The Educational Attainments of the “Second Generation”: A Comparative Study of Britain, Canada, and the United States
by Catherine Rothon, Anthony Heath & Laurence Lessard-Phillips - 2009
Background: This analysis compares the educational attainments of the “new” second generation in Britain, Canada, and the United States using three nationally representative datasets.
Objective: To assess how the second generation has fared within Western educational systems. The study examines the achievements of seven minority ethnic groups: Africans, Caribbeans, Chinese, Filipinos, Indians, Irish, and Pakistanis.
Setting: Britain, Canada, and the United States.
Research Design: Secondary data analysis
Conclusions: The study suggests that there is a strong association between the educational level of the parental generation and that of the second generation. There is substantial intergenerational progress (measured relative to the majority population in the country of destination), especially among women. Most groups perform as well as or better than members of the majority population of the same age and similar parental background. Chinese of both sexes are notable for their high performance. Indians also tend to make strong intergenerational progress; for Caribbeans, Africans, and Filipinos, this is more muted. The performance of the second generation in Britain is slightly poorer than that in the other countries. This is probably explained by the lower selectivity of the first generation in Britain rather than by institutional features.
This article compares the educational attainments of the new second generation in Britain, Canada, and the United States. The great migration to affluent Western countries from a wide diversity of origin countries in the last third of the 20th century has been one of the most striking developments of recent social history. As a result of these migrations, a growing number of second-generation children are now moving through the Western educational systems and into the labor market. Their experience is of great significance and has important implications for equality of opportunity and social integration. Pessimists foresee problems for the members of the new second generation, with the possibility of downward assimilation into a disadvantaged underclass for some minorities. Optimists hope for upward mobility, especially through education. In both accounts, education is key, both in the form of human capital, with its valuable payoffs in the labor market, and as a crucial mechanism of social integration. Shared experiences of higher education in particular are often believed to promote tolerance and the blurring of group boundaries.
The key question that we ask in this article is: How has the new second generation fared within Western educational systems? Is there evidence of disadvantage, perhaps arising from discrimination or other social processes that lead to downward assimilation? Or have minorities been able to take advantage of Western educational opportunities, often far superior to those in their parents countries of origin? Do some national contexts, perhaps as a result of affirmative action or multicultural policies, provide for more favorable educational achievements?
This article will focus on three countries: Britain, Canada, and the United States. Canada often proves to be distinctive in cross-national research, partly because it has had an unusual point system for immigration that is believed to have resulted in positively selected immigrants who may therefore provide particularly favorable socioeconomic backgrounds for their children, and partly because Canada has had a long history of multicultural policies designed to promote civic integration. The United States and Britain, on the other hand, have not been as selective as Canada in their immigration policies and would not therefore be expected to see such favorable educational outcomes among the children of immigrants. However, the United States is notable for its affirmative action policies within the higher education sector, and although these were primarily instituted to help African Americans, they could well have benefited other groups such as the Black Caribbeans and Africans. In contrast, Britain has traditionally focused on class inequalities and on policies to redress disadvantages associated with social class and has not as yet developed strong policies on the ethnic side. Britain also continues to have an educational system that, although moving toward a North American model of contest mobility, still retains some features of its older sponsored system (for the distinction between sponsored and contest systems of education mobility and its application to Britain and the United States, see the classic paper by Turner, 1960). It may well be that a contest system will provide greater opportunities for ethnic minorities to advance through education.
We focus on seven key minority ethnic groups in our three countries: Africans, Caribbeans, Chinese, Filipinos, Indians, Irish, and Pakistanis. These groups have been chosen on the basis that their members can be found in at least two of our countries, thus enabling detailed comparative analysis of the same groups. We also include Mexicans in the United States who can perhaps be regarded as structural equivalents of Pakistanis in Britain. Our comparison of these minorities also gives an interesting range of ethnic groups. The Irish provide an example of a long-established group of European ancestry that is similar in many ways to the other classic migrant European groups, such as Poles and Italians, that have been highly successful in joining the American (and Canadian) mainstream in the past. Afro-Caribbeans, Filipinos, Mexicans, and Pakistanis are representative of the new migrant groups from less developed countries and are largely composed of labor migrants who, in the first generation, tended to enter less skilled work (and thus to provide working-class environments for the second generation). In contrast, Indians and Chinese represent two minority ethnic groups that have performed highly in terms of educational attainment and are often regarded as being the success stories of the new immigration. Finally, Black Africans are an interesting group, largely coming from less developed countries but often themselves being quite highly educated. We compare the educational experience of these minority groups with that of the White majority populations in each country.
Broadly speaking, our expectations, on the basis of previous research, are that the children of the labor migrants will be relatively disadvantaged in comparison with the native-born members of the White majority population and will therefore be most at risk of downward assimilation; that the Irish will achieve at least as highly as the majority population and possibly better; and the Indians and Chinese will be relatively high-achieving groups that have outstripped the White majority.
Following two introductory sections describing general patterns of migration to the three countries and key features of the three educational systems, the third section of the chapter provides basic descriptive statistics on the educational achievements of these second-generation minority ethnic groups. It compares their educational attainments with those of the respective majority populations. This will involve analysis of three key data sets: the pooled General Household Surveys (GHS) in Britain, the Public Use Microdata Files (PUMF) for the 2001 census in Canada, and the pooled Current Population Surveys (CPS) for the United States. These three data sources provide us with the most authoritative and up-to-date sources on the educational achievements of the adult populations of the three countries and also are sufficiently large in size for us to identify samples of our main second-generation groups.
The fourth section then considers the extent to which these differences between the educational achievements of the second generation and their respective majority populations can be explained by the social background of minority ethnic people. Social background is a major factor, perhaps the major factor, in explaining the educational fortunes of young people generally. Because minority ethnic groups typically, although not always, have disadvantaged social backgrounds (reflecting to some extent the lack of opportunities in the countries of origin and the difficulties experienced by the migrant generation in the labor market), it is likely that a substantial part of minority ethnic disadvantage in education can be explained by general processes of social reproduction (see, e.g., Heath & Brinbaum, 2007; Kao & Thompson, 2003). Data on class origins are not directly available from the data sets, but if it is assumed that the class distribution of the early arrivals in the first generation gives an indication of the social origins of the second generation, it is possible to make some estimates regarding the extent to which social background might explain the pattern of ethnic inequalities in education.1 Because the three countries have had rather different rules for immigration, it is also quite possible that some of the cross-national differences in the second generations educational outcomes can be explained by the varying level of selectivity of the first generation.
Finally, we will consider possible explanations for any remaining cross-national differences that remain after taking account of differences in selectivity and social background.
THE MINORITY ETHNIC POPULATIONS OF BRITAIN, CANADA, AND THE UNITED STATES
As with most developed countries in Europe, Britain experienced a major influx of immigrants in the 20th century. The great majority of this migration was what might be termed postcolonial, consisting of migrants coming from the territories of the former British Empire. In fact, the 1948 British Nationality Act gave an unrestricted right to settle and work in Britain to 800 million members of the colonies and commonwealth countries worldwide (R. Hansen, 2000).
A substantial group, mainly Catholics, from Ireland arrived in the first half of the century. These were originally British citizens before the partition was formalized in 1922. They continued to have free entry to Britain after partition and continue to do so through their membership of the European Union. Traditionally, they were discriminated against, largely on religious grounds. This was particularly the case in Northern Ireland. At the time of the partition of Ireland, two thirds of the regions population were Protestants. The first prime minister of Northern Ireland, Sir James Craig, described the state as having a Protestant Parliament for a Protestant people. The state effectively discriminated against Catholics in housing and jobs as well as in political representation. Membership of the Orange Order, which was often essential for progress in politics and business, was forbidden to Catholics. Most local councils were dominated by unionists (Fitzduff, 1996). Discrimination in Northern Ireland continues to remain a pressing issue.
The years following the World War II witnessed the entrance of sizeable groups of visible minorities to Britain. Black Caribbeans started arriving in Britain in large numbers in the 1950s. There was some direct recruitment, particularly for nurses to work for the National Health Service (NHS), but the migration was mainly voluntary. Women generally went into the more highly skilled jobs, with men taking up less skilled positions. There were significant numbers of female-headed households, perhaps a remnant of slavery and plantation life. Early research on the experiences of the first-generation migrants suggested that they experienced discrimination in housing and were often unable to secure employment appropriate to their qualifications and skills (Castles & Kosack, 1973; Daniel, 1968; for research on the first generation of postwar migrants to Britain, also see Chiswick, 1980; Heath & Ridge, 1983; McNabb & Psacharopoulo, 1981; Stewart, 1983).
