Enumerating Australia’s “diverse”: ethnicity and raciology in census and workplace diversity surveys

Introduction: what is Australia’s “ethnicity”? Footnote¹

In June 2022, Australia’s Minister for Immigration, Citizenship, and Multicultural Affairs, Andrew Giles, announced that from 2026, the Australian Census, administered by the Australian Bureau of Statistics (ABS), would enumerate Australians by “ethnicity.” Until 2021, the five-yearly Census asked Australians their “ancestry” – a term the new decision’s advocates argue lacks precision for allowing respondents to choose unrelated descriptors, including national labels like “Australian” (Wales Citation2022). Such advocates, including Mohammad al-Khafaji of the Federation of Ethnic Community Councils of Australia (FECCA), have argued for this change in progressive, social justice terms (Lattouf, Malhi, and al-Khafaji Citation2022). In these terms, improving the “ethnic” legibility of Australia’s population will support state and other agencies to design more inclusive services and address inequalities that complicate Australia’s claim to be a successful multicultural society (Yussuf and Walden Citation2022). Other proponents, such as Allen and Fozdar, argue that in addition to such benefits, the new measure will contribute to an improved sense of inclusion across the Australian community (Allen Citation2021a; Citation2021b; Fozdar Citation2022). This “sense” would follow the measure by enabling Australians to “see themselves” reflected in the nation’s data, and by extension, in its institutions – enabling better service provision and the growth of an inclusive, pluralist, civic nationalism (Chang Citation2023). According to Soutphommasane, the resulting inclusion will support diverse Australians to contribute to Australia’s “national success and prosperity” (Soutphommasane Citation2018).

In such arguments, advocates present ethnic enumeration as an exercise in gathering “data” – true and incontestable evidence reflecting the nation’s diversity that will support policy advocacy for improved social justice outcomes. In this vein, the Australian Government will seek to steer the exercise through a working group of appointed “data collection and demography experts” and multicultural advocates, who it claims will create “whole of government standards for measuring Australia’s diversity.” Such claims signal that the working group will write the new census question and steer how the resulting data is treated, including by other agencies that will replicate the ABS’ new method (Wales Citation2022; Yussuf and Walden Citation2022).Footnote² In the process, the working group will engage with a genuine analytical problem, in which multicultural advocates know, but lack data to demonstrate, that racial inequality exists in Australian society, because national datasets do not gather it. Jakubowicz cites Australia’s Notifiable Diseases Database (NDD) as an example of this problem, as COVID-19 or SARS-COV-2 likely infected recent arrivals from South Asia disproportionately. This outcome is likely due to their concentration in “essential” employment sectors like hospitality, cleaning, security, and delivery driving. The database, however, does not record patients’ backgrounds, denying Australian policymakers a clear understanding of Australia’s “political economy of ethnic groups,” by which members of certain groups are “clumped into certain occupations” (Jakubowicz Citation2020).

The decision’s advocates are arguing in good faith. Their arguments, however, raise serious problems that they and the government should address before proceeding. Most importantly, “ethnicity” defies categorisation and advocates, who assume “ethnicity” is an uncontroversial category, are in fact on unstable definitional ground. Advocates do not acknowledge, for example, that “ethnic” enumeration cannot be distinguished from “racial” classification, nor can claims about “culture” that are built on such schemas – although their own practice of conflating such terms should signal clearly that this problem exists (Brubaker Citation2016, xiii, 2; Chang Citation2023; Yussuf and Walden Citation2022). Further, advocates also present specifically “ethnic” enumeration as the best choice for Australian society, despite the proliferation of ethnic categories that Australians use to describe themselves, not all of which can encapsulate clear single, or even dual, affiliations. This proliferation suggests that Australians’ ethnic illegibility is not simply a data problem but a reflection of their genuine “superdiversity” (Vertovec Citation2022). In addition, proponents ignore long-standing traditions of critical study in sociology and history showing that census operations do not reflect existing social categories but rather, create, socialise, and reify them, racialising populations and interpellating “ethnic” groups in the state's own interest. As a great proportion of Australia’s “diverse” population can attest, the categories such states created in Asia now feed political disputes that cannot be resolved, and structure their experiences even after they migrate elsewhere. Instead of recognising these problems, however, proponents limit their comparative horizons to other white majority nations, eliding their colonial settler histories while also ignoring the histories of former European colonies. Indeed, social justice is not the only rationale for categorising diverse populations by ethnicity – a term often used to obscure the issue of race even as debates about classification reinforce the persistence of “raciology,” in Gilroy’s terms (Gilroy Citation2000).

Instead of pushing forward, therefore, advocates should pause to study existing attempts to work through these problems. One such attempt is the Diversity Atlas (DA) tool created by the Australian social enterprise Cultural Infusion, which has recently failed to resolve them. Regardless, it is used in enumeration exercises in Australian workplaces, such as the “Counting Culture” initiative managed by the Diversity Council of Australia (DCA) (DCA Citation2022). DA’s experience provides crucial lessons for the census, including the possibility that the ABS’ existing methodology – of using the broad term “ancestry” to acknowledge the slipperiness of “ethnic” categories – is the least objectionable solution available.

