Social inequalities experienced by children of immigrants across multiple domains of life: a case study of the Windrush in England and Wales

It is well known that children of immigrants experience inequality. Less is known about how inequalities compare across multiple life domains and multiple generations. We conduct a case study of England and Wales, focussing on children of Caribbean immigrants (the ‘Windrush generation’). We use large-scale census data to compare inequalities across five domains of life—education, employment, occupation, housing, and health—separately for women and men across three distinct generations: the one-point-five generation, second-generation, and two-point-five generation. The children of the Windrush generation experience social inequality in all life domains, relative to comparable groups of the White British population, although there is considerable variation according to sex and generation. Men of all generations are uniformly disadvantaged; children of the Windrush are more disadvantaged if they belong to the two-point-five generation. Inequality is pervasive, persistent, and strongly indicative of segmented adaptation.

& Nee, 2005). We address this by conducting a quantitative case study of England and Wales. We focus upon children of Caribbean immigrants-more specifically children of the 'Windrush generation' .
The term 'Windrush' is synonymous with the history of international migration to England and Wales. The 'Empire Windrush' was a vessel that arrived in England in 1948, carrying many of the first Caribbean immigrants who settled in England after World War II as full British citizens. In the years since, the term 'Windrush generation' has come to refer to the more than 300,000 men, women and children from the Caribbean who arrived in Britain between 1948 and 1971. Many were educated, employed and taxed in Britain and having established multi-generational families there, considered themselves full citizens of the 'Mother country' (Hewitt, 2020). More recently, the term has become synonymous with a public scandal in 2017, after it emerged that hundreds of British citizens-overwhelmingly the children of the Windrush generation had been wrongly detained, deported and denied legal rights through the hostile environment policy (Hewitt, 2020;McKee, 2018;Wardle & Obermuller, 2019). This scandal not only highlighted the mistreatment of the children of the Windrush by the government, but the barriers that their parents faced over years of immobilization, separation and exclusion (Wardle & Obermuller, 2019).
Despite the historical importance of their parents and their current public and political relevance (Gentleman, 2019), we know little about the social inequalities that the children of the Windrush have experienced, especially based upon large-scale quantitative data representative of the entire resident population. Previous studies have focused on a wider group, people with Black Caribbean ethnicity, finding that they experience social inequality in different domains of life, for example in their education (Chapman & Bhopal, 2019;Strand, 2012) and employment (Li & Heath, 2020). However, the experiences of this broader ethnic group-which includes Black Caribbeans who were born in, or migrated to, Britain after 1971 [the so-called "Air Jamaica generation" who did not arrive as citizens and were subject to immigration control (Jolly, 2019)]-is unlikely to be entirely representative of the unique lived experiences of the children of the Windrush.
By evaluating social progress among the children of the Windrush, it is possible to gain a deeper understanding of intergenerational inequalities and the challenges that they face. Here, we carry out a quantitative analysis of social inequalities across five different domains of life: education, employment, occupation, housing, and health. Our aim is to provide an holistic overview of the social inequality experienced by the children of the Windrush. We use secondary census data and contribute a comparison within and across these domains of life, investigating whether the children of the Windrush experience social inequality compared with the White British cohort. We distinguish between-and compare-children born in the Caribbean who arrived with adult members of the Windrush generation (G1.5), children born in Britain to two members of the Windrush generation (G2), and children born in Britain to one Windrush parent and a White British parent (G2.5). By conducting a multi-generational, multi-outcome analysis, we offer one of the most detailed overviews of social inequality among the children of immigrants in Britain, for the Windrush generation or any other group. Importantly, in disaggregating the analysis by generation and sex, our approach also represents a comprehensive test of adaptation theories.