Indian immigrants to Britain came predominantly from urbanized settings and were often highly skilled. South Asian Indians were divided into two groupsSikhs (mainly from the Punjab) and Hindusbut there was also an important group of Indians who came from East Africa (the twice migrants) following independence of these East African countries and consequent policies of Africanization. Many of these East African Indians had achieved higher level occupations in business or government, and the high skill levels of these migrants from East Africa and their experiences in an urban setting also meant that they arrived in Britain with a considerable command of mainstream skills such as language, education, and familiarity with urban institutions and bureaucratic processes (Bhachu, 1985). These skills were likely to have contributed to a smoother transition into educational institutions for the second and third generations as well as a greater ability on the part of the parents to offer educational support at home. In contrast, Pakistani and Bangladeshi migrants to Britain arrived somewhat later and were more likely to come from rural and less educated backgrounds.
The Chinese population in Britain originates from a wide number of countries. According to Cheng (1994), the single largest group of foreign-born Chinese in Britain is from Hong Kong, at 35% of the sample that she used. China and Vietnam make up about 11% each. The rest come primarily from Southeast Asian countries such as Malaysia, Singapore, and the Philippines. The Chinese migration into Britain before World War II was very limited. The migration following this was classically economic; the demand on the part of the British for ethnic cuisine and the deterioration of economic conditions in rural Hong Kong were the major factors. There has also been more recent migration from Hong Kong, which has been highly selective and composed of persons with high human and financial capital. Of all the minority ethnic groups in Britain, the Chinese are the most diverse in terms of background characteristics.
Africans are the most recent group of migrants to Britain, having arrived in the country in significant numbers over the last decade. Their relatively high qualifications might be explained by the fact that many came to Britain for higher or even secondary boarding education; they have been termed students who stayed (Daley, 1996). Their migration can be characterized as the selective migration of the highly educated.
In total, according to the 2001 census, around 8% of the adult British population is now of Black or ethnic minority origin (including the growing group of mixed origin). However, this is considerably less than in the classic immigration countries of the United States and Canada.
With the exception of the aboriginal population, most ethnic groups in Canada are relatively recent arrivals. The first colonizers, the French, began to settle in Canada at the beginning of the 17th century. A century and a half later, in 1760, New France was conquered by Britain,2 and by the mid-19th century, largely through the control of migration, the British became the majority of the Canadian population. Although politically correct politicians commonly refer to those two groups as charter groups, many critical analysts have characterized their power relationship as unequalespecially until 1969, when the equality of status of French and English as official languages was finally officially recognized (although their actual inequality persists). The two nations within the Canadian state have also developed a different historical relationship with migration, which has evolved over time, a reality reflected by the fact that the only province with a French majority (Québec) exercises a quasi-constitutional responsibility in the selection and integration of immigrants coming to its territory.3
The first non-English or non-French speaking immigrants came from Germany; they arrived between about 1780 and well into the 1800s (Angelini & Broderick, 1997; Herberg, 1989). Irish farmers followed from the 1830s. The Irish wave peaked between 1846 and 1854, bolstered by the potato famine (18451846). Finally, with the 1850s Gold Rush in British Columbia, Chinese and Sioux Indians arrived in significant numbers, initially from the United States (Isajiw, 1999).
Following the 1867 British North America Act, a priority was to expand the labor force through immigration from Europe. The government sought to do this actively with offers of free land and, in some cases, help with transportation. These immigrants were welcomed as permanent future citizens (Kelley & Trebilcock,1998; Reitz, 1988; Richmond, 1990). A second influx occurred in 19011914 following aggressive recruitment from Northern, Central, Eastern, and Southern Europe. Ukrainians and Italians were prominent groups of immigrants at this time. A number of groups from outside Europe also started to arrive, including approximately 5,000 Sikhs from the Punjab between 1905 and 1908 (Kurian, 1991) and over 8,000 Japanese in 1907.
Immigration was low during and between the two World Wars. It was the economic boom following World War II that reopened Canada to migrants on a large scale. Groups arriving in the years following the end of the war included Poles, Hungarians, Italians, Portuguese, and Greeks.
As outlined, migrants to Canada before the 1960s came primarily from European countries. From the mid-1960s, recruitment was widened to include groups from less developed countries. For example, large number of West Indians started to come to fill domestic, nursing, and factory jobs (Kalbach & Kalbach, 1999). Indian and Chinese migration largely dates from this period. From the 1970s, Central and South American migrants arrived in large numbers, and there was an influx of Vietnamese in the late 1970s and 1980s after the Vietnam War.
An important feature of the Canadian immigration system since the 1960s has been the selection of a significant number of applicants for migration on the basis of a point system in which employability and skills are central. Before then, immigrant selection was mostly based on national origin; immigration policies focused on selecting immigrants who were deemed socially (and racially) acceptable for the country. There existed a hierarchy of preferred nations for immigrant selection, with individuals from Western Europe and the United States being preferred over individuals from Northern, Central, Southern, and Eastern Europe. Individuals from less preferred countries were usually admitted to the country to perform less desirable work (Castles & Miller, 1998; Li, 2003). Pressures to shift Canadian immigration policy away from the nationality-based criterion were brought about by postwar economic and political changes. In 1962, immigration policy shifted its admission criterion toward a skills-based approach. This change was standardized in 1967 by the introduction of the point system (Kelley & Trebilcock, 1998).
This point system, differs slightly between English Canada and Québec, applies only to the independent class (people selected for their likelihood of integrating successfully), which represents roughly half (60% in Québec) of the annual immigration flux. The two other categories (family reunification and refugees), which account for the rest, are not selected. Caribbeans, for example, have usually come under the family reunification or refugee class more often than Chinese, Africans, and Indians. As will be seen in the analysis that follows, this system may have led to the recruitment of a larger proportion of well-qualified groups of migrants.
The longest resident minority ethnic groups in the United States are indigenous Native Americans and persons of Mexican origin who lived in the Southwest before its conquest. The ancestors of African Americans, who now make up about 14% of the population, arrived before 1808, the year in which importing slaves became illegal. Prior to 1860, most voluntary immigrants were from Germany, Ireland, the Netherlands, and the United Kingdom. They were later joined by Scandinavians and, from the 1880s, growing numbers of migrants from Southern and Eastern Europe, including Italy, Russia, Greece, and Poland. A small number of Chinese also came in the late 19th century to work in agriculture and in mining, but Chinese migration was restricted by the Chinese Exclusion Act of 1882, and a similar law limited Japanese migration in 1924 (Edmonston & Passell, 1994).
Immigration tailed off during World War I, only to rise again thereafter. The largest flows continued to be from Europe, but a number of groups of West Indian origin migrated to the United States during this period (Kasinitz, 1992), and there was a certain amount of seasonal migration from Mexico. However, in 1924, the Immigration Act limited legal immigration to 2% of the foreign-born population of any given country of origin based on the 1890 census, thus favoring immigration from Northern and Western Europe (Edmonston & Passell, 1994). These limitations, combined with economic depression and war, suppressed migration to 10% of its earlier level until the mid-1960s, when it began to rise again.
Change came with the 1965 Hart-Celler Act, which ended discriminatory national quotas and made family reunification the main criterion for immigration; it now accounts for three quarters of admissions. As a result, the majority of migrants now come from South and Central America, Asia, and Africa rather than from Europe. Although some highly skilled migrants are admitted on work visas, there is no point system as in Canada, so migrants tended to possess a lower level of skills on average. The United States has also admitted considerable numbers of refugees from countries including Vietnam, Laos, Cambodia, Cuba, Haiti, and the former Soviet Union. The skill level of these migrants varies considerably (Bean & Stevens, 2003).
Mexicans are now by far the largest single group of immigrants to the United States. Mostly involving seasonal and temporary agricultural work, Mexican migration was relatively low between the 1930s and 1960s, but from the 1970s, larger numbers of Mexicans began to migrate and take jobs in the service, retail, and construction sectors in cities, and in agriculture. A series of amnesty programs in the 1980s and 1990s enabled many unauthorized Mexicans to legalize their status. By 2000, there were an estimated 20.6 million Mexicans living in the United States, the vast majority of them low-skilled labor migrants with relatively low levels of education (Bean & Stevens, 2003).
In addition to Mexicans, sizeable numbers of immigrants come from Peru and Ecuador, as well as Colombia, Argentina, and other South American countries. Although labor migrants dominate these flows, financial crises and political tensions in South America have also led to substantial middle-class migration to the United States. Migration from the Caribbean has also swelled, not only from the English-speaking former British colonies (Jamaicans are the most numerous) but also from the Dominican Republic and French-speaking Haiti (Rumbaut & Portes, 2001). Immigration from Africa has further diversified Americas Black population. In this article, however, we focus on the children of migrants from the English-speaking Caribbean rather than from the Spanish- or French-speaking countries to maximize comparability.
