FindArticles > Journal of Population Research > May, 2001 > Article > Print friendly
A decomposition of immigrant divorce rates in Australia - Research Note
Siew-Ean KhooCrude divorce rates of overseas-born Australians by birthplace are not comparable because different birthplace groups have different age structures and marriage patterns that affect the rates. This paper decomposes the crude rates into components due to the effects of age structure, marriage patterns and the divorce rate of married men and women. This allows for a better comparison of the level of divorce by birthplace. The range in the standardized divorce rates by birthplace is smaller than that indicated by the crude rates. Nonetheless, there remains considerable variation in immigrant divorce rates in Australia.
**********
The demographic behaviour of immigrants can be an important indicator of their adjustment to their new social and economic environment. Studies examining immigrant fertility and mortality rates have usually attempted to find out whether these rates differ from or are converging to those of the native-born population, because such trends can indicate whether the migrants are adjusting to their new environment. While there have been many studies investigating the fertility and mortality rates of immigrants in Australia (see, for example, Yusuf and Rockett 1981; Young 1987, 1991; Khoo and Shu 1996), there has been little focus on their divorce rates.
In recent years, the Australian Bureau of Statistics has published crude divorce rates for males and females for the 15 largest birthplace groups in Australia. In 1996, the rates show considerable variation by birthplace, ranging from the lowest of 3.2 per thousand recorded among women born in Italy to the highest of 19.4 per thousand among women born in China (Australian Bureau of Statistics 1997). The crude divorce rate, however, is only a rough measure of the actual level of divorce. It is the ratio of the number of men or women who have divorced during a given period to the total population of men or women at mid-period, and does not take into account that only married people are at risk of divorce or that younger couples are at a higher risk of divorce than older couples.
Migrant groups can have very different age structures and marriage patterns due to their migration histories and diverse cultural backgrounds, and these can affect their crude divorce rates. Different birthplace groups have very different age structures depending on when they migrated to Australia. Those who migrated mainly in the 1950s and 1960s have a much older age structure than those who arrived in the 1970s and 1980s. Thus more recent migrant communities such as those from Asian countries have younger age structures than older migrant communities such as those from Southern European countries. The age structures of their married populations are also likely to be younger. Since younger couples generally have higher divorce rates than older couples, different age distributions contribute to very different crude divorce rates. Also, some migrant groups have higher proportions of married people than others, because of cultural patterns of earlier marriage or more family or marriage migration. A population w ith proportionately more married adults will have a higher crude divorce rate, all other things being equal, than a population that has proportionately fewer married adults. Because of these effects of age and the proportion married, the crude divorce rate is not a good indicator of the real risk of divorce faced by each migrant group.
This paper proposes to decompose the crude divorce rate into three demographic components: the proportion of married people in the population, the age composition of the married population, and the divorce rate for the married population. This makes it possible to compare the divorce rates of immigrant groups without the distortions due to differences in the proportion of married people in different birth-place groups and their age distribution. While it is also relevant to consider the effect of marriage duration on the divorce rate (because the divorce rate in general is higher for marriages of shorter duration than for marriages of longer duration, longer marriages being selected for greater stability), it is not possible to do so in this paper. This is because the data required, the distribution of married men and women by birthplace and duration of marriage, are not available. However, it can be argued that the effect of duration of marriage is substantially addressed through the age effect because there is usually a high correlation between a married person's age and marriage duration.
Decomposing the crude divorce rate
As a conventional demographic method, decomposition has been widely used in comparative studies of fertility, mortality, and to a lesser extent household formation (Coale 1965; Butch 1980; Smith 1992). The same principle can be applied to the comparison of divorce patterns in different populations.
The level of the crude divorce rate is determined by three demographic factors: the proportion of married people in the population, the age composition of the married population and their age-specific divorce rates. Accordingly, the crude divorce rate can be decomposed through the following steps.
Let D denote the crude divorce rate. (1) Then D = din, where d is the number of people who divorced during a given year and n is the mid-year population. (2) Since only married people can divorce, the crude divorce rate D can be decomposed into the proportion who are married and the divorce rate of married people: D = P x M The proportion who are married, P, is equal to m / n, where m is the mid-year number of married people. The divorce rate of married people, M, is equal to d / m. It can also be expressed as M = [SIGMA] [R.sub.i] [F.sub.i], where [R.sub.i] is the divorce rate among married people aged i and [F.sub.i] is the proportion of all married people who are aged i. Hence D = P x [SIGMA] [R.sub.i] [F.sub.i]. That is, the crude divorce rate is the product of two factors. The first is the proportion of married people in the population. The second is the sum of the products of the divorce rate of married people of age i and their proportion of the total number of married people in the population.