The Windrush and socio-political context
The dialogue and decisions of government play a crucial role in shaping the institutions and lives of those who encounter them (Rhodes, 2016). In this section, we outline the historical and political policy context within which the Windrush arrived and settledor in the case of their children, grew up-because it represents an important source of the inequality that they might face.
Following the end of World War II, the British Government encouraged immigration from its colonies through the British Nationality Act of 1948. This act enshrined the rights of British Commonwealth subjects and accompanying children to move to Britain as full citizens of the country (Hewitt, 2020). This act also provided citizenship by descent to children born in Britain. Yet, rather than an act of benevolence, this was an act borne out of necessity-a result of severe labour shortages. The British government never considered that the Windrush generation would stay for so long (Hewitt, 2020). Indeed, even the arrival of the Empire Windrush from the (then) West Indies was controversial, with the Prime Minister trying to prevent its arrival altogether (Hewitt, 2020). From the beginning, the mere presence of immigrants and their families from the West Indies in Britain was contested and socio-politicized and they were 'othered' within a system in which stereotyping, discrimination and marginalisation were commonplace (Reddie, 2020).
By the late 1950s-the end of the post-war economic boom-saturated job markets and larger-than-expected waves of immigration led to growing social unrest and the consolidation of anti-immigrant sentiment in Britain (Hewitt, 2020;Pager & Shepherd, 2008). In response, the 1962 Commonwealth Immigrants Act sought to limit arrivals from the Commonwealth to those with government-issued vouchers (Hewitt, 2020). The 1968 Commonwealth Immigrants Act further amended this act, restricting entry only to (white) individuals with a parent or grandparent born in Britain. The Immigration Act of 1971 went on to end Commonwealth migration altogether, albeit granting 'right of abode' (i.e. unconditional right to remain in Britain) for Windrush already resident in the country. This was followed by the British Nationality Act of 1981 and provision of citizenship by 'virtue of the law'-rather than the provision of identity documents. This would have serious ramifications for the children of the Windrush after the rollout of 2012 Hostile Environment Policy. The government sought to cut off undocumented migrants from access to public services and make working or renting impossible for those without adequate paperwork (Griffiths & Yeo, 2021). For Hewitt (2020), the ongoing Windrush scandal merely reflects a continuation of concerted efforts to renege on prior commitments to the Windrush generation.
On to specific domains-the children of the Windrush largely experienced their education during the integrationist and assimilist approach of the government in the 1960s and 70s, which sought to minimise and conditionalize cultural diversity (Race, 2005). Official agenda framed underachievement as the problem of 'black children' and 'ethnic failure' . This became the basis for intervention, leading to alienation and exploitation in the classroom and wider society (Race, 2005).
Regarding the labour market, the children of the Windrush started to enter the labour market at a time when reforms started to deregulate it in order to foster a more flexible market (Robinson, 1997). In 1989, legislation was passed for the first time that prohibited ethnic discrimination in the labour market (Robinson, 1997). Despite this, continued inequality in the labour market has led to various policies aimed directly at closing the gap between ethnic minorities and the White British. Phung (2011) cites the 1998 New Deal, designed to increase employment in socially disadvantaged groups. The policy, he believes, achieved some success, but overlooked ethnic minorities-including Black Caribbeans-facing multiple disadvantages (Phung, 2011). Clark and Shankley (2020) highlight the lack of success of the 2003 Ethnic Minority and Employment Task Force (EMETF) to close ethnic minority-White British gaps. They also highlight the 1999 National Minimum Wage and how employer non-compliance is an issue for ethnic minorities concentrated in informal economic activities (Clark & Shankley, 2020)-a consequence of deregulation.
With respect to housing Lukes et al. (2019) highlight government's 1974 Right to Buy policy-which permitted tenants of social housing-in which Black Caribbeans are overrepresented-to buy the house they lived in. In the absence of new social housing, this led to a widespread shortage within Britain; tenants who could not afford to buy were driven to the precarity of the private market. Such were the scale of the housing challenges facing ethnic minorities in the 1970s and 1980s-particularly Black Caribbeans who were concentrated in the social renting sector that specific policies and ethnic minority-led housing associations were set up to increase their access to better quality housing and more desirable neighbourhoods (Shankley & Finney, 2020). Lukes et al. (2019) also argue that the government's promotion of gentrification in low-income neighbourhoods has only sought to displace ethnic minorities elsewhere.
Regarding health, general perception about Commonwealth families from the 1950s to 1970s was that they would fit into the British system without requiring bespoke support. Health authorities had little contact or consultation with ethnic minorities about their specific health needs (Bahl, 1993). Following limited health campaigns targeting specific subgroups in the 1980s, ethnic minority health was officially incorporated into the review system by the Department of Health (Bahl, 1993). Even so, significant barriers persist in access to healthcare for ethnic minorities, driven by confusing information on health care systems and settings, a lack of affordable and reliable transport to reach health care setting where ethnic minorities live (another possible consequence of gentrification) and cultural insensitivity among (certain) health care employees (Jayaweera, 2021).

