting responsibility to key external subject matter agencies to lead with statistical support. The model is now being applied in the decision-making process around the future of the Australian and New Zealand Standard Classification of Occupations (ANZSCO) and how New Zealand might take a different pathway but within the lessons learnt, from and the governance model applied to ANZSRC.

The ongoing challenge however, from the New Zealand perspective for joint statistical classifications is that much of the decision-making on when and how a classification review is undertaken is determined by Australian needs. As noted in section 1 of this paper, MBIE in 2016, in conjunction with Stats NZ, the New Zealand Ministry of Education, and Tertiary Education Commission released the 2016 Research, Science and Innovation Domain Plan. This domain plan highlighted the need to revise ANZSRC and lead to a wider engagement between the ABS and Stats NZ on the how and why for a classification review. This then led to greater internal and trans-Tasman collaboration, building on the existing collegial relationships between the two standards and classifications teams, but expanding it wider and across and between both agencies. This has had a flow-on effect for other engagement between the statistical agencies, particularly in regional contributions to international standards, and also in the Pacific.

ANZSRC continues to provide a useful case study on the practicalities of undertaking a classification for the right reasons, that there is more than one way to deliver an outcome and that collaboration and building on past experiences is an invaluable guiding tool going forward.

[top of entry]

7. Conclusion

The revision of any statistical classification is generally a time-consuming and expensive process for national statistical offices, hence the cyclical nature of the revision process. That said, the revision of the Australian and New Zealand Standard Research Classification (ANZSRC) has led to an updated, accurate statistical classification, which has sufficient robustness to allow for long-term usage and implementation. There are significant changes to the structure and content within the classification from that which was originally published in 2008 but that reflects reality. ANZSRC (2020) provides the Australian and New Zealand research community with a classification that is contemporary and reflects the changes in research practices that have occurred in the intervening years since the first release of the joint classification. A high level of engagement from the research communities combined with the collaborative inter-agency project team enable a successful outcome to be achieved. Coupled with a clear revision process, governance model and robust stakeholder consultation, the application of sound principles and ensuring that stakeholder voices were listened to has resulted in a classification which is fit for purpose and which enables the production of better quality R&D statistics to meet national and international requirements.

ANZSRC 2020. Australian and New Zealand Standard Research Classification (ANZRC): A statistical classification used for the measurement and analysis of R&D in Australia and New Zealand. Belconnen, Australia: Australian Bureau of Statistics. https://www.abs.gov.au/statistics/classifications/....

Australian Bureau of Statistics. 1998. Australian Standard Research Classification (ASRC). 1998. https://www.abs.gov.au/ausstats/abs@.nsf/0/2D3B6B2B68A6834FCA25697E0018FB2D?opendocument.

Barité, Mario. 2018. ‘Literary warrant”. Knowledge Organization 45, no. 6: 517-536. Also available in ISKO Encyclopedia of Knowledge Organization, ed. Birger Hjørland, coed. Claudio Gnoli. https://www.isko.org/cyclo/literary_warrant

Curtis, Asher, Sarah E. McVay, and Sara Toynbee. 2020. “The Changing Implications of Research and Development Expenditures for Future Profitability”. Review of Accounting Studies 25, no. 2: 405–37. https://doi.org/10.1007/s11142-019-09528-6.

Hancock, Andrew. 2013. Best Practice Guidelines for Developing International Statistical Classifications. United Nations Department for Economic and Social Affairs. Statistical Division. ESA/STAT/AC.267/5 of 6 May 2013. Available at: https://unstats.un.org/unsd/classifications/bestpractices/Best_practice_Nov_2013.pdf.

Hancock, Andrew. 2017. Approval Process for an International Statistical Classification. Meeting of the Expert Group on International Statistical Classifications New York, 6-8 September 2017. New York: United Nations Department of Economic and Social Affairs, Statistics Division. https://unstats.un.org/unsd/classifications/bestpractices/Approval_Process_Sep_2017.PDF.

Hjørland, Birger and Claudio Gnoli. 2022. “Research Classification System“. Preliminary editorial placeholder article. In ISKO Encyclopedia of Knowledge Organization, eds. Birger Hjørland and Claudio Gnoli, https://www.isko.org/cyclo/research.

ISO/IEC Guide 2, 1996: Standardization and related activities – General vocabulary. https://www.iso.org/standard/24887.html (This standard has been revised by ISO/IEC GUIDE 2:2004, https://www.iso.org/standard/39976.html.)

