Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

4.1 Introduction

Individuals often choose partners with social and cultural backgrounds that are similar to their own. The extent of intermarriage in Australia is an important measure of the social distance between Indigenous and non-Indigenous people, and between those of Anglo-Celtic descent and those of other ethnic backgrounds. Intermarriage may be interpreted as a measure of the successful integration of minority groups or, conversely, as a threat to cultural identity. Either way, increasing intermarriage over time implies the erosion of social boundaries between ethnic groups.

Previous studies have shown that while mid-twentieth century immigrants were often partnered with compatriots, the rate at which their children and grandchildren form exogamous relationships has increased with each generation. Are more recently arrived migrant communities from Asia and the Middle East replicating this pattern? Are the barriers to intermarriage between Indigenous and non-Indigenous Australians eroding at a similar pace?

Using customised data from the 2011 Census, and comparing this with data from earlier censuses, this analysis investigates the extent of intermarriage (both formal and informal) within Australian society according to Indigenous status, country of birth and ancestry. Patterns of intermarriage are examined using percentages and with log-linear models that control for the size of various ethnic groups within the population.

4.2 Review of the Literature

Characteristics such as birthplace, ancestry and Indigenous status are traditionally strong determinants of partner choice. While many individuals choose to partner with someone of a similar background, others cross ethnic barriers in their choice of a spouse. Intermarriage – defined here as formal or informal (de facto) heterosexual marriage between two people who differ by country of birth, ancestry or Indigenous status – has been a subject of much interest to social scientists.

American research dominates this field, with studies of marriage between the majority white population and one or more ethnic or racial minority groups (Tucker and Mitchell-Kernan 1990; Kalmijn 1993; Hwang et.al. 1997; Crowder and Tolnay 2000; Tzeng 2000; Bratter and Zuberi 2001; Rosenfeld 2008). However, the subject of interethnic partnering is equally salient in the context of Australia, with its large and diverse migration program continually adding to the nation’s cultural mix. It is all the more pertinent given the pre-existing divide between Indigenous and non-Indigenous Australians, and the official endorsement of multiculturalism as a cherished feature of Australian society.

4.2.1 Intermarriage as a Measure of Integration or Assimilation

At the most basic level, intermarriage is considered to be the outcome of close social interactions between members of different ethnic groups (Kalmijn and Flap 2001). The extent of partnering across ethnic groups is therefore considered a key indicator of social integration. This view is well expressed by Alba and Nee (2003: 90): ‘A high rate of intermarriage signals that the social distance between the groups involved is small and that individuals of putatively different ethnic backgrounds no longer perceive social and cultural differences significant enough to create a barrier to long-term union’.

Historically, there has been significant ‘social distance’ between Indigenous and non-Indigenous Australians, and between the Australia-born of Anglo-Celtic background and those of other ethnic backgrounds. This has resulted from prejudice within the mainstream community towards ‘others’, and/or from within minority ethnic or Indigenous communities themselves. Some ethnic groups have traditionally discouraged or proscribed marriage outside the group boundaries – known as ‘exogamy’, in the language of sociology – and, conversely, have encouraged or prescribed marriage within the group, known as ‘endogamy’ or ‘homogamy’. In Australia, some commentators have expressed concern that multiculturalism may encourage endogamy and thereby perpetuate group boundaries (Blainey 1994).

However, if rates of intermarriage are high or increasing, it implies that concerns about the social segregation of migrant groups in Australian society are unfounded. Similarly, the extent to which Australians are partnering across Indigenous/non-Indigenous lines is an important indicator of whether past social or cultural divisions between the Indigenous and non-Indigenous communities have dissipated.

Intermarriage across ethnic groups may also mean that these groups are becoming more similar with regard to other social and demographic characteristics. People tend to look for partners with similar educational and class backgrounds to themselves (Kalmijn 1998). Where minority groups are socially or economically disadvantaged relative to the rest of society, exogamy is less likely, since prospective marriage partners are unlikely to bridge this gulf. Conversely, the sociological literature suggests that intermarriage will be relatively high where members of a minority group achieve upward social mobility. Indeed, classic assimilation theory holds that intermarriage occurs only after minority groups achieve equality or near-equality on other dimensions (structural assimilation). Relatively high levels of education, in particular, are often found to facilitate intermarriage (Kalmijn 1993, 1998). This has been verified by past analyses with regard to patterns of intermarriage among Indigenous Australians: rates are high among those who have achieved relatively high levels of education and incomes, even outside the nation’s capital cities where exogamy is otherwise low (Heard et al. 2009a).