Theories predicting inequalities among immigrants and their descendants
The life course of immigrants after arrival is typically theorized as a process of adaptation, which may or may not occur to varying degrees (Zhou & Gonzales, 2019). Although the concepts of adaptation, assimilation and integration are often used synonymously, there are notable differences (which themselves are the subject of debate) (Alba & Nee, 1997;Brubaker, 2001). Adaptation is typically conceptualised as a process of convergence leading toward a lack of inequality or disparity between foreign-born and native-born groups at the population-level (Portes et al., 2009;Zhou & Gonzales, 2019). Integration has been defined as the degree to which immigrants have the knowledge and capacity to build a successful and fulfilling life (Harder et al., 2018), while (classical) assimilation has been defined as the decline and the disappearance of ethnic distinctions and the cultural or social differences that relate to this distinction (Alba & Nee, 1997). Inequalities in socioeconomic outcomes, in comparison to the majority population, are therefore indicative of the extent to which immigrants are able to adapt (Alba & Nee, 2005;Massey, 1981;Waters & Jiménez, 2005), while also forming part of the evidencebase about integration and assimilation. Literature on immigrant outcomes highlights the inter-relationship between conditions on arrival-i.e. how, when, why and where immigrant groups arrive and settle-and the factors that influence their behaviour after arrival (Massey, 1981;Portes & Zhou, 1993;Waters & Jiménez, 2005;Zhou, 1997). Importantly, the sending and receiving context work to shape immigrants' assumptions and expectations, both initially and after arrival, for example with respect to job security, housing, and the health and wellbeing of family members (Alba & Nee, 2005;Negy et al., 2009). This is particularly relevant for the Windrush who arrived with expectations of equivalent rights and opportunities granted to full British citizens (Hewitt, 2020). For many, these expectations did not materialise, and when coupled with numerous other social and structural barriers-such as racism and discrimination-their experiences are highly likely to have hindered their adaptation and become a source of adversity.

Inequalities experienced by the children of immigrants
To better understand the persistence of inequalities over time, theories of adaptation have not only focussed on immigrants, but also their children. Indeed, some researchers suggest the only way to understand the legacy of immigration is to examine the lives of immigrant's descendants (Crul & Vermeulen, 2003;Hirschman, 2006;Portes & Rumbaut, 2001;Thomson & Crul, 2007). Many researchers expect that children of immigrants will be less likely to experience inequalities-and more likely to adapt toward the destination average-than their parents do because of greater exposure to the destination (including people and institutions) (Alba & Nee, 2005).
A fundamental question is therefore whether the descendants of immigrants are experiencing disadvantage, as compared with descendants of the majority population. Where adaptation was once considered inevitable for subsequent generations (i.e., 'straightline' adaptation) (Alba & Nee, 2005;Zhou & Gonzales, 2019), contemporary theories of adaptation distinguish between the mechanisms of adaptation for the children of immigrants and the mechanisms of adaptation for their parents (for early discussion see Gordon, 1964) and the contrasting and often conflicting social demands that play out against a backdrop of hostility and discrimination towards descendants with a migration background (Drouhot & Nee, 2019;Portes & Zhou, 1993).
It has become increasingly common to challenge the notion that the descendants of immigrants are less likely to experience inequalities than their parents. Portes and Zhou (1993)-among others-argue that a significant number of children of immigrants will face risks of persistent inequality, downward social mobility and negative outcomes across multiple life domains, a process often referred to as 'segmented adaptation' (or 'segmented assimilation' , see: Zhou & Gonzales, 2019). Segmented adaptation also problematises 'classical' theories of adaptation (such as straight-line adaptation) by emphasizing heterogeneity. Some children of immigrants will be better able to integrate than others (Portes & Zhou, 1993). However, many groups may be less willing or able to adapt, leading to pervasive poverty and inequality (Drouhot & Nee, 2019). This vulnerability varies according to family background and structural factors such as neighbourhood characteristics, and the presence of opportunities for social mobility (Crul & Vermeulen, 2003;Heath & Ridge, 1983;Platt, 2005). Yet researchers have also shown that individual factors-language, aspirations, and experiences of prejudice-play an equal part in determining social inequalities (Heath et al., 2008). Among descendants, adaptation is determined by circumstances, conflict, and negotiation (Drouhot & Nee, 2019;Parker & Song, 2007).
Empirical research from the US is equally ambiguous regarding whether or not adaptation is inevitable for the children of immigrants. There is considerable evidence of heterogeneity, such that certain groups (e.g., Asians) are experiencing an advantage over US-born non-Hispanic Whites, while others (e.g., Blacks and Hispanics) suffer persistent social disadvantage (Drouhot & Nee, 2019;Portes & Zhou, 1993;Tran et al., 2019;Zhou & Gonzales, 2019). Similarly, there is considerable variation in Western Europe. Children with parents from less-developed non-European origins (except for some East and South Asian groups) are disadvantaged with respect to educational, occupational attainment and access to labour markets relative to majority populations and the children with parents of European origin (Heath et al., 2008). Such limited upward mobility has often been attributed to social inequality, rather than a systematic ethnic penalty. For example, Drouhot and Nee (2019, p. 183) argue that "overall trends in the second generation's labour market outcomes are social reproduction in existing structures of inequality". Indeed, in Britain, Li and Heath (2020) found patterns of social reproduction in occupational attainment between generations to be comparable between immigrants and White British population. However, despite convergence in many of the markers of adaptation for the second generation, an ethnic penalty for entry into the labour market has been found (Bertrand & Mullainathan, 2004;Heath & Di Stasio, 2019;Zschirnt & Ruedin, 2016). Irrespective of ongoing debates about the determinants of social inequality, there is now a substantial body of evidence that the children of immigrants in high income countries experience disadvantage across domains of life (Alba & Holdaway, 2013;Heath & Di Stasio, 2019;Tran et al., 2019).
In light of socio-political context of the children of the Windrush and the previous theoretical and empirical literature outlined above, we undertake an investigation of the social inequalities experienced by the children of the Windrush. We present four research questions to achieve our aim: 1. In which domains of life-if any-do the children of the Windrush generation experience inequality relative to the White British population? 2. Do patterns of inequality vary between women and men? 3. How does inequality compare across generations-is there evidence of straight-line or segmented adaptation? 4. Do inequalities persist when we compare with members of the White British population who have similar social characteristics?

