Intergenerational Mobility in the Post-1965 Immigration Era: Estimates By An Immigrant Generation Cohort Method

Abstract

The new second generation of the post-1965 immigration era is observed as children with their parents in 1980 and again as adults 25 years later. Intergenerational mobility is assessed for both men and women in four major racial/ethnic groups, both in regard to children’s status attainment relative to parents and with regard to the rising societal standards proxied by native-born non-Hispanic whites. A profile of intergenerational mobility is prepared using multiple indicators of status attainment: high school and college completion, upper white-collar occupation, poverty, and homeownership. The immigrant generation cohort method we introduce accounts for four distinct temporal dimensions of immigrant progress, clarifying inconsistencies in the literature and highlighting differences in mobility between racial/ethnic groups and with respect to different outcome measures. The immigrant generation cohort method consistently finds greater intergenerational mobility than suggested by alternative approaches. Our analysis also shows that the intergenerational progress of women is greater than that of men and provides a more complete record of immigrant mobility overall. Findings for individual racial/ethnic groups accord with some expectations in the literature and contradict others.

The children of post-1965 immigrants to the United States are regarded as the “new” second generation (Massey 1995; Portes and Zhou 1993; Rumbaut 1997) to distinguish them from the second generation of the great wave of immigration during the earlier part of the twentieth century. As the children of this earlier era of immigration entered adulthood, theories of acculturation and assimilation were created and tested to understand the adaptation processes of immigrants and their descendants. Aspects of these theories are still called upon today in reference to the new second generation, but new theories have also emerged.

Why are there new theories of immigrant adaptation or assimilation for the new second generation? Many scholars, such as Gans (1992) and Portes and Zhou (1993), have explained that there have been substantial changes in immigration flows as well as the U.S. context in such a way as to warrant these new theories. First and foremost, post-1965 immigrants are more racially and economically diverse than past immigrants, which may substantially influence the likelihood of intergenerational progress. Second, the economic context has been restructured from being heavily dominated by solid middle-class, blue-collar jobs to an hourglass economy that has bifurcated economic opportunity. Furthermore, the social context of rising equality for women in educational and occupational attainment has also warranted a new narrative of gendered paths of socioeconomic progress.1 Finally, the advent of much richer microdata sets, including historical Public Use Microdata Samples (PUMS) files, has enabled more nuanced investigation than was possible before 1980.

The diversity of immigrants and the changing context of reception have prompted pessimistic theoretical assertions for immigrant mobility (Perlmann and Waldinger 1997). Gans (1992) identified second-generation “decline” as a possible description of what will occur when second-generation expectations are not met and immigrants’ children are not willing to take the kinds of low-wage jobs that their parents had. Followed by many, Portes and Zhou (1993) described divergent destinies for different ethnic groups in their “segmented assimilation” theory. Within a changing economic context, the central question remains: Will the new second generation no longer merge into the mainstream as the second generation of the past did?

The changing U.S. context in the past few decades certainly impacts the socioeconomic prospects for the new second generation, but it also impacts those in the mainstream. Is there a method to capture both the intergenerational progress of immigrant generations as well as the changing U.S. context? In this article, we offer the immigrant generation cohort method that simultaneously assesses how the second generation has progressed relative to their parents and how the mainstream has changed. We use the 1980 PUMS and 2005 Current Population Survey to observe actual parents of the second generation and their children when they reach the same age some 25 years later. The socioeconomic status of both generations is within a societal context that is concurrently changing, as scholars have emphasized, so the tracking of the mainstream is an integral part of this method. We attempt to be temporally specific so that we can be more explicit about what changes we are actually measuring.

In this article, we use the immigrant generation cohort method to address the following questions: Does the new second generation have higher socioeconomic status than their immigrant parents, or are they in fact experiencing decline? Is this progress or decline more or less than is experienced by members of the mainstream who are of a similar age? Does the level of progress or decline remain consistent across multiple indicators of socioeconomic status? Beyond gender differences in socioeconomic attainment, is the intergenerational mobility from immigrant mothers to second-generation daughters more than what was experienced by mothers and daughters in the mainstream?

BACKGROUND

Defining Immigrant Mobility

The concept of immigrant mobility hinges on tracking changes over time. However, scholarship on immigrant mobility has not always been clear about the dimension of change being analyzed. Early scholarship on the turn-of-the-century immigrants and their children relied mostly on the conception of melting into the American mainstream (Zangwill 1908). Mid-century scholars of immigrant assimilation theorized about this pace of change without recourse to any data (Gordon 1964) or based only on what was observable at the present moment (Glazer and Moynihan 1963). More explicit status comparisons of the first and second generations began with Warner and Srole (1945) and continued in the post-1965 era but were hampered by the absence of generational data collected on a large national scale between 1970 and 1994.2 Data for earlier decades also were very limited, required resourceful use of zones of residence and address directories, and identified occupations of fathers (Thernstrom 1964). Mid-century quantitative studies of intergenerational mobility of fathers and sons, ignoring nativity, relied on retrospective surveys of parents’ status (e.g., Blau and Duncan 1967; Duncan, Featherman, and Duncan 1972). While the concept of social mobility in these studies is measured by indicators of educational and occupational achievement, Gans (2007) emphasized that economic mobility differs from social mobility, and others have observed that earlier scholars of immigrant assimilation often conflated mobility and assimilation (Alba 1985; Kasinitz, Mollenkopf, and Waters 2004). The current availability of more detailed data on generational changes now requires more explicit choices about relevant comparisons and their interpretation. Measurement of mobility on different social and economic dimensions certainly should be kept separate from interpretation of assimilation, even if that mobility can be used to inform those interpretations.

In a systematic conception, progress or mobility with respect to status attainment can be defined in four distinct ways. First, upward mobility, or lifetime mobility, occurs during an individual’s lifetime. Second, each subsequent generation may achieve higher socioeconomic status than its predecessor, or what can be referred to as intergenerational mobility. Third, convergence (or divergence) with respect to the mainstream is often termed assimilation. And fourth, as backdrop to these individual changes, societal progress occurs as the prevailing norms or standards in the American mainstream rise or fall. Thus, an individual could make substantial lifetime progress but still end up below her or his parents’ level, or she or he could achieve mobility that exceeds the parents’ attainments but still lag behind the rising societal standards.