Intermarriage can be both cause and consequence of upward social mobility for minority group members. There is little doubt that intermarriage can assist minority groups to assimilate with a majority group, or to adopt cultural characteristics of that group (Gevrek et al. 2011). However, the relationship between ethnic intermarriage and ethnic inequality is not unequivocal, depending on the minority group in question and on the characteristics of their ‘mainstream’ partners (Okun and Khait-Marelly 2010; Song 2010). For example, educational achievement is positively associated with exogamy among US Hispanics, but not US Asians (Gonsoulin and Fu 2010). At the micro level, marriage is an institution in which ethnic differences may be resolved, or it may be an arena in which cultural models compete (Lomskey 2010).

4.2.2 Preferences and Opportunities

Intermarriage is a social phenomenon that is open to several interpretations. It may result from individual preferences for a specific marriage partner, or from structural constraints in the marriage market (Bull 2005: 44). Using log-linear models, scholars have attempted to separate demographic effects from the effects of changing preferences or social norms (Kalmijn 1993; Harris and Ono 2005; Walker 2010; Qian and Lichter 2011). Log-linear models remove the effect of variation in the relative sizes of the different ethnic groups from the analysis of the rates of intermarriage between these groups. This effectively allows a disentangling of opportunity from preference in patterns of spousal choice (Uunk et al. 1996).

Such models confirm that rates of intermarriage are partly determined by opportunity (Blau 1977; Alba and Golden 1986). Intermarriage is less likely the larger the size of one’s own group and, therefore, the availability of potential spouses from that group (Chiswick and Houseworth 2011; Feng et al. 2010). For example, Choi and Mare (2012: 449) show that migrants in the US are more likely than non-migrants to be exogamous, because the relatively small size of their group compels them to expand their pool of potential spouses to include non-migrants.

Despite the importance of opportunity, several recent studies have found that preferences remain the more powerful factor with regard to ethnicity (e.g. Hitsch et al. 2010). Kalmijn and Van Tubergen (2010) reveal large differences in endogamy across 94 ethnic groups in the US, finding that although ‘both structural and cultural group-level factors have significant effects on endogamy’, ‘cultural explanations (which focus on the role of norms and preferences) play a more important role than structural explanations (which focus on meeting and mating opportunities).’ Similarly, in Britain, some ethnic groups have higher propensities to form endogamous partnerships, even after controlling for factors such as education and length of residence in the country – yet all are equally responsive to opportunity structures (Muttarak and Heath 2010).

4.2.3 Social Change and Increasing Intermarriage

It is tempting to assume that intermarriage will inevitably increase over time, the longer different ethnic groups live side by side. In western societies, the forces of individualization, secularization and globalization have diminished the influence of parents and of religious institutions over partner choice. Along with the cultural weight given to romantic love, these forces have increased the autonomy of young people in choosing partners and point to ever-increasing opportunities for intermarriage.

Further, intermarriage itself facilitates the erosion of group boundaries by binding families and communities of different ethnic backgrounds together. This can become a recursive process, as partnering choices are shaped by those of the preceding generation. Children of mixed ethnicity couples are less likely to identify as belonging to a single ethnic group, further reducing cultural distinctions (Stephan and Stephan 1989).

Yet change is not necessarily unidirectional. The growth of ‘identity politics’ (Appiah 2006) or the ‘politics of recognition’ (Connolly et al. 2007) may imply a greater propensity to take pride in group identity, and a greater interest in the preservation of ethnic subcultures. Any economic or political circumstances which limit social mobility are also likely to perpetuate barriers to intermarriage.

Homogamy has decreased in Britain (Muttarak and Heath 2010). According to Rosenfeld (2008), racial endogamy in the US has also declined sharply over the twentieth century, but race is still the most powerful division in the marriage market.Footnote 1 In both countries, as in Australia, intermarriage is higher among the second generation of migrants than the first (Khoo and Birrell 2002; Muttarak and Heath 2010), and higher still where migrant communities have produced third and subsequent generations (Giorgas and Jones 2002; Alba and Nee 2003; Khoo et al. 2009; Heard et al. 2009b).