Why this question now?

After erasing much of its Indigenous diversity through “white” settlement, Australian society is now diverse to an extent not imagined when it abolished its White Australia Policy (WAP) in the 1960s, and created its first multicultural policies in the 1970s (Carey and McLisky Citation2009). These policies envisaged that Australia’s white majority – membership of which was sometimes open to others with “aspiration[s]” to cultural dominance – would learn to live alongside and tolerate non-white migrants (Hage Citation2000, 20). These migrants would be permitted to maintain their cultural practices, which would enrich Australian society and culture, on the implied basis that together, they constituted a small demographic minority overall. This minority, in turn, would consist of several discrete and distinct ethnic groups, each with its own culture and interests. The Australian Government would manage these groups’ claims by consulting with recognised “community leaders” from within their ranks – a notion that has informed FECCA’s work as Australia’s peak ethnic advocacy body since its foundation in 1979. Since then, FECCA has structured and mediated often clientelist relationships between successive Australian Governments and “ethnic communities,” helping sustain the illusion that Australia’s minorities select “representatives” to negotiate outcomes with the state (FECCA Citationn.d.; Kwok Citation2008). Insofar as it represents minorities, however, FECCA also limits the space for representation within its own consultative structure, keeping white majority privileges safely apart from minorities and their claims. In this way, Australian multiculturalism sections off the work of managing minorities as a special project of statecraft, while the desirability of minority inclusion remains a question open for debate at any time. Whether white Australians support or oppose diversity is a second order concern, as both “good” (tolerant) and “bad” (racist) white nationalists should realise that either way, they remain in control (Hage Citation2000).

The assumptions underlying this sense of control, however, have eroded as Australia has diversified enormously in the twenty-first century. Non-white migration continued through the 1990s, sparking occasional moral panics about “Asians,” which, as Australia’s Asians were recast as professional achievers, gave way to new debates about refugees and Muslims, especially during the “War on Terror.” Yet even as asylum seeker policy became more punitive from the 2000s, the volume and composition of “regular” migration increased and diversified, so that net overseas migration has since been Australia’s leading cause of population growth (Gothe-Snape Citation2018). In this context, Asian migration has boomed, facilitated by Australia’s international student intake. Reflecting this boom, at the census in 2021, 27.6 per cent of Australians reported they were born overseas, while almost half, or 48.2 per cent of Australians, reported at least one parent born overseas. Within this cohort, annual arrivals from China, India, and Southeast Asia now surpass those from “traditional” sources of mostly white migration such as the United Kingdom (UK) and New Zealand (NZ). As a result, numbers of Australians who report India and China as their birth countries now exceed the number from NZ, leaving only Australia and England with greater population shares (ABS Citation2022a; Citation2022b). Meanwhile, as refugee arrivals have also continued despite the bitter debates, Australia is now also home to more “multiple-origin, transnationally connected, socio-economically differentiated and legally stratified” refugees than ever before (Vertovec Citation2022, 6). Many Afghans, Uyghurs, and Sri Lankans, for example, arrived in Australia as refugees on temporary visas, often passing through third countries and serving periods of detention on arrival. Due to these developments, like other historically white nations transformed by migration, Australia is now increasingly “superdiverse,” a condition Vertovec argues follows not only increased diversity but also the “diversification of diversity.” This phenomenon refers to growing numbers of identity groups – based on ethnicity, religion, shared cultures, languages, citizenships, and migration histories for example – as well as increasing diversity within these groups themselves (Vertovec Citation2022, 9).

Australia’s diversifying diversity is obvious in its cities, and Australians carry layers of nested, multiple, and sometimes contradictory, identities, contained within overlapping and intersecting categories of difference that state agencies struggle to disentangle. Accordingly, the state and its ethnic intermediaries no longer manage interactions between diverse Australians. Instead, these interactions are more organically “intercultural” than in the past, in the sense that identity groups cannot operate within self-contained clusters, with “leaders” speaking for them. As newer and older generations of non-white migrants live and work in close contact with each other – and with white Australians – members of all groups have had little choice but to become more adept at navigating their similarities and differences. Many Australians are supported in this effort by experiences of growing up in diverse nations overseas, and/or by decades of multicultural education in Australian schools, which has grown more sophisticated, albeit unevenly, since its introduction. Accordingly, many schools have moved on from simply hosting “spaghetti and polka” days to mainstreaming theories of interculturality, including those which deconstruct ethnic categories and their boundaries, alongside the notion of white identity and its privileged majority positioning (Watkins and Noble Citation2021, 2). In addition, in line with their rising numbers and population share, Australia’s “diverse” are increasingly politically assertive and economically influential. Holding a growing and hard-fought share of the nation’s social capital, they are pushing power holders to work harder to cater to their needs and aspirations, fuelling growth in the diversity, equity, and inclusion (DEI) industry and creating a new electoral calculus in many parliamentary seats.Footnote³