Any assessment of mobility necessarily involves some combination of these temporal dimensions, even if it is not explicitly recognized. Rather than seize on a single dimension, it is preferable to track all four. The greater information obtained also guards against the risk of making misleading interpretations because of omitted temporal dimensions. A further advantage of introducing greater clarity of temporal analysis is that this may help to reduce ambiguity and inconsistencies in how “mobility” is treated in the immigrant literature.

Identifying the Status Attainment of Immigrant Parents and Grown Children

The mobility of the second generation cannot be assessed unless we first develop a measurement of the status achieved by the first generation. In the first great immigration wave of the twentieth century, the socioeconomic status of immigrant parents was assumed to be low and fixed (Lieberson and Waters 1988). However, there are now well-known differences between immigrants from different origins. In fact, the post-1965 era of immigration witnessed great diversity not only in the countries of origin for immigrants but in their educational, occupational, and economic status as well as their proficiency in English (Alba and Nee 2003; Bean and Stevens 2003; Perlmann and Waldinger 1997; Portes and Rumbaut 2006). This diversity among immigrants requires a more exact measurement of the first generation, against which children’s mobility can be assessed.

The immigrant mobility literature has almost wholly focused on the progress of immigrants and their children in their country of new settlement. This focus usually disregards immigrants’ status relative to others in the country of origin and the selective nature of immigration, even though it is an important context for understanding mobility globally. As Zeng and Xie (2004) pointed out, many immigrants finish their education in their country of origin, and the place of education has an impact on earnings in the United States. Feliciano (2005) has also raised the important point that immigrants are often drawn from higher education classes in their home country, whereas their children are relatively less advantaged in the United States. This could have the implication of downward mobility for the second generation, although that also requires assessing the education differences between countries. Though these are important considerations, we focus here on the mobility that occurs within the United States and therefore use the U.S. mainstream, rather than source countries, as the reference group. The question of appropriate reference group is crucial, and we return to it in a later section.

Measuring generational progress requires some precision about the age when generations are compared. Some aspects of status attainment are relatively fixed over the life course (e.g., education and occupation), but others vary more substantially (e.g., earnings and homeownership). Because of this variability, it is essential to compare generations when they are the same approximate age. Given that the second-generation children of post-1965 immigrants are now well established in early adulthood, it is possible to compare them to their parents when they were of similar age.

At the same time, prevailing societal standards for some attainments have changed substantially from 1970 into the twenty-first century. A comparison of two different generations observed at the same age must also take account of changes in average attainments in society. Among the clearest examples is the rising college education norm for women (Stoops 2006). Another is the changes over the decades in homeownership attainment that are influenced by rising prices and new credit arrangements. Accordingly, it is important to assess intergenerational mobility relative to ongoing societal progress.

Multiple indicators surely provide a better measure of status attainment of generations than reliance on only one or two. It may not be reasonable to equate status attainment with a single indicator, such as education or earnings, nor is it reasonable to assume that men and women, or different ethnic groups, can be equally well assessed on one single indicator. Further, whether intergenerational mobility proceeds in tandem with respect to all indicators deserves to be tested. For all these reasons, comparing outcomes of status attainment on multiple indicators is preferable.

Variations in Mobility by Race/Ethnicity

The racial/ethnic as well as economic diversity of immigrants and their children is a distinctive aspect of the post-1965 era (Gans 1992; Portes and Zhou 1993). As of 2005, 51.8% of the second generation is Latino and 17.9% is Asian, and most scholarly attention has been devoted to these groups. Distinctive lessons for scholarly understanding of immigrant progress have been drawn also from black immigrants, as well as from specific ethnic groups among Latinos and Asians. For example, some findings from research on Mexican or Puerto Rican intergenerational progress lend support to the thesis of “second-generation decline” (Gans 1992), while research on Cubans in Miami has bolstered a different notion of the importance of enclave economies (Portes 1987; Waldinger and Der-Martirosian 2001). Meanwhile, many of the post-1965 Asian immigrants arrived with high occupation levels and education levels that were higher than the U.S. average. However, it is important to note that there is considerable diversity within the Asian group and that some Asian immigrants do not have high socioeconomic status. In general, will second-generation Asians maintain these high status attainments, achieve even higher levels, or converge to societal standards?

Non-Hispanic blacks compose only 4.8% of the second generation, but they also have made an imprint on scholarly notions of second-generation mobility. As the children of black immigrants navigate through the entrenched racial climate of the United States, the intrinsic value of assimilation is challenged (Waters 1999). A question we can begin to address is, How do second-generation blacks fare socioeconomically as compared with their immigrant parents? Remarkably, little attention in recent years has been given to white immigrants and their children, even though the children compose 25.5% of the second generation. Tracking this group, although they are of many different national origins, can be particularly telling because they evidence less racial difference from the established, native-born white population, and yet they also are learning to adapt to the United States like other immigrants. The lifetime mobility and intergenerational progress of white immigrants provides a useful reference for gauging the progress of other immigrant groups. Similarities in progress observed across racial/ethnic groups would substantiate a general story of immigrant intergenerational progress. Alternatively, if the white second generation experiences greater progress than the second generation of other groups, that would corroborate the importance of race.

Overall, what is the evidence of second-generation decline for any of the groups of immigrant children? Comparing children and parents at the same stage of life, we can control the variation in lifetime mobility. That allows us to focus on tracking the other three dimensions of change. Surely, if immigrant children achieve levels of status attainment that are higher than their parents and higher than the societal standard, which is also rising, that is not second-generation decline. But what is the assessment when they exceed their parents but fail to keep up with the rising standard? Or what is the assessment when immigrant children exceed the societal standard but fall short of their parents’ attainments? All of these outcomes may be expected in the data analysis to follow. With an accumulated set of findings, we can then discuss better the meaning of second-generation decline.

Mobility of Women and Men

Gender has received much less attention in the literature on intergenerational mobility than race and class. Among others, Feliciano and Rumbaut (2005) have voiced concern that gender is also an important factor in understanding how mobility is experienced. Especially in the areas of educational, occupational, and earnings attainment, most scholars recognize that there can be vast differences between men and women. Some studies of the new second generation’s educational attainment have revealed that women tend to have higher educational attainment then men (Feliciano and Rumbaut 2005; Kao and Tienda 1995; Portes and Rumbaut 2001; Rumbaut 2005; Zhou and Bankston 2001). Many point out that this is particularly noteworthy because of the sharp departure from the first generation, who have educational attainment levels reflecting their more traditional home countries (Feliciano and Rumbaut 2005; Lopez 2003; Zhou and Bankston 2001). Many like Feliciano and Rumbaut (2005) have focused on the educational expectations of immigrant parents and then subsequently on the differences between men and women in actual educational attainment.