Migrants from Asia are extremely diverse, including refugees from Vietnam and Cambodia, and also highly skilled professionals from India who have, on average, higher levels of education than native-born Whites. The Philippines, followed by India, is the top sending country in Asia, but immigration from China has grown rapidly since limitations on emigration were eased in the late 1970s. There are now over a million foreign-born Chinese in the United States, or 3.2% of the population. Chinese make up one of the most diverse immigrant populations, including both low-skilled labor migrants who enter under the family reunification policy or without documentation, and highly skilled professionals from Taiwan, China, and the Chinese diaspora.
Table 1 summarizes the current position of in our three countries, focusing on the proportions of our main minority groups in the three destination countries. We restricted the table to men and women aged 2544 because this is the age range that we study in more detail in the remainder of this article. (For details of the measurement of the White majority population and of the minority groups, see the appendix.)
Table 1: Second-Generation Ethnic Minorities in Britain, Canada, and the United States: Percentage of Adult Population Aged 2544
Sources: British GHS 19912000, Canadian Census PUMF 2001, American CPS 19942003. Weighted data.
Overall, we can see that Canada and the United States have very substantial proportions of immigrants and children of immigrants. Combined, they make up over one fifth of the population of this age range in the United States and over a quarter in Canada. Britain, in contrast, has seen less immigration, with over 80% of the population being White British.
Britain also stands out in that its main groupsAfricans, Caribbeans, Chinese, Indians, Irish, and Pakistanisconstitute a larger proportion of the second generation than they do in Canada or the United States. In this respect, then, Britain is less diverse than Canada and the United States in its ethnic composition (many of the other second-generation group, especially in Britain, came from developed Western countries such as Western Europe, the United States, and the old commonwealth [Canada, Australia, and New Zealand]).
THE EDUCATIONAL SYSTEMS OF THE THREE COUNTRIES
We turn next to the nature of the educational systems in our three countries. Although there are some strong similarities in their systems, there are also important differences that may have implications for the education of the second generation. We should note that in all three countries, education is the responsibility of local rather than national bodies, and there are many differences of detail within each country. However, these local arrangements tend to be variants on a dominant national system, and it is not a gross simplification to talk of national educational systems. However, we should be aware that there are some important differences between the system in Quebec and those in the rest of Canada,4 and similarly, there are some important differences between Scotland and the rest of Britain.5
All three countries have, broadly speaking, comprehensive systems of education with a period of compulsory education lasting from 5 or 6 years of age until 16.6 Some students terminate their schooling at this point and enter the labor market, but many (most students in Canada and the United States) have an 2 additional years of education in secondary school, until the age of 18, at which point they transition either to college/university or into the labor market. Therefore, a simple student classification that applies reasonably well to all three countries is among incomplete secondary (the students who drop out at age 16), completed secondary (those who continue until at least age 18), and completed tertiary (those who complete a university degree). In the United States and Canada, there is also an important distinction between those who have some postschool tertiary education and those who have completed a university degree. We therefore distinguish four levels of education. (In Britain, we can also identify people with some postschool qualifications below degree level. However, this has a rather different meaning than in Canada or the United States and more often consists of vocational qualifications. For details of the measurement in the three countries, see the appendix.)
Although many institutional details vary across the three countries (e.g., the roles of private and church schools), perhaps the key one for our purposes is that Britain has a system of public examinations at ages 16 and 18 that has no real equivalent in Canada or the United States. The majority of students in England and Wales take the General Certificate of Secondary Education (GCSE) examinations or equivalents at age 16.7 A large proportion of students leave school after completing these qualifications. This sort of public examination in Britain has a long history and goes back to the prewar period. These examinations were an integral aspect of the old selective system of education that Britain had before comprehensive reorganization took place after 1965, and although the details of the examinations have been changed to cope with the increased numbers entering for them, they can be regarded as an important legacy of Britains historical sponsored system of education.
A similar scenario does not exist in the United States or Canada. This structural difference has implications for the position of a 16-year-old in the labor market in the three countries. In the United States, Model and Fisher (2007) employed a binary logistic model of avoidance of unemployment and found that the coefficients for those with no qualifications and those with schooling up to lower secondary level were identical. In Britain, by contrast, there is a significant difference in the chances of being unemployed for those with no qualifications and those with lower secondary qualifications such as GCSE taken at age 16 (Cheung & Heath, 2007). The difference between the coefficients for entry to the salariat is even greater.
At the upper secondary level, the United States and Canada are again very similar, with a high school diploma (usually at age 17 or 18) representing completed secondary education. In England and Wales, students at this age take the GCE Advanced Level examination, which is the usual entry qualification for university. These A levels are awarded by national examination bodies, unlike the high school diplomas, which are certified by the schools. Secondary school graduation requirements vary between states and provinces, which have specific grade and credit prerequisites associated with secondary school graduation but no national examinations (Council Ministers of Education Canada [CMEC], 2006; U.S. Department of Education, 2007).
After secondary education, there is a wide diversity of possible routes open to students in all three countries. In the United States, there are various junior colleges that provide 2-year courses. Many students may also move on to some form of tertiary education without actually completing their degree. In Britain, there is a wide diversity of further education courses, many leading to vocational qualifications. One such qualification is the Higher National Diploma, usually a 2-year course (which in some cases can be converted to a degree with further study). These qualifications may be seen as broadly comparable with junior college in the United States and are included within the incomplete tertiary category. It is argued here, however, that the completion of a degree has the clearest identifiable meaning in all three countries. We therefore focus on that as a benchmark for attainment.
Tables 2a and 2b show the overall picture for men and women in each of the three countries, again focusing on those aged 2544.
Table 2a: Highest Level of Education: Men Aged 2544
Table 2b: Highest Level of Education: Women Aged 2544
Source: British GHS 19912000, Canadian Census PUMF 2001, American CPS 19942003. Weighted data.
As we can see, Britain stands out from the other two countries, with a much higher proportion of students who fail to complete secondary education (over 50% compared with 20% or less in Canada and United States). However, as we suggested earlier, this may be in part due to the option to complete school at 16 with some qualifications, an option that is not available in the United States or Canada. An American or Canadian pupil who dropped out at 16 would be at a great disadvantage relative to British students, who finish compulsory schooling with formal qualifications. There is therefore less impetus to drop out at this stage in Canada and the United States.
At the top end of the spectrum, the United States has more university graduates than Britain and Canada, although the differences are somewhat less marked than at the lower end of the spectrum. Note that one implication of this pattern is that among students who do complete secondary education, in Britain, a rather higher proportion (over one third) actually go on to complete a degree, whereas in Canada and the United States, the proportion of high school graduates who go on to complete a university degree is less than one third. This may be a relic of the old sponsored system of educational mobility in Britain: Those sponsored for elite education had high chances of progressing right through the system, whereas those who were not sponsored were expected to make an early exit into the labor market.
Given the variation in the educational qualifications of the overall populations by country, it is expected that the individual minority ethnic groups will also display cross-national differences in the proportion of their number attaining a qualification at higher secondary or tertiary level across the countries considered here. It is therefore important to take account of the performance of a given minority group relative to the majority population (third or higher generation Whites) in the country in question. This is what we focus on in the next section.
THE EDUCATIONAL ATTAINMENTS OF THE SECOND GENERATION
The discussion now turns to the differential educational attainments of minority ethnic groups within the three countries. We compare these attainments with those of the majority population in the respective countries and with other members of the second generation.
As we noted earlier, the classification of British schooling is largely based on qualifications. Previous research has shown that Black Caribbean boys (not girls) tend to have rather lower rates of securing GCSEs than do British Whites of the same age, whereas Indian boys and girls have higher rates (Connolly, 2006; Rothon, 2007), but there is less evidence on our other main groups. Tables 3a and 3b show the overall patterns in our data set. We ordered the different groups according to the proportions obtaining university degrees. (The ordering by degrees is somewhat different from the order that we would obtain if we ordered by proportions failing to complete secondary education, with Pakistanis in particular being quite polarized.)
Table 3a: Highest Level of Education of the Majority Population and of the Second Generation in Britain: Men Aged 2544
Table 3b: Highest Level of Education of the Majority Population and of the Second Generation in England and Wales: Women Aged 2544
Source: GHS 19912000, weighted data.
The picture is very similar for men and women, although there are some important differences of detail, with a considerable gender difference between Pakistani men and women. The key story, however, is that apart from the Black Caribbeans and the Pakistani women, all other minority groups are more highly educated than the majority population. There is considerable ethnic diversity, however, with the most highly educated groupsthe Indianshaving twice as many members with degrees as the majority population. This is clearly a story of very considerable educational success for the children of immigrants.
