Assimilation Theories in the 21st Century: Appraising Accomplishments and Future Challenges

Over a quarter century has passed since theoretical debates surrounding competing assimilation models emerged and durably structured research on immigrants and their descendants in America and beyond (Portes and Zhou 1993; Alba and Nee 1997). Much work since then has contributed to these debates (e.g., Portes, Fernández-Kelly and Haller 2009; Alba, Kasinitz and Waters 2011; Haller, Portes and Lynch 2011a, 2011b). In this article, I offer a three-pronged reflection on the contemporary state of assimilation research. First, I aim to take stock of the relative merits of segmented and neoassimilation theories and their ability to explain major empirical trends in both the US and Europe. I argue that there is now an emerging empirical consensus about the second generation in the US and Western Europe primarily experiencing upward mobility and intergenerational progress in socioeconomic terms (Waters and Jiménez 2005; Kasinitz et al. 2008; Alba and Foner 2015; Waters and Pineau 2015; Drouhot and Nee 2019), which is broadly aligned with neoassimilation rather than stagnation or downward assimilation as envisioned by segmented assimilation perspectives.

Secondly, I identify six analytical challenges facing further theory building for assimilation scholarship, both substantive and methodological. First, I argue that it is important to clarify the role of race and racial difference in influencing assimilation among non-White immigrants and their children. Such a clarification entails a move away both from earlier, “cost of race” approaches derived from population studies in terms of life chances among African Americans as well as contemporary, critical race theoretical approaches focused on white supremacy. A second substantive challenge includes better specifying the importance of culture and cultural difference between immigrants and natives, ¹ particularly in the context of the growing politicization of Muslim populations and cultural diversity in previously homogeneous European nation-states. Third, we need to reconsider the theoretical relationship between social mobility and belonging. While such relationship has been largely assumed to be nonproblematic and straightforward in past assimilation theories, recent studies of the “integration paradox” (Schaeffer and Kas 2023) have shown that immigrants perceive more exclusion and discrimination as they experience upward mobility, not less. Fourth, we need to progressively switch our research agenda toward the study of the third generation, the grandchildren of immigrants—without which a holistic assessment of assimilation is impossible.

Methodological challenges include, first, the acknowledgment of immigrant selectivity in shaping assimilation trajectories. The accurate assessment of the social status of parents in their origin countries leads to a reevaluation of what may first look like intergenerational mobility (e.g., Ichou 2014; Feliciano and Lanuza 2017; Engzell and Ichou 2020; Feliciano 2020) and a previously noted “second-generation advantage” (Kasinitz et al. 2008). Another set of methodological challenges revolve around issues of sample attrition based on processes which are, themselves, endogenous to assimilation. Ethnic attrition in samples based on self-identification is an important methodological problem insofar as individuals undergoing certain assimilation-related processes, such as intermarriage in the parental generation and intergenerational upward mobility, may be less likely to identify with their ethnic origin groups (Alba and Islam 2009; Duncan and Trejo 2011; Emeka 2018). As a result, they may “exit” such samples entirely, and downwardly bias measurements of assimilation predicated on ethnic “stayers.” Conversely, survey research based on samples of individuals who are not institutionalized will, by design, miss those who are institutionalized—for example, the imprisoned. Indeed, despite the lack of official statistics on national origins groups in US prisons, extant longitudinal research suggests that significant proportion of US-born males among some origin groups are incarcerated in adulthood (Rumbaut et al. 2006; Morín 2009). Whether and how much the master trend of assimilation I discuss in the first part of the paper hinges upon the upward bias induced by the positive selectivity of those not imprisoned remains unclear.

I conclude by looking at the future. I reflect on methodological divisions and the risk of epistemic drifts into separate research communities. I compare and contrast existing data infrastructures in the United States and Western Europe and outline future data requirements to enable further theory building. I evaluate the potential and pitfalls of computational approaches and social research in the digital age to advance substantive knowledge on assimilation processes in the twenty-first century. Specifically, I argue that new sources of highly granular data featuring many dimensions of assimilation at once should help us move past “categorical intuitionism,” or the tendency to irreflexively approach assimilation and immigration-driven diversity through taken-for-granted categories such as race, ethnicity, and religion. Instead, data richness may enable “categorical agnosticism,” whereby researchers may focus on within-category heterogeneity and rely on data-driven approaches to identify the empirical appropriateness of certain categories rather than prenotions or theoretical obedience. In a new era of unprecedented data availability, I argue that future theory building about assimilation will increasingly rely on abductive, post hoc theorizing putting data and empirical discovery first rather than deductive theories traditionally proceeding from general propositions to hypothesis testing (Goldberg 2015; Salganik 2018).

The paper follows the broad structured implied thus far, with a brief history of assimilation theories, contemporary tensions between different approaches and what the last three decades of research suggest by way of theoretical adjudication. I then outline outstanding and newly emerging challenges, both from the American and Western European scholarship, before proceeding to reflect on the future of our field in terms of methodological divisions, data requirements, and the potential of digital social research. Given the American origins of contemporary assimilation theories, much of my assessment and reflections are primarily written from the point of view of American assimilation research. I nevertheless aim at productively bringing the rich new Western European empirical scholarship to bear on how we can and should think further about assimilation, now and in the future ² .

Assimilation in Europe and America: What Have We Learned Since the 1990s?

The Ebbs and Flows of Assimilation

Before reflecting upon the empirical state of things in contemporary migration societies and what it means with regard to existing theories, it is useful to locate the current moment within the larger intellectual history of assimilation—and the Ebbs and flows of its popularity as a tool for social science research. What do we refer to as “assimilation” today? A synthetic reading of existing definitions and various empirical literatures pertaining to various aspects of immigrant incorporation in multiple contemporary societies yields the following: assimilation designates a two-way process of mutual convergence between immigrant and nonimmigrant populations, whereby ethnic origins and other differences associated with it (linguistic, racial, religious, etc) progressively cease to matter in the lives of immigrants and their descendants (Alba and Nee 2003). Assimilation has a socioeconomic dimension, where it is signaled by increasing parity between immigrants and natives. It has a relational dimension, where it is signaled by blending between populations, in terms of social and marital ties. Finally, it has a cultural dimension, where it is signaled by decreasing social costs of immigrant-specific practices, identities and beliefs, and ultimately a progressive broadening of the social groups and cultures included in the imagined community of the nation state ³ (Drouhot and Nee 2019, 178–9; see also Alba and Nee 2003; Alba and Foner 2015; Schachter 2016; Jonsson, Kalter and van Tubergen 2018; Statham and Foner 2024).

How did we get there? We can distinguish three eras of assimilation theories, marked by an increasingly stronger theoretical precision and compatibility with empirical test (see Brubaker 2001; Alba and Nee 1997, 2003; Jung 2009; Stepick and Stepick 2010; Kivisto 2015; Zhou and Bankston III 2016 for further discussion of the history of assimilation theories). During the classical era of the Chicago school of sociology, scholars influenced by Park or working under his tutelage produced empirical, largely qualitative studies of European-origin groups in American cities (e.g., Thomas and Zaniecki 1920; Warner and Srole 1945). In our collective scholarly memory, these efforts are associated with the diffuse influence of Park's (1950) race relations cycle, which famously stipulated “assimilation” as the endpoint of intergroup relations in ethnically diverse societies (see also Park and Burgess 1969[1921]). Past research has often criticized Park's theory for being teleological in that it could not envisage anything but assimilation as its endpoint (Lyman 1968).