Rates of intermarriage between Indigenous and non-Indigenous Australians appear to be high and increasing wherever there are opportunities for mixing: the great majority of partnered Indigenous persons living in Australia’s capital cities are exogamous (Heard et al. 2009a). Trends in intermarriage by members of migrant groups are less predictable, because the Australian migration program is characterized by large waves of migrants coming from particular regions of the world at different times. Studies from the US, Britain and Sweden all show that rates of intermarriage can vary widely among different ethnic groups within the same multi-ethnic setting, even accounting for the respective sizes of these populations, and that cultural similarity or dissimilarity plays a significant role (Muttarak and Heath 2010; Kalmijn and Van Tubergen 2010; Dribe and Lundh 2011). Some minority groups may be more resistant to exogamy than others, but the attitudes held by members of the majority ethnic group are also important. US research shows that while Americans have become more accepting of interracial relationships in recent decades (McClain 2011), members of the majority white population are more willing to form relationships with members of some ethnic minorities than others (Herman and Campbell 2012). Thus, though ever-increasing intermarriage appears to have been inevitable for European migrant communities to Australia, the partnering behaviour of more recent waves of migrants from Asia and the Middle East will not necessarily follow the same pattern if their real or perceived cultural distance from the Anglo-Celtic majority is greater (Jones and Luijkx 1996; Giorgas and Jones 2002).

Early work on ethnic intermarriage among first- and second-generation Australians used marriage registration data, which included information on country of birth (Price 1982, 1993; Gray 1987; Jones 1991; Young 1991). However, registration data no longer includes information on the birthplaces of the parents of marriage partners, so that it is now impossible to examine intermarriage patterns among the second generation using this source. Instead, more recent studies of intermarriage in Australia have used birthplace and ancestry data from the quinquennial Australian census (Penny and Khoo 1996; Roy and Hamilton 1997; Giorgas and Jones 2002; Khoo 2004; Khoo et al. 2009). The census provides information on Indigenous status as well as on birthplace and ancestry, enabling a more detailed study of intermarriage across all these sub-group boundaries in Australia (Birrell and Hirst 2002; Heard et al. 2009a).

4.3 Methodology

Using customised data from the 2011 census purchased from the Australian Bureau of Statistics (ABS), this paper assesses the extent of ethnic intermarriage in Australia. Descriptive statistics are the starting point for the examination of partnership patterns, examining the percentage of marriages both within and between ethnic groups. Where possible, data from previous censuses are used to assess the direction of change in patterns of intermarriage in Australia. The analysis methodology then extends previous Australian research in this field through the use of log-linear models.

The data presented relate to partnered persons only, living in married or de facto relationships with a partner of the opposite sex. There were inevitably some partnered census respondents who did not state their Indigenous status, birthplace or ancestry. Those for whom these characteristics were not stated are excluded from the calculations in the following analysis. However, couples are included in the calculations if one partner stated his or her marital status but the other did not.

The measures of ethnicity used are ancestry, birthplace, and Indigenous status. There are advantages and disadvantages to each measure. Individuals have a single birthplace, whereas they may identify with multiple ancestries. However, ancestry may tell us more about the self-perceived cultural identity of an individual. Although “Australian Aboriginal” is one of the options in the ancestry question in the census, there is also a separate question on Indigenous status, which asks whether the respondent is of Aboriginal and/or Torres Strait origin. We use both variables in our analysis of intermarriage, and compare the results.

The analysis of percentages is extended by examining odds-ratios and log-linear models of the partnership data. Log-linear modelling allows comparisons to be made between the levels of homogamy for groups of different sizes. In particular, the quasi-independence (QI) model provides coefficients that model the number of homogamous partnerships. The cell frequencies m ij of the marriage tables are modelled by:

$$ \log\;{\mathrm{m}}_{\mathrm{i}\mathrm{j}}=\mu +{\lambda_{\mathrm{i}}}^{\mathrm{m}\mathrm{Eth}}+{\lambda_{\mathrm{j}}}^{\mathrm{fEth}}+{\delta}_{\mathrm{i}}\mathrm{I}\left(\mathrm{i}=\mathrm{j}\right) $$

where I(.) is the indicator function for the diagonal of the frequency table.

$$ \begin{array}{ccc}\mathrm{I}\left(\mathrm{i}=\mathrm{j}\right)& =1,& \mathrm{i}=\mathrm{j}\\ {}& =0,& i\ne j\end{array} $$

The δi parameters represent the number of homogamous relationships above and beyond those predicted by ancestry or birthplace. Exponentiating the δi parameters provides a factor which indicates how many times greater (or less) is the number of couples expected to have a homogamous partnership, over and above the independence model. Thus the parameters presented in our results provide a measure of how many times greater is the number of homogamous couples than would be expected by chance, given the total number of males and females in the relevant ethnic groups. The larger the quasi-independence parameter, the stronger the pattern of homogamy. For a more detailed explanation of quasi-independence models and other relevant statistics, see Goodman (2007).