Meanwhile, much of the research on occupation and earnings has sidestepped this issue by limiting the sample to men (Borjas 1985, 1995; Card 2005; Duleep and Regets 1997; Duncan et al. 1972; Smith 2003). When gender has been addressed, research most often entailed comparisons between men and women at a given point in time (Waldinger and Feliciano 2004). Those differences in attainments are important, but they do not get at the actual progress made from one generation of women to the next. The conception of gendered paths for intergenerational mobility (Feliciano and Rumbaut 2005; Waldinger and Feliciano 2004) could be concretely tested by comparing the attainment levels of immigrant mothers with that of their second-generation daughters. More specifically, how much better off are second-generation women than their immigrant mothers? How does this pattern of intergenerational mobility differ from that of their male counterparts?

METHODOLOGY AND DATA

Alternative Temporal Frameworks for Immigrant Mobility

The four dimensions of immigrant mobility identified above are not readily extracted from data commonly available. Instead, scholars have mined data sets for what generational information they contain, and the inherent limitations necessitate various compromises. The most common approach is comparison of all generations at a single point in time (Bean et al. 1994; Grogger and Trejo 2002; Kao and Tienda 1995; Zsembik and Llanes 1996). This approach of a generation cross-section within one period can be faulted because today’s first generation cannot also be the parents and grandparents of today’s second and third generations. At least, they cannot all be young adults at the same moment in time. Furthermore, today’s first generation has arrived in an economic and social context far different from the context that received the first generation decades earlier.

A second approach tries to solve these problems by comparing an older cohort from the first generation (e.g., 50 years and older) to a younger cohort of the second generation. Farley and Alba (2002) compared the status of different generations in a single cross section by selecting progressively older age groups to represent the earlier generations. This lagged birth cohorts approach using cross-sectional data is widely accepted as a way to measure intergenerational progress (Reed et al. 2005). The drawback in this approach is that the age groups’ current status attainments are being compared, not their status when they occupied similar adult ages at an earlier point in time. In addition, when all generations are compared at a single point in time, it is impossible to compare their mobility relative to the societal standard that has been changing over the last several decades. We prefer to adopt a third alternative temporal framework, an alternative that compares the first generation observed in one decade with the second generation observed in a later period, employing an historical time spread that matches the approximate spacing between generations. We can describe this general design as a generation lagged across periods. Major examples of this design, which was only recently developed as an analytical strategy, are offered by Smith (2003) and Card (2005). They not only matched staggered age groups, as did Reed et al. (2005), but also compiled an historical time series of censuses and CPS surveys. Thus, Smith matched parent birth cohorts from earlier eras to birth cohorts born 25 years later and observed during their adult years. While this approach avoids the artifice of the method of lagged birth cohorts in a single period, which compares generations at the same point in history, it must overcome an additional problem. Essentially, generations are being compared from different eras, under different economic conditions, and so we are comparing both generations and periods of history. Some means to separate out those period effects is needed. Smith’s solution was to compare Mexican and Hispanic status relative to the status of native-born whites in the same birth cohort in the same period.

Immigrant Generation Cohort Method

In this article, we introduce a new model design for tracing immigrant mobility that employs the general strategy of generations lagged across periods. We term this new method the “immigrant generation cohort” method. Generations are specifically defined in this method by observing foreign-born parents living with U.S.-born children in 1980. Parents and children are selected from specific age groups in 1980, and the status of grown children who report foreign-born parents is assessed in 2005 when their cohort reaches the same approximate age as that observed for the parents’ generation in 1980. The model employs a white native-born or third- and higher-generation reference group to represent changes in societal standards over the decades, and all groups are compared at age 35.

The model structure efficiently captures multiple dimensions of immigrant mobility in a compact design adapted from the double cohort model of Myers and Lee (1996) and Myers and Cranford (1998). The foundation for the model is in the data selection to represent generations. The analysis focuses on the second generation observable in the Current Population Survey in recent years. Accordingly, parents are observed a generation earlier in the decennial census of 1980. For this intergenerational mobility analysis, in place of arrival cohorts, we have generational status (G), which comprises four pooled groups: first-generation parents from the 1980 sample and second-generation grown children from the 2005 sample, both coded G = 1; and the native-born from both 1980 and 2005, both coded G = 0. What might be ideal, consistent with Alba and Nee’s (2003) notion of a merged mainstream, is a pooled sample of all the native-born combined who are third or higher generation. However, limitations in the 1980 data permit us only to observe all the native-born (second generation and higher), and to better approximate the changing societal standard in the United States, as well as to provide a more specifically defined reference group, we restrict this native-born reference to non-Hispanic whites in both 1980 and 2005. This group offers a fairly stable measurement of mainstream societal attainment since more than three quarters of the third and higher generations, according to the 1970 census, is non-Hispanic white. This is discussed in greater detail in the Data section.

Changes over time and between the generations are represented by Year and its interactions. The main effect of Year represents period change in outcomes for the “mainstream” reference group between 1980 (Year = 0) and 2005 (Year = 1). The differential effect of passage between immigrant generations is represented by Year × G. The resulting intergenerational model is represented thus:

$$(\text{O})=\text{Year}+\text{G}+(\text{Year}\times \text{G})+\text{Age}+\mathbf{\text{X}},$$

where (O) is the outcome variable of interest; Year is the observation year (1980 = 0 and 2005 = 1), capturing period effects for the “mainstream” reference group; G represents generation, represented by second generation in 2005 and first generation in 1980, contrasted with a reference group proxy for the “mainstream” (native-born whites); (Year × G) is the differential effect of passage of time between first and second generations over and above changes for the “mainstream” reference group; Age is measured in exact years, center-coded to age 35; and X is a vector of covariates (gender, marital status, education, area contextual factors, or other).