Materials and methods
We use a representative individual-level 5% sample of the England and Wales population from the most recent Census in 2011. We use the regional safeguarded microdata file downloaded from the UK Data Service (Office for National Statistics, 2019), a national data service that provides free access to a range of social and economic data. This data contains all of the relevant information required to define the children of the Windrush and the intended outcomes. The 2011 Census is not only the latest census available, and the only data source that is large enough and comprehensive enough to carry out this research, but it also directly precedes the onset of the Windrush scandal that began in 2012. As such, our analysis contributes a timely assessment of inequalities experienced by children of the Windrush, prior to any additional impact of the scandal. Table 1 presents the populations of interest and the criteria used to define them. As we select the G1.5 based on those arriving 1945-71, which results in an age profile of 40 to 65 in 2011, we condition all groups on birth cohort . Aside from the methodological advantage of standardising age profiles across groups (age is also adjusted in our models), this restriction has an additional benefit. In the absence of direct information on parental characteristics it helps ensure that we only include G2 and G2.5 whose parents arrived during the Windrush era and that few (if any) grandchildren of the Windrush (G3) are included in our sample. Our year of arrival conditions are 1945 to 1971, not 1948 to 1971, due to how arrival year is banded in the data.

Target population
In accordance with prior work (Alba & Nee, 2005;Bélanger & Gilbert, 2006), we differentiate by generation to reflect varying degrees of exposure to destination. The foreign-born children of immigrants (G1.5) spend at least some of their childhood in both the origin and destination countries. As a group, they will receive some early life exposure (most notably to educational institutions) in both contexts, but unlike the second generation, they will also have some experience of migration. Relative to G1.5 and G2, G2.5 have the highest level of exposure to the destination by virtue of having both a foreign-born and a native-born parent. This increasing exposure gradient across these parallel generations may be crucial in the establishment of status, awareness of rights, and access to opportunities for children of Windrush in England and Wales and in the ability of these different generations of Windrush to adapt, or not, to From a starting sample of 2.8 million people, we remove those outside of our target population (i.e. immigrants and their children from countries other than Caribbean ones, members of the Windrush generation who arrived as adults (18+), those arriving from Caribbean countries after 1971, the children of Caribbean immigrants born in England and Wales after 1971, all of those outside of the target age range 40-64 in 2011, and individuals who were economically inactive due to early retirement or were still in (mature) education). This left an eligible population of 685,977, all of whom were retained for final regression analyses; n for the populations are in Table 1.

Outcomes
Inequality is generally conceptualised as the state of not being equal, particularly in respect of status, rights, and opportunities (UN, 2015). Here, we focus on socioeconomic inequality across five domains: education, employment, occupation, housing, and health. Conceptualising inequality in this way captures the dynamic and multi-faceted nature of the concept (Shaw et al., 2007) allowing us to make rounded conclusions about the inequalities experienced by the children of the Windrush in England and Wales. Our indicators capture inequalities in outcomes reflective of the interrelations between the different domains and the inherent power structures and social processes that maintain inequality. Each domain includes a positive and a negative life outcome to avoid limiting the analyses to a particular part of the sample's distribution and to avoid conceptualising inequality in a single direction. Table 2 provides both the domains and outcomes.