4.4 Results

The examination of homogamy is divided into three sections on Indigenous partnering, intermarriage by birthplace, and intermarriage by ancestry.

4.4.1 Indigenous Status

This part of the analysis uses data derived from the Indigenous status question on the census, which asks whether the respondent is of Aboriginal and/or Torres Strait Islander origin. The self-identification of Indigenous status raises some unique measurement issues. The number of Australians identifying as Aboriginal or Torres Strait Islander has more than doubled over 25 years, reaching 548,370 in 2011. Over and above natural increase, more Australians think of themselves as Indigenous and/or are more inclined to declare this identity on their census forms (ABS 1999).

Table 4.1 shows the percentage of Indigenous men and the percentage of Indigenous women in homogamous partnerships, for each region of Australia. The patterns shown in the table for 2011 are consistent with those of 2001 and 2006, where the percentage of Indigenous people in homogamous relationships was lower in capital cities and higher in the remainder of each state (Heard et al. 2009a). Regional Northern Territory and Western Australia had the highest percentages of Indigenous people in homogamous partnerships, with their capitals Darwin and Perth leading the cities.

Table 4.1 Indigenous couples and individuals in homogamous relationships by Australian region, census years
Full size table

Overall there is a trend towards exogamy. There is a decrease in the percentage of individuals in homogamous relationships across Australia, even where there was increase between 2001 and 2006. The few exceptions are the stable percentages in the Northern Territory, and the increase in regional Tasmania.

Across Australia, 40.9 % of partnered Indigenous women and 43.3 % of partnered Indigenous men were in homogamous partnerships. By comparison, the non-Indigenous partners of the 59.1 % of Indigenous women who intermarried represented only 0.8 % of non-Indigenous partnered men, and the non-Indigenous partners of the 56.7 % of Indigenous partnered men who intermarried represented 0.7 % of partnered non-Indigenous women. With only two groups represented in the data (Indigenous and non-Indigenous) there are insufficient groups to parameterise a log-linear model. However, we can examine the odds ratios for Indigenous and non-Indigenous males and females. Converting the percentages to odds ratios, we find that a non-Indigenous male is 162 times more likely to have a non-Indigenous partner than an Indigenous male is to have an Indigenous partner (compared to 159 times in 2006). The odds for a non-Indigenous female are 197 (204 times in 2006) times that of an Indigenous female.

The odds, like the percentages, vary considerably by location. With the data described here it is not possible to disentangle attraction from availability. It is likely that the higher rates of homogamy outside of the capital cities, and particularly in Western Australia and Northern Territory, are due to greater social interaction with and availability of potential Indigenous partners, but the variation could also indicate different social norms or differences in ethnic identification.

Table 4.2 includes an additional variable relating to educational attainment. In most regions, the association between education and homogamy remains straightforwardly negative: the percentage of individuals in homogamous partnerships is highest for those with minimal education (Year 10 or lower), and lowest for those who have completed a post-school qualification.

Table 4.2 Indigenous males and females in homogamous relationships (per cent) by region and highest qualification, 2011
Full size table

The table suggests that education increases intermarriage, particularly in regional areas where Indigenous homogamy is otherwise strong. In regional Western Australia, for example, rates of homogamy are 28 percentage points lower for Indigenous men and women with post-school qualifications than for their counterparts whose highest qualifications were ‘Year 10 or lower’. A clear educational gradient also applies in most of the capital cities. Yet the data also show that in the bigger cities, high rates of intermarriage prevail regardless of educational attainment. In Greater Sydney, Greater Melbourne and Greater Brisbane, 80 % or more of Indigenous men and women are exogamous, even among those with the lowest qualifications.

4.4.2 Birthplace

Table 4.3 shows the various combinations of birthplaces for couples. The degree of change between 2006 and 2011 is small. There has been a small increase in the percentage of couples where both partners are Australia-born, and an increase in the percentage of couples where both partners are overseas-born. This is possible due to a lower percentage of couples where one or both birthplaces are unknown, indicating an improvement in the validity of the data from the 2011 census.

Table 4.3 Couples by birthplaces of partners, census years
Full size table

Overall, 18 % of couples included an Australia-born partner and an overseas-born partner. Many more couples in which both partners were born overseas are also exogamous. Table 4.4 shows the proportion of individuals partnered to someone from the same birthplace, and the proportion partnered to someone born in Australia, for countries with sufficient migrant populations to create meaningful percentages. These proportions vary widely by birthplace. Since birthplace does not necessarily reflect ancestry, the foreign born groups with high percentages partnered to Australia-born individuals may still represent intermarriage between individuals with the same ancestry (for example an Australia-born Chinese person partnered with a China-born Chinese person).