To compare the divorce rates of different populations, a and s, the ratio of the two crude divorce rates can be expressed as
[D.sub.a]/[D.sub.s] = [P.sub.a] x [SIGMA] [R.sub.ia] [F.sub.ia]/[P.sub.s] x [SIGMA] [R.sub.is] [F.sub.is]
Multiplying the right side of the equation by
[SIGMA] [R.sub.is] [F.sub.ia]/ [SIGMA] [R.sub.is] [F.sub.ia]
and rearranging gives
[D.sub.a]/[D.sub.s] = [P.sub.a]/[P.sub.s] x [SIGMA] [R.sub.is] [F.sub.ia]/ [SIGMA] [R.sub.is] [F.sub.is] x [SIGMA] [R.sub.ia] [F.sub.ia]/ [SIGMA] [R.sub.is] [F.sub.ia]
If a denotes the population of interest or the comparison population and s is the standard or reference population, then the term on the left of the equation is the ratio of the crude divorce rate of the comparison population to the crude divorce rate of the standard population. The right side of the equation comprises three factors. At the left, [P.sub.a]/[P.sub.s] is the ratio of the percentage of married people in population a to that in the standard population. The middle term
[SIGMA] [R.sub.is] [F.sub.ia]/[SIGMA] [R.sub.is] [F.sub.is]
compares the age structures of the two married populations and measures the effect of the age composition of married people in population a on the ratio of the crude divorce rates. The remaining factor
[SIGMA] [R.sub.ia] [F.sub.ia]/[SIGMA] [R.sub.is] [F.sub.ia] compares the two sets of age-specific divorce rates of married people and measures the effect of the age-specific divorce rates for married people in population a on the ratio of the crude divorce rates.
To simplifY the expression, the equation can be written as: [I.sub.c] [I.sub.m] x [I.sub.a] [I.sub.r] where [I.sub.c] is the comparative index of the crude divorce rate; [I.sub.m] the comparative index of the proportion married; [I.sub.a] the comparative index of the age composition of the married population; and [I.sub.r] the comparative index of the age-specific divorce rate of the married population. If [I.sub.c] is greater than 1, the comparison population has a higher crude divorce rate than the reference population. If [I.sub.m] [I.sub.a] or [I.sub.r] is larger than 1, then respectively the proportion of married people in the comparison population, their age structure, or their age-specific divorce rate is more favourable to a high crude divorce rate than in the reference population. If an index is less than 1, the corresponding factor is less favourable to a high crude divorce rare than in the reference population.
It should be noted that [I.sub.r] is the standardized divorce ratio in the indirect standardization of the divorce rate. It can be conceptualized as the ratio of the actual number of divorces in the comparison population to the expected number of divorces in the population if it had the same age-specific divorce rates as the standard population. This ratio is used as a scale factor. The product of the crude divorce rate of the reference population and the standardized divorce ratio is the indirectly standardized divorce rate. Thus the crude divorce rate of the comparison population can be expressed as the product of [I.sub.m] [I.sub.a] and the indirectly standardized divorce rate.
In applying the decomposition to a series of comparison populations, the following points should be emphasized. First [I.sub.m] and [I.sub.a] are more comparable than [I.sub.r] when they are calculated for different comparison populations. In measuring the effect of marriage patterns, [I.sub.m] compares the proportion currently married in all comparison populations against that in the reference population. Similarly, the same set of standard age-specific divorce rates is used in computing [I.sub.a] for all comparison populations. In contrast, the calculation of [I.sub.r] is based on the principle of indirect standardization. Since [I.sub.r] indices for different populations are based on different age compositions, the influence of different age structures is not completely eliminated. Secondly, as in any standardization, the value of the indices can be affected by the selection of the reference population. However, in most cases changing the reference population has only a marginal effect on the outcome.
The comparative indices by birthplace
The method described above has been used to decompose the 1996 crude divorce rates of all immigrant groups in Australia for which divorce data are available. The data are the number of divorces by birthplace of husbands and wives (Australian Bureau of Statistics 1997), and 1996 Census data on the number of males and females by age, marital status and birthplace. The reference population is the 1996 total population of Australia. The age-sex-specific divorce rates of the reference population are those computed for all married men and women in Australia in 1996 (Australian Bureau of Statistics 1997). As shown in Figure 1, the divorce rate is highest in the age group 25-29 for both men and women and then declines rapidly with age.