It is very important to recognize that our four categories of education are rather broad educational groupings and that there may be some important heterogeneity within each level. In particular, Boliver (2006) has shown that although ethnic minorities are generally overrepresented in tertiary education relative to the White British, they tend to be found more in the lower status institutions, notably the post-1992 universities. These universities have rather lower status than the older research universities such as Oxford and Cambridge, and the other members of the so-called Russell Group. Bolivers research shows that this holds particularly for the Caribbean and African groups, although less so for Indians and Chinese. The educational success of the Africans and Pakistanis may therefore be somewhat less than is shown in Table 3, and the disadvantage of the Black Caribbeans may be rather greater. Nonetheless, it is doubtful that this kind of problem would fundamentally alter the general picture displayed in Table 3.
We now move on to Canada, where we find that the picture is in many respects very similar to that in Britain. Once again, the children of Chinese, Indian, Pakistani, Filipino, and African migrants outperform the majority population by a considerable distance.
Table 4a: Highest Level of Education of the Majority Population and the Second Generation in Canada: Men Aged 2544
Table 4b: Highest Level of Education of the Majority Population and the Second Generation in Canada: Women Aged 2544
Source: PUMF 2001, weighted data.
The most striking finding, however, is that the majority population comes right at the bottom of the order, below the Caribbeans and indeed below the broad residual category of other second generation. The picture of second-generation educational success is thus even clearer cut than it is in Britain.
Although these results may seem very surprising, we should note that they have also been found by other researchers using different data sets. Thus, J. Hansen and Kucera (2004), using the Survey of Labour and Income Dynamics, concluded that the second-generation immigrants actually appear to be better rather than worse off in terms of educational attainment than similarly-aged native Canadians (p. 8). Similar results overall have been found by Aydemir and Sweetman (2006), Boyd and Grieco (1998), Boyd (2002), Davies and Guppy (1998), Guppy and Davies (1998), and Bonikowska (2007). Lessard-Phillips (2007) found that in Canada, Black second-generation individuals had more favorable odds of attainment at the postsecondary level below university but were not different from the majority population with regard to university-level attainment.
In the United States, the story is very similar to that in Britain. The Indians, Chinese, and Black Africans come at the top, a long way ahead of the majority population. Filipinos and Irish also outperform the White majority. However, Black Caribbean men (but not women) are a little behind the White majority, whereas Mexican men and women are quite a long way behind.
Table 5a: Highest Level of Education of the Majority Population and of the Second Generation in the United States: Men Aged 2544
Table 5b: Highest Level of Education of the Majority Population and of the Second Generation in the United States: Women Aged 2544
Source: CPS 1994-2003, weighted data.
These results are broadly in line with those found by previous researchers. In their valuable review of the field, Kao and Thompson (2003) reported that Asians are the most likely to complete college (exceeding the national average), with Asian Indians outperforming Chinese. Blacks, Hispanics, and Native Americans have lower rates of college completion than Whites. Similarly Farley and Alba (2002) found that second-generation individuals of Caribbean origin in the United States had a higher proportion of individuals with a 4-year college degree and a lesser proportion of individuals without high school diplomas. However, as with Britain (and very probably Canada too), it is clear that some ethnic minorities, particularly Hispanics and African Americans, are more likely to be found in lower prestige tertiary institutions (Hearn, 1991; Karen, 2002). This pattern might also apply to Caribbeans. Possibly, then, the results in Table 5 may slightly exaggerate the success of some groups.
Tables 3, 4, and 5 show a remarkably similar picture of second-generation success in all three countries, with the children of Indian and Chinese migrants consistently outperforming the majority population of the same age and the Irish also outperforming the majority population, although by a lesser amount. The position of the Caribbeans varies more between the three countries, but only in Britain do they fall well behind the majority population.
Given the rather different overall educational profiles in our three countries, it is not easy to draw any firm conclusions about cross-national, as opposed to ethnic, differences from Tables 3, 4, and 5. However, odds ratios are a useful statistical measure for assessing relative performance, and they have the useful property that they are independent of the overall totals achieving, say, degree-level qualifications. They are thus a good way of making cross-national comparisons that take account of differences in the overall levels of education in the three countries.
In Table 6, we therefore show log odds ratios (log odds ratios being centered on 0) comparing each ethnic group with the majority population. Log odds ratios with positive signs thus indicate that the group in question is more successful than the majority population, whereas a negative sign indicates that it is less successful. We also show the standard errors. (If we multiply these by 1.96, we can obtain the 95% confidence intervals.) In calculating these log odds ratios, we have also controlled for the year of the survey (in Britain and the United States, setting 2001 as the reference year) and for age (because there may be some variations within our broad age band of 2544 between the different groups).
Table 6a: Second-generation Attainment of Tertiary Education, Relative to the Majority Population: Men Aged 2544
Log odds ratios
Table 6b: Second-Generation Attainment of Tertiary Education, Relative to the Majority Population: Women Aged 2544
Log odds ratios
Note: Log odds ratios are derived from logistic regressions controlling for age and, where appropriate, year. Reference age is 25 and reference year is 2000. Standard errors are given in brackets.
A number of possible odds ratios can be computed from the previous tables, but for simplicity, we focus on those contrasting university degrees (completed tertiary) with lesser educational levels (i.e., the three lower levels combined). Because the nature of a degree is reasonably comparable in all three countriescertainly more comparable than the other qualificationsthis seems the wisest strategy.
One conclusion emerges very clearly from Table 6: Ethnic minorities educational success relative to that of the majority population is much more marked in Canada and the United States than it is in Britain. Although Indians, Pakistanis, Irish, and Africans all have positive log odds ratios in Britain (at least for men), they are noticeably smaller than their equivalents in Canada or the United States. Second-generation success is thus somewhat more muted in Britain than in the other two countries. However, we should not neglect the special disadvantaged position of second-generation Mexicans in the United States.
It is not clear, on the other hand, that any straightforward conclusion can be drawn about the relative success of ethnic minorities in Canada and the United States. In the case of the Chinese, Caribbeans, and Irish, the differences are small and probably would not reach statistical significance if we carried out a formal test. Indians seem to fare better in the United States, but Filipinos fare better in Canada. It would also be unwise to read anything into the log odds ratios for the residual group of other second generation. This residual group is more successful in Britain and least successful in Canada. However, its make-up is very different in the three countries.
Nevertheless, there are some reasonably clear overall conclusions. First, in all three countries, there is substantial evidence of second-generation success, with several of our groups greatly outperforming their respective majority populations. Second, we have clear indications that this success is even greater in the United States and Canada than it is in Britain. Third, as other researchers have found, Indians and Chinese are the more successful groups, with Caribbeans (and Mexicans in the United States) being the least successful. It is important, however, to remember the caveats as well. In particular, there are some methodological differences in the way in which ethnicity has been operationalized in the three countries, and for some of our groups, conclusions are based on rather small sample sizes.
THE SIGNIFICANCE OF SOCIAL BACKGROUND
How, then, might we explain these interethnic and cross-national patterns? A first possibility that we must explore is that they can be explained by the social background of minority ethnic people. The association between social class and educational attainment has been extensively documented in the sociology of education (see for example Breen & Jonsson, 2004; Floud, Halsey, & Martin, 1956; Halsey, Heath, & Ridge, 1980; Heath, 2000; Jonsson & Mills, 1993a, 1993b; Shavit & Blossfeld, 1991). Because the different second-generation groups may well come from more or less advantaged social backgrounds, it is possible that a substantial part of the interethnic differences in attainment that we have just seen can be explained by general processes of class reproduction.8 It is also quite possible that some of the cross-national differences can be explained in this way. For example, the Canadian point system may have led to the recruitment of better educated migrants than in Britain, and in this way, too, general processes of social reproduction might lead to greater second-generation success in Canada than in Britain.
Hence, the question is whether the higher achievements of the Indian and Chinese groups compared with the Caribbeans or Mexicans can be explained by their more advantaged backgrounds and whether the greater success of the second generation in Canada and the United States can be explained by the selection of more highly educated migrants than in Britain. Furthermore, it is only after controlling for social background that we can get a clear picture of whether certain institutional arrangements, such as the affirmative action policies in the United States, provide more favorable environments for minority educational progress.
The ideal strategy for exploring whether social background can explain the differences between groups and countries would be to have individual-level data (of the sort used in traditional sociological inquiries on educational attainment). Such individual-level data would enable us to control statistically for each individual respondents social background. To do this, we would need to have data sets that included, for each respondent, details about their mothers and fathers social background (such as their educational and occupational attainment).