Table 4.4 Males and females by region and country of birtha, partner born in Australia or partner born in same country, 2011
Full size table

Eighty-six percent of Australia-born men and 85.1 % of Australia-born women had a partner also born in Australia. These high percentages are not surprising for two reasons. Firstly, birthplace alone does not account for ancestry or Indigenous status, so some of these couples are potentially “intermarried” across other measures of ethnicity. Secondly, Australia-born individuals represent the vast majority of people living in Australia. Therefore, when Australia-born individuals intermarry, their numbers represent a relatively small percentage of the Australia-born majority, but much larger percentages of the minority groups they intermarry with. This issue will be addressed later in this section, where a log-linear model is used to compare the level of homogamy, while controlling for the relative sizes of the different groups.

A high percentage of males and females born in Afghanistan, Bangladesh, Iran, Iraq, Pakistan, and Sri Lanka have a partner who was born in the same country. Although this could be interpreted as a lower rate of integration into Australian society, it is more likely a reflection of couples migrating to Australia together. The census data does not indicate where the relationship formed. This means that inferences can be made about the patterns of partnered Australian residents, but not the formation of relationships within the Australian marriage market. Although this information cannot be directly determined, the examination of ancestry by generation (see Sect. 4.4.3) gives some indication of intermarriage by successive generations of ethnic groups in the Australian context.

In contrast, the percentage of homogamous partnerships among those born in Anglo-Celtic and other English speaking countries is much lower (meaning intermarriage is higher). However, this does not necessarily mean that those in exogamous relationships are with an Australia-born partner. It is common for those born in New Zealand, South Africa and England to have a partner born in one of the other two countries rather in Australia (data not shown).

There is a strong asymmetric pattern in the partnering of people born in Thailand, the Philippines, Japan and Korea (although small asymmetries exist among many other groups), where a large percentage of women born in each of these countries have an Australia-born partner, but only a small percentage of men born in these countries have an Australia-born partner. Among those born in Thailand, 15 % of men have an Australian-born partner compared to 51 % of women.Footnote 2 Gender asymmetries have been attributed to differing gender roles in Asian families (Penny and Khoo 1996). In some cases, such asymmetries point to ethnicity-specific marriage markets between Asia-born women and Australia-born men. In particular, the excess of Filipino women over men as a consequence of patterns of spouse sponsorship is a long-recognized phenomenon in Australia (Hagan 1989; Ethnic Affairs Commission of New South Wales 1992; Iredale 1994; Holt 1996; Robinson 1996; Khoo 2001). In other cases, asymmetry is due to gendered partnering patterns between overseas-born groups, rather than gender differences in rates of partnering with the Australia-born. For example, Afghanistan-born men are more likely than Afghanistan-born women to have a partner born in Pakistan.

The next step is to analyse the parameters of a quasi-independence model. That is, for each birthplace group, how do rates of homogamy compare to what would be expected by chance, once the relative sizes of the groups are taken into account? The p-values for the model are not shown for two reasons. Firstly, given the size of the counts, every p-value for every parameter is very small, giving no real indication of significance. Conversely, goodness of fit measures become very large.

Table 4.5 shows the ten highest and ten smallest homogamy parameters from the quasi-independence model for country of birth. Controlling for the sizes of the groups, the patterns of homogamy are in keeping with the analysis of percentages. Countries such as Bangladesh, Iraq and Afghanistan still show a high degree of homogamy once the relative sizes of the various groups are controlled for by the log-linear model. Since these groups are predominately from recent waves of immigration, it is difficult to distinguish preferences from couple migration patterns. However, a partial solution to this is to examine generation in conjunction with ancestry (see Sect. 4.4.3).

Table 4.5 Homogamy by country of birth, ten highest and ten lowest quasi-independence parameters, 2011
Full size table

Of the ten least homogamous birthplace groups, the majority are Anglo-Celtic, and all showed low percentages of homogamous partnership in Table 4.4. Those born in the Netherlands and Germany also recorded low rates of homogamy. Although Papua New Guinea may seem like an incongruous birthplace in a table otherwise dominated by Anglo-Celtic and European countries, it is geographically close to Australia, and a large percentage of individuals born in Papua New Guinea have an Australia-born partner.