We define the second generation as U.S.-born children who do not have a U.S.-born parent. Our definition is more specifically targeted than what has been common practice. Some scholars use an overly inclusive definition of the second generation to include children with one foreign-born and one U.S.-born parent.3 However, Ramakrishnan (2004) has shown that there are significant differences in status attainments between the true second generation and those with one U.S.-born parent, who he terms the 2.5 generation. Another practice uses an over-inclusive definition of the first generation to include all immigrants over the age of 50 as a proxy for the parents of the second generation (e.g., Farley and Alba 2002). However, there are several downsides to this method. First, many older immigrants are not truly parents of the second generation. They may have lived in the United States since 1980 but immigrated with foreign-born children or they had no children at all. In addition, this broad definition also includes those who immigrated to the United States more recently and could not possibly be the first-generation parents of today’s second generation. In some ways, this definition is not inclusive enough because not all of the parents of today’s second generation had survived to be surveyed in 2005 either due to emigration or mortality. The immigrant generation cohort method avoids these potential pitfalls by selecting only immigrant parents who are living with second-generation children in 1980.

The treatment of age deserves explanation because it differs from the cross-sectional controls commonly employed in many models. Although most authors have applied some form of age control, this effort to hold age constant does some violence when other time factors are allowed to proceed. A good illustration of the difference between cross-sectional age and lifetime aging is provided in Myers and Cranford (1998), who showed how immigrant women in the Los Angeles region moved briskly out of low-wage factory jobs as they grew older and resided longer in the United States (the cohort trajectories steeply declined from 1980 to 1990). Nonetheless, despite their decline, older women retained a much higher likelihood of holding these jobs than did young women. This was a relic of the older women’s earlier job choices, something not being followed by young women. In a static picture, the age effect is positive on this choice of low-wage job, but in a dynamic view, the effect of aging is negative. Simple insertion of a control for age effects could not capture the correct direction of lifetime mobility and might lead to erroneous conclusions. Thus, a model that performs well by controlling age in the case of education, which is invariant over the life course, such as tested by Smith (2003), should not be assumed to perform equally well for other outcomes.

Data

The data set used in this analysis is constructed from the 1980 census Public Use Microdata Samples (PUMS) and the pooled data from the 2003, 2005, and 2007 Current Population Survey (CPS) data (from here on referred to as “2005”).4 The sample drawn from the entire United States is designed to repeatedly observe second-generation birth cohorts in 1980, when they were ages 0 to 16, and again in 2005, when they were 25 years older (i.e., ages 25 to 41). The presence of these children in 1980 is the marker for locating their parents, the first generation in 1980.5 We identify them as foreign-born parents living with native-born children who are between the ages of 0 and 16.6 This sample of parents, restricted to ages 25 to 44 for comparability, is then compared with a separate sample of second-generation children in 2005, drawn from cohorts now grown 25 years older and approximating the same age as the parents were in 1980.7 It deserves note that the sample is constructed from repeated cross sections and does not longitudinally trace kin between generations.

Separately, there is the issue of defining the group to represent the American “mainstream” for use in tracking the rising societal standards. We stated earlier that a third-generation reference group is desired as a proxy for the mainstream in both 1980 and 2005. Given the parental nativity questions in the CPS, it is possible to identify the third and higher generations in 2005. However, there is a long lapse between the 1970 decennial census and the 1994 CPS, during which questions about parents’ nativity or other means of determining third-generation status were not included in large federal data sets (Gibson and Jung 2006). Accordingly, we cannot identify the third generation in 1980 separately from all native-born. Our challenge is to define a reasonable proxy for the third-generation mainstream. Our solution is to select the white, non-Hispanic native-born between the ages of 25 and 44, just as for the first generation. Although these native-born adults include some who are second-generation, analysis of this cohort in 1970, when it was 10 years younger, shows that 77.4% are third or higher generation. For consistency’s sake, in 2005, we again define a proxy for the mainstream as white, non-Hispanic, native-born persons. Due to the extremely large sample size of the “mainstream” in 1980 that far outweighed the first generation in statistical analyses, we randomly selected 2% of the sample for our analysis so that the sample size would be more comparable to that of the first generation. Because the 2005 CPS sample is much smaller than the decennial PUMS, we randomly selected 10% of the native-born non-Hispanic whites to again be more comparable to the second generation.

We carry out this analysis for the total population as well as for Hispanics, non-Hispanic Asians and Pacific Islanders, non-Hispanic blacks, and non-Hispanic whites. The (mostly) third-generation sample of native-born whites is used as a consistent reference group for gauging mobility of all ethnic groups, not as a goal for assimilation but as a marker of changing socioeconomic standards across periods. To further refine the age comparisons, analyses will be controlled to age 35, a point at which adult status attainment should be fully revealed for the new second generation.

Key selected outcome variables are used to measure socioeconomic characteristics of the generations: educational attainment is determined by measuring the percent of the population that has completed a high school diploma or higher and a bachelor’s degree or higher. Occupational attainment is determined by the percent of all workers in upper white-collar occupations (those in professional and managerial occupations).8 Poverty status is determined by the percentage of persons who fall below the federally determined 100% poverty level. Finally, homeownership is measured by the percentage of householders and spouses who live in homes that are owner-occupied.

COMPARISON OF CROSS-SECTIONAL AND IMMIGRANT GENERATION COHORT APPROACHES

Table 1 shows the socioeconomic status of the first and second generations measured by alternative methods. The central question of interest is, how different are the conclusions about intergenerational mobility that are reached with the cross-sectional and immigrant generation cohort approaches? First, the simplest method for measuring socioeconomic status for immigrant generations is to take a cross section of the first and second generation at the same point in time for all adults age 25 years and older. Next, because we understand that the first generation necessarily has to be older than the second generation, the lagged birth cohort cross-section also takes the first and second generation from the same point in time but with the first generation being limited to ages 50 to 69 and with the second generation being limited to ages 25 to 41. Lastly, the immigrant generation cohort method observes the first generation (immigrant parents of the second generation aged 0 to 16) between the ages of 25 and 44 in 1980 and the second generation in 2005 between the ages of 25 and 41. The socioeconomic status of the second generation is exactly the same in both approaches (observed at ages 25–41 in 2005). What differs is the characterization of the first generation. In the cross-sectional approach, this is represented by an age group that is approximately the parents’ generation in 2005 (those between the ages of 50 and 69). For the immigrant generation cohort approach, the first generation is specifically observed to be the parents of the second generation in 1980 and was between the ages of 25 and 44 at that time.

Table 1.