These developments have pushed leaders in state and other agencies to work harder to “read” Australia’s identity transformation, only to find their efforts frustrated by the genuine complexity of categories and the resulting confusion about leaders, numbers, and population shares. These leaders continue to seek out ethnic intermediaries, but they often cannot determine which constituencies they speak to, or whether they enjoy broad-based legitimacy among members of the identity groups they purport to lead. Footnote⁴ Further, owing to growing differentiation within many identity groups, established “leaders” may not understand others’ experiences or views, and can propound essentialist claims about “their” communities, while “new” leaders can also create embarrassing incidents for credulous audiences. Footnote⁵ It is, therefore, increasingly difficult for institutions to claim legitimacy via consultations with ethnic leaders, and some have responded with top-down attempts to create new ethnic identities such as “Asian Australian,” along with new leaderships to speak for them. These attempts also lack clarity, however, especially in terms of who is included and why they should subsume their complex affiliations within such broad labels.

Why target the Australian census?

The Australian Census is an important institutional target for the multicultural policy and advocacy sector, for three reasons. First, as the nation’s leading data agency, the ABS could quickly set an example for other organisations by adopting the working group’s “national standards.” As a result, the NDD and other datasets like it could also incorporate the new ethnicity question into its surveys as part of a “whole of government” approach. Second, the census decision “aligns” Australian data practices with important external precedents, including in countries the Australian Government frequently uses as comparators, namely the UK and its other settler colonies. Indeed, Giles flagged his new approach to the Australian Census at the 2019 FECCA Conference in exactly these terms while serving in opposition, pointing out that the UK, the United States (US), Canada, and NZ all have ways of enumerating their populations by ethnicity and/or race (Davidson Citation2019). In these countries, census forms ask people directly if they belong to one or more ethnic and/or racial categories. Census agencies then report on population aggregates with labels like “Asian,” “Latino,” “Black,” or “Māori,” who are sometimes grouped together under more expansive labels such as “Asian and Pacific Islander” (AAPI) or “Pacific Peoples,” as required for comparative purposes. In other cases, such large categories are aggregated further still, into hold-all containers for all non-white people, such as “Black, Indigenous, and People of Colour,” (BIPOC), “Black and Minority Ethnic (BAME),” “Middle Eastern/Latin American/African (MELAA),” and so on. These categories have developed to match the logics employed by the states that use them, not logics of sameness or difference that matter to members of these categories themselves, which vary enormously.

Third, the data the Australian census generates does not support the multicultural sector’s efforts to make the nation’s diverse population legible to its leaders. As a result, Australian advocates cannot emulate social justice advocacy methods popular in the UK and its other settler colonies, whose proponents argue that categories like “BIPOC,” “MELAA,” or “BAME” are essential to demonstrating the persistence of unequal social outcomes. Australian advocates promoting ethnic enumeration are also working towards a similar category to hold all groups they consider non-white. Soutphommasane, for example, has noted that without one, it is complicated to deduce how many non-Indigenous Australians have a “non-European” [that is, non-white] background, making it difficult to identify how they might experience racism (Soutphommasane Citation2018, Citation2016).Footnote⁶ The DCA is promoting a purported solution to this problem, namely by creating a new category for disaggregating non-white Australians from the older label many agencies use now, that is, “Culturally and Linguistically Diverse.” This label includes southern and eastern Europeans descended from the post-war immigration boom that stretched the boundaries Australia once placed around “whiteness,” before it ultimately abolished the WAP. These Australians, however, have arguably since “graduated” into whiteness in comparison with later groups of migrants who arrived after the WAP was abolished (Bahr Citation2023). Accordingly, the DCA is proposing “Culturally and Racially Marginalised” as Australia’s new hold-all non-white category (Mapedzahama et al. Citation2023).Footnote⁷

Certainly, the ABS’ “ancestry” question has not provided multicultural advocates with the data they require to make their desired comparisons. The only current census category that functions as “ethnoracial” – the term Brubaker prefers because ethnicity and race cannot be separated – is “Aboriginal and Torres Strait Islander.” Beyond this category, it is impossible to precisely quantify second-, third-, or sixth-generation non-white migrant Australians, especially if they were born here and speak English at home, leaving no statistical trace of their non-white heritage beyond their “ancestry” answers (Chang Citation2023). There is no certain way of predicting if those answers correspond normatively with their “ethnicity.” Further, it is not clear which marker of identity one should prioritise when naming an “ancestry” – country of birth, preferred language, religious identification, surname or kinship group name, “race,” or indeed any other. That determining what “ancestry” means is impossible is clear from the very list of examples provided alongside the question, which asks people to nominate up to two answers. The form explicitly lists the categories “English,” “Irish,” “Scottish,” “Italian,” “German,” “Chinese,” and “Australian,” the last of which allows respondents to opt out of locating their origins overseas. Alternatively, respondents can write in any other answer they prefer, albeit likely influenced by the list of examples provided, including “Greek” (both a national and ethnic designation), “Kurdish,” and “Hmong” (ethnic identities that do not correspond to a single associated nationality). It also offers “Australian South Sea Islander” (a geographical term spanning diverse ethnicities). Respondents also can, and do, nominate national (such as “Pakistani,” “Indonesian,” or “South African”) and religious (such as “Sikh”) labels as well as “ethnic” ones (such as “Kadazan,” “Tamil,” or “Zulu”) (ABS Citation2022b). It can therefore be complicated to disaggregate white from non-white South Africans, nor is there a reason for Bangladeshi British respondents not to enter “British,” or for people born in NZ with grandparents from Vanuatu not to write “New Zealand” instead of ni-Vanuatu (Aspinall Citation2007, Citation2012) (Figure 1).