The second era of assimilation theories came with Gordon's Assimilation in America (1964), famously outlining assimilation as a multidimensional process unfolding in a distinct causal order. Compared to Park's race relations cycles, Gordon's approach to assimilation was more precise and elaborate, skillfully considering aspects of culture, identity, social relations, and politics in shaping the assimilation process. It also reflected the influence of prevailing theoretical perspective at the time, namely structural-functionalism associated with the work of Talcott Parsons. Gordon's account of assimilation met a scholarly and political context marked by the Civil Rights, and more broadly the “differentialist turn” (Brubaker 2001) of the 1960s. Such an epistemological and political turn emphasized minority rights and experiences, while turning the concept of assimilation in a lightning rod for criticisms regarding its normative undertones and failures to account for the social experiences of African Americans (Metzger 1971). While Gordon's account remained influential in broad theoretical discussions of immigration (Hirschman 1983), a concern for assimilation as a multidimensional social process was largely displaced in the 1970s and 1980s in favor of a narrower focus on ethnic identity (Glazer and Moynihan 1970[1963]; Yancey, Ericksen and Juliani 1976; Gans 1979; Waters 1990) leading to the “melting pot vs pluralism” debate (Lieberson and Waters 1988; Gans 1997). Meanwhile, in the wake of the Civil Rights movement, scholars became increasingly focused on comparing the socioeconomic destiny of European (White) immigrants with that of African-Americans to study the endurance of ethnic and racial inequality. Somewhat remarkably, Gordon's model itself paid little attention to the role of socioeconomic mobility in the assimilation process (Alba and Nee 1997, 835–6), which left empirical scholars interested in these issues without a guiding theoretical model. Indeed, pioneering quantitative studies of socioeconomic attainment among immigrant-origin populations (e.g., Lieberson 1980) did not engage with Gordon's framework.

The emergence of the “new” second generation—the children of immigrants who arrived after the 1965 Hart-Celler act—gave rise to a third moment in the history of assimilation theories, one of renewed interest and vitality after decades of relative darkness, having led Nathan Glazier (1993) to provocatively ask whether assimilation was, perhaps, simply “dead” (see also Alba and Nee 1997, 826–7). In their seminal article outlining “segmented assimilation,” Portes and Zhou (1993, 76) noted that the last study explicitly focused on the second generation was Child's (1943) study of the children of Italian immigrants, published some 50 years earlier. The “new” second generation, Portes and Zhou observed, was different from earlier, European migration waves in at least two major respects: first, it was highly heterogeneous in ethnic and racial composition due to national origins in Central and South America, East and Southeast Asia, as well as the Caribbean; second, it faced a changing economic landscape with a rising knowledge-intensive sector necessitating high human capital, and a shrinking pathway of upward mobility into the middle class from unskilled, industrial jobs as had been the case for previous generations of (European) immigrants (Portes and Zhou, ibid). Consequently, and in conjunction with a rich theoretical framework outlining the role of governmental reception, societal attitudes, and strength of coethnic communities together with human capital and family structures (Portes and Rumbaut 1990, 91), Portes and Zhou famously predicted that we would observe new, heterogeneous trajectories of second-generation adaptation: upward mobility into the white middle-class for some, “downward assimilation” into an urban “rainbow” underclass of impoverished minorities for others, or “selective acculturation” consisting in a mixture of upward mobility and ethnic retention (see also Gans 1992 for an analogous discussion of various “second-generation scenarios”). Importantly, Portes and Zhou used the concept of assimilation in a loose, intuitive way centered around similarity between immigrants and natives, and eschewed an analytical definition.

Just a few years later, Alba and Nee (1997) proposed a theoretical rescue of assimilation, which involved distinguishing between an empirical process worthy of social analysis from prescriptive aspects often associated with the term. “As a state-imposed normative program aimed at eradicating minority cultures,” they wrote, “assimilation has been justifiably repudiated. But as a social process that occurs spontaneously and often unintendedly in the course of interaction between majority and minority groups, assimilation remains a key concept for the study of intergroup relations.” (Alba and Nee 1997, 827). Thus, on a conceptual level, Alba and Nee's defense of the contemporary value of assimilation involved recognizing “intellectual sins” (ibid) of certain earlier works, such as the ethnocentric reification of White, middle-class culture as a reference point for immigrants to tend toward (and as found for instance in Gordon's account). Empirically, Alba and Nee (1997, 841–9) argued that the older European-origin populations, like their contemporary counterparts from Latin America, the Caribbean, and Asia, faced relatively adverse contexts of reception and economic conditions. In particular, and contra Portes and Zhou, they argued that many European-origin groups (such as the Italians and the Irish) experienced ethnoracial discrimination due to not always being seen as White, and that many of their children came of age during the Great Depression (ibid). Nevertheless, they experienced gradual acceptance and assimilation over generations. Based on a review of recent studies of socioeconomic and spatial attainment, Alba and Nee argued we were likely to witness another period of gradual assimilation in the twenty-first century, that rather than a strong bifurcation along ethnoracial lines (Alba and Nee 1997, 849–61). Later, in their book-length account, they provided a now widely used definition of assimilation—“the decline of an ethnic distinction and its corollary cultural and social difference” (Alba and Nee 2003, 11) in the lives of immigrants and their children—as well as further empirical evidence of second-generation attainment and inclusion, for instance in terms of language use, socioeconomic attainment, and intermarriage (Alba and Nee 2003, chapter 6).

The Second Generation in America and Western Europe: What Have we Learned Since the 1990s?

These two texts structured assimilation research for the following decades. For purposes of research and teaching about immigration in the contemporary period, it has become commonplace to oppose Portes and Zhou's segmented assimilation with Alba and Nee's new assimilation theory in an archetypical (and admittedly tired) way—the latter being sometimes lumped together with exponents of “straight-line assimilation” associated with Gordon (1964) and Warner and Srole (1945). Nevertheless, both approaches converge on several important points, such as the importance of the stocks of (human) capital immigrant families bring with them, the role of social capital within the ethnic community, and the importance of formal (state) institutions in shaping assimilation or the lack thereof, for instance in the case of undocumented immigration (see also Alba, Kasinitz and Waters 2011, 763–6). Despite remaining interrogations regarding their status as scientific theories from the standpoint of philosophy of science (Karimi and Wilkes 2023), the influence and merits of both sets of writings are palpable in their transatlantic reach. Indeed, they have also contributed to organize the enormous boom in European research on “integration” since the turn of the century, which has been close to assimilation in spirit but with a stronger focus on social networks and cultural difference, and that is discussed further below (Heath, Rothon and Kilpi 2008; Jonsson, Kalter and van Tubergen 2018; Drouhot and Nee 2019).

Such overlap notwithstanding, we can contrast segmented and new assimilation theories, and assess their relative merit, by zeroing in on the clearest area of tension between them—race, and its weight in shaping the social destiny of the new second generation. While new assimilation theory did not envisage race as a master factor for the fate of the second generation in the post-Civil Rights era (Alba and Nee 2003, 14–5), segmented assimilation provided a clearly more pessimistic outlook (Portes and Zhou 1993; Portes and Rumbaut 2001, 55–6) through its emphasis on “downward assimilation”—which undoubtedly proved to be its most controversial element over the years (Stepick and Stepick 2010). At its core, “downward assimilation” revolves around the effect of intense racialization and how much one's life chances, networks, identities, and (non)belonging may be shaped by preexisting racism in the destination society. There exists some ambiguity in the extent to which racialization operates across key writings of the segmented assimilation tradition, however. I distinguish between a “strong form” of downward assimilation, in which race alone operates as a master status shaping the outlook of the children of non-White immigrants regardless of their class background—as implicitly argued by Portes and Zhou (1993, 91–92) in the case of West Indians in Miami for instance—and a “moderate” form in which race operates in conjunction with lower-class status and low human capital. The latter relates to broader issues of race, urban segregation, and concentrated poverty (Wilson 1978; Massey and Denton 1993) and assumes that poor non-White immigrants come into contact with marginalized native Black populations in low-income urban areas (and hence adopt their “adversarial stance” to the mainstream, Portes and Zhou 1993, 81). This “moderate form” is also present in Portes and Zhou's (1993, 83–5) original account and features prominently in later writings of this research tradition (e.g., Portes and Rumbaut 2001, 56–60; Portes, Fernández-Kelly and Haller 2009; Haller, Portes and Lynch 2011a; Zhou and Bankston III 2016, 73–7, Zhou and Gonzales 2019). Western European nation-state contexts differ from the United States and each other in myriad ways, but the postwar migration waves originating in North Africa, Turkey and later Sub-Saharan Africa share characteristics making them theoretically liable to downward assimilation—namely visible ethnoracial differences, low stocks of human capital at the first generation, and urban concentration in low-income areas (Thompson and Crul 2007; Vermeulen 2010; Alba, Sloan and Sperling 2011; Alba and Foner 2015; Drouhot and Nee 2019).