Alternative Measures of Intergenerational Mobility for the Total Population (percentages)

Variable	Approach	First Generation	Second Generation	Intergenerational Mobility
High School Completion
	Simple cross sectiona	66.3	82.4	16.1
	Lagged birth cohort cross sectiona	69.0	90.3	21.3
	Immigrant generation cohortb	58.1	90.3	32.2
College Completion
	Simple cross sectiona	25.9	28.0	2.1
	Lagged birth cohort cross sectiona	28.9	36.8	8.0
	Immigrant generation cohortb	20.3	36.8	16.5
Upper White-Collar Occupation
	Simple cross sectiona	27.5	42.5	15.0
	Lagged birth cohort cross sectiona	31.6	40.5	8.9
	Immigrant generation cohortb	22.5	40.5	18.0
Above Poverty
	Simple cross sectiona	85.8	91.8	6.0
	Lagged birth cohort cross sectiona	88.8	90.9	2.1
	Immigrant generation cohortb	84.9	90.9	6.1
Homeownership
	Simple cross sectiona	59.4	76.2	16.8
	Lagged birth cohort cross sectiona	71.3	63.7	–7.6
	Immigrant generation cohortb	56.7	63.7	7.0

^aSimple Cross-Section is derived from 2005 data for those aged 25 and older. Lagged birth cohort cross section is derived from 2005 data. The first generation was aged 50–69 and second generation was aged 25–41.

^bImmigrant generation cohort is derived from 1980 and 2005 data. The first generation was aged 25–44 in 1980. The second generation was aged 25–41 in 2005.

The simple cross section broadly misestimates attainments because of the broad age category and the many decades of experience pooled together. It underestimates educational attainment for the second generation because it includes older adults who have lower education that was completed decades earlier. On the other hand, the inclusion of older adults then overestimates occupational, economic, and homeownership attainments because older adults have had more time to achieve upward mobility in their lifetimes. Alternatively, the lagged birth cohort cross section corrects for life cycle spacing between generations by controlling the first generation to only those aged 50 to 69 and narrows the ages of the second generation to younger adults. However, it does not adequately control for life cycle factors since the generations are not compared at comparable stages in their lives, nor is there any adjustment for changing societal standards. In contrast, the findings from our preferred immigrant generation cohort approach correct for both these age and generation mismatches. The results reveal generally greater intergenerational mobility than estimated by the other methods and remain consistent across different racial/ethnic groups.

FINDINGS FROM STATISTICAL MODELS

The foregoing comparison is descriptive and without controls save those of our sample definitions. A series of logistic regression models are estimated according to the specifications detailed above. Five different outcome variables are estimated for four major racial/ethnic groups, with the estimated coefficients reported in Tables 2–6. As explained above, the coefficient for Year in the statistical models indicates the rate of change for the “mainstream” reference group, while the coefficient for Year × Generation indicates the relatively greater change when the second-generation adults in 2005 are compared with the status of the parent generation in 1980. For example, in the case of Hispanic college completion (Table 2), the logistic coefficient for Year is 0.5014, while that for Generation (–1.4598) indicates that Hispanic immigrant parents were greatly disadvantaged in 1980. The subsequent intergenerational mobility measured by Year × Generation is given by the logistic coefficient of 0.8583. This represents the children’s mobility in excess of the changing mainstream standard.

Table 2.

Logistic Regression Results for Educational Attainment, 1980 and 2005

	High School Completion				College Completion
Variable	Hispanic	Non-Hispanic White	Non-Hispanic Black	Non-Hispanic Asian	Hispanic	Non-Hispanic White	Non-Hispanic Black	Non-Hispanic Asian
Intercept	1.7345***	1.7170***	1.7035***	1.7118***	–1.1765***	–1.1869***	–1.1993***	–1.1708***
Year
1980 (ref.)
2005	0.9751***	0.9820***	0.9882***	0.9843***	0.5014***	0.5046***	0.5087***	0.4996***
Generation
Parents of second generation	–2.2623***	–1.1099***	–0.5764***	0.1737***	–1.4598***	–0.3639***	–0.4086***	1.3211***
Year × Generation
2005 × Second generation	1.2313***	1.3653***	0.4430**	0.1789	0.8583***	0.9175***	0.6502***	–0.3235***
Age (age 35 = 0)	–0.0133***	–0.0279***	–0.0437***	–0.0333***	–0.0009	–0.0070***	–0.0137***	0.0026*
Number of Observations	94,371	82,138	61,633	74,513	94,371	82,138	61,633	74,513
–2 Log-Likelihood	91,866	75,004	48,953	56,898	84,600	88,137	68,158	88,102

^*p < .05;

^**p < .01;

^***p < .001

Table 6.

Logistic Regression Results for Being Above Poverty, 1980 and 2005

Variable	Hispanic	Non-Hispanic White	Non-Hispanic Black	Non-Hispanic Asian
Intercept	3.2806***	3.3216***	3.3261***	3.3235***
Year
1980 (ref.)
2005	–0.0626	–0.0738	–0.0662	–0.0680
Generation
Parents of second generation	–1.6378***	–0.6842***	–1.2273***	–0.5067***
Year × Generation
2005 × Second generation	1.4423***	0.8627***	1.2940***	0.8440***
Age (age 35 = 0)	0.0108***	0.0241***	0.0113***	0.0161***
Gender
Male (ref.)
Female	–0.2965***	–0.3732***	–0.4330***	–0.3985***
Marital Status
Married (ref.)
Unmarried	–1.2400***	–1.1649***	–1.1689***	–1.1818***
Number of Observations	93,996	81,763	61,258	74,138
–2 Log-Likelihood	62,749	42,038	30,918	35,812

^***p < .001

Patterns of Intergenerational Mobility

Intergenerational mobility proceeds differently in each case, but a strong overall pattern is shared. To make this evidence more accessible and to facilitate comparison across outcomes and groups, we provide graphic summaries of the expected values from models with significant effects.9 The expected values are computed at the mean value for all variables in the specific race-generation group. A separate subplot of the resulting mean status attainment is presented for each of the four major racial/ethnic groups, and within that subplot is a depiction of changes estimated for each of the five outcome measures (Figure 1). Those changes depict the shift in the reference standard from 1980 to 2005 (grey line), and against that background they display the intergenerational mobility from first to second generation (black line), all controlled to status estimated at age 35. The advantage of this inter generational mobility profile chart is that it not only presents the fitted values for both immigrant generations and the mainstream at two different points in time but also presents the slopes of progress. This format allows for the evaluation of intergenerational progress, not only in absolute terms, but also in relation to societal progress.10

Figure 1.

Intergenerational Mobility Profile Compared With Native-Born Non-Hispanic Whites, 1980 and 2005

Notes: Fitted values were derived from statistically significant coefficients. Nonsignificant coefficients were set to zero. Educational and occupational attainment models control for age. Above poverty status controls for age, sex, and marital status. Homeownership status controls for age, sex, marital status, and area homeownership.