The three main data sets we have used for investigating the educational achievements of the second generation do not include these data. There are some data sets that do include these data in all three countries, but they are relatively small in size and do not enable us to investigate the social background of ethnic minority respondents with any accuracy. In addition, they cannot be used to make sensible comparisons among our three countries because the nature of the relevant data sets is very different.
In Britain, the main available data set is the Youth Cohort Study, which Rothon (2007) and Connolly (2006) have used to show that even after controlling for parental social class, Chinese and Indian students have higher achievements at GCSE than do White students, whereas Black students (a category that unfortunately combines the very different groups of Black Caribbean and Black African ancestry) performed significantly worse than White students at GCSE. In other words, with an important caveat about Black Africans, the overall rank ordering that we saw using the GHS in Table 3 persists even after controlling for social class. More specifically, Rothon showed that although social class does explain part of the Caribbean disadvantage in Britain, it actually masks some of the Indian success. In other words, the figures in Table 3 tend to underestimate the extent to which Indians outperform Whites because Indians tend to come from somewhat more disadvantaged backgrounds than the comparison group of British-born Whites.
In Canada, some research has also been able to examine the social background of second-generation students and see how far social background can explain ethnic differences. J. Hansen and Kucera (2004) used Statistics Canadas Survey of Labour and Income Dynamics (SLID) to examine the educational achievement of the second generation, controlling for parental education. They found that parental education does not explain second-generation success. Unfortunately, the small sample size means that they are unable to look at the detailed groups that we are using (although they did find a significant Irish advantage with respect to postsecondary and university levels). Boyd (2002) used the SLID to examine the educational achievements of the second generation. She distinguished the visible minorities (grouping together Black, Chinese, and Indian groups in a single category) and found that after controlling for parental education, the second generation visible minorities had significantly higher rates of high school graduation and of number of years of completed education than did the population as a whole. However, as she emphasized, it is likely that there will be some considerable stratification within the visible minority group.
In the United States, small sample sizes also make it difficult to say a great deal about the educational achievements of specific groups after controlling for social background. Most studies have used the National Education Longitudinal Study (NELS), but this requires aggregation into a few very broad categories of ethnicity. For example, Kao, Tienda, and Schneider (1996), using the NELS of 1988, found that second-generation Asians had the highest GPA, 3.24, versus 2.96 for Whites, 2.74 for Hispanics, and 2.73 for Blacks (a group that would be predominantly composed of African Americans). After taking a variety of measures of social background into account, Kao and her colleagues found that the mean GPA of Hispanics was no longer significantly different from that of Whites, whereas that of Asians remained significantly higher and that of Blacks significantly lower (see also Lutz, 2007).
Given the difficulties that scholars have had in undertaking detailed analysis using individual-level data of second-generation educational achievements controlling for social background, our own strategy in this article is to use aggregate level data for each of our ethnic groups in each country. As we noted, it is not possible to measure the social background of the second generation directly in the three large-scale data sets that we have been using. However, if it is assumed that the social profile of older members of the first generation who arrived 20 or more years ago gives some indication of the social origins of the second generation, it is possible to make some estimates of the extent to which social background might explain the pattern of ethnic inequalities in education. That is to say, we used our large-scale data sources to obtain measures of the social class and education of a notional parental generation. We could then plot the relationship between the characteristics of this parental generation and the education of the second generation to see how far the characteristics of the parental generation could explain either cross-national or intergroup differences in achievement.
The parental generation was defined as those who were born abroad, who migrated before 1975, and who were aged between 45 and 64 at the time of the survey. The selection of this age range and period of immigration was designed to capture those persons who were likely to be the parents of the second generation, as opposed to more recent immigrants. Of course, this was at best a rough approximation, but we checked the results against data on the actual parents in the previously cited data sources (Youth Cohort Study, GHS, SLID, and NELS).9 The major drawbacks with this method are that the different groups have rather different rates of intermarriage with the White population (which may also vary by educational level) and have different rates of return migration. However, they did provide us with a reasonable idea of the scale and nature of generational change.
We should also mention that there is a potential advantage too from our use of aggregate level data because this will cover contextual and individual-level effects of social background. The usual methods for controlling for social background consider only the characteristics of the individual respondents family (usually father and/or mother) because these are the major influences on a childs educational attainment. However, there is a large literature suggesting that, particularly among ethnic minorities, the social composition of the wider community may also be relevant to educational achievement. Notable contributions here are those of Wilson (1987) and Borjas (1995). Our method in effect took account of both family and community background (albeit at a rather high level of aggregation).
We next present the log odds ratios, comparing each parental group to the majority population, analogous to the log odds ratios shown in Table 6 for the second generation. We focused on parental education rather than on parental social class (although the results for social class were very similar). Parental education and social class were correlated with each other (although not perhaps so strongly among the migrant generation as among the native born) and both had strong relationships with childrens educational attainment. However, parental education is perhaps more relevant to our concerns with the selectivity of immigrants to the different countries, whereas parental social class will, to a large extent, reflect the treatment of the minorities after arrival in the Western labor markets.10
We used the relative measure of the log odds ratio rather than a conventional measure, such as the percentage completing tertiary education, partly for symmetry with our treatment of the second generation. However, it has the further advantage that it does not assume strict comparability of the classification system across countries. Furthermore, it can be argued that sociologically, it is not the absolute level of education that matters but whether one has more or less than ones competitors in the education and labor markets. To have highly educated parents in the United States may not be such an advantage as it might be in Britain, where fewer of ones competitors come from educated backgrounds.
Table 7a: Parental generation attainment of tertiary education, relative to the majority population: men aged 45-64
Log odds ratios
Table 7b: Parental generation attainment of tertiary education, relative to the majority population: women aged 45-64
Log odds ratios
Note: Parental generation is defined as those aged 4564 who arrived in the country of destination before 1985. Log odds ratios are derived from logistic regressions controlling for age and, where appropriate, year. Reference age is 45 and reference year is 2000.
Turning to Table 7, the first point to note is that these coefficients tend to be rather lower than the corresponding ones for the second generation. In other words, relative to the majority populations of the same age, the parental generation was less advantaged educationally than we found the second generation to be. This further implies that relative to their respective majority populations, these minority groups have made substantial intergenerational progress. Again, this has been noted by other researchers (Boyd, 2002; Heath & Yu, 2005; Kao & Thompson, 2003).
It is also noteworthy that in Table 7, the coefficients for the mothers tend to be rather lower than for the fathers in general, with the important exceptions of Caribbeans and Filipinos. This contrasts with Table 6, where we found that there was rather little difference overall between men and women in the second generation in their educational achievements (relative to their respective majority populations). The clear implication is that women ethnic minorities have made greater intergenerational progress than men. In effect, what we found is that in the migrant parental generation, there was considerable gender inequality, whereas this gender inequality has been largely removed in the second generation.
These two patterns of intergenerational progress and reduced gender inequality appear to be common ones to all three countries. We should, however, note some cross-national differences. In particular, the log odds ratios in Britain for the parental generation seem to be less favorable than those in the United States or Canada. This contrast with Canada is not surprising, but the fact that the American parental generation tends to be more advantaged educationally than the British is somewhat more puzzling because the American system of immigration rules is generally assumed to be less selective. However, as we noted at the beginning of this article, in the early postwar period (when many of the parental generation would have been migrating to Britain), there was free access to Britain for members of the British Commonwealth. Britain subsequently and progressively tightened up on its entry rules, but these increasing restrictions will have had more effect on later migrants.
Whatever the detailed explanations for them, the pattern of national differences in the selectivity of the first generation suggests that our earlier findings about the second generation in particular, the rather lower educational success of the children of immigrants in Britainmight be explained by the degree of selectivity in the first generation. It begins to look as though differential selectivity of the migrants, combined with the standard processes of social reproduction, might explain the cross-national differences in second-generation success.
To explore this, we plotted the log odds ratios for the first- and second-generation groups in the three countries against each other. Figure 1 shows the pattern for men, and Figure 2 shows the pattern for women. (For Britain, we also included estimates for the Chinese based on data from the Labour Force Survey [LFS]).
Figure 1: Parental Generation and Second-Generation Attainment of Tertiary Education, Relative to the Majority Population: Men
Figure 2: Parental Generation and Second-Generation Attainment of Tertiary Education, Relative to the Majority Population: Women
Note: Second-generation log odds ratios are derived from logistic regressions controlling for age and, where appropriate, year. Reference age is 25, and reference year is 2000. Parental generation is defined as those aged 4564 who arrived in the country of destination before 1985. Log odds ratios are derived from logistic regressions controlling for age and, where appropriate, year. Reference age is 45 and reference year is 2000.