So—to what extent has the new second generation in Europe and America experienced downward assimilation? Almost 20 years ago, Portes, Fernández-Kelly and Haller (2005, 1019) had noted that “[for] the segmented assimilation hypothesis to be disproved,” research would have to demonstrate that “downward assimilation does not exist or affects only an insignificant number of second-generation youths.” It is far beyond the scope of this article (or that of a dozen book-length manuscripts) to provide an exhaustive answer, and I refer interested readers to the multiple reviews summarizing the empirical state-of-the-art published over the years (Zhou 1997; Waters and Jiménez 2005; Heath, Rothon and Kilpi 2008; Alba, Sloan and Sperling 2011; Alba and Foner 2015; Waters and Pineau 2015; Drouhot and Nee 2019; Zhou and Gonzales 2019; Flippen and Farrell-Bryan 2021) as well as specific empirical tests (Waldinger and Feliciano 2004; Portes, Fernández-Kelly and Haller 2005; Alba, Kasinitz and Waters 2011; Haller, Portes and Lynch 2011a; Xie and Greenman 2011). It is also important to keep in mind that comparing different origin groups inherently hides intragroup heterogeneity and myriad nonethnic factors shaping assimilation trajectories at the individual (rather than the ecological and group) level.

Despite the enormous complexity inherent to such a broad question, the last 30 years of social science research on both sides of the Atlantic sketch a general answer regarding downward assimilation. Simply put, there exists altogether limited evidence of downward assimilation among children of immigrants in Western societies, and in the words of Alba, Kasinitz and Waters (2011), “the kids are (mostly) alright.” There exist vivid, ethnographic descriptions of impoverished immigrant-origin populations adopting adversarial stances toward the perceived mainstream and criminal behaviors both in the U.S. (e.g., Vigil 2002) and Western Europe (Sauvadet 2006; Bucerius 2014), which are suggestive of downward assimilation in its “moderate” form. While these works evoke stalled or even failed assimilation, they reflect exceptional situations at the extreme of the socioeconomic spectrum rather than central empirical tendencies at the population level. Altogether, the expectation that the bulk of the second generation would join the ranks of a “rainbow underclass” in major American and European cities has not materialized. ⁴

Instead, research at large suggests the master trend across immigrant groups and country contexts to undoubtedly be one of intergenerational progress and inclusion in terms of schooling and labor market attainment, network integration and cultural identities, and analogous in spirit to expectations derived from new assimilation theory (Drouhot and Nee 2019). Beyond a sole focus on the second generation, Alba (2020; see also Alba and Maggio 2022) has shown that racially mixed unions are progressing in younger cohorts in the United States, and that the top of the occupational structure is becoming increasingly diverse—all signals of dissolving boundaries preventing non-White, immigrant-origin minorities from accessing the U.S. mainstream. As envisioned by both segmented assimilation and new assimilation perspectives, intergenerational progress at the second generation remains moderated by the relatively humble social origins of many immigrant families, who start at the bottom end of the social ladder in the country of destination (Pichler 2011; Bucca and Drouhot 2024) and envisage migration as multigenerational, family projects (Foner and Dreby 2011). The emergence of immigrant (upper) middle classes of varied origins bear witness to such a broad dynamic of immigrant upward mobility (Crul, Keskiner and Lelie 2017; Drouhot 2024). Crucially, and contra the “strong form” of downward assimilation implied in Portes and Zhou (1993, 81), there exists to my knowledge virtually no study systematically demonstrating that the life chances and assimilation outcomes of non-White members of the second generation are structurally shaped by race or ethnic origins alone (Drouhot and Nee 2019, 181–2).

These broad results do not definitely settle the intellectual debate that has structured assimilation research since the 1990s. Rather, they suggest that Portes and other authors may have misconstrued the selectivity and social composition of migration flows in the postwar era. Additionally, they may have underestimated the ability of immigrant families to navigate ethnoracial barriers in Western liberal societies—that is, in destination contexts where such barriers subsist informally, but where formal ethnoracial equality under the law is the prevailing institutional regime. The literature of Black migration from the West Indies and Sub-Saharan Africa to the U.S. offers a critical example. In the first half of the twentieth century, Black immigrants from the Caribbean in the New York metropolitan area saw their occupational attainment quickly converge with those of native-born Blacks due to the intensity and pervasiveness of racism (Hamilton 2019). From the 1960s onward, however, Black immigrants have, like African Americans, reaped the benefits of Civil Rights legislation increasing the significance of social origins in shaping their mobility prospects (Wilson 1978; Hout 1984). Indeed, migration flows from Sub-Saharan Africa and the Caribbean feature strong educational selectivity, as many arrive in the US with a college education or pursue one at destination—for instance, 63% of Nigerian immigrants in the U.S. have at least a bachelor's degree while only 7% of the Nigerian population do (Hamilton 2020, 303). Given such socioeconomic profiles, Nigerian-origin populations in the US are well positioned to take advantage of the American knowledge economy, and unlikely candidates for downward assimilation. In fact, recent empirical evidence suggests second-generation Nigerian Americans’ educational attainment surpasses that of third-generation plus Whites and even second-generation Asians (Sakamoto et al. 2021). Relatedly, qualitative evidence suggests that educational attainment is crucial to the upbringing and cultural identities of many second-generation Nigerians, for whom “it is un-Nigerian not to go to college” (Imoagene 2017, chapter 3). Such a culture of educational attainment helps them achieve an ethnic distinction vis-à-vis African Americans, in a way reminiscent of earlier research on the management of Blackness among West Indian immigrant families and middle-class African Americans (Neckerman, Carter and Lee 1999; Waters 1999).