Latinos have experienced substantial mobility across all socioeconomic outcomes from the first to the second generation. They have more than doubled their share completing a high school diploma, but they have not reached the high school completion rate of the mainstream reference group, which has also increased over time. (All other racial/ethnic groups have reached this standard). The same pattern emerges for college completion, but in this case, grown children of the white and black second generation have both exceeded the white third-generation standard. In the case of upper white-collar occupational attainment, there is a sizable gap between Latinos and the reference group that persists in both 1980 and 2005 (see Table 3). Progress was made between the first and second generation that was in addition to the rising societal standard, so the gap with the mainstream closed moderately.11 Latino progress with respect to rising above poverty and entering homeownership is even more substantial because the Latino second generation not only achieved intergenerational mobility but also converged with the mainstream (which fell slightly on those two indicators).

Table 3.

Logistic Regression Results of Occupational Attainment, 1980 and 2005

	Upper White-Collar Occupations
Variable	Hispanic	Non-Hispanic White	Non-Hispanic Black	Non-Hispanic Asian
Intercept	–0.9537***	–0.9566***	–0.9630***	–0.9474***
Year
1980 (ref.)
2005	0.5608***	0.5620***	0.5638***	0.5589***
Generation
Parents of second generation	–1.2971***	–0.2918***	–0.4563***	0.5959***
Year × Generation
2005 × Second generation	0.9203***	0.6298***	0.7302***	–0.0177
Age (age 35 = 0)	0.0226***	0.0207***	0.0165***	0.0270***
Number of Observations	80,728	70,281	54,260	64,943
–2 Log-Likelihood	80,788	81,808	63,875	79,688

^***p < .001

The most noteworthy issue with regard to Asians and Pacific Islanders is their exceptionally high college completion in the first generation, due to the selective nature of Asian immigration prior to 1980. The subsequent college completion rates of their children are only marginally higher, so one could say that there is very little intergenerational progress. In fact, the parents’ educational advantage over the mainstream (coefficient of 1.3211 in Table 2) has been eroded by their children (coefficient of −0.3235). Nonetheless, this loss is relative to a rising mainstream standard that is even higher (coefficient of 0.4996), and so the children’s generation still managed some absolute progress. In the end, the second generation of Asians and Pacific Islanders retain a college completion rate that is nearly twice as high as the mainstream standard (Figure 1).

The high educational attainment of Asians and Pacific Islanders has not been matched by corresponding levels of occupational attainment. In fact, Asian and Pacific Islanders have a lower level of occupational attainment relative to their college completion rate for both the first and second generation, which is opposite the case for native-born whites and the other racial/ethnic groups. The unique case of Asians and Pacific Islanders warrants further investigation into the impact of educational attainment on occupational attainment.

Table 4 presents model results with and without education controls for Asians and Pacific Islanders. Higher education is shown to have the expected strong effect on holding a higher-status occupation. Once educational attainment is controlled in Model 2, the first generation has a lower level of occupational attainment than the mainstream (coefficient of −0.1539), showing that the first generation’s occupational advantage was due to higher education, but at the same time indicating that the first generation did not receive as much occupational benefit from its higher education than did the mainstream. Among their children, however, we find substantial occupational gains, controlled for education (coefficient of 0.2558), relative to the rising occupational achievement in the mainstream. This indicates that the second generation of Asians and Pacific Islanders has more than overcome its parents’ earlier low attainment relative to their education. As a result, even though educational attainment of the children may have fallen off from the previous high standard, they have succeeded in converting that education into occupational rewards much more effectively.

Table 4.

Logistic Regression Results of Asian and Pacific Islander Occupational Attainment, 1980 and 2005

	Upper White-Collar Occupations
Variable	Model 1	Model 2
Intercept	–0.9474***	–2.8067***
Year
1980 (ref.)
2005	0.5589***	0.3699***
Generation
Parents of second generation	0.5959***	–0.1539***
Year × Generation
2005 × Second generation	–0.0177	0.2558***
Age (age 35 = 0)	0.0270***	0.0391***
Education less than high school (ref.)
High school and some college		1.2457***
College degree or more		3.4901***
Number of Observations	64,943	64,943
–2 Log-Likelihood	79,688	63,296

^***p < .001

A final issue of interest with regard to Asians and Pacific Islanders is homeownership attainment. Substantial variation in the prevalence of homeownership across the states could bias findings, especially analysis of changes over time, given that the first generation was more concentrated in the low homeownership states of California and New York (especially true of Asians) and if the second generation dispersed to states with higher prevalence of homeownership. In order to measure more accurate progress in homeownership net of area effects, the location effects of both generations and the reference group must be controlled. The statistical model introduces a control for state area homeownership rates (the prevailing percent of white native-born households in each state that remain as renters). Models with and without this geographic adjustment are reported in Table 5. Results for all groups show that the parent generation had lower homeownership than the mainstream (although this was especially lower in the case of Hispanics and blacks), but over time, the mainstream prevalence of homeownership has risen, and the second-generation children have achieved even greater progress into homeownership than the rising standard. Even though the area prevalence of homeownership has the expected effect, it does not appear to alter any of the observed differentials or trends for the first or second generation among either Asians or other groups.

Table 5.

Logistic Regression Results of Homeownership Attainment, 1980 and 2005

	Hispanic		Non-Hispanic White		Non-Hispanic Black		Non-Hispanic Asian
Variable	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Intercept	1.4023***	2.5533***	1.4162***	2.7276***	1.4558***	2.5935***	1.4395***	2.2881***
Year
1980 (ref.)
2005	0.2640***	0.2271***	0.2595***	0.2179***	0.2773***	0.2388***	0.2720***	0.2423***
Generation
Parents of second generation	–1.5150***	–1.2879***	–0.3220***	–0.1694***	–1.5836***	–1.3544***	–0.6965***	–0.5289***
Year × Generation
2005 × Second generation	0.9618***	0.9792***	0.3625***	0.3968***	0.4916***	0.5203***	0.2514***	0.2972***
Age (age 35 = 0)	0.0943***	0.0942***	0.1020***	0.1034***	0.0960***	0.0977***	0.0978***	0.0988***
Gender
Male (ref.)
Female	0.2041***	0.2023***	0.2262***	0.2254***	0.1651***	0.1661***	0.1933***	0.1960***
Marital Status
Married (ref.)
Unmarried	–1.6903***	–1.6683***	–1.7115***	–1.6983***	–1.8176***	–1.7921***	–1.7880***	–1.7621***
Area Homeownership
Renter rate of non-Hispanic whites aged 25–44 by state		–0.0416***		–0.0472***		–0.0411***		–0.0308***
Number of Observations	86,691	86,691	75,228	75,228	55,035	55,035	67,351	67,351
–2 Log-Likelihood	97,485	96,711	77,317	76,455	56,588	56,136	71,536	71,077

Note: The sample for homeownership is householders and spouses.