Statistical methods
We fit logistic regression models to estimate the odds of each life outcome for our populations of interest compared to the White British born in England and Wales. Models are specified as follows: Our regression models allow us to compare and contrast different subpopulations within the same life domains, estimate baseline age-standardised inequalities and further estimate age and SES-standardised inequalities relative to the White British reference, differentiated by sex and generation. Table 3 shows the distribution of our study population according to the core demographic and geographic variables available in our data. The largest Windrush generation is the G2, which is nearly three times larger than the G1.5 and the G2.5, but (expectedly) much smaller than the White British group. While the White British are fairly evenly distributed across the age groups, the G1.5 have higher shares in their fifties, while the G2 and G2.5 have higher shares in their forties. G2 and G2.5 Windrush are highly concentrated in London and the Midlands, while the G2.5 are more evenly distributed across England and Wales, albeit with higher shares in London and the Midlands than the White British. For civil status, all Windrush groups have higher shares of single and lower shares of individuals in registered unions compared to the White British.

Results
Turning to the socioeconomic variables in Table 4, Windrush men have higher shares in lower education levels and lower shares in higher education levels than White British men; we see the opposite for women (the G2.5 aside). For occupation, smaller shares of Windrush men occupy managerial occupations compared with White British men while larger shares occupy routine or semi-routine occupations, are long-term unemployed (especially the G2 and G2.5) or have never worked (notably the G2.5). Windrush men also have higher shares of economic inactivity.
Conversely, both G1.5 and G2 Windrush women have smaller shares in routine occupations, larger shares in managerial jobs and similar levels of economic activity compared with White British. All Windrush men and women have lower shares that own their own homes and higher shares in deprived housing. For health, Windrush men and women (aside from the G2) have higher shares reporting a limiting longterm illness than White British men and women. For self-reported health, however, Windrush women (generally) have higher shares with bad to very bad health and lower shares with good to very good health; the distribution for men is mixed. With the possible exception of the housing domain, these descriptives suggest stark differences between female and male children of the Windrush and between the G1.5 and G2 compared to the G2.5. Figures 1 and 2 offer a summary of the odds ratios for each combination of generation, sex, and domain of life. Figure 1 displays the negative outcomes; Fig. 2 displays the positive outcomes. Full regression tables relating to each specific outcome can be found in Additional file 1: Tables S1-S20. We interpret the regression results domain-by-domain with respect to our four research questions.

Education
In this domain, we observe inequalities for women and men, in particular among the G2.5. In Fig. 1, G1.5 and G2 women are advantaged in the baseline model (Model 1); they have lower odds of having attained no academic or professional qualifications. They retain this advantage in the adjusted model (Model 2), when standardising to compare them with the White British women with similar characteristics. These two groups also have persistently higher odds with respect to the positive educational outcome in Fig. 2 of having attained a degree. By contrast, G2.5 women have a baseline disadvantage relative to White British women in both the positive and negative education outcomes, which is absent from Model 2. The results for men are less encouraging. All three generations have lower odds of having attained a degree, both at baseline and in the adjusted model. The situation is better with respect to no qualifications, where they are on par with White British men in the baseline models and then advantaged in the adjusted models.

Employment
For employment, we again find more equal outcomes for women and less equal outcomes for men. In both the baseline and adjusted models in Fig. 1, all male generations have persistently higher odds of being long-term unemployed and persistently lower odds of actively employed. In the baseline models, G2.5 men experience the largest relative inequality, but they are also the generation for whom the ratio changes the most in the adjusted models. The size of these inequalities for men are worth noting. For every male generation, the baseline odds of being long-term unemployed are around 2 times higher than for White British men and this odds ratio remains 1.5 or above in Fig. 1 Summary panel of logistic regression models for negative outcomes in the five life domains for the children of the Windrush relative to White British population resident in 2011. Notes Baseline models adjust age and subpopulation, final models adjust region of residence, civil status, level of highest qualification, NS-SEC, housing tenure, and self-reported general health, minus the control relating to the specific domain i.e., we do not adjust for the highest level of qualification if our outcome is having obtained a degree level+). Source: Authors' calculations based upon 5% microdata sample of resident population of England and Wales the adjusted models. Similarly, the baseline odds ratio for being actively employed is below 0.5 for all generations (Fig. 2), and this remains below 0.7 in the adjusted models. Women are less likely to suffer employment inequalities. Compared with White British women, G1.5 and G2 women have similar odd ratios of being long-term unemployed ( Fig. 1) or actively employed (Fig. 2) in the baseline models, and exhibit evidence of an advantage in the adjusted models. Echoing the results for education, G2.5 women fare worse than the G1.5 and G2, indicative of a relative disadvantage in the baseline models, but not in the adjusted models.