Beyond the absence of large-scale patterns of downward assimilation, various aspects of segmented assimilation theory have been the object of empirical tests over the years, often with unconvincing results. This includes the assumed importance of the type of acculturation for socioeconomic attainment—whether “selective” or “dissonant” (Waters et al. 2010) -, the assumed importance of the socioeconomic context within the neighboring community (Xie and Greenman 2011) and whether key elements of the “modes of incorporation” typology (e.g., governmental reception, strength of the coethnic community, Portes and Zhou 1993, 84) are actually predictive of distinct assimilation trajectories (Waldinger and Catron 2016). ⁵

Despite such limited empirical support, it bears emphasizing that past research confirms the validity of certain theoretical elements common to both segmented and new assimilation theories, such as stocks of human capital at the family level and governmental receptions. Additionally, a more primitive reading of segmented assimilation writings would emphasize the importance of ethnoracial heterogeneity in shaping assimilation trajectories—something that the literature to date, showcasing important bifurcation between ethnoracial groups, largely bears out. More generally, the segmented assimilation approach has had intellectual merit in offering a theoretical framework amenable to empirical tests, and thus cumulative social science. A reasonable criticism of new assimilation theory is that, despite correctly predicting a master trend of assimilation at the second generation, it has arguably been less theoretically precise in specifying a variety of empirical outcomes. As such, it might therefore have been less liable to falsification and negative empirical results (Karimi and Wilkes 2023). In Haller, Portes and Lynch's (2011a, 736) words, “When ‘mainstream’ can signify anything from the upper class to the minority poor, and assimilation may or may not happen across generations […]” the theory is at risk of being relegated “into a benign expectation that immigrants and their descendants will eventually join, in one way or another, a multifaceted American mainstream.” Inferring broad trends from triangulation of empirical evidence across multiple dimensions (as done in e.g., Alba and Nee 2003; Alba 2020) provides a useful narrative but does not amount to a strict empirical test of neoassimilation as a social scientific theory. In recent years, new assimilation theory has been amenable to falsification and conceptual refinements based on novel results, but has often required qualified interpretation to devise precise empirical tests regarding the influence of ethnic and immigrant origins on socioeconomic attainment, networks, acculturation, belonging, and their interrelation (Schachter 2016; Bucca and Drouhot 2024; Zhao and Drouhot 2024; see also Jiménez 2010).

Contemporary Analytical Challenges

The theoretical debates from the 1990s and 2000s are now aging, and in fact have somewhat stalled in the 2010s. ⁶ It is noteworthy that different theoretical positions have relied on different data sources for empirical support: for instance, Portes, Rumbaut, and other collaborators have mostly relied on the Children of Immigrants Longitudinal Survey (Portes and Rumbaut 2018) carried in San Diego and Miami while advocates of new assimilation have relied on the Immigrant Second Generation in Metropolitan New York (Mollenkopf, Kasinitz and Waters 2011) as well as public statistics sources derived from the U.S. Census. To a significant extent, different statistical instruments, each with their specific regional coverage, sampling strategy and dimensionality, may have shaped theoretical disagreements to date. Revitalizing assimilation theories thus surely involves gathering new empirical data. However, in the intervening period since the emergence of contemporary assimilation theories, new substantive and empirical challenges have surfaced and older ones have remained, all contributing to determine current and future data requirements. I now outline such challenges.

The Role of Race and Racial Inequality in Shaping Assimilation Remains Unclear

Segmented assimilation envisioned the possibility of downward assimilation through systematic racism leading non-White minorities into the “underclass,” while neoassimilation provided a more optimistic account of mobility for non-White minorities, reasoning that immigrant origins should not exert a strongly negative influence on life chances in the post-Civil Rights era. In that respect, both theories reflect the prevailing zeitgeist about race at the time of their inception. Alba and Nee (2003) have structurally conceived of assimilation through a “cost of ethnic origins” approach in terms of life chances influenced by the debates on the relative significance of race and class in shaping African American mobility before and after the Civil Rights (e.g., Wilson 1978; Hout 1984). Culturally, Portes and coauthors have approached the effect of race on assimilation prospects through the “culture of poverty” lens and surrounding public concerns for the urban underclass prevailing in the 1980s and 1990s (Jung 2009; Wacquant 2022). As noted by Jung (2009, 383), the assimilation and race scholarships in the United States largely overlapped until the advent of the racial formation perspective (Omi and Winant 1991[1986]), which shifted the race scholarship toward a more autonomous path—one largely repudiating the notions of ethnicity and assimilation (Omi and Winant 1991[1986], 14–24). Both strands of scholarship have since then drifted apart (Jung 2009), with important developments in the race literature (for instance, colorism and skin tone stratification, e.g., Monk 2021; Bucca 2024) having been seemingly ignored by migration scholars.

Such an intellectual separation leads to a larger analytical problem. Empirically, large-scale patterns of downward assimilation have not materialized, which indeed suggests that racial barriers for the non-White second generation may not be as strong as they have historically been for African Americans in the US. Yet, empirical studies suggest that minority origins remain highly salient in the social experiences and self-understandings of the non-White second generation in the U.S. (Waters 1999; Frank, Akresh and Lu 2010; Vasquez 2011; Vallejo 2012; Clerge 2014; Imoagene 2017) and Western Europe (Beaman 2015; Yanasmayan 2016; Simonsen 2018). How are we to reconcile large-scale trends evoking assimilation with other (often microlevel, qualitative) accounts suggesting persistent racialization? Large-scale studies of racialization (e.g., Telles and Ortiz 2008) showcase patterns of inequality persisting along ethnoracial lines long after migration (in that case older cohorts of Mexican Americans) but do not theorize the racialization process, while existing definitions (Omi and Winant 1991; Hochman 2019) tautologically designate racialization as a causal process and an outcome (Uyan 2021). Simply put, a middle-range theory of racialization specifying social mechanisms and lending itself to deductive hypothesis testing remains to be written.

In recent years, critical race perspectives arguing for the centrality of “structural racism” in the US and elsewhere have gained traction in some provinces of migration scholarship, and invariably sought to attack assimilation theories for their alleged lack of attention to race and racism, and their putative reflection of, and participation in, white supremacy (Romero 2008; Jung 2009; Bashi Treitler 2015; Saenz and Douglas 2015; see also Kasinitz and Waters 2024). However, the issue with critical race perspectives is that they are just that—critical. Their primordial ontological commitment to the analytical supremacy of race makes them unsuitable for an inductive approach attempting to do just the opposite, namely precisely appreciate how much race and racial categories shape assimilation. Relatedly, critical race scholarship typically shuns epistemological realism in favor of standpoint subjectivism (Bonilla-Silva 2019). As such, it is unlikely to be helpful for theory-building among empirically minded scholars. Conceptual developments around “racialized assimilation” suggesting that structural aspects of assimilation (such as time and generation since migration) go hand in hand with the increasing significance of race for certain empirical aspects such as self-identification, social relations, residential location, and mental health (e.g., Golash-Boza 2006; Lee and Kye 2016) appear promising (see also Beaman and Clergé in this issue). Nevertheless, the concept remains undertheorized, and empirical practices remain a priori organized along with racial lines within American social science. In particular, the recursiveness between the salience of racial categories within empirical data sources and researchers’ prenotion that race matters implies that documenting racial differences in social outcomes pertaining to assimilation hardly constitute evidence of the weight of race per se. Detangling race from other life chances-shaping processes with which race correlates thus remains an outstanding analytical and empirical challenge for future assimilation scholarship.

Cultural Difference, and Especially Religious Difference Matter for Assimilation But Contemporary Theoretical Models Do Not Take Culture as a Causal Variable

In a now classic essay, Zolberg and Woon (1999) famously asked if Islam was “like Spanish”—or in other words, if the politicization of cultural difference around religion and Muslim minorities in Western Europe was analogous to that of Spanish language use in the U.S. Over the years, migration scholars have expanded this categorical comparison to compare religion in Europe to race (Foner 2015) and legal status (Drouhot and Nee 2019). Regardless of the appropriate ground for comparison, there exists an emerging consensus on the salience of the Muslim–non Muslim boundary for the second generation in Western Europe (Alba 2005; Alba and Foner 2015; Drouhot and Nee 2019), and the long shadow cast by Muslim religious affiliation, identities and practices on dynamics of labor market discrimination (Adida, Laitin and Valfort 2016; Di Stasio et al. 2021), friendship segregation (Kretschmer and Leszczensky 2022), national identification and subjective experiences of belonging (Leszczensky, Maxwell and Bleich 2020; Drouhot 2024), and worth and status in the ethnocultural hierarchies of European nation-states (Bleich 2009). Together, these works and others (see Drouhot and Nee 2019, 187–9) have firmly established religious difference as an autonomous analytical angle on assimilation—that is, one operating through (at least) partly independent causal channels. ⁷