^***p < .001

Analysis of white immigrant generations highlights the immigrant effects distinct from race because they share the same race (if not ethnicity) with the white native-born reference group used to proxy the mainstream. Second to Latinos, white immigrants have a very low rate of high school completion in 1980. By 2005, the second-generation whites have converged with the mainstream. In college completion, the second-generation whites have surpassed the educational and occupational attainment of the mainstream. In escape from poverty, the white second generation has converged to the mainstream and has surpassed the mainstream in homeownership attainment.

Finally, the pattern of intergenerational mobility for black immigrant generations is somewhat distinct. Black immigrants came in with a relatively high level of high school completion, and the second generation closed the gap with the mainstream. In college completion, the black second generation has surpassed that of the societal mean. We also see that the second generation has caught up to and surpassed the mainstream’s standard of employment in upper white-collar jobs (Figure 1). Finally, we observe that second-generation blacks have achieved intergenerational mobility relative to their parents’ achievement of homeownership, but they remain with the lowest homeownership rates of all second-generation young adults.

Gender Differences in Intergenerational Mobility

The foregoing analysis has treated the first and second generations irrespective of gender. In status attainments often shared by men and women living together, such as homeownership or poverty, it is difficult to attribute household or family-level attainments to individual characteristics. However, education and occupation are individual attainments that vary substantially between men and women, and as a result, women and men may experience and contribute to the overall intergenerational change in very different ways. The substantial increases in women’s status in recent decades makes it likely that mobility for women is substantially greater than for men, and so the intergenerational progress we have observed could be mainly due to increases in women’s educational attainment and growing participation in paid employment. It bears emphasis that the common approach of focusing on men alone could be especially misleading because that leaves out an entire half of the second generation; indeed, the half that might be experiencing the greatest changes.

For these reasons, it is important to separately examine the intergenerational mobility of men and women with regard to their individual status attainments. We focus here on the likelihood of college completion and of holding an upper white-collar job (i.e., employed in a professional or managerial occupation). Introducing a mainstream reference standard specific to women (and to men) will reflect period changes, and thus a key question is whether immigrant women have kept up with or exceeded these societal changes. A second question is then how great is the difference between the estimated intergenerational mobility of men and women. Does the previously estimated overall intergenerational progress more closely reflect the mobility of women than of men?

The gendered paths of intergenerational mobility are reported in Figure 2, detailing the expected educational and occupational attainment by race and sex. As before, these expected values are computed from mean characteristics of each race-sex group, with age adjustment to 35 and significant coefficients in the estimated models (Table 7). Separate lines for men and women depict the level of attainment and mobility estimated for each racial/ethnic group. For reference, the first group plotted is the native-born white population. The results for high school completion are very similar for both men and women, and so we present results only for college completion. Among the native-born whites in Figure 2 (top graph), women in 1980 had a lower college completion rate than men. However, these women in 2005 had completed college at approximately the same rate as their male counterparts.

Figure 2.

Gender Differences in Intergenerational Mobility in Educational and Occupational Attainment, 1980 and 2005

Notes: Fitted values were derived from statistically significant coefficients. Nonsignificant coefficients were set to zero. Both models control for age.

Table 7.

Logistic Regression Results of Educational and Occupational Attainment, by Gender, 1980 and 2005

	College Completion				Upper White-Collar Occupation
Variable	Hispanic	Non-Hispanic White	Non-Hispanic Black	Non-Hispanic Asian	Hispanic	Non-Hispanic White	Non-Hispanic Black	Non-Hispanic Asian
a. Women
Intercept	–1.4361***	–1.4510***	–1.4659***	–1.4337***	–1.0215***	–1.0218***	–1.0269***	–1.0125***
Year
1980 (ref.)
2005	0.7739***	0.7788***	0.7840***	0.7731***	0.8068***	0.8069***	0.8088***	0.8035***
Generation
Parents of second generation	–1.4047***	–0.4707***	–0.5769***	1.3364***	–1.4163***	–0.4463***	–0.4457***	0.2984***
Year × Generation
2005 × Second generation	0.8142***	1.0667***	0.9471***	–0.3208***	1.0253***	0.7912***	0.5565***	0.1671*
Age (age 35 = 0)	–0.0117***	–0.0194***	–0.0267***	–0.0104***	0.0092***	0.0091***	0.0062**	0.0144***
Number of observations	49,264	43,387	31,618	38,049	37,276	33,095	25,589	30,296
–2 Log-likelihood	40,133	41,557	31,885	42,569	36,596	37,345	29,804	36,436
b. Men
Intercept	–0.9457***	–0.9573***	–0.9650***	–0.9416***	–0.9021***	–0.9075***	–0.9122***	–0.9016***
Year
1980 (ref.)
2005	0.2513***	0.2552***	0.2577***	0.2500***	0.3310***	0.3327***	0.3342***	0.3308***
Generation
Parents of second generation	–1.4921***	–0.1961***	–0.1923***	1.3526***	–1.2217***	–0.1634***	–0.4592***	0.8457***
Year × Generation
2005 × Second generation	0.8673***	0.7142***	0.1900	–0.3705***	0.8360***	0.5230***	0.8509***	–0.1481*
Age (age 35 = 0)	0.0080***	0.0008	–0.0036	0.0105***	0.0331***	0.0288***	0.0253***	0.0334***
Number of observations	45,107	38,751	30,015	36,464	43,452	37,186	28,671	34,647
–2 Log-likelihood	43,902	45,587	35,660	44,799	43,974	44,193	33,963	42,805

^*p < .05;

^**p < .01;

^***p < .001

In general, women exhibit steeper advancement toward college completion than do the men in the same racial/ethnic group. Against the background of rising “mainstream” status (coefficient of 0.7739), the estimated coefficients for Year × Generation show very substantial added mobility for Latinas (0.8142), whites (1.0667), and blacks (0.9471). Among Asian and Pacific Islanders, however, the second generation experienced less progress than the reference group (–0.3208). However, women of the first generation in that group held college completion rates far above the mainstream in 1980 (coefficient of 1.3364), and the second-generation women still attained college completion at higher rates than any other group. In contrast, Latinas held the greatest disadvantage in 1980 and also achieved the least progress (other than Asians), leaving them with the lowest college attainment in 2005.