Table 1 shows how immigrant groups differ in their age composition and proportions currently married. Most of the European-born groups had between 18 and 25 per cent of their population aged 15-44 years in 1996. In contrast, the more recently arrived Asian born groups had between 50 and 75 per cent of their population aged 15-44. The proportion married also varied widely by birthplace. More than 70 per cent of the European-born migrants were currently married in 1996 compared with about 50 per cent or less among groups from Asian countries. The Australian-born group is included for comparison.
The results of the decomposition are presented separately for males and females in Table 2. The comparative index of the crude divorce rate [I.sub.c] varies considerably. For both men and women, the highest indices are recorded among China-born migrants, 2.79 for males and 3.06 for females. In contrast, the lowest indices are found among those born in Italy, 0.81 and 0.60 respectively.
The comparative index of the proportion married Im is greater than 1 for almost all immigrant groups. This indicates that the proportion married is generally higher in the immigrant groups than in the total Australian population. Other things being equal, this would lead to a higher crude divorce rate. As expected, high indices (above 1.5) are recorded for migrants born in Greece, Italy, Malta, Turkey, Netherlands, Egypt, Lebanon, former Yugoslavia and China. These populations have relatively few single people compared to the reference population because they either have an older age structure (the European-born who migrated in the 1950s and 1960s), or have a higher proportion of people married at younger ages (migrants from Lebanon and Turkey). In 1991, more than 70 per cent of women aged 20-24 who were born in Lebanon or Turkey were already married compared with 23 per cent of Australian-born women of the same age (Khoo and Shu 1996). Low indices (about 1.0 or less) are observed among people born in Hong Ko ng, Singapore and Thailand, and male migrants from Thailand in particular. These populations have a larger proportion single in the 20-24 year age group, many of whom are likely to be students at Australian tertiary institu-tions (Andressen 1997).
The comparative index [I.sub.a] shows that the effect of the age composition of married people on the crude divorce rate also varies considerably by birthplace, ranging from 0.61 for Hungary-born males to 1.34 for Thailand-born females. High indices are also shown for migrants born in Fiji, Turkey, Vietnam, Philippines and Hong Kong. These are largely new migrants, and the age structure of the married population is compara-tively young in these groups. This contributes to higher crude divorce rates because the divorce rate is usually higher among younger adults, as shown in Figure 1. The low indices for migrants from European countries, between 0.6 and 0.8, reflect the older age compositions of their married populations, which would lower their crude divorce rates.
The variation in the comparative index of the age-specific divorce rates for married people [I.sub.r] from 0.44 for women born in Sri Lanka to 1.92 for women born in Thailand, is much smaller than that in the comparative index of the crude divorce rate [I.sub.c] While in most cases, groups with a high value of [I.sub.c] also have a high value of [I.sub.r] there are some noticeable exceptions. These include migrants born in Turkey and Lebanon for whom is greater than 1, but [I.sub.r] is less than 1. This indicates that the high crude divorce rates in these two groups are the result of a high proportion of married people and a high proportion of young people among the married population. The divorce rate of the married population in these groups is actually lower than that of the married total population after standardizing for age and proportion married. A substantial positive difference between [I.sub.c] and [I.sub.r] is also observed for migrants born in Vietnam, although their [I.sub.r] is still greater tha n 1. In contrast, for migrants from the former USSR and Hungary, and for men born in Thailand, the [I.sub.r] index is greater than the [I.sub.c] index. Their actual divorce level is therefore higher than that indicated by the crude divorce rate.
Standardized divorce rates by birthplace
Indirectly standardized divorce rates per 1000 married men or women by birthplace were obtained from the product of the standardized divorce ratio and the divorce rate of married people in the reference population, which was 12.9 per 1000 married men or women (Australian Bureau of Statistics 1997). Figure 2 shows the indirectly standardized divorce rates by birthplace ranked by female rate. The rates vary from less than 10 per 1000 married men or women for migrants born in Sri Lanka, Greece and Italy to just under 25 per 1000 for migrants from Thailand, China and Hungary.
There are also sex differences by country of origin, with the male rate appreciably higher than the female rate for migrants from Austria, Egypt, former Yugoslavia, Turkey and Sri Lanka, and the female rate distinctly higher than the male rate for migrants from Thailand and the Philippines. These sex differences are due to the fact that not all men and women are divorcing spouses who are born in the same country A study of intermarriage between immigrants and Australian-born or other overseas-born persons shows that the rate of intermarriage is higher for some groups than others and that it can also vary by sex within each group (Penny and Khoo 1996). The intermarriage rates were much higher for women than men from Thailand and the Philippines, for example. The observed sex differences in the standardized divorce rate would suggest a higher rate of divorce for marriages between women born in Thailand or the Philippines and men not born in these countries compared with marriages where both spouses are born in Thailand or the Philippines.