The social significance of religious difference for assimilation resides in a dual, endogenous dynamic of cultural distance and stigmatization. There is, first, the import (via migration), and later maintenance, of different religiosity levels between immigrant and non-Muslim native populations resulting in real and perceived cultural differences across populations on the basis of ethnic origin. Indeed, European nation-states contexts overwhelmingly tend to be characterized by secularism and attenuated forms of religiosity after a long period of modernization (Norris and Inglehart 2011). Meanwhile, much of the new second generation in Western Europe originate from Muslim majority countries in North and Sub-Saharan Africa and the Middle East where organized religion has played, and continues to play, a much stronger role in collective life. Empirical studies across national contexts show that the Muslim second generation inherits the religious affiliation of their parents at high rates (Fleischmann 2022) and exhibit high levels of religiosity (Drouhot 2021). Accordingly, they are likely to hold conservative moral attitudes—for example, toward gender equality and homosexuality (Röder and Spierings 2022; Karim 2024)—compared to non-Muslim, native populations. In turn, Muslim populations are stigmatized and discriminated against because of such high perceived religiosity (Adida, Laitin and Valfort 2016; Helbling and Traunmuller 2018) by natives, who also tend to overestimate the size of Muslim populations (Mulders and van Tubergen 2023). Relatedly, high religious transmission among Muslim immigrant families imply that natives probabilistically infer religious affiliations from ethnic backgrounds and engage in ethnic qua religious discrimination, regardless of actual religious affiliation or religiosity among those perceived to be Muslims (Di Stasio et al. 2021). Such perceived hostility from non-Muslims fuel reactive religiosity among Muslims, in a circular fashion evoking a negative equilibrium (Adida, Laitin and Valfort 2016; Drouhot 2021). These double dynamics of cultural transmission, stigmatization, and differentiation, in which real and perceived differences endogenously feed off one another, may be compounded by legal and socioeconomic inequalities between both populations (Wimmer and Soehl 2014; Drouhot 2021).

Both new assimilation and segmented assimilation theories have difficulties accounting for the case of Muslim populations in Western Europe because they primarily consider culture and cultural difference (e.g., language or ethnic identity in much past research) as outcomes rather than causes within assimilation. They implicitly assume that cultural similarity between immigrants and natives, and the recognition of the former by the latter as “one of us” in a broader imaginary community, are subordinated to dynamics of upward and downward mobility (Schachter 2016; Drouhot 2024). It is ironic that Gordon's classic model considered acculturation as the initial dimension of assimilation allowing others to take place, an emphasis that appears to have been largely forgotten in later work (unlike approaches predating Gordon's, see Gordon 1964, 62–7). Indeed, a wariness to take culture and cultural difference seriously became de rigueur in American sociology following debates on culture and poverty pertaining to African-Americans in the 1960s and the infamous Moynihan report, published a year after Gordon's account. In recent years however, multiple developments in cultural sociology have offered renewed purchase on the role of culture (e.g., beliefs, norms, values, narratives, repertoires) in shaping cognition and behavior (Vaisey 2009; Small, Harding and Lamont 2010; Patterson 2014; Valentino and Vaisey 2022). Migration scholars are now arguably better equipped than in the past to study how differences in culture across populations may matter for assimilation. Importantly, and with regard to major concerns among scholars who repudiated the study of culture in the past for fear of pathologizing or essentializing the culture of minority groups (Small, Harding and Lamont 2010), taking religion and other aspects of culture seriously in assimilation can be done without essentializing immigrant and native populations as irrevocably different from each other or ignoring their internal heterogeneity (Drouhot 2021).

The Integration Paradox Challenges Assumptions About How Social Mobility Relates to the Lived Experience of Belonging and Exclusion among Immigrants and Their Children

As noted by Schachter (2016), new assimilation theory, in the lineage of the Park and Burgess (1921) designates the endpoint of assimilation as a form of collective belonging, whereby “individuals on both sides of the boundary see themselves more and more as alike” (Alba and Nee 2003, 10). While neoassimilation identifies immigrant mobility of out ethnically segmented neighborhoods and labor markets as the “siren call to assimilation” (ibid, 67), it has to date remained unclear how measurable intergenerational progress on “hard” indicators such as socioeconomic attainment and intermarriage lead to belonging and the inclusion of immigrant-origin others as “one of us” in the larger cultural community (Schachter 2016; Drouhot 2024). Rather, there has been an implicit assumption that immigrant socioeconomic attainment and mobility would necessarily lead to a subjectively felt acceptance in the imaginary community of belonging (Gans 2007; Schachter 2016). ⁸

A rising new literature in Western Europe knows as the “integration paradox” has firmly challenged this assumption and forces us to rethink the supposedly seamless link between immigrant socioeconomic achievement and belonging within contemporary migration societies (see Schaeffer and Kas 2023 for a meta-analysis). Altogether, the integration paradox literature shows that immigrant minorities report more discrimination as they undergo structural aspects of the assimilation process, such as time and generation since migration, and rising socioeconomic status—particularly educational attainment (Schaeffer and Kas 2023). Some studies show that higher educated immigrants also report lower levels of belonging and subjective inclusion within the national community at destination (Geurts, Davids and Spierings 2021; Drouhot 2024). Elucidating the mechanisms and scope of the integration paradox is the object of ongoing research. For instance, recent findings suggest that the paradox may be circumscribed to groups that are more easily identifiable as foreign in virtue of their appearance or name (Flores 2015; Diehl, Liebau and Mühlau 2021), and that it is more clearly undergirded by growing awareness (for instance, through higher news consumption) than “exposure,” that is, interpersonal encounters with hostile natives (Diehl, Liebau and Mühlau 2021). Additionally, while most research to date has focused on reports of personal discrimination, recent findings suggest that rising education, better language skills, and generation-since-migration are also predictive of perceptions of societal discrimination—whether ethnic, racial, or religious discrimination is widespread in society at large (Van Tubergen 2024).

The integration paradox has theoretical significance insofar as it offers a way to tackle the first challenge I described above, namely reconciling scholarship emphasizing assimilation—as manifested in upward mobility—and racialization—as manifested in the enduring significance of race and ethnoracial stigma in the lived experience of immigrants and their children. From a general theoretical point of view, the integration paradox suggests that empirical trends of structural inclusion may be accompanied by protracted periods of “brightened” ethnic boundaries during which claims to discrimination proliferate. A strong disjunction between objective metrics of assimilation and the subjective experience of exclusion signals that the lived experience of immigrant minorities cannot be ignored (Lee and Sheng 2024) and calls for revised theoretical linkages between socioeconomic assimilation and belonging (Schachter 2016; Drouhot 2024).

The Precise Measurement of Immigrant Selectivity Complicates Dominant Conceptions of Second-Generation Assimilation and Mobility

Recent decades have witnessed major development in the literature on immigrant selectivity—that is, how immigrants systematically differ from nonmigrants in their country of origins—and its effects on assimilation trajectories at destination (Feliciano 2020). The selectivity literature has particularly focused on educational selectivity and repeatedly showed that immigrants tend to be positively selected—to have higher educational attainment than nonmigrants in the country of origins (Feliciano 2020, 316–319). Even origin groups often regarded as having low average levels of education in absolute terms such as Mexican immigrants compare well to nonmigrants in Mexico, illustrating the importance of premigration characteristics in context. For instance, someone with a high school degree obtained in the 1970s and 1980s in Mexico is highly educated compared to other members of the same educational cohort, in which very few completed secondary education—similarly to the example of Nigerian immigrants discussed above. Using a measure of relative educational attainment in terms of percentile rank by cohort originally proposed by Ichou (2014), Engzell and Ichou (2020) show that the majority of immigrants in Western Europe are positively selected in terms of education. Focusing on the French case, Ichou (2014) showed that relative educational attainment at the first generation (i.e., within the country of origin) predicts the educational attainment of their children in France above and beyond absolute measures of parental socioeconomic status, such as parental occupations. Ichou's results suggest that the relative educational position of parents correlate with aspirations for status attainment transmitted to the second generation at destination.