Figure 1. The “ancestry” question in a sample Australian Census form (ABS Citation2023).

Display full size

As census data does not support the comparisons advocates hope to make, it is not surprising that the multicultural sector is the group most animated by the “diversity data” problem. This sector aside, political parties, too, are finding it increasingly difficult to determine which voters they should prioritise for advocacy work and electoral messaging. The contest for non-white voters is intense, and truisms from the 1990s, when the Australian Labor Party (ALP) presented itself as the “party of multiculturalism,” are no longer credible. “Diverse” voters also support Labor’s competitors now, and indeed, 2021 polling by Resolve Political Monitor found that 44 per cent of “those identified as non-Anglo Saxon [that is, non-white]” voted for the Coalition at the 2019 election, while only 31 per cent voted for Labor (Crowe Citation2021). In another example, before Australia’s 2022 election, the Indian Link newspaper polled its readers to find their loyalties were evenly split between Labor and the Liberal-National Coalition (Luthra and Luthra Citation2022). Recognising the now competitive nature of “ethnic politics,” parties are working to modernise their approaches, including by testing ways to identify (and mobilise) second- and third-generation, non-white constituencies. Labor, for example, has established a Multicultural Engagement Taskforce and is investing in constituency-building work among Asian Australians, including in partnership with The Australian National University (ANU) (ALP Citationn.d.; ANU Citationn.d.). Here, its outreach functions through a Centre for Asian Australian Leadership (CAAL) led by Jieh-Yung Lo, previously a staffer to former ANU Chancellor and Labor Foreign Minister Gareth Evans. In its search for data to deploy, CAAL collaborates with the ANU Centre for Social Research Methods, which conducts polling at regular intervals. This approach suggests that Labor is ready to move on from older methods, including scanning the electoral roll for supposedly “ethnic” names or using social media campaigns to target voters via awkward gestures (Malhi Citation2017; Rachwani Citation2022a).

Until Labor’s victory at the 2022 election, however, the ABS was unwilling to entertain its criticisms, arguing instead that its previous methodology allowed Australians to enter labels they found most salient, showing their “active association” with their own preferred categories (and not others). Further, this methodology allowed the ABS to adopt a definition of “ethnicity” not as a single, clearly identifiable trait that is innate in individuals, but a socially constructed, “multi-dimensional concept” pertaining to a shared identity based on mutable characteristics. These characteristics might include “a long, shared history,” “a cultural tradition,” “a common geographic origin,” “a common language,” “a common religion,” “being a minority,” and/or “being racially conspicuous.” In addition, the ABS already sorts Australians’ “ancestry” answers into ethnicity-like categories, albeit behind the scenes, using a database of identity labels known as the Australian Standard Classification of Cultural and Ethnic Groups (ASCCEG) (ABS Citation2019). Scholars of enumeration have pointed out that there are various problems with such methods, and yet at the same time, they are not uncommon (Aspinall Citation2007; Citation2012; Bochsler et al. Citation2021).

Cultural infusion’s diversity atlas as a lesson for the census

The decision to enumerate Australians by ethnicity will reduce the flexibility afforded both to respondents and to the ABS. It will require all parties to understand the term “ethnicity,” along with how it differs from any other identity category the census has previously permitted. In determining how to proceed, the decision’s proponents should therefore take stock of previous experiences, including in the case of Diversity Atlas (DA), an enumerative platform released in 2017 by the Melbourne social enterprise Cultural Infusion (CI), led by CEO, Peter Mousaferiadis. CI, which Mousaferiadis founded in 2003, has typically worked with schools and youth organisations, mostly in Australia and Asia, to promote intercultural experiences through exposure and involvement in music and the arts. More recently, CI has also developed a line of gamified educational technology products that teach world cultures, for which it has won a United Nations Alliance of Civilisations Intercultural Innovation Award and recognition as a UNESCO partner. In 2015, Mousaferiadis became a board member of the Ethnic Communities’ Council of Victoria, a position that placed CI at the heart of debates around diversity and multicultural policy, in Australia’s fastest growing and diversifying capital city. These debates have informed CI’s entry into the enumeration technology marketplace.