Legendre, Ariadne. 2019. “The Development of the Canadian Research and Development Classification”. Knowledge Organization 46, no. 5: 371-379. Also available in ISKO Encyclopedia of Knowledge Organization, eds. Birger Hjørland and Claudio Gnoli, https://www.isko.org/cyclo/crdc.

Millerand, Florence and Geoffrey C. Bowker. 2009. “Metadata Standards: Trajectories and Enactment in the Life of an Ontology”. In Standards and Their Stories: How Quantifying, Classifying, and Formalizing Practices Shape Everyday Life, edited by Martha Lampland and Susan Leigh Star. Ithaca, NY: Cornell University Press, 149–165.

Ministry of Business, Innovation and Employment, Statistics New Zealand, Ministry of Education and Tertiary Education Commission, New Zealand. 2016. 2016 Research and Innovation Domain Plan. https://www.mbie.govt.nz/assets/....

National Science Foundation. National Center for Science and Engineering Statistics. 2022. Definitions of Research and Development: An Annotated Compilation of Official Sources. https://ncses.nsf.gov/pubs/ncses22209.

Netherlands Data Archiving and Network Systems, National Academic Research and Collaborations Information System (NARCIS). https://www.narcis.nl/about/Language/en.

OECD. 2015. Frascati Manual Guidelines for Collecting and Reporting Data on Research and Experimental Development. https://www.oecd.org/publications/frascati-manual-2015-9789264239012-en.htm.

OECD. 2018. Oslo Manual Guidelines for Collecting, Reporting and Using Data on Innovation, 4th Edition. https://www.oecd.org/science/oslo-manual-2018-9789264304604-en.htm.

Scharnhorst, Andrea and Peter Doorn (eds.) 2019. “Special Issue: Research Information Systems and Science Classifications; including papers from Trajectories for Research: Fathoming the Promise of the NARCIS Classification”. Knowledge Organization 46, no. 5: 337-97. doi.org/10.5771/0943-7444-2019-5.

Statistics Canada, Canadian Research and Development Classification (CRDC). 2020. https://www.statcan.gc.ca/eng/subjects/standard/crdc/2020v1/index.

Statistics New Zealand, Australian and New Zealand Standard Classification of Occupations (ANZSCO). 2006. http://aria.stats.govt.nz/...

Statistics New Zealand, Australian and New Zealand Standard Industrial Classification (ANZSIC). 2006. http://aria.stats.govt.nz/...

Statistics New Zealand, Australian and New Zealand Standard Offence Classification (ANZSOC). 2008. http://aria.stats.govt.nz/...

Statistics New Zealand, Australian and New Zealand Standard Research Classification (ANZSRC). 2020. Fields of Research: http://aria.stats.govt.nz/a...

Statistics New Zealand, Australian and New Zealand Standard Research Classification (ANZSRC). 2020. Socio-Economic Objectives: http://aria.stats.govt.nz/...

Statistics New Zealand, Australian and New Zealand Standard Research Classification (ANZSRC). 2020. Type of Activity: http://aria.stats.govt.nz/...

United Nations. 2022. The Sustainable Development Goals Report 2022. https://unstats.un.org/sdgs/report/2022/.

United Nations Educational, Scientific and Cultural Organisation. 2011. International Standard Classification of Education (ISCED). https://unstats.un.org/unsd/classifications/Family/Detail/1045.

United Nations Fundamental Principles of Official Statistics. 2014. https://unstats.un.org/unsd/dnss/gp/fundprinciples.aspx.

United Nations Guidelines for the Template for a Generic National Quality Assurance Framework (NQAF). Prepared by the Expert Group on National Quality Assurance Frameworks. 2012. https://unstats.un.org/unsd/statcom/doc12/BG-NQAF.pdf.

United Nations Statistics Division. 2007. International Standard Industrial Classification of All Economic Activities (ISIC Rev.4). New York: United Nations. https://unstats.un.org/unsd/classifications/Family/Detail/27.

United Nations Statistical Division. 2012. Guidelines for the Template for a Generic National Quality Assurance Framework (NQAF). https://unstats.un.org/unsd/statcom/doc12/BG-NQAF.pdf.

United Nations Statistics Division. 2015. Central Product Classification (CPC) V2.1. Series M: Miscellaneous Statistical Papers, No. 77 Ver. 2.1, New York: United Nations. https://unstats.un.org/unsd/classifications/Family/Detail/1074.