Immigrant selectivity matters because it influences how the starting position of the first generation is measured, and whether second-generation outcomes in fact capture a movement of progress, stagnation, or decline once such starting positions position are accounted for. In their influential study of the second generation in New York, Kasinitz et al. (2008) noted the large educational progress among their respondents compared to their parents, who often had much lower schooling in absolute terms. More recently, Abramitzky et al. (2021) show that the second generation in the US has had a consistently higher income than natives of comparable income origins, both in the nineteenth century and among the new second generation. Together, these findings suggest a “second-generation advantage” in intergenerational mobility within immigrant compared to native families. However, Feliciano and Lanuza (2017), using the measure of relative educational attainment at the parental generation developed by Ichou (2014), demonstrate that immigrant selectivity drastically alter the optimistic picture of a second-generation advantage in the US: whereas the second-generation indeed appears to overperform given absolute levels of parental education in terms of years of schooling completed, it actually experiences intergenerational downgrading in educational ranks once parental relative attainment is taken into account. For instance, East Asian immigrants parent on average rank in the 88th percentile at origins, but their children rank in the 77th percentile only. More dramatically, Black immigrant parents rank in the 96th percentile of educational attainment at origins, while their children rank in the 55th percentile in the United States educational distribution (Feliciano and Lanuza 2017, 230). Mexican families experience a similar pattern of intergenerational educational downgrading.

Theoretically, better measuring selectivity matters beyond the substantive reframing of a putative second-generation advantage it induces. Indeed, relative position within educational or income rank correlate with subjective social status. Research in Western Europe demonstrates how positive selectivity among immigrants translates into status loss and relative deprivation in the country of destination (Engzell and Ichou 2020). Measurements of selectivity thus open a potentially fruitful link with the lived experience of status, belonging and worth among immigrants and their children ⁹ (Feliciano and Lanuza 2017, 233). In that regard, there appears to be a clear potential link between immigrant selectivity and perceptions of discrimination, which are sensitive to parental education and unfulfilled aspirations for mobility (Schaeffer 2019).

Samples of Noninstitutional Populations Based on Self-Identification May Under- and Overestimate Assimilation

While survey research has been and continues to be the workhorse of quantitative research into migration and assimilation, survey samples based on self-identification remain limited by important issues of coverage. Self-selection into and exit out of samples of the noninstitutional population may occur based on social processes endogenous to assimilation itself, such as ethnic attrition and incarceration. Together, these endogenous sources of bias may downwardly or upwardly impact assimilation estimates. ¹⁰

Ethnic attrition designates the process through which immigrant-origin individuals stop identifying with their ethnic ancestry. For instance, the child of a Mexican-born parents may stop identifying as having Mexican or Hispanic ancestry and simply identify with a broader national (“American”) or racial (“White”) category. Surveys of the Mexican-origin population relying on self-identification for eligibility would miss that individual by design. The issue of potential bias stems from the selective nature of ethnic attrition: for instance, among the Mexican-origin populations, ethnic attrition appears more likely in later generations, among the children of mixed couples, and among the highly educated and those speaking only English (Alba and Islam 2009; Duncan and Trejo 2018)—all central tenets of assimilation. Thus, in 2000 approximately 4% of those of Mexican origins in the US did not identify as either Hispanics or Mexicans, and instead favored a White racial identification (Alba and Islam 2009, 118). Conversely, and relatedly, those who favor a “Latino” panethnic identification only are more likely to be first generation, bilingual, and have less than a high school degree (Golash-Boza 2006; Martínez and Gonzales 2021). Ethnic attrition among Mexican-Americans is positively selective whereby English-only language use, nativity, educational attainment all chip away at ethnic identification. Among second-generation Nigerian-Americans by contrast, Emeka (2018) estimates a 26% decline in the number of self-identified Nigerian-origin individuals between 2000 and 2010, largely in favor of “African-American” as their only ethnic origin in the US Census. Ethnic exit among Nigerian-Americans is more likely among those with lower family income and parental education (ibid)—reflecting the tight linkage between Nigerian ethnic identity and university education (Imoagene 2017). While positive ethnic attrition may lead scholars to underestimate assimilation among ethnic stayers for Mexican-origin populations, negative ethnic attrition may lead them to overestimate assimilation among Nigerian-origin populations.

There is less work to date on the ethnoracial selectivity of incarceration among descendants of immigrants, in large part because there exists no official statistical source about incarcerated populations in terms of national origins and ancestry. Above, I noted that the pessimistic prophecy of downward assimilation (Portes and Zhou 1993) failed to materialize. Downward assimilation would nevertheless lead to widespread incarceration at the second generation. How much of our optimistic assessment rests upon shaky foundations induced by selective exit from samples of the noninstitutional population? Based on the CILS survey (Portes and Rumbaut 2018), Haller, Portes and Lynch (2011a, 742) presented unweighted estimates suggesting that approximately 15–18% of Haitian, Jamaican, West Indian, and Mexican second-generation males had been incarcerated in the 5 years preceding the survey, which is less than but comparable to figures for African-American males in 2000 (approximately 21%, Haller, Portes and Lynch 2011, 741). The ISGMNY (Mollenkopf, Kasinitz and Waters 2011) survey show lower, but nevertheless substantial rates of arrest and incarceration for the second generation in New York (Kasinitz et al. 2008, 183–188). In their comment, Alba , Kasinitz and Waters (2011, 767) urged caution about these numbers due to certain aspects of survey design in CILS, but the situation remains unclear over a decade later. Meanwhile, a rich new literature on the intergenerational consequences of paternal incarceration and neighborhood policing on educational attainment and other outcomes among minority youths has developed (e.g., Haskins 2014; Legewie and Cricco 2022) and makes the assessment of incarceration and encounters with the criminal justice system at the second and later generation all the more important.

These issues of selection in and out of samples of the noninstitutional population relying on self-identification are particularly vexing for the third generation—arguably the next frontiers in the study of assimilation (Jiménez, Park and Pedroza 2018; see below). Using some of the rare official data sources allowing to reliably identify foreign-born grandparents, parents, and their third-generation grandchildren, Duncan and Trejo (2011, 604–5) show that approximately 30 percent of third-generation Mexican youths no longer identify as Mexican, and that those who do are more likely to be high school dropouts, and less likely to come from ethnically mixed background in the parental and grandparental generations. Mixed background induces ethnic attrition at even higher rates among Asian-origin members of the third generation (Duncan and Trejo 2018). However, as in the case of Nigerian-Americans, ethnic attrition among Asian-origin groups appear negatively selected, as de-identifier are less educated than ethnic stayers on average (ibid, 136).

The Grandchildren of Immigrants are a Crucial Yardstick in Assimilation Theories But Are Difficult to Study

The grandchildren of immigrants have long been identified as a critical test within theories of acculturation and identity change (Hansen 1938; Gans 1979; Alba et al. 2002), for it has historically been “only with the third […] generation that the powerful undercurrent of assimilation came unmistakably to the surface” (Alba and Nee 2003, 215). In that sense, the current literature on the “new” second generation remains only suggestive of longer term trends, which can only be decidedly characterized by three-generational patterns. There exist two broad scenarios for the social destiny of the third generation in Western Europe and the United States. On the one hand, we could see persistent intergenerational progress and social blending, as has historically been the case for European-origin populations in the United States for instance (Lowrey et al. 2021). On the other hand, we could see enduring disadvantage among the most stigmatized origin groups, and possibly a third-generation decline, as has been the case for earlier cohorts of Mexican Americans in the United States for instance (Telles and Ortiz 2008; Ortiz and Telles 2017). Emerging findings suggest both scenarios may apply, with a master trend of assimilation coinciding with enduring exclusion and blocked assimilation among certain low-status populations across national contexts (Pupaza, Harber-Aschan and Wilson 2023; Zhao and Drouhot 2024).