DA builds on the premise that Australian organisations are increasingly under pressure to adopt, demonstrate their commitment to, and improve their performance against, DEI targets across their workforces, while also responding to the growing superdiversity of customers and clients. Aiming to support organisations to respond to these conditions, CI is pitching DA as a “unique diversity data-analysis platform that provides insight into cultural and demographic diversity within an organisation. It enables organisations to understand the diverse richness of their teams and to better measure, understand, acknowledge, and act on their diversity, inclusion, and development strategies” (Cultural Infusion Citationn.d.b). Further, CI argues that using DA to improve staff diversity will enable them to better match and reflect the diversity of their external environments. By using DA, CI argues, organisations will be able to adopt a “continuous improvement” approach to diversity, progressively improving their performance when assessed at reviews that their management teams would conduct at regular intervals. Ultimately, they should reach an end point that CI refers to as “workforce mutuality,” or a condition in which “the diversity of an organisation or a sector reflects the diversity of the community served by the organisation” (Cultural Infusion Citationn.d.a). By implication, therefore, although CI does not market DA for this purpose, the platform could also measure external diversity to support mutuality assessments. In short, DA could enumerate sample groups from audiences, markets, and constituencies in a manner that foreshadows the direction of the government’s census working group, as CI has effectively pre-empted Giles’ decision.

Accordingly, DA flips the ABS’ previous approach – from using a tool like ASCCEG to code respondents’ varied “ancestry” answers, to socialising a fixed notion of “ethnicity” directly with the Australian public and asking respondents to work with it. Indeed, DA includes an “ethnicity” question alongside a range of others, such as:

1. Country of birth

2. Parents’ countries of birth

3. Grandparents’ countries of birth

4. Languages spoken (i.e. this question is not limited to “language spoken at home”) and

5. Religion or “worldview” (to account for secular orientations), an idea that DA elaborates by offering a range of political and philosophical labels.

In addition to these categories, DA can also ask respondents to volunteer their “race,” although CI intends to offer this option only in markets where the use of this category is normalised and consistent with existing statistical practices. Importantly, CI has assumed that respondents can conceptually distinguish between these categories, and to assist them to do so, DA guides their choices more actively than the ABS has. Accordingly, instead of allowing free text entries, DA offers respondents drop-down menus containing pre-loaded answers, delineating country of birth (say, “India”), from ethnicity (“Punjabi,” for example, but not “Sikh”), from language (possibly “Hindi,” but not “Indian”), and religion (“Sikh,” perhaps, but not “Punjabi”). To pre-empt criticisms that the menu is stifling respondents’ choices, or offering them too few or perhaps inappropriate answers, CI has elected to include as many answers as possible in each menu. As a result, if one’s parents or grandparents were born in Rhodesia, British India, or the Ottoman Empire, and not Zimbabwe, Pakistan, or Türkiye for example, these options are available for selection, along with a staggering number of ethnic, religious, and linguistic labels. Indeed, in 2020, DA offered almost 8,000 language categories, 10,000 ethnic groups, and 650 religions and “worldviews.” CI sourced these categories from a wide range of existing databases – created by organisations as diverse and varied in credibility as Ethnologue and the United Nations – which it spliced together, removing duplicate labels. According to CI, this method allows it to represent each respondent’s profile more “accurately” than the ABS has to date, as its questions do not conflate identity categories such as language or nationality with others such as ethnicity or religion.

Once respondents have entered their data, organisations can analyse it via charts arranged on a central DA dashboard. The display is attractive and intended to be easy to understand, including by organisational managers or human resources professionals with no specific training in culture, history, languages, or politics. According to CI, such professionals should nevertheless be able to visualise the entire organisation as well as specific departments, enabling them to see if all their “diverse” staff are in accounts and information technology, while all their policy managers are white, and so on. Further, CI intends for these displays to be attractive to children of all ages who might be taking social studies, Asia Pacific Studies, or world culture subjects at school. The attractiveness of DA’s visualisations means they also appeal to adults, including in DCA’s workplace diversity surveys and at conferences, where they offer participant cohorts glimpses of their own diversity.