Occupation
In common with the results for education and employment, we observe more equal occupational outcomes for women. G1.5 and G2 women have lower odds of working in a routine occupation in the baseline model (Fig. 1). This remains the case for the G2 in the adjusted model, but not the G1.5, whose conditional risk of working in a routine occupation does not differ from White British women. The same is true of G2.5 women in the baseline and adjusted models. Similarly, all female generations exhibit similar odds of working in a managerial or professional occupation as White British women in the two models (Fig. 2). For men, the G1.5 and G2 have persistently higher odds of working in routine occupations in the baseline model and even the adjusted models. The G2.5 differ in that they are most likely to work in a routine occupation in the baseline models, Notes Baseline models adjust age and subpopulation, final models adjust region of residence, civil status, level of highest qualification, NS-SEC, housing tenure, and self-reported general health, minus the control relating to the specific domain i.e., we do not adjust for the highest level of qualification if our outcome is having obtained a degree level+). Source: authors' calculations based upon 5% microdata sample of resident population of England and Wales but then least likely to experience inequality in the adjusted models (as compared to the other two generations). This picture is complicated further by the results for managerial or professional occupations in Fig. 2, where all three male generations exhibit inequality compared with White British men, but there is no evidence of a generational gradient (in either model). Despite this, it is clear that occupational inequalities exist and that they are gendered.

Housing
Housing is the domain in which we find the greatest similarity between Windrush women and men. In the baseline models, all three generations of both sexes have much higher odds of living in deprived housing (Fig. 1), combined with much lower odds of owning their own home (Fig. 2) compared with White British women and men. The odds ratios are also substantial (above 2 for deprived housing and below 0.5 for home ownership). For both outcomes, while relative levels of inequality smaller in the adjusted models, they still persist. The only exception is G1.5 women, a group that has neither higher nor lower odds of living in deprived housing in. We observe some evidence of a generational gradient with respect to home ownership for both sexes and housing deprivation among women; the odds of owning one's own home diminish across the generations (G1 to G2 to G2.5), such that the most disadvantaged generation are the G2.5.

Health
Health is the only domain in which women appear to fare worse than men, although this is more apparent for good general health (Fig. 2) than it is for limiting long-term illness (LLTI; Fig. 1). In the baseline models for LLTI, G2.5 women are more likely to report an LLTI than White British women, and the same is true for men. However, there is no difference in LLTI for the other generations of women or men, and no difference in the adjusted models for G2.5 women or men. Yet for general health, there is evidence of inequality for all generations of women, in both the baseline and adjusted models. They are less likely to report good or very good health as compared with White British women. This contrasts with the results for men, for whom the only evidence of a similar health disadvantage is for G2 and G2.5 in the baseline models. There is some evidence of a generational gradient for both of the health outcomes in the baseline models for both women and men, where the G2.5 are once again most likely to experience inequality.