The quasi-independence model is useful in revealing a low parameter for Australia-born couples. Although 85 % of Australia-born women and 86 % of Australia-born men have an Australia-born partner, this high percentage is largely due to the Australia-born individuals comprising a large proportion of the population. The quasi-independence parameter shows that once the relative sizes of the different birthplaces are controlled for, the number of homogamous couples is only 4.6 times higher than would be expected by chance, given the size of the group. This is the second-lowest rate of homogamy across the birthplaces analysed.

4.4.3 Ancestry

Ancestry is a variable that represents a self-nominated identity. In the Australian Census individuals may nominate multiple ancestries in any order, making the data difficult to interpret. This analysis focuses on individuals who nominated a single ancestry (representing 83.4 % of partnered men and 85.6 % of partnered women). This also provides mutually exclusive groups for modelling.

Ancestry can provide more information than birthplace in the sense that knowing that someone is born in Australia (for example) does not provide a complete picture of his or her cultural or ethnic identity. Ancestry also makes distinctions that birthplace cannot, particularly where there may be several separate ethnicities or cultures within a single country of birth (such as Assyrian and Arab, or Tamil and Sinhalese). This does mean that there are a greater number of smaller groups than in the birthplace analysis.

Table 4.6 shows the ancestries with the highest rates of homogamy, the lowest rates of homogamy and the largest differences between male and female rates. There are clear similarities to the analysis by birthplace, with many of the same patterns of high, low, or asymmetric partnering that were present in the birthplace data appearing in the data for the corresponding ancestry (or ancestries). The most homogamous ancestry groups overlap with the most homogamous birthplace groups (Bangladeshi, Afghan, Iraqi).

Table 4.6 Lowest homogamy, highest homogamy and largest gender difference in rate of homogamy, by ancestrya, 2011
Full size table

Interestingly, however, the rate of homogamy recorded by those with Australian Aboriginal ancestry (93–94 %) is more than twice the rate of Indigenous homogamy suggested by the earlier analysis of Indigenous status, which is separately measured in the Census. This highlights a stark social difference between Indigenous Australians with some Aboriginal and/or Torres Strait Islander heritage, and those whose sole nominated ancestry is Australian Aboriginal. The difference may relate to cultural preference, but it is also likely that opportunities for intermarriage are limited in the more remote locations of those whose sole ancestry is Australian Aboriginal. The 2011 census data confirms the regional concentration of these individuals: 72 % of individuals nominating Australian Aboriginal as their sole ancestry were living in Very Remote or Remote Australia, according to the ABS’ remoteness classification, rising to 84 % if Outer Regional areas are included. A state-by-state analysis shows that 40 % of sole-ancestry Australian Aborigines reside in the Northern Territory outside of Greater Darwin, 21 % in Queensland outside of Greater Brisbane and 15 % in Western Australia outside of Perth (data not shown).

The ancestry groups with the lowest percentages of homogamous individuals (and therefore the highest rates of intermarriage) are still the Anglo-Celtic ancestries (English, New Zealander, Canadian), followed by other European ancestries (French, German, Czech). Whilst birthplace does mask some cultural heritage, ancestry does the opposite. The interpretation of ancestry varies, but many respondents nominate the ancestry of parents or grandparents. As a result, over half of Australia-born men and women claim English ancestry, whereas only a third nominate Australian ancestry (data not shown).

As with birthplace, the largest gender asymmetry is seen in some of the Asian ancestries (notably Thai, Japanese, Filipino and Indonesian), where a much larger percentage of women than men have intermarried, predominantly with men of Australian or English descent (data not shown). Table 4.6 also shows the ancestries where a greater percentage of males are intermarried (Samoan, Tongan, Egyptian and Arab). However, the gender asymmetry is much smaller in these cases.

Table 4.7 shows the ten highest and ten lowest quasi-independence parameters for ancestry. With the quasi-independence model controlling for group size, the Australian Aboriginal ancestry group has the highest rate of homogamy, with the number of homogamous partnerships being 280 times greater than we would expect under an independence model, and nearly 69 % greater than the next highest groups. The remaining groups in the table are predominantly the same groups prominent in the percentages (Table 4.6), with the exception of South Africans (although there are a number of countries with only slightly lower QI parameters outside of this top ten). The ten ancestries with the lowest rates of homogamy are all European or Anglo-Celtic, with the exception of Syrian. The reason for this is a high rate of intermarriage with those of Lebanese ancestry, with approximately 900 Syrian/Lebanese intermarried couples compared to only about 100 Syrian/Australian intermarried couples (data not shown).