Steeped in raciology

These glimpses, however, are illusions. They owe their attractiveness to the way they speak to top-down, statist impulses to impose order upon the “superdiverse.” Indeed DA – like Giles’ census decision – responds to the complex, shifting nature of identity by seeking to stabilise categories that innately resist stabilisation, requiring elaborate contrivances to distinguish and essentialise them. Determining which contrivances are acceptable for use in Australian society is a problem the Australian Census working group – and multicultural advocates – should recognise and acknowledge, including by engaging critiques of enumeration, which they have ignored so far. Brubaker, for example, has critically discussed such impulses in the context of a more general contemporary mood felt across Britain and its settler colonies. In these increasingly superdiverse societies, ethnoracial categories are experiencing a “moment of cultural flux, mixture, and interpenetration” linked to “burgeoning discussions of hybridity, syncretism, creolisation and transnationalism” (Brubaker Citation2016, 5). At the very same time, and perhaps as a result, they also host re-invigorated forms of “boundary work,” in which appeals to such categories – whose boundaries are often portrayed as “natural,” given, essential, and objective – are gaining in salience (Brubaker Citation2016, 31–39). The result is that even as these boundaries shift, their proponents also frame them in a positivist or “objectivist” light, as simple facts of “data,” enabling their renewed and uncritical use not only as social constructions, but also as social “facts.” These categories are, as Bourdieu has argued, “adjusted to the divisions of the established order (and, therefore, to the interests of those who dominate it)” – and not only by states but also by activists, as Brubaker has pointed out. And yet, precisely as Bourdieu has criticised, those seeking to represent complex social realities should not uncritically reproduce the structures and categories that have underpinned local and global inequalities (Bourdieu and Wacquant Citation1992, 29–30).

Instead of engaging such critiques, however, Australian advocates have presented ethnic categories as essential to improving the government’s diversity statecraft, including, for example, by claiming that agencies could read Australians’ food preferences from their census data (Chang Citation2023). This framing enables policy-realist arguments that appeal to positivist sensibilities, which privilege “data-driven” solutions without interrogating the assumptions that inform the creation of the data in question. The way DA coaches respondents to answer its questions, however, demonstrates that many such assumptions are profoundly unsound, as they rely on constant boundary work to maintain their coherence. Indeed, by committing to maintain so many categories and boundaries between them, DA is instituting a form of “raciology” – as Gilroy has called the “ontology of race,” or a “distinctive regime of truths [and] discourses” fixing respondents’ “irrevocable otherness” (Gilroy Citation2006, 185–186). It also affirms an indefensible connection between raciology and statecraft by performing the ontological groundwork for national agencies to shape and structure future knowledge about the nation’s population and its purported ethnic fault lines and cleavages. CI has received consistent market and adviser feedback that its category boundaries cannot fail to be contested, and in response, it has repeatedly reaffirmed its commitment to categorisation, resulting in enormous category lists that might in fact make no sense to users. For example, in its decision to offer as many categories as it can, DA offers “Pakistani Punjabi” as an option to Punjabi speakers. Yet the spoken Punjabi dialect continuum is not clearly organised in terms of the 1947 India-Pakistan border (Masica Citation1993). The result is that DA proffers a national category as a linguistic one, undoing its efforts to distinguish categories of difference from each other.

Additional category problems also arise. One such problem is that of nested identities. DA cannot address categories that exist both within other groups and outside them, such as those that nest within the term “Malay,” which has a different meaning in Malaysia than it does in Indonesia. In Malaysia, the term defines an aggregate “race” category in which many ethnoracial groups – such as Javanese or Bugis – must slot themselves, despite not doing so in Indonesia, where each group holds a separate ethnic identity, distinct from that held by “Malays.” How should “Malay” Javanese born in Malaysia refer to themselves in Australia – as “Malays,” or as “Javanese,” as Javanese born in Indonesia might prefer to do? Further, as members of “Indonesian” identity groups list themselves as “Sundanese,” “Madurese,” or “Acehnese,” for example, what does the residual category, “Indonesian,” mean? What of the Sundanese who recognise this ethnic identity at family and community gatherings, but who elect to list “Indonesian” on their official documents (see Cribb and Narangoa Citation2004)? Further, what of nested categories that some, but not other, users might contest? For example, how will DA treat the category “Ahmadi” – as a sub-category of “Islam,” potentially upsetting some Muslim users, or as a separate religion, potentially upsetting some Ahmadis? How will it treat Sikhs – as a separate group or as a sub-set of Hindus, as many Hindu nationalists claim they are? And how will it address religious identities that are also emerging as “ethnicities,” such as “Sikh,” “Muslim,” and “Hindu,” as some members of these groups assert or demand (Evans et al. Citation2023; Sandhu Citation2021; Kaur Citation2007, chap. 8)? How will it categorise “Fiji Indians,” many of whom identify themselves as a distinct group from “Indians” at census time – as Asians, or as Pacific Islanders (RNZ Citation2021)? What of emerging identities, such as “Hongkonger,” which some activists argue should be distinguished from “Chinese” (Handley and Rizmal Citation2021)? Is “Chinese” an ethnicity at all, or a national descriptor? Moreover, what of potential melding or separation? For example, the category “Asian Australian” might become salient over time, and it could begin to function as an “ethnic” aggregate category. If it does, who will be Asian Australian? There is a chance the label might include only East Asians, as Australians often conflate these categories. Meanwhile they often consider South and West Asians as distinct and separate groups from other Asians, a perception likely fuelled by the government’s adoption of the term “Indo-Pacific” to describe much of Asia in its foreign policy discussions. Further, what about hybrid identities? For example, an “Asian Australian” might also recognise their ethnicity via any combination of national, ethnic, and linguistic labels, any (or none) of which might be salient in their circumstances at any given time. In what circumstances might a respondent pick “Asian” instead of “Khmer” or “Bhutanese?” Why?