Discussion
This study has investigated the social inequalities experienced by three generations of children of immigrants across multiple domains of life. We used census data to carry out a quantitative case study, focussing on children of the Windrush generation (Caribbean migrants who arrived between 1948 and 1971). Our aim was to contribute an holistic overview of the social inequality experienced by this group. In addition to a theoretical interest in intergenerational adaptation, and whether this was segmented or straightline, our study was motivated by the historic and ongoing discrimination experienced by the Windrush generation and their children (Gentleman, 2019). We address four research questions using analyses of a representative 5% sample of the 2011 Census for England and Wales. Our first question asks whether inequality exists, and if so, then in which domains of life. Although we did not find evidence of systematic inequality (i.e. for all sexes and all generations in all outcomes), we did document evidence of inequality in every domain for at least some specific groups (i.e., particular generations and/or sexes). Housing was one domain in which we documented uniform disadvantages for all groups; men and women of every generation were less likely to own their own homes and more likely to live in deprived housing than the White British population. We also found the largest relative inequalities in this domain.
Our findings can be positioned in the context of previous research on the children of immigrants in Britain. Regarding education and the labour market, children of-particularly Indian and Chinese-immigrants tend to be better educated than their parents and White British population. To that end, the poor educational outcomes of male children of the Windrush is interesting. Yet, all children of immigrants are less likely to be employed; some earn lower wages than the White British, notably Pakistanis, Bangladeshis and Black Caribbeans (Dustmann et al., 2011). Our finding of large, persistent housing inequality reflects wider ethnic disadvantages in housing (Shankley & Finney, 2020). Specifically, rising levels of housing precarity, higher levels of social renting in groups like other Black, Black Caribbean and Black African, variation in home ownership-with higher levels among Indians and Pakistanis and lower levels among Black African, Arab and the other Black group-and overcrowding among Bangladeshis, Pakistanis and Black Africans (Shankley & Finney, 2020). Our findings for health add to a small body of evidence of poor health among children of South Asians, Black Caribbeans and Black Africans-but not Chinese (Harding & Balarajan, 2000), alongside of excess adult mortality among the children of Black Caribbean, Black Other and Pakistanis and Bangladeshi immigrants (Wallace, 2016).
Our second research question focusses on the differences between women and men. Here, our findings are definitive. Male children of the Windrush are systematically disadvantaged in all life domains in the baseline models. The only exceptions (i.e., the only cases in which they did not experience baseline inequality relative to White-British men) were the G1.5 and G2 for both of the education outcomes and for self-reported health (5 models out of 30). Conversely, female children of the Windrush face much lower odds of inequality. In some domains, they even have better outcomes than White British women in the baseline models (e.g., lower odds of having no qualifications and working in a routine occupation, and higher odds of having a degree for the G1.5 and G2). More often, Windrush women have similar odds to White-British women in the baseline models but still have a disadvantage in some domains (e.g., housing and health). Health is the only domain in which women fare worse than men, and housing is the only domain where we find a broad consistency between sexes. For the remaining life domains-education, employment, and occupation-we find clear and consistent evidence that male children of the Windrush fare worse than females. In general, male-female differences persist in the adjusted models.
To interpret these gender differences, it is essential to consider the intersection between gender, ethnicity and class. These may combine to create distinctive, disparate outcomes among male and female children of the Windrush (Browne & Misra, 2003). With respect to education and the labour market-in which we observe lower relative inequality among female children of the Windrush-previous explanations of gendered outcomes among the children of immigrants focus on suggestions that women are less likely to be criminalised, have higher educational and occupational aspirations and are positively affected by gender-role socialisation (Feliciano & Rumbaut, 2005;Mickelson, 1989). In Britain, previous work has highlighted lower aspirations among Black Caribbean boys (Francis & Skelton, 2005). With respect to health-the domain in which we find lower relative inequality among male children of the Windrush-differences might be explained by increased prevalence of certain diseases among women. A recent scoping review found that Black Caribbeans-especially women-had twice the risk of developing type 2 diabetes compared to the White British (Bennett et al., 2015). Curtis and Lawson (2000) found that the reporting of psychosocial illness varied between Black Caribbean men and women, that they recognised different health risks and that gender differences in self-reported health were more pronounced in the ages we study, a result of traditional, fixed gender expectations. Gender aside, the better health of the G1.5 compared to the G2 and G2.5 may relate to some residual healthy migrant effect (i.e., its presence in the G1.5 and its absence in G2 and G2.5) (Wallace, 2016).
Our third question focuses upon generational gradients in inequality to test adaptation. Our intergenerational comparison provides evidence in support of segmented adaptation for the children of the Windrush. To elaborate, we show evidence that inequalities not only persist but also increase across generations. This is particularly notable in the baseline models, for example home ownership and both health outcomes for women and men. We also find several instances where there is no clear gradient (i.e., increasing relative levels of inequality from the G1.5 to the G2 to the G2.5), yet the level of inequality is worse among the G2.5 relative to the G1.5 (e.g., no qualifications for men and women and both employment outcomes). At the same time, the adjusted models do reduce differences in the relative inequality experienced over all generations.
Our findings therefore have theoretical implications, in particular with respect to theories of segmented adaptation. Our results not only demonstrate that adaptation can be segmented, and is far from guaranteed (e.g. in line with Zhou & Gonzales, 2019), but also that this segmentation is far from homogenous, varying according to gender, generation, and life domain, as well as the interconnectedness between different domains. Moreover, this heterogeneity is evident even though we analyse a specific case, thereby removing multiple sources of heterogeneity with respect to ethnicity and parental migration. We therefore recommend that theories of segmented adaptation acknowledge this potential heterogeneity, even when focussed on specific ethnic groups.