The differences in immigrant divorce rates by birthplace, after standardizing for differences in age and marriage compositions, may be related to their different cultural and religious backgrounds and may be indicative of the divorce rates in their countries of origin. However, divorce rates are not available for many countries of origin of the immigrant groups included in this paper so that a systematic comparison is not possible. It is also possible that the high divorce rates of some birthplace groups are related to aspects of their migration and adjustment to a new environment. Examination of this hypothesis would require individual-level data on immigrants' migration and settlement experiences and their marriage outcomes, which are also currently not available.
Conclusion
Crude divorce rates of the immigrant population in Australia are not comparable by birthplace because the different birthplace groups have different age structures and marriage patterns. In this paper, we have suggested applying the decomposition method of conventional demographic analysis to the crude divorce rate to obtain indices of the effects of age structure, marriage patterns and the level of divorce. The index of the divorce rate of the married population is both a better indicator of the level of divorce and a more comparable measure. It is equivalent to a standard divorce ratio and can be multiplied by the divorce rate of married men and women in the reference population to obtain the age-standardized divorce rate of the married population in each birthplace group.
The results of the decomposition show that the level of divorce is much lower for most immigrant groups than that indicated by their crude divorce rates. However, considerable differences in the divorce rates among the married population by birthplace still remain, with some birthplace groups having divorce rates that are twice as high, and others only half as high, as that for the total Australian population.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
Table 1
Percentage of population aged 15-44 and percentage currently married by
sex and birthplace, 1996
Males Females
Country of birth
% currently
% aged 15-44 married (a) % aged 15-44
Australia 46.0 39.0 44.9
Austria 20.4 70.3 23.7
Canada 58.9 50.3 61.4
Chile 56.4 59.6 57.1
China 54.2 71.0 54.0
Egypt 33.2 75.3 33.2
Fiji 64.1 58.7 67.3
Former USSR 22.2 66.4 21.6
Former Yugoslavia 37.3 73.2 41.6
Germany 24.4 70.1 24.8
Greece 23.1 84.9 22.5
Hong Kong 66.4 42.7 67.9
Hungary 18.2 70.3 21.6
India 51.9 65.5 49.2
Indonesia 65.9 45.0 65.2
Italy 18.8 84.5 19.1
Lebanon 57.3 74.3 64.2
Malaysia 64.6 47.7 65.5
Malta 26.4 79.3 27.7
Netherlands 21.3 78.0 23.4
New Zealand 61.7 46.7 62.6
Philippines 60.6 49.0 64.6
Poland 32.3 67.2 34.1
Singapore 67.7 40.0 68.2
South Africa 54.3 54.7 54.3
Sri Lanka 53.9 64.1 53.8
Thailand 62.8 21.1 69.9
Turkey 62.2 74.7 64.5
UK & Ireland 39.1 68.5 39.0
USA 49.9 52.0 55.7
Vietnam 74.5 54.3 73.3
Females
Country of birth
% currently
married (a)
Australia 39.6
Austria 60.7
Canada 51.2
Chile 60.9
China 60.4
Egypt 66.9
Fiji 60.3
Former USSR 52.3
Former Yugoslavia 70.7
Germany 61.9
Greece 78.2
Hong Kong 47.8
Hungary 57.9
India 65.9
Indonesia 49.6
Italy 74.4
Lebanon 74.4
Malaysia 51.3
Malta 75.5
Netherlands 67.7
New Zealand 47.5
Philippines 63.4
Poland 56.4
Singapore 45.2
South Africa 54.2
Sri Lanka 62.6
Thailand 42.4
Turkey 74.2
UK & Ireland 62.5
USA 52.3
Vietnam 55.3
Note: (a)Including those currently separated.
Source: ABS, 1996 Census Customized Matrix Table CS068.