As these questions show, DA seeks to flatten an enormous variety of identity categories, that each become salient in specific historical, political, and locational contexts, into a series of free-floating and decontextualised indicators of difference. Indeed, as a purportedly universal tool, local and expert contextual knowledge is precisely what it cannot offer, causing it to reflect the problems of “indicator culture” that Merry has criticised, in which a false trust is built in the “capacity of numbers to render different social worlds commensurable” (Merry Citation2016). Ultimately, CI’s attempts to impose order on the thousands of categories cannot lead to a decisive or permanent resolution, and CI could benefit from testing other options, such as allowing respondents to write in their own preferred identity labels, so that answers simply reflect how respondents see themselves at any given point in time. Such an approach would, of course, bring DA closer to the Australian Census’ “ancestry” approach until 2021, which Aspinall points out reflects the direction other census agencies are moving in after encountering such problems themselves (Aspinall Citation2009). The alternative is to make essentialist claims about ethnoracial “types,” perhaps in consultation with “community leaders,” taking the matter back full circle to the growing difficulty in identifying legitimate leaders in a superdiverse society. For these reasons, CI’s experience in building DA is an important lesson for the Australian Government. At the heart of the problem is the assumption that states can resolve the constant state of flux that identities display and pin them down decisively, for the last time, when in fact ethnoracial categories exist in a wide range of dynamic social and political processes. As the government’s working group will soon learn, there is no straightforward social consensus, leave aside a “national standard,” prescribing what “ethnicity” means.

The problem of raciology in census operations means that even the nations Giles named as examples are sites of serious, long-running debates about the categories they employ, along with their value and meaning. In the US and UK, for example, the use of terms like “Asian” or “Latino” is the subject of growing criticism because they group large and diverse groups of people together based on logics that seem frozen in time, using categories that might no longer be relevant. Alternatively, they might be understood only by the state and not by people themselves, whose cultural, ethnic, linguistic, national, and migration history affiliations are extremely varied. The US census therefore does not reflect the existence of existing racial categories, but rather, it is a “race-making instrument,” as Thompson has put it. In her discussion of US, UK, and Canadian census operations, she demonstrates how the state’s very decision to select, create, or promote specific categories creates powerful incentives for social groups to adopt these categories, precisely to structure their interactions with the state (Thompson Citation2016). At the same time, such decisions can also provoke contestation and rejection. On the origin of the race-making practices embedded in the US Census, Perlmann has shown the intellectual connections between ideas of “race” and the US List of Races – later the List of Races and Peoples – on the one hand, and its adoption of the “ethnicity” category in the 1960s on the other. Some Jewish organisations protested the existence of a “Hebrew” category, insisting they were a religious group, while others, professing cultural nationalism, condemned the protests. Some Slavs, on the other hand, refused to be classed in terms of the empires that ruled them, and insisted on a freestanding category of “Slavs”. Whites and non-whites were classed separately, and migrants arriving in later waves were classified in terms of racial aggregate labels, such as “Hispanic” and “Asian” (Perlmann Citation2018). There is no reason, therefore, as Thompson and Perlmann show, to believe the US categories to be natural or unproblematic, or indeed useful. The UK Census is no less problematic. There, the use of the category “BAME” has been criticised for obscuring the different experiences felt within this aggregate identity set, many of whom dislike its use in contexts other than statistical reporting (Katwala Citation2021). In short, in many contexts, ethnic identification does not only encapsulate identifiable racial and cultural attributes but can involve assertions of power to change groups’ real or perceived positions within political hierarchies (Kertzer and Arel Citation2002). Equally, such assertions entrench the racialisation of consciousness that such hierarchies can engender, often generating dissatisfaction with the constraints they create for many people (Mamdani Citation2020).

The present and the colonial past

It is essential that Australian multicultural advocates also expand their comparative horizons beyond Britain and its settler colonies. One reason is that Asia is now the source of several large and politically important diaspora communities living in Australia, and the most important source of Australia’s superdiversity. A second reason is that many of the identity labels that Asian migrants carry with them are the direct outcomes of previous conflicts, contests, and processes of identity production based on colonial census categories that continue to organise public life in postcolonial societies. Creating a more expansive frame for Australia’s census discussion should enable it to move beyond its Anglosphere presentism, including by engaging critical histories of modern census operations’ origins in processes of imperial state making in Europe and its colonies. Indeed, the global spread of census operations took place in concert with European imperial expansion, alongside racialising practices aimed at managing diverse populations by sorting them into categories. For this reason, Cohn described them as constituting a central “investigative modality” of colonial rule, which Anderson argues not only constructed “ethnic-racial classifications” but also gave them “real social life” despite their origins as imperial “fantasies” (Anderson Citation2006, 168–171; Cohn Citation1996, 8). In many postcolonial contexts, the resulting categories have become reified and now perpetuate a wide range of inequalities, although there is often widespread political opposition to changing or removing them, particularly from their beneficiaries. Even among non-beneficiaries, racial and raciological thinking is a feature of institutional and everyday life. Many of Australia’s superdiverse have therefore already experienced ethnoracial census enumeration, along with how it has helped to harden identity categories and accentuate differences between them, instead of supporting efforts to equalise opportunities across their boundaries.