While the third generation may hold the key to understanding assimilation in Western migration societies, its study will pose certain empirical challenges, such as issues of synthetic generation designs. In synthetic generation designs, unrelated individuals are compared by generational status—as opposed to comparing within a same family, for instance the native-born child of foreign-born parents. Indeed, synthetic generation designs necessarily confound any generational differences with cohort differences, and comparing three synthetic generations may compound the problem. One viable solution is within-family designs, whereby grandparents, parents, and children are reliably identified. Administrative data, such as register data, often allows for within-family, cross-generational linkage (Pupaza, Harber-Aschan and Wilson 2023). In countries where such data are not available or not suitable, retrospective and prospective approaches in which focal survey respondents report on their parents and children appear promising (Song and Mare 2015; Ferry et al. 2024). These type of data remains rare as of yet, and solely exist in Western Europe (Jiménez, Park and Pedroza 2018; Tran 2018).

Finally, increasing mixedness between natives and immigrants over generations complicates the study of the third generation. In Western Europe for instance, the large majority—between 75 and 90+%—of the third generation has either one or two immigrant grandparents (Drouhot et al. 2023; Zhao and Drouhot 2024). The pathways and processes by which such limited migration background influence social trajectories and identities two generations after migration remains unclear. While the endogeneity of mixed unions to generational change within immigrant populations is in line with expectations of population blending within assimilation theories (Kalmijn, 2015; Alba, Beck and Sahin 2018), certain studies show that lower educated natives are more prone to partnering with immigrants (Elwert 2020). While undoubtedly important, the implications of negative selectivity of natives engaged in mixed unions for assimilation among immigrant-origin population remain largely unknown.

Looking Ahead and into the Future of Assimilation Research

Conversations Across Methodological Traditions and Substantive Subfields Will be Necessary to Avoid Epistemic Drift and Encourage Theoretical Progress

It will not be lost on attentive observers that the bulk of what we know about assimilation stems from survey research—that is, descriptive research gauging how much and among whom assimilation is occurring in terms of socioeconomic equality, blending networks and intermarriage, and culture and belonging (Drouhot and Nee 2019). Yet, a renewed understanding of assimilation involves an explanatory focus on the theoretical causal processes underlying empirical trends documented in survey research. For that analytical task—not understanding how much but why assimilation occurs or not—survey research remains of limited help, and notable contributions leading to theoretical progress in recent years have come from other methodological traditions. For instance, Tomas Jiménez's (2010, 2017) qualitative research on ethnic boundaries, identity and belonging among later-generation immigrants in California and other immigrant-receiving regions has highlighted the theoretical importance of continuing migration in complexifying acculturation processes and problematizing what counts as the “mainstream” in majority-minority contexts. ¹¹ At the other end of the methodological spectrum, Schachter (2016) used conjoint experiments to study the determinants of perceived similarity of various immigrant groups among American Whites, highlighting the salience of race and undocumented status in pushing immigrants outside of the imaginary community of belonging. Together, these works further identify potential mechanisms and scope conditions for assimilation and point to unspoken assumptions and gaps within existing theories.

In recent years however, interventions aimed at productively criticizing, extending, or falsifying assimilation theories have slowed down, along with the emergence of new literatures pursuing different theoretical goals, such as the scholarship on new immigrant destination (Flippen and Farrell-Bryan 2021) or legal status marginality (Zhou and Gonzales 2019, 392–394). Perhaps more pointedly, we have witnessed increasing methodological specialization and segregation of quantitative and qualitative approaches in the social sciences. Tellingly, contemporary scholarship explicitly centered on assimilation overwhelmingly tends to be survey-based and quantitative. Meanwhile, mixed methods design productively combining quantitative and qualitative approaches—for example, Kasinitz et al. 2008; Portes, Fernández-Kelly and Haller 2009—remain rare. Given the track record of qualitative and experimental research for fruitful theoretical interventions, fragmentation into distinct methodological communities threatens theoretical progress. Better understanding the causal mechanisms undergirding assimilation and tackling the future challenges, I have outlined above critically depends on overcoming self-segregation into sophisticated yet separate methodological “cultures” (Goertz and Mahoney 2012) if we are to usher in a new era of theorizing on assimilation.

More generally, there exists a risk of epistemic drift going beyond methodological boundaries. Assimilation, one of the last remnants of earlier eras of structural-functionalist, “grand” theorizing, ¹² is unusual in its multidimensionality within the landscape of social science concepts. Simply put, assimilation may involve the study of inequality between immigrants and natives in schools, labor market and neighborhoods, blending or segregation in their respective social networks, and change in both immigrants’ and natives’ culture and identities as a result of migration, combined with dynamics of transmission over time and generations, and across countries of origins and destinations. The analytical complexity of assimilation as a social process on the one hand, and the increasingly intensive division of scientific labor pushing toward specialization on the other hand, imply that future scholarship is at a risk of fraction into separate epistemic communities, each too focused on discrete empirical aspects to make unified theoretical progress. I contend that future theoretical progress about assimilation crucially depends upon intellectual trades across specialist subfields and disciplines to reach a more holistic understanding of the assimilation process.

Existing Data Infrastructures and Future Data Requirements

Theoretical progress will depend on intellectual trade across methodological traditions and specialist subfields, but precise empirical description of assimilation will continue to rely on survey research in the future to test specific theories. Clearly, the stagnation of assimilation debates in the 2010s is a partial byproduct of the absence of new survey data since the gathering of CILS (Portes and Rumbaut 2018) and ISGMNY (Mollenkopf, Kasinitz and Waters 2011) surveys. Meanwhile, a major issue with other publicly available data sources, such as Census related surveys, is that they do not contain variables pertaining to cultural and attitudinal dimensions within assimilation. Additionally, and crucially, they do not allow to reliably identify the third generation due to missing questions on the birthplace of grandparents (Duncan and Trejo 2018; Tran 2018). Research on ethnic attrition clearly points to the need to reliably identify ancestry, while other substantive developments outlined above (e.g., regarding the importance of immigrant selectivity, racialization and cultural difference, the emergence of the third generation) similarly call for appropriate and up-to-date measurements within new survey instruments.

Yet, and as much as American-born theories of assimilation have influenced migration research in the United States and Western Europe, the American data landscape remains characterized by scarcity. In fact, there exist no nationally representative samples of the adult second generation in the US (Waters et al. 2015, chapter 10). By contrast, new survey instruments in Western European countries have been rapidly developing in the last 25 years, and have now firmly outstripped their American counterparts in terms of breadth and depth. In part, this discrepancy reflects a surge in public interest about the children of immigrants in Europe in the 2000s, leading to major funding efforts. The “Norface” Research Programme on Migration is particularly noteworthy in that regard: through an open call for projects in 2009 and with a budget of 28.6 million euros, it funded 12 major research projects. One such project, the Children of Immigrants Longitudinal Survey in 4 European Countries (CILS4EU 2016), a survey of immigrant-origin and native youths in classrooms within European schools and partly inspired by the American CILS, featured a unique sociometric modules of friendships in European classrooms which pioneered and gave rise to a new literature on friendship formation among ethnically and religiously diverse youth (e.g., Smith, Maas and van Tubergen 2014). The 2000 Families survey (Guveli et al. 2017), a survey of Turkish migrant families in multiple European countries along with a survey of nonmigrant families in Turkey and the two Trajectoires et Origines surveys in France (Beauchemin et al. 2016; Beauchemin, Ichou and Simon 2023), which are unique in their thematic breadth and coverage of the immigrant and nonimmigrant-origin adult populations, are also noteworthy. The European Social Survey, a biannual, cross-sectional fielded in dozens of European countries with rotating core themes since 2002, has also featured substantial subsamples of migrant-origin populations and proven to be a valuable research asset over the years. By contrast, the main purveyor of resources to fund specific survey instruments pertaining to the second generation in the United States—such as the CILS and the ISGMNY surveys—has been the Russell Sage Foundation, a private nonprofit organization focused on social science research with a keen interest in immigration research (Waters et al. 2015, chapter 10).