Figures 1 and 2 show a very striking pattern. First, as we can see, there is a strong positive relationship between the log odds ratios for the first generation and those for the second generation. In other words, the educational level of the parental generation (relative to their respective majority populations) predicts that of the second generation rather well. The groups that tended to be better educated in the migrant generation than the majority population, such as Indians and Chinese, also tend to be relatively successful in the second generation, whereas groups such as the Caribbeans or Mexicans, which tended to be less qualified in the migrant generation, are also less successful than the other groups in the second generation. This is, of course, what we would have expected from the general theory of social reproduction.
Second, we can see that the line of best fit passes well above the origin (the 0,0 point on the diagram), cutting the vertical axis at the point +0.5 in the case of men and even higher for women. This represents the intergenerational progress that the minorities have made relative to their majority populations. (If the minorities had remained at the same distance relative to their majority populations, the line would have passed through the origin.) What this means is that the second-generation groups are in general achieving more highly than would have been expected from the normal processes of social reproduction that obtain in their majority populations.
Third, it is interesting to see which groups and which countries tend to fall above or below the line of best fit. If a particular group or country constitutes a more favorable context for educational progress, we would expect them to be above the line of best fit, and conversely if they constitute a less favorable context.
Considering the different minorities, we see that all three groups of Chinese lie above the line of best fit, indicating that in all three destination countries, they have outperformed what might have been expected (given their parental background and the general process of second-generation advance). None of the other groups showed such a clear-cut pattern of intergenerational progress, although Caribbeans, Africans, and Filipinos all tended to follow below the line of best fit. But just because Mexicans tended to fall at the bottom left of our figure should not lead us to suppose that they make less progress than would have been expected given their parental background. It is position relative to the line of best fit that tells us whether a group is making more or less progress than other groups, and Mexican men and women were both very close to the line.
What about country differences? Our previous analysis of the second generation suggested that minorities in Britain were less successful educationally than those in Canada and the United States. However, we now have to modify this conclusion. In the case of men in Britain, only 1 out of the 6 minorities fell below the line, whereas in the case of women, 4 out of 5 did. Overall, this strongly implies that Britain is little different from the other two countries in the opportunities that it gives for educational progress. This is precisely what might be expected if the differences are to be explained by differential selectivity in migration to the destination country rather than by educational contexts within the destination. Although it would be unwise to draw any strong conclusions from this cross-national comparison, it does not appear that there are any clear-cut differences between countries. These findings need to be replicated by systematic comparisons with data on actual parents education and social class. But in the meantime, we have to reject the hypothesis that the different institutional arrangements, such as affirmative action or multicultural policies, make a substantial difference.
Furthermore, it is worth noting that, as regards the individual minorities, our results are in line with the limited amount of individual-level analyses that have been reported by other authors. As noted earlier, both Rothons (2007) and Connollys (2006) work suggest that, controlling for social background, Caribbeans in Britain perform less well than the Chinese or Indian second generation, and this is exactly what our analysis also shows. Similarly, Boyds (2002) and Lessard-Phillipss (2007) research on Canada shows that, controlling for individual-level social background, visible minorities in general outperform native White Canadians, and this too is exactly what our analysis shows. Similarly, in the United States, a number of studies have confirmed that Indians and Chinese considerably outperform White Americans even after controlling for social background, again exactly in line with our results.
We have, then, four major findings: (1) There is a strong association between the educational level of the parental generation and that of the second generation. (2) There has been substantial intergenerational progress (measured relative to the majority population in the country of destination), especially among women, with most groups performing as well as or better than members of the majority population of the same age and similar parental backgrounds. (3) The Chinese are notable for their outperformance of what might have been expected, among both men and women. Indians also tend to make greater educational progress, but Caribbeans, Africans, and Filipinos somewhat less progress. (4) The poorer performance of the second generation in Britain as compared with the other countries can most likely be explained by the lower selectivity of the first generation in Britain rather than by institutional features of the British context.
We must remember the various methodological caveats involved in our investigation: Ethnicity has been measured in slightly different ways in the different countries; the first generation identified here is only a rough proxy for the parental generation and may be a poorer proxy for any groups that have intermarried with the majority population or that have high rates of return migration; and our measure of parental background needs to be supplemented by other measures such as social class. We also need to see if our conclusions are confirmed by examination of additional ethnic groups and by checks against other data. Nevertheless, our results are consistent with those of other scholars working on the individual countries and provide a coherent and plausible story.
First, the strong association between parental and second-generation education is not at all surprising given the well-known sociological findings about social reproduction. It would have been more surprising if we had failed to find a positive association between the two sets of log odds ratios. However, our findings do run counter to some predictions of downward mobility. It is true that in all three countries, some minorities make less progress than others, but in the great majority of cases, the log odds are more favorable in the second generation than in the first, parental generation. The one minority for whom this does not appear to apply is the men of Black African origins, but we need to be particularly cautious here because of the large standard errors associated with these estimates. To be sure, there will be considerable stratification within each minority, and it is certainly possible that some members experience downward assimilation. However, we should be careful not to attribute this to the minority as a whole.
Furthermore, even the notion of straight line assimilation is perhaps inaccurate. If there were such assimilation, we would expect to see our line of best fit passing through the origin, but it passes well above the origin. Why might this be the case?
Positive selection of migrants for unmeasured attributes such as drive and ambition is the natural answer. Migrants are a self-selected group who will surely be unusual in their drive and high aspirations. For some, a better future for their children will have been one of the primary reasons for migration. For others, migration will not have led to the hoped-for benefits for themselves (perhaps because of discrimination in the destination labor market or because of lack of the relevant human capital) and so they may, as a result of these blocked mobility aspirations, transfer their aspirations to their children.
There is, however, one important caveat that must be mentioned about this picture of intergenerational progress. We have necessarily had to compare the educational attainments of the parental generation with those of the majority population in the country of destination. But if we were to compare them with those of the populations in the countries of origin, we might get a less optimistic picture of intergenerational progress. As Feliciano (2005) showed for migrants to the United States, migrants tend to be better educated than nonmigrants in the country of origin. For example, in our data, Mexican migrants in the first generation have much lower educational levels than do native-born White Americans, but Feliciano showed that they are nonetheless slightly better educated than nonmigrant Mexicans. The implication of this is that if we measured parental education relative to that in the country of origin, we would find that the second generation had made less intergenerational progress in terms of their relative ranking.
Finally, what of the role of cross-national differences in institutional structures and in the contexts of reception? Given that we have rather weak evidence at present for any clear cross-national differences after controlling for the profile of the parental generation, any remarks on institutions need to be very tentative.
It should also be noted that the three countries examined here exhibit far less variation in the broad outlines of their educational systems than many European countries. All three of our countries have comprehensive systems of secondary education and have moved to mass, but highly stratified, systems of tertiary education. This contrasts with many European countries, where classic versions of sponsored systems of education remain to this day, with students tracked at an early age into quite distinct school types with important consequences for educational careers.
A typical finding in the sociology of education is that the later the key decision points (i.e., the more a system approximates to a contest system), the lower the social class inequalities (Breen & Jonsson, 2004), and a number of writers have suggested that this might apply to ethnic inequalities too. In Britain, Canada, and the United States, however, there are few such decision points early in the school career, although the presence of the public examination of the GCSE at age 16 does mean that there is an important decision point in Britain that comes 2 years earlier than in Canada and the United States. This would have accounted nicely for the pattern we noted in Table 6, in which the British second generation appeared to be less successful than in the other two countries. But because this pattern is not nearly so evident now that we have controlled for parental background, we are inclined to discount this suggestion (at least until better evidence in its favor is forthcoming). From these data, then, we cannot support the argument that institutional arrangements make a difference. That will have to wait until we include a wider range of countries with greater institutional variation.
1. Analysis of this type has the potential to be complicated by the presence in the data sets of recent younger first-generation arrivals. This problem is dealt with here by restricting the first generation to those between 45 and 64 years of age who had arrived in the destination country at least 20 years earlier.
2. New France was officially ceded to Britain in 1763.
3. Québec has had its own department of immigration since 1968 and, through various agreements with the federal government, was actively involved in the selection process until 1991, when the province was granted responsibility for the selection, reception, and integration of immigrants (Young, 2004).
4. In Quebec, high school occurs after completion of Grade 11 (compared with Grade 12 in other provinces), but entry to university requires a further 2 or 3 years of enrollment in a preuniversity program in Cégep (Collège denseignement professionnel et général). Cégep is also where most postsecondary professional programs are offered (CMEC, 2006).
5. In Scotland, there is a different system of secondary school examinations, and university degrees take 4 years rather than the 3 years typical of degrees in England and Wales.