Table 4.7 Homogamy by ancestry, ten highest and ten lowest quasi-independence parameters, 2011
Full size table

It can be useful to consider ancestry in conjunction with birthplace variables in order to get some sense of how well established in Australia are individuals and groups with different ethnic backgrounds. The census provides information on the birthplaces of individuals and of their parents. We derived an additional ‘generation’ variable using the following definitions: a person born overseas and with one or both parents born overseas is ‘first generation’; a person born in Australia with one or both parents born overseas is ‘second generation’; and a person with two Australia-born parents is ‘third generation or later’.

In Table 4.8, the percentage of same ancestry partnerships among first generation Australians can be compared to rates in the second, third and subsequent generations. All ancestries show a decrease, often dramatic, in the percentage of endogamous partnerships between the first and second generations, and most also between the second and third or later. For some groups with a shorter migration history to Australia, there are insufficient numbers of individuals of the third generation (or later) to calculate a meaningful percentage. Data for some of the more recent migrant groups from Africa and the Middle East relate to the first generation only, and are not included in the table.

Table 4.8 Females and males in homogamous partnerships by ancestry and generation, 2011
Full size table

An anomaly is the increase in the rates of homogamy for English, Irish, Scottish and German individuals from the second generation to the third or later. These are among the most common ancestries nominated in the Australian census. Although we cannot be certain of the reason for this pattern, it may be simply that so many Australians claim some distant (‘third generation or later’) Anglo-Celtic or Anglo-Saxon heritage, making homogamy (in its very broadest sense) very likely.

There is great variation by ancestry in the proportion of homogamous partnerships reported by first generation immigrants. Again, this is likely to reflect variations in the extent to which overseas-born individuals are partnered prior to migration. Clearly, for example, the majority of migrants from New Zealand have arrived unpartnered, and record low rates of homogamy even in the first generation. By the second and subsequent generations, only very small proportions (3–4 %) are homogamous.

However, the ‘speed’ with which homogamy decreases in successive generations also varies greatly by ancestry, and may provide a useful measure of integration. For example, from similar levels of homogamy (91 %) in the first generation, homogamy among men of Bangladeshi descent decreased far more in the second generation (to 38 %) than did homogamy among men of Afghan descent (71 % in the second generation). The same is true for women with these ancestries. Comparisons can also be made between men and women of the same ethnic backgrounds. From the second to the third and subsequent generations, homogamy decreased by more than 20 percentage points among men of Turkish descent, but by less than 10 percentage points among women of Turkish descent.

4.5 Discussion

Intermarriage provides a way of examining social distance between groups (Kalmijn and Flap 2001). In the Australian census data we see different patterns of endogamy and exogamy for different Indigenous, ancestry and birthplace groups. These patterns may have resulted from individual preferences for a specific marriage partner, or from structural constraints in the marriage market (Bull 2005). Patterns vary widely among different ethnic groups within the same multi-ethnic setting, even accounting for the respective sizes of these populations, and in keeping with the view that cultural similarity or dissimilarity plays a significant role (Muttarak and Heath 2010; Dribe and Lundh 2011; Kalmijn and van Tubergen 2010).

In Australia, one key question is whether multiculturalism may encourage endogamy and thereby perpetuate group boundaries (Blainey 1994). The partnering behaviour of more recent waves of migrants from Asia and the Middle East will not necessarily follow the same pattern as earlier waves of European migrants if their real or perceived cultural distance from the Anglo-Celtic majority is greater (Jones and Juijkx 1996; Giorgas and Jones 2002). Our data suggests a high degree of homogamy among those born in the Middle East and Indian subcontinent.

While these high levels of homogamy may be partly due to ethnic preferences, it is probable that individuals from these groups are also more likely to be in homogamous partnerships before they come to Australia. It may be decades before it is possible to fully examine the integration of all the recent migrant groups from the Indian subcontinent and the Middle East into Australian society, based on the intermarriage patterns of the second and third generations of these groups. However, the generational ancestry data does show that among the longer-established groups, the second and third generations are much more likely to intermarry. Therefore, as for communities originating from earlier waves of European migration, all signs are that intermarriage steadily increases the longer these groups are present in Australia, albeit at a faster or slower pace depending on the group in question.

Muttarak and Heath (2010) have described a pattern of segmented assimilation in the UK, where individuals from some groups enter into exogamous relationships more than others. In particular, they found that those from Indian, Pakistani and Bangladeshi backgrounds were less likely to intermarry. However, when considering other factors, they concluded:

It appears that members of the three South Asian groups do indeed respond to opportunity structures in much the same way as other groups do and that the solidary community might not be quite as powerful in inducing conformity as strong versions of the theory [segmented assimilation] would suggest.