The research boom on the second generation in Western Europe within the last few years has undoubtedly led to new insights, including some of the emerging analytical challenges I discussed above. Such progress bears witness to the importance of robustly funding research instruments for the advancement of theory-driven empirical research. New data need not generate new theories, but further theoretical debates crucially depends on new, high-dimensional data with up-to-date measurements, lest assimilation scholarship falls into intellectual obsolescence. In particular, we need high-dimensional data to analyze how socioeconomic, relational, and cultural dimensions of assimilation relate to one another—including data on perceptions of discrimination, belonging, and educational attainment among parents and grandparents to measure educational selectivity in order to tackle the new challenges I have outlined above. While official statistical sources and administrative data (e.g., population registries) may offer rich way to describe patterns of socioeconomic attainment and intermarriage for instance, they more rarely afford researchers opportunity to study these dimensions in conjunction with culture, identity, and attitudes. In the future, tailor-made survey instruments will remain indispensable due to the multidimensional nature of assimilation, and the assumed theoretical relationships between different dimensions calling for empirical tests.

Possibilities of the Digital Age: Data-Driven Theorizing in an Era of Data Plenitude

Traditional surveys based on probability sampling remain expensive. Face-to-face interviewing always involved great costs, but even telephone surveys—historically the work horse of survey data collection—have dramatically risen in costs due to declining responses rates in recent years, which in turn is associated with potential measurement error (Couper 2017). Just like telephone survey and random-digit-dialing dramatically altered survey research in the 1960s, the digital age is precipitating the decline of telephone surveys and probability sampling, ushering in a new era of dramatically cheaper, self-administered survey based on online, nonprobability samples (Stern, Bilgen and Dillman 2014; Couper 2017; Salganik 2018, 85–7). Proprietary, representative web panels are now well-established in social science research, and more recently, online recruitment of survey respondent via targeted advertising campaigns has shown promise for the study of migrant-origin populations (Pötzschke and Braun 2017).

Outlining the potential of web-based surveys for migration research is hardly new, as attested by several distinguished contributions to our field made through them (e.g., Bansak, Hainmueller and Hangartner 2016; Schachter 2016). However, their potential importance has not been fully appreciated by migration and assimilation scholars to date, whose attention may have been diverted by the effervescence around “big data”—large-scale sources of data emanating from naturally occurring online behavior of digital users as well as the records of governments and private companies (Salganik 2018, chapter 2). Big data hold promise for migration and integration research due to its behavioral rather than attitudinal nature, high granularity, and for digital sources, its “always-on” nature allowing to record human behavior over time (ibid). For instance, large-scale digital data are particularly well-suited to study changing discourses and attitudes pertaining to migration and integration (Drouhot et al. 2023). In spite of such strengths, big data remain hampered by missing demographic variables that are standard in surveys (Salganik 2018, 24–7). The absence of variables identifying generational status, place of birth, and ethnicity are particularly vexing and will continue to constitute an unsurmountable barrier until reliable of ways of inferring them emerge (Drouhot et al. 2023, 397).

Given these limitations, I contend that future theoretical progress about assimilation will hinge upon fielding rich web surveys allowing the study of multiple assimilation dimensions at once, with reliable information on ancestry, nativity, and generational status at a fraction of the cost of traditional surveys. Specifically, I expect that large, and high-dimensional samples of second- and third-generation respondents help us get away from “categorical intuitionism” or the tendency to take for granted the relevance and analytical primacy of certain categories within assimilation. Categorical intuitionism is a long-standing issue in migration research at large, whereby different strands of scholarship primarily approach the study of immigrants and their descendants through a focus on either race, ethnicity, national origins, religion, social class, legal status, and other potential categories depending on their substantive and theoretical inclination. In the American literature, it is commonplace to use racial and national origin categories and engage in “ethnoracial Olympics” pitting different groups against one another, whereas the European literature has seen a progressive displacement of ethnic categories in favor of religious ones, particularly that of “Muslims” (Brubaker 2013)—although a generic and loose usage of “ethnicity” looms large in the quantitative European scholarship on “ethnic penalties” for instance (Wimmer 2009).

Over time, and across research communities and national contexts, different categories fluctuate in analytical attractiveness along with their degree of politicization in the public sphere and entrenchment in both public statistical infrastructures and common sense (Wimmer and Glick-Schiller 2003). Altogether, we are left with a “categorical cacophony” (Türkmen 2024) blurring both our scholarly gaze and the internal heterogeneity of immigrant-origin groups, and deductively imposing certain categories as analytically appropriate. In the future, I expect that new high-dimensional, high-quality digital surveys should help us get past categorical intuitionism and closer to categorical agnosticism—an analytical stance emphasizing within-category heterogeneity and cross-categorical intersectionality, and inductively letting the most appropriate categories emerge from the data source at hand rather than a priori imposing them on it, for instance using data-driven techniques when deciding what groups are relevant analytical categories in assimilation (Molina and Garip 2019, 35–8). National origin, race, social class, religion, gender, destination region, and their intersections all may or may not form relevant lines of differentiation within the landscape of assimilation. A categorically agnostic approach puts the data first, since the salience of different categories of differentiation are “a key part of what we want to explain, not what we want to explain things with; it belongs to our empirical data, not to our analytical toolkit” (Brubaker 2004, 9).

Secondly, and in closing, a new era of data plenitude can usher in new ways to conceptualize how assimilation dimensions relate to one another. Indeed, some of the analytical challenges I noted above involve just that—for instance, how socioeconomic mobility may lead to more, not less, perceived discrimination. However, there may be strong heterogeneity in that relationship—upward mobility may lead to stronger perceived discrimination among some members of some immigrant-origin groups, while mobility may indeed lead to higher perceived belonging among others, along with the lines hypothesized by assimilation theories. In other words, they may be both heterogeneity in the strength of association between different assimilation dimensions and in the direction of these associations. Large enough sample sizes—which web surveys are generally able to yield—and data-driven techniques for the inductive identification of shared structures in responses to survey data can enable the discovery of new migration and assimilation pathways (Garip 2016; Drouhot 2021; Drouhot and Garip 2021). Altogether, and in conjunction with categorical agnosticism, data plenitude enabled by cheaper and more easily deployable survey instruments may allow for data-driven theorizing, whereby relationships between different assimilation dimensions do not exclusively derive from hypotheses within middle-range theory but also from empirical discovery of surprising or anomalous patterns leading to post hoc theorizing (Goldberg 2015; Salganik 2018, 61–2). In the context of aging assimilation theories primarily developed in one social–cultural context, namely the United States, I expect abductive reasoning—the “inferential creative process of producing new hypotheses and theories based on surprising research evidence” (Timmermans and Tavory 2012, 170) in other national contexts to play an important role in future theorizing within our research field (Goldberg 2015). In the long run, post hoc explanations of new empirical patterns may aggregate in more general predictions, closer in spirit to middle-range theories and the deductive hypotheses and typologies that have made the heydays of new assimilation and segmented assimilation theories.