6. Mandatory school age, or compulsory schooling, varies across Canada but is generally between ages 5 and 7, and 16 and 18. The period of compulsory education is generally from the ages of 6 to 16. In the United States, education is compulsory, beginning at age 7 in 21 states, at age 6 in 18 states, at age 5 in nine states, and at age 8 in two states. Because of the different state laws governing the structure and organization of education programs, the length of compulsory school attendance also varies: 9 years in 16 states; 10 years in 19 states; 11 years in seven states; 12 years in six states; and 13 years in two states (World Education Services, Canada; http://www.wes.org/ca).
7. Some students of this age complete vocational courses leading to the award of nationally assessed vocational qualificationsfor example, General National Vocational Qualifications (in England, Wales, and Northern Ireland) and Scottish Vocational Qualifications.
8. It should be noted that this is not always the case. In addition, some of the explanation for the apparently lower social class origins of some migrant groups largely reflects the discrimination that they experienced on arrival.
9. In the case of Britain, we can also use the Longitudinal Survey, which links results from the 1991 and 2001 censuses, to check the results. Using this source, we can link the educational attainments of the second generation in 2001 with the characteristics of their parents in 1991, providing they were living with their parents at the earlier date.
10. However, we should note that some members of the first generation will have acquired further education after arrival in the country of destination. We are unable to distinguish in our data sets when the education was acquired.
Alba, R. D. (1985). The twilight of ethnicity among Americans of European ancestry: The case of Italians. Ethnic and Racial Studies, 8, 134158.
Alba, R. D. (1990). Ethnic identity: The transformation of White America. New Haven, CT: Yale University Press.
Angelini, P. U., & Broderick, M. (1997). Race and ethnicity: The obvious diversity. In P. U. Angelini (Ed.), Our society: Human diversity in Canada (pp. 105129). Toronto, ON: Nelson.
Aydemir, A., & Sweetman, A. (2006). First and second generation immigrant educational attainment and labor market outcomes a comparison of the United States and Canada (IZA Discussion Paper No. 2298). Retrieved June 15, 2007, from http://www.iza.org/en/webcontent/publications/papers/viewAbstract?dp%5Fid=2298
Bean, F., & Stevens, G. (2003). Newcomers and the dynamics of diversity. New York: Russell Sage Foundation.
Bhachu, P. (1985). Twice migrants: East African Sikh settlers in Britain. London: Tavistock.
Boliver, V. (2006). Social inequalities of access to higher status universities in the UK: The role of university admissions decisions (Department of Sociology Working Paper No. 2006-05). Retrieved June 15, 2007, from http://www.sociology.ox.ac.uk/research/workingpapers/2006-07.pdf
Bonikowska, A. (2007). Explaining the education gap between children of immigrants and the native born: Allocation of human capital investments in immigrant families (Working Paper, University of British Columbia). Retrieved June 15, 2007, from http://web.uvic.ca/econ/bon.pdf
Borjas, G. (1995). Ethnicity, neighborhoods and human capital externalities. American Economic Review, 85, 365390.
Boyd, M. (2002). Educational attainments of immigrant offspring: Success or segmented assimilation? International Migration Review, 36, 10371060.
Boyd, M., & Grieco, E. M. (1998). Triumphant transitions: Socioeconomic achievements of the second generation in Canada. International Migration Review, 32, 853876.
Breen, R., & Jonsson, J. (2005). Inequality of opportunity in comparative perspective: Recent research on educational attainment and social mobility. Annual Review of Sociology, 31, 223244.
Castles, S., & Kosack, G. (1973). Immigrant workers and class structures in Western Europe. Oxford, England: Oxford University Press.
Castles, S., & Miller, M. J. (1998). The age of migration: International population movements in the modern world. London: Macmillan.
Cheng, Y. (1994). Education and class: Chinese in Britain and the U.S. Aldershot, England: Avebury.
Cheung, S. Y., & Heath, A. (2007). Nice work if you can get it: Ethnic penalties in Great Britain. In A. Heath and S. Y. Cheung (Eds.), Unequal chances: Ethnic minorities in Western labour markets (Proceedings of the British Academy 137, pp. 507550). Oxford, England: Oxford University Press.
Chiswick, B. (1980). The earnings of White and coloured male immigrants in Britain. Economica, 47, 8187.
Connolly, P. (2006). The effects of social class and ethnicity on gender differences in GCSE attainment: A secondary analysis of the Youth Cohort Study of England and Wales 19972001. British Educational Research Journal, 32, 321.
Council Ministers of Education Canada. (2006). General Overview of Education in Canada. Retrieved July 23, 2007, from http://www.cmec.ca/
Daley, P. (1996). Black-African: students who stayed. In C. Peach (Ed.), Ethnicity in the 1991 Census, Volume 2: The ethnic minority populations of Great Britain (pp. 6694). London: HMSO.
Daniel, W. (1968). Racial discrimination in England. London: Penguin.
Davies, S., & Guppy, N. (1998). Race and Canadian education. In V. Satzewich (Ed.), Racism and social inequality in Canada: Concepts, controversies and strategies of resistance (pp. 131156). Toronto, ON: Thompson Educational.
Edmonston, B., &Passell, J. (1994). Immigration and ethnicity: The integration of Americas newest arrivals. Washington, DC: Urban Institute Press.
Farley, R., & Alba, R. (2002). The new second generation in the United States. International Migration Review, 36, 669701.
Feliciano, C. (2005). Educational selectivity in U.S. immigration: How do immigrants compare to those left behind? Demography, 42, 131152.
Fitzduff, M. (1996). Beyond violence: Conflict resolution processes in Northern Ireland. Tokyo: United Nations University.
Floud, J. E., Halsey, A. H., & Martin, F. M. (1956). Social class and educational opportunity. London: Heinemann.
Guppy, N., & Davies, S. (1998). Education in Canada: Recent trends and future challenges. Ottawa, ON: Statistics Canada and the Minister of Industry.
Halsey, A., Heath, A., & Ridge, J. (1980). Origins and destinations: Family, class and education in modern Britain. Oxford, England: Clarendon Press.
Hansen, R. (2000). Citizenship and immigration in post-war Britain: The institutional origins of a multicultural nation. Oxford, England: Oxford University Press.
Hansen, J., & Kucera, M. (2004). The educational attainment of second generation immigrants in Canada: Evidence from SLID. Unpublished manuscript, Concordia University.
Hearn, J. C. (1991). Academic and non-academic influences on the college destinations of 1980 high school graduates. Sociology of Education, 64, 158171.
Heath, A. (2000). The political arithmetic tradition in the sociology of education. Oxford Review of Education, 26, 313331.
Heath, A. F., & Brinbaum, Y. (2007). Explaining ethnic inequalities in educational attainment. Ethnicities, 7, 291305.
Heath, A., & Ridge, J. (1983). Social mobility of ethnic minorities. Journal of Biosocial Science, Suppl. 8, 169184.
Heath, A., & Yu, S. (2005). Explaining ethnic minority disadvantage. In A. F. J. Heath, J. Ermisch, & D. Gallie (Eds.), Understanding social change (pp. 187224). Oxford, England: Oxford University Press for the British Academy.
Herberg, E. N. (1989). Ethnic groups in Canada: Adaptations and transitions. Scarborough, ON: Nelson Canada.
Isajiw, W. W. (1999). Understanding diversity: Ethnicity and race in the Canadian context. Toronto, ON: Thompson Educational.
Jonsson, J., & Mills, C. (1993a). Social class and educational attainment in historical perspective: A Swedish-English comparison. Part I. British Journal of Sociology, 44, 213247.
Jonsson, J., & Mills, C. (1993b). Social class and educational attainment in historical perspective: A Swedish-English comparison. Part II. British Journal of Sociology, 44, 213247.
Kalbach, M. A., & Kalbach, W. E. (1999). Becoming Canada: Problems of an emerging identity. Canadian Ethnic Studies, 31, 116.
Kao, G., & Thompson, J. S. (2003). Racial and ethnic stratification in educational achievement and attainment. Annual Review of Sociology, 29, 417442.
Kao, G., Tienda, M., & Schneider, B. (1996). Racial and ethnic variation in educational achievement. Research in Sociology of Education and Socialization, 11, 263297.
Karen, D. (2002). Changes in access to higher education in the United States: 19801992. Sociology of Education, 75, 191210.
Kasinitz, P. (1992). Caribbean New York: Black immigrants and the politics of race. Ithaca, NY: Cornell University Press.
Kelley, N., & Trebilcock, M. (1998). The making of the mosaic: A history of Canadian immigration policy. Toronto, ON: University of Toronto Press.
Kurian, G. (1991). South Asians in Canada. International Migration, 29, 421433.