Our final question examines whether inequalities persist when we compare with members of the White British population who have similar social characteristics. The adjusted models make comparisons that control for region of residence, marital status, and variables representing the four other life domains. This comparison also provides some sense of the inter-dependencies between inequalities across the different domains. Our findings suggest a complex picture of inequality that is not easily summarised. Generally, we can say that inequality-where observed in the baseline models-is often reduced, but rarely fully explained, with the addition of the controls. Even compared to White British people with similar characteristics, the children of the Windrush generation often experience relative disadvantage. There are also instances in which the level of inequality is unchanged after the addition of controls (e.g., both employment outcomes among G1.5 men). In general, adjustment tends to have the greatest impact upon the G2.5.
For cases in which inequalities persist after standardising for variables that represent inequality in other domains of life, it may be that our choice of control does not adequately capture exactly what aspect of a given domain-or domains-account for the inequality that we have observed. Case in point, if our outcome is education, it may be that the choice of tenure-as opposed to, say, overcrowding-to represent the housing domain may have weaker explanatory power because tenure is not as relevant to educational success as, say, the physical and/or mental space that one has to study in. There are also other factors that generate inequality that simply cannot be captured in the census. Not least, the experience of personal and institutional discrimination and racism, the role of government policy in specific life domains, residential context (notably segregation), and the impact of characteristics of the parents of the children of the Windrushincluding any inequalities that they experienced and whether they can be transmitted across generations.
We might consider housing-given this is where we find large, persistent inequality relative to White British even after standardisation for disadvantage in the other domains. Recent research has linked the inability of ethnic minorities to find affordable, good quality housing to factors that we cannot capture with the census. These include discrimination in the labour market, low levels of personal wealth, the adverse impact of recent hostile immigration policies, and a social security system that disproportionately penalises ethnic minority groups (Rhogaly et al., 2021). Shankley and Finney (2020) discuss how Black Caribbeans are concentrated in the social renting sector as a consequence of their immigration history and socioeconomic profile at the time of arrival.
Taking the findings of all of the research questions together, we show that the children of the Windrush generation experience pervasive and persistent social inequality relative to similar groups of the White British population, albeit with considerable variation by domain, outcome, sex, and generation, as indicated by our analysis. We also provide strong evidence of segmented adaptation.
Nevertheless, we note several limitations. First, our analysis says little about the processes that generate inequality, including the mechanisms that determine adaptation, which is a dynamic process that unfolds over time. Second, by examining inequalities among the children of the Windrush in 2011, we only focus on those alive and resident in England and Wales in 2011. As such, we say nothing about inequalities among individuals who emigrated or died before 2011, groups that may well be selected upon our outcomes. Data are not yet available for the next census (which was in March 2021). Third, we use country of birth and ethnicity to define our generations. Ethnicity is a fluid concept that can change depending on self-identification and how the question and categories are defined (Burton et al., 2010). A further limitation concerns the generalisability of our findings. In our case study, we have found strong evidence in support of segmented adaptation across multiple outcomes. However, our evidence is limited to children of the Windrush generation. Although there may be comparable inequalities experienced by other groups-within and beyond Britainthis remains to be tested using comparable data and a similar research design. Finally, a consequence of our quantitative approach means that we say nothing of the deeply personal experience of inequality of the children of the Windrush in Great Britain, how they attempt to cope it with it and how this might vary from person-to-person.
Mindful of these limitations, we indicate directions for future research. It is recommended that future analysis look to complement our findings with some in-depth qualitative interviews. This would offer a deeper, richer understanding of the unique and lived experiences of the observed inequalities in this paper and how they are produced, reproduced and maintained over the life course and between generations. This would, in turn, offer greater potential to address these persistent inequalities via public policy intervention. Regarding further quantitative research, when the 2021 Census data are available, it would be insightful to see how inequalities have changed over the last decade for the children of the Windrush. We also recommend that one of the UK's longitudinal studies Is used to examine how the social inequalities that we have found developed over time, in particular within households, families and individuals. Our case study also focused exclusively on children of the Windrush generation. It would be interesting to examine whether similar social inequalities-and similar differences by sex and generation-exist for other groups, in particular for the children of immigrants who arrived prior to 1971 from other Commonwealth countries. Directly, extending our findings, future research could also look at why baseline inequalities are worse among men and why inequalities differ across generations.
Our results represent a renewed evaluation of intergenerational adaptation for the descendants of Caribbean immigrants in Britain and one of the most detailed multigenerational, multi-outcome studies of descendants of migrants anywhere. Our findings are alarming; we document inequality across most life outcomes alongside evidence of rising inequality across generations suggestive of segmented adaptation. In the aftermath of Britain's departure from the European Union (EU), the ongoing plight of the Windrush generation looms large as social and political commentators fear for the future status of both EU citizens resident in Britain and British citizens who have settled in EU member states (BBC, 2019;Hinsliff, 2020). Our results show that inequality among children of the Windrush generation is pervasive. Despite having settled in Britain as British citizens with all the associated rights and opportunities of being British, the Windrush generation and their children have faced marked disadvantage and inequality. Our findings can contribute to, and suggest a need for, policies to address inequality across multiple domains of life. The ongoing Windrush scandal and the persisting elements of the 2012 hostile environment policy may further exacerbate the ingrained disadvantage that we find here. In designing future social policies, it will be crucial to harness the potential for the children of the Windrush-and the grandchildren of the Windrush-so as to experience greater levels of social equality.