Table 2
The four comparative indices for males and females by birthplace, 1996
Males Females
Country of birth
[I.sub.c] [I.sub.m] [I.sub.a] [I.sub.r] [I.sub.c]
Australia 0.90 0.87 1.03 1.03 0.93
Austria 1.62 1.58 0.73 1.45 1.11
Canada 1.61 1.13 1.14 1.30 1.84
Chile 1.64 1.33 1.16 1.10 1.88
China 2.79 1.59 1.08 1.68 3.06
Egypt 1.68 1.69 0.87 1.18 1.17
Fiji 2.49 1.32 1.25 1.56 2.76
Former USSR 1.16 1.49 0.61 1.31 1.08
Former Yugoslavia 1.18 1.64 0.92 0.80 0.91
Germany 1.72 1.57 0.80 1.40 1.45
Greece 1.00 1.90 0.73 0.97 0.68
Hong Kong 1.01 0.96 1.23 0.89 0.93
Hungary 1.74 1.58 0.61 1.88 1.51
India 1.28 1.47 1.07 0.84 0.97
Indonesia 0.93 1.01 1.10 0.86 0.97
Italy 0.81 1.89 0.66 0.66 0.60
Lebanon 1.72 1.67 1.15 0.92 1.57
Malaysia 0.87 1.07 1.14 0.74 0.97
Malta 1.19 1.78 0.85 0.81 0.95
Netherlands 1.44 1.75 0.75 1.15 1.29
New Zealand 1.54 1.05 1.16 1.31 1.58
Philippines 1.27 1.10 1.25 0.96 1.96
Poland 1.37 1.51 0.74 1.27 1.25
Singapore 1.13 0.90 1.13 1.15 1.22
South Africa 1.21 1.23 1.08 0.94 1.20
Sri Lanka 0.90 1.44 1.11 0.58 0.69
Thailand 0.95 0.47 1.32 1.58 2.41
Turkey 1.97 1.67 1.22 0.99 1.63
UK and Ireland 1.61 1.53 0.89 1.21 1.42
USA 1.66 1.16 1.12 1.32 1.53
Vietnam 2.03 1.22 1.26 1.37 2.00
Females
Country of birth
[I.sub.m] [I.sub.a] [I.sub.r]
Australia 0.89 1.02 1.04
Austria 1.37 0.70 1.18
Canada 1.16 1.16 1.41
Chile 1.38 1.12 1.24
China 1.50 1.10 1.89
Egypt 1.51 0.87 0.91
Fiji 1.36 1.28 1.62
Former USSR 1.18 0.65 1.44
Former Yugoslavia 1.60 0.95 0.61
Germany 1.40 0.74 1.44
Greece 1.77 0.70 0.56
Hong Kong 1.08 1.21 0.73
Hungary 1.31 0.65 1.82
India 1.49 1.09 0.62
Indonesia 1.12 1.16 0.76
Italy 1.68 0.63 0.58
Lebanon 1.68 1.23 0.78
Malaysia 1.16 1.15 0.75
Malta 1.71 0.80 0.71
Netherlands 1.53 0.70 1.23
New Zealand 1.07 1.15 1.31
Philippines 1.43 1.24 1.13
Poland 1.28 0.80 1.25
Singapore 1.02 1.15 1.06
South Africa 1.23 1.06 0.95
Sri Lanka 1.42 1.13 0.44
Thailand 0.96 1.34 1.92
Turkey 1.68 1.28 0.78
UK and Ireland 1.41 0.87 1.18
USA 1.18 1.17 1.13
Vietnam 1.25 1.25 1.30
(1.) Upper case letters are used for rates and proportions. Lower case letters are used for numbers of people. Subscripts are used to refer to a particular age group or population.
(2.) Sex is not specified in the discussion of decomposition procedures.
References
Andressen, C. 1997. The location and characteristics of Asia-born overseas students in Australia. Pp.75-98 in J.E. Coughlan and Dj. McNamara (eds), Asians in Australia: Patterns of Migration and Settlement. Melbourne Macmillan.
Australian Bureau of Statistics. 1997. Marriages and Divorces Australia 1996 Catalogue no. 3310.0. Canberra.
Burch, T.K. 1980. The index of overall headship: a simple measure of household complexity standardized for age and sex. Demography 17:15-37.
Coale, AJ. 1965. Factors associated with the development of low fertility: a historic summary. Pp.205-209 in United Nations, World Population Conference Belgrade, Vol.2.
Khoo, S.E. and J. Shu. 1996. Immigration Family Formation Patterns in Australia. Canberra Department of Immigration and Multicultural Affairs.
Penny, J. and S.E. Khoo. 1996. Intermarriage: A Study of Migration and Integration. Canberra: Australian Government Publishing Service.
Smith, D. 1992. Formal Demography. New York Plenum Press.
Young, C.M. 1987. Migration and mortality: the experience of birthplace groups in Australia. International Migration Review 21(3):531-554.
Young, C.M. 1991. Changes in the demographic behaviour of migrants in Australia and the transition between generations. Population Studies 45(1):67-89.
Yusuf, F. and I. Rockett. 1981. Immigrant fertility patterns and differentials in Australia, 1971-76. Population Studies 35(3):413-424.
Siew-Ean Khoo +
+ Address for correspondence: Demography Program, Research School of Social Sciences, The Australian National University, Canberra ACT 0200, Australia.
COPYRIGHT 2001 The Australian Population Association
COPYRIGHT 2003 Gale Group