Lessard-Phillips, L. (2007). Ethnic educational inequalities amongst the second generation in four receiving societies. Unpublished DPhil thesis, University of Oxford, Oxford, England.
Li, P. S. (2003). Destination Canada: Immigration debates and issues. Oxford, England: Oxford University Press.
Lutz, A. (2007). Barriers to high school completion among immigrant and later-generation Latinos in the United States: A look at language, ethnicity and family structure and socioeconomic status. Ethnicities, 7, 306322.
McNabb, R., & Psacharopoulos, G. (1981). Racial earnings differentials in the U.K. Oxford Economic Papers, 33, 413425.
Model, S., & Fisher, G. (2007). The new second generation at the turn of the new century: Europeans and non-Europeans in the U.S. labour market. In A. Heath & S. Y. Cheung (Eds.), Unequal chances: Ethnic minorities in Western labour markets (Proceedings of the British Academy, pp. 591638). Oxford, England: Oxford University Press.
Reitz, J. (1988). The institutional structure of immigration as a determinant of inter-racial competition: A comparison of Britain and Canada. International Migration Review, 22, 117146.
Richmond, A. H. (1990). Race relations and immigration: a comparative perspective. International Journal of Comparative Sociology, 31, 156176.
Rothon, C. (2007). Can achievement differentials be explained by social class alone? An examination of minority ethnic educational performance in England and Wales at the end of compulsory schooling. Ethnicities, 7, 306322.
Rumbaut, R., & Portes, A. (2001). Introduction: Coming of age in immigrant America. In R. Rumbaut & A. Portes (Eds.), Ethnicities: Children of immigrants in America (pp. 119). Berkeley: University of California Press.
Shavit, Y., & Blossfeld, H.-P. (1991). Persistent inequality: Changing educational attainment in thirteen countries. Boulder, CO: Westview Press.
Stewart, M. (1983). Racial discrimination and occupational attainment in Britain. Economic Journal, 93, 521541.
Turner, R. (1960). Sponsored and contest mobility and the school system. American Sociological Review, 25, 855867.
U.S. Department of Education. (2007). Education in the U.S. Retrieved July 23, 2007, from U.S. Network for Education Information Web site: http://www.ed.gov/about/offices/list/ous/international/usnei/us/edlite-index.html
Wilson, W. J. (1987). The truly disadvantaged. Chicago: University of Chicago Press.
Young, M. 2004. Immigration: the Canada-Quebec accord (Parliamentary Research Branch Background Paper BP-252E). Retrieved July 15, 2007, from http://www.parl.gc.ca/information/library/PRBpubs/bp252-e.pdf
DATA AND METHODOLOGY
For Britain, the General Household Survey (GHS) is the main source, although results have been checked with the Labour Force Surveys. To obtain a sufficiently large sample, the GHS data for the period 19912000 are pooled. The GHS is a continuous household survey representative of private households in the population of Great Britain, excluding Northern Ireland.
For Canada, the 2001 Census Public Use Microdata Files (PUMFs) were employed. These contain data based on a sample that represents approximately 2.7% of the population enumerated in the census.
For the United States, the Current Population Survey was used. This monthly survey of the population uses a sample of households designed to represent the civilian noninstitutional population of the United States. In this article, a combined file of the surveys taken from March 1994 to March 2003 was used. The Current Population Survey is widely recognized as a source for data on a wide variety of demographic, social, and economic characteristics of the population.
CONSTRUCTING THE ETHNIC CLASSIFICATIONS
In Britain, the GHSs contain a question on self-reported ethnic identity and further questions on the respondents and the respondents parents countries of birth. We defined the first generation as those born outside the United Kingdom, the second generation as those with one or both parents born outside the United Kingdom, and the third (or higher) generations as those with both parents born in the United Kingdom.
The standard British ethnic identity question used in the GHS at the time distinguished White, Black African, Black Caribbean, Chinese, Indian, Pakistani, and Bangladeshi ethnicities (strictly speaking, cultural backgrounds). The reference category of third-generation White British consists of third- or higher generation individuals who gave a White ethnic identity. The Black African second generation was defined as those with one or both parents born in the East African Commonwealth or rest of Africa Commonwealth and who gave a Black African identity. The Black Caribbean second generation was defined as those with one or both parents born in the Caribbean Commonwealth and who gave a Black Caribbean identity. The Chinese second generation was defined as those with one or both parents born in the Far East Commonwealth or in the rest of Asia & Oceania and who gave a Chinese identity. The Indian second generation was defined as those with one or both parents born in India or in the East African Commonwealth and who gave an Indian identity. (Many Indian families were twice migrants who had migrated first to East Africa.) The Pakistani second generation was combined with the Bangladeshi and defined as those with one or both parents born in India, Pakistan, or Bangladesh and who gave a Pakistani or Bangladeshi identity. (Note that some respondents may have been born in what is now Pakistan or Bangladesh but before the partition of India.) The first generation was defined similarly but on the basis of own country of birth rather than parents countries of birth. The Irish second generation were those coded as White, with one or both parents born in the Republic of Ireland.
In the United States, we used variables on race, nativity, mothers birthplace, and fathers birthplace to construct our measure of ethnic group. The majority group was defined as White respondents (code 100 of the race variable) who are native-born with native-born parents (code 1 of the nativity variable). In essence, this is a category composed of third- or higher generation Whites, but note that it will include Hispanics, who are counted as White in the race variable. This is therefore closely comparable with the category of White British used in the British census and, as in the British case, will include longstanding White ethnic minorities. As Alba (1985, 1990) has argued, there has been substantial blurring of the boundaries between these White groups of European ancestry.
The specific ethnic categories were constructed from the measures of race and mothers and fathers birthplace. We included respondents either of whose parents came from the specified origin countries. In the case of Indian ethnicity, we included people classified as Asian or Pacific Islander, or Asian only (codes 650 or 651), if one parent was born in India. In the case of Chinese ethnicity, we included people classified as Asian or Pacific Islander, or Asian only, if one parent was born in Hong Kong or China. In the case of Filipino ancestry, we include people classified as Asian or Pacific Islander, or Asian only, if one parent was born in the Philippines. In the case of Irish ethnicity, we included people classified as White if one parent was born in Ireland. In the case of Mexican ethnicity, we included people classified as White if one parent was born in Mexico. In the case of Black African ethnicity, we included people coded as Black/Negro on the race variable (code 200) if one parent was born in the West Indies (not otherwise specified), Belize, Jamaica, Barbados, Bahamas, Grenada, Guyana, and Trinidad and Tobago, but we excluded Cuba, Dominica, and Haiti. In the case of African ethnicity, we included those classified as Black/Negro if one parent was born in Africa (not otherwise specified), Ethiopia, Kenya, Nigeria, Ghana, or South Africa. In this way, we attempted to construct categories that broadly corresponded to the ones that have typically sent migrants to Britain and Canadathat is, English-speaking former British colonies.
In the case of Canada, we constructed our ethnic classification on the basis of variables measuring ancestry (self-reported), visible minority status, and nativity. As with the other countries, we defined the first generation as those born in Canada, the second generation as those with one or both parents born outside Canada, and the third generation as those with both parents born in Canada. We included in the majority group respondents who were native born with both parents native born and who were classified as not a visible minority. In effect, this gave us a category of third-generation (or higher) White respondents and is thus broadly equivalent to the definition of the majority group in Britain or the United States. Note that this will include many White groups of European ancestry, such as Italians, Poles, Germans, and Ukrainians, as well as the standard charter categories of British or French descent.
For our specific minority groups, we defined the Black African group as those coded as Black on the visible minority indicator and who gave single African ethnic origins. The Black Caribbean group consists of those coded as Black on the visible minority indicator with Jamaican or Caribbean single-ethnic origins. The Chinese were those coded as Chinese on the visible minority indicator and who gave a single-ethnic origin of Chinese. The Indians are those coded as South Asian on the visible minority indicator and who gave single East Indian ethnic origins. The Pakistanis are those coded as South Asian on the visible minority indicator and who reported that they belonged to a Muslim religious group. The Filipinos are those who were coded as other visible minority on the visible minority indicator and who gave single Filipino ethnic origins. The Irish were those coded not a visible minority on the visible minority indicator and who gave single Irish ethnic origins.
CONSTRUCTING THE EDUCATIONAL CLASSIFICATION
A number of potential problems need to be highlighted in the context of cross-national educational research. At the most basic level, it is extremely difficult to construct a comparable measure of highest educational qualification from diverse data sets because the national systems of education differ from each other in important ways, as do the detailed categories available in our data sets.
Our measure was constructed as shown in Table A1.
Table A1: Categorization of Educational Qualifications for Britain, Canada, and the United States