The same would appear to hold in the Australian context. For individuals belonging to the longer-established of the South Asian migrant communities in Australia, such as those of second and third generation Indian and second generation Pakistani and Bangladeshi ancestries, there is a pattern of greatly increased intermarriage when compared to first generation migrants. Indeed, homogamy drops particularly quickly between the first and subsequent generations among those of Indian and Bangladeshi descent, and even more so among those of Sri Lankan ancestry. The shift between generations is less dramatic among Australians of Middle Eastern descent, but intermarriage nevertheless increases steadily.

The asymmetry in the Australian-Asian partnerships is very similar to that observed in New Zealand Census data, particularly in 2001 and 2006, where Asian women were more likely to have a New Zealand-born European male partner than vice-versa (Walker 2010). In some cases, this is likely due to the continuing phenomenon of Australian men sponsoring spouses from specific countries for intermarriage (Khoo 2001), although visa data is required to verify this. The sponsorship of foreign spouses has little to do with the integration of diverse ethnic groups within Australia, but shows that intermarriage is a phenomenon that transcends national borders.

The final pattern of interest in this data is that of intermarriage between Indigenous and non-Indigenous Australians. Data from the 2006 Census showed that rates of intermarriage between Indigenous and non-Indigenous Australians were high and increasing in the capital cities, but that Indigenous homogamy remained strong outside of the capital cities, particularly in the Northern Territory and Western Australia. For Australians who identified themselves as having Indigenous heritage, this remained the case in the 2011 Census. However, among those who nominated a sole ancestry of “Australian Aboriginal”, there was a particularly high degree of homogamy. At 93–94 %, this level of homogamy for Australian Aboriginals (as defined by ancestry) is comparable only with the rate of homogamy recorded by Indigenous Australians (as defined by the separate Indigenous status question) living in the remote Northern Territory outside of Darwin. Indeed, we suggest that this phenomenon is largely due to the concentration of sole-ancestry Australian Aborigines in more remote locations. Similar patterns are seen among the Indigenous Maori population in New Zealand (Walker 2010): those in urban centres are much more likely intermarry than those in rural areas, and those who identify as Maori only are less likely to intermarry than those who nominate multiple ethnicities.

Indigenous Australians with higher levels of education are more likely to be intermarried. Educational differences are particularly evident outside the major capitals, where homogamy is otherwise high. This pattern may partly reflect opportunity, due to the mix of Indigenous and non-Indigenous Australians in educational institutions and in employment. More fundamentally, it suggests that intermarriage by Indigenous Australians is facilitated by equality or near-equality on socio-economic dimensions, lending support to theories of structural assimilation.

4.6 Conclusion

This analysis of the ethnic partnering patterns in the 2011 Australian Census shows similar patterns to those seen in the 2001 and 2006 censuses. The analysis extends previous research with Australian data by complementing the raw percentages with quasi-independence log-linear models.

The patterns of Indigenous partnership show very slight increases in intermarriage in most regions. The high degree of variability remains consistent with the 2001 and 2006 censuses, where intermarriage in the capital cities is much higher than in the regional areas of each state. The Northern Territory and Western Australia have the highest rates of homogamous Indigenous partnering. The Aboriginal Australian ancestry group, which was measured using a separate variable, and counted those who solely identified their ancestry as Australian Aboriginal, showed very high rates of homogamy. Although cultural factors may contribute, the concentration of this group in remote and regional areas of Australia suggests an explanation centred around opportunity.

The examination of both birthplace and ancestry showed that those who were born in, or identified their ancestries as belonging to, European or Anglo-Celtic countries were far more likely to intermarry. The log-linear models showed that once group size was adjusted for, this was the case for the Australia-born too. Those from the Middle East, the Indian subcontinent, and other non-English speaking countries were less likely to intermarry. From census data alone it is difficult to determine whether this is an indicator of preference, or merely reflects patterns of immigration, where couples of the same ancestry or birthplace migrate to Australia together. However, for groups whose migration can be traced across multiple generations, the second generation invariably records greater rates of intermarriage than the first, while third (and later) generations tend to have higher percentages of intermarriage still. The other notable pattern in the birthplace and ancestry data is the asymmetry of partnering seen in some Asian groups (in particular Thailand and the Philippines), with a much higher percentage of Thai and Filipino women partnered to Australian men than vice-versa.

Future research in the area of intermarriage in Australia could usefully involve matching data from other sources. Whilst the Census provides information about all Australians, it can only provide limited detail. Information about partnership status upon immigration, lengths of relationships, divorce and repartnering could supplement the census data and provide a more complete picture of interethnic partnering in Australia.