Examples abound in the former British colonies of Asia, including Myanmar, Malaysia, and India, where racial, religious, and caste categories first introduced in colonial census operations gradually became entrenched in regimes for allocating basic rights, including citizenship. In Myanmar, for example, Cheesman has shown that the notion of “national races” developed alongside a “juridical project” to exclude even citizens who are not members of these national races, namely Rohingya. This exclusion positions Rohingya within a minority forced to plead for its readmission into the national community, even as their exclusion justifies murderous violence against them, causing their mass displacement, for example (Cheesman Citation2017). In Malaya, an Emergency declared at the point of decolonisation racialised support for communism as a “Chinese” characteristic, and anti-communism as the natural politics for members of the aggregate Malay category, to the point that communist Malays were expelled from Malaya (Malhi Citation2021). In its turn, the very category “Malay” was, as Kahn has shown, only produced from 1901 via the colonial census, a process in which migrants from the Netherlands East Indies were both welcomed as frontier settlers and also rendered invisible by their classification as “Malays” (Kahn Citation2006). In India, while “race” was not the central category, religious and caste categories were deployed by census operations in ways that “unyoked” them from specific communities, spaces, and places, and made them universal. Grouped up in enormous new categories that might once have been flexible but that hardened from the 1870s, Indians became enmeshed in national-scale dynamics of social competition organised in terms of inclusion and exclusion (Appadurai Citation1993; Pandey Citation2006). So powerful are these categories that their far-reaching effects extend not only to competition over jobs and scholarships, but even to the notion that separate national territories were and are required for communal groups. The effects of both communal and caste enumeration are now felt in the migrant nations that many Indians live in today, even feeding open street fighting between ethnoracial or religious groups in Sydney, for example (Rachwani Citation2022b).

It is not obvious, therefore, that introducing an “ethnicity” question to the Australian Census will naturally support improved social justice outcomes for Australia’s “superdiverse.” Rather, it is more likely that it will open complex political and ethical questions that cannot be resolved in terms of boundary work or statistical methods. These questions will include whether – and how – Australians might define a shifting concept like “ethnicity” in the first place, and whether policymakers should adopt it to support discussions of Australian “culture,” as they already have in discussions of the census decision to date. They will also include whether and how to distinguish between “ethnicity” and “race” – a term carefully excluded from Australian discussions, but which carries no practical difference in meaning from the term “ethnicity.” Additional questions will likely include whether – and how – Australians should draw boundaries between specific “ethnic” groups, how to account for old and new categories, and how to assess how respondents might respond to the new census question when it launches. Foremost, this decision will – or should – also raise the question of whether it is politically acceptable to classify diverse populations in terms of such categories at all, given the colonial roots of such exercises and their complicated consequences. These consequences have included open contestation, in which identity groups mobilise to redefine the categories census operations create for them. Conversely, they have also included hardened identity categories that resist being shed or changed – or that make other categories invisible by grouping them within ill-fitting aggregate labels – and that nevertheless mobilise their bearers in contests for social and political benefits. There is no reason to assume that Australia will be exempt from these possibilities, and policymakers should recognise them as potential outcomes that could flow from the census decision.

Conclusion

The Australian Government’s census decision derives from policymakers’ statist impulse towards instituting raciology as a response to illegible superdiversity in the population. This raciology is also evident in CI’s DA. The government and multicultural advocates should therefore pause to acknowledge the DA experience, along with the political and ethical problems associated with ethnic enumeration. While improved social justice outcomes are desirable, it is a mistake to presuppose that formalising ethnic categories in Australian public life will build a more inclusive civic nationalism. Australia’s multicultural advocates are also assuming that Australia will somehow avoid the problems evident in DA, in other settler colonies’ census operations, and in the history of census operations in Europe’s former colonies in Asia. Indeed, in finding illegible superdiversity a “problem,” contemporary Australian arguments resemble those from the colonial past, in which policymakers understood shifting identities as problems requiring state-driven solutions, namely enumerative regimes that are now politically impossible to dismantle. Judging from the contests that have accompanied or followed the decisions embedded in these regimes, along with the resulting benefits for some and social harms for others, it is inappropriate for advocates to present ethnoracial enumeration as an uncomplicated social good. Rather than pressing forward on this basis, the Australian Government should openly consider the pitfalls discussed above, an awareness of which has informed the ABS’ approach to date. While the “ancestry” category has presented its own issues as discussed, the Australian Government would do well to heed the warnings the ABS has issued on “ethnicity” and its many problems.