The pattern of increasing college attainment among men is generally much weaker than for women. Increase is less in the reference group (0.2513 in Table 7) and in each of the immigrant groups save Latino. The black second-generation men did not achieve any significant progress over their fathers’ generation, contrasting sharply with the progress of women. Further, men among the Asian and Pacific Islander second generation are the only ones among all racial/ethnic groups to actually achieve a lower rate of college completion than the first generation. Their slippage relative to the parents’ generation (coefficient of −0.3705) exceeds the background increase of the mainstream reference group. Thus, the estimated absolute level of college completion among men actually declines, even if it still remains above all other groups.

Turning to the achievement of upper white-collar occupation, the considerable educational gains of second-generation women may not necessarily translate into gains in occupational attainment, which would contradict some recent studies in this area (Feliciano and Rumbaut 2005). For the “mainstream” in 1980, young full-time working women and men had approximately the same share in upper white-collar occupations. By 2005, occupational attainment substantially increased for young women, with 44.4% in high occupations compared with 35.3% for their male counterparts. This result is consistent with the overall U.S. occupational distribution by sex (Fronczek and Johnson 2003) in that men are more evenly distributed across different occupation categories while women are more concentrated in management/professional or sales/office occupations. That pattern is accentuated for women with a college education.

Within this context, intergenerational mobility in occupational attainment is not the same across all racial/ethnic groups. Of Latina second-generation workers, 34.5% are in upper white-collar occupations, which is much higher than expected given their educational attainment (22.6% college completion). Asian and Pacific Islander second-generation women have slightly lower occupational attainment (55.1%) than they do in college completion (59.2%). But because their immigrant mothers had a much lower share in high occupations, the second-generation women’s intergenerational progress in occupations is greater than what was observed in educational attainment. The occupational attainment of white women and men looks similar to the patterns of the “mainstream,” except that the first generation started off lower and the second generation attained higher. On the other hand, black women and men are almost identical in occupational attainment, which is unexpected given the findings for educational attainment. Although the educational attainment of all black second-generation men is lower than it is for black women, black full-time working men have achieved parity with their female counterparts.

DISCUSSION AND CONCLUSION

In this article, we put forth a model for estimating immigrant intergenerational mobility that adheres to more detailed and rigorous temporal criteria in order to better assess the socioeconomic mobility experienced by today’s second generation. We offer a method that not only measures mobility across actual immigrant parents of the new second generation but also measures it in comparison with the mobility of the mainstream at comparable ages.

Our proposed method allows for more specific measurements of change. However, it is not insulated from certain limitations that warrant discussion. First, our definition of the first generation has been refined to include only actual immigrant parents of the second generation, but we cannot trace actual kin. This means that we cannot analyze the relationship between specific parental attributes and outcomes for children. Our analysis is limited to comparisons of average differences between generations. Second, our method requires large-scale data from both the current period and a period some 25 or 30 years earlier. Effectively, this limits us to using census data from 1980, and accordingly, we lack variables that measure attitudes, or cultural and political behavior. The 1980 data set also lacks questions on parental nativity that would allow us to more narrowly define a third- or higher generation reference group. Lastly, for this article, we track intergenerational mobility for broad racial/ethnic groups without regard to specific ethnic differences within each racial group.

The findings from this immigrant generation cohort method inform the literature on the new second generation in several ways. As a first observation, we find evidence of greater intergenerational progress for the new second generation than is commonly reported in the literature (Farley and Alba 2002). This is especially important for outcomes that are expected to rise over the life cycle, such as economic well-being and homeownership. A second contribution is that the use of multiple outcome indicators reveals that not all aspects of second-generation socioeconomic status rose at the same rate. This is an indication that reliance on a single indicator to infer overall mobility can be misleading. Our intergenerational mobility profile chart enables clear visual comparisons across indicators and groups of expected values from significant effects.

Third, we find that the Latino and black second generations have attained higher socioeconomic status than their immigrant parents, a finding that challenges the “second-generation decline” hypothesis. However, our multipart measurement of progress over time unveils some deeper insights into the meaning of progress. We find that the second generation, particularly in the case of Latinos, has not reached parity with the mainstream, which is an alternate basis for the decline hypothesis. Separately, the Asian second-generation has higher educational and occupational attainment than the mainstream, a fact that is used to perpetuate the “model minority” hypothesis. At the same time, however, second-generation Asians have lower status attainments than their immigrant parents on several outcomes. Therefore, should we conclude that this indicates second-generation decline for Asians?

These observations seemingly support theories that are diametrically opposed to one another. The critical question that must be explicitly addressed is, To whom is the second generation being compared? Indeed, our fourth contribution is a method of tracking both immigrant generations and the mainstream that enables multiple ways of measuring mobility, whether conceived as the amount of progress from immigrant parents to the second generation, or as the amount of intergenerational progress relative to mainstream progress, or as the changing difference between immigrant generations and the mainstream. Our method of multiple measurements of change on multiple indicators for different racial/ethnic groups forces attention to the competing emphases often left unclarified in the literature. The literature about the socioeconomic mobility of the new second generation would be improved by making explicit which measurement of change informs which particular theory of immigrant assimilation.

A final contribution and set of findings concern gender, a topic that is frequently ignored in the measurement of intergenerational mobility (Feliciano and Rumbaut 2005). A prime example of a rapidly changing aspect of the U.S. context is in the socioeconomic attainments of women in the latter part of the twentieth century. It is difficult to assess intergenerational mobility and immigrant assimilation when the societal standards are so rapidly rising. However, focusing only on men’s mobility risks error of neglecting a substantial portion of the immigrant experience. Our analysis reveals that women experienced substantially more intergenerational mobility than their male counterparts. The attainments of second-generation women have all but closed the gender gap once observed between mothers and fathers in the first generation. In fact, much of the intergenerational mobility observed overall can be attributed to the vast socioeconomic progress of women. Scholars underestimate intergenerational progress when they leave out those who have experienced the most mobility.

As the new second generation of the post-1965 immigration enters into adulthood, we are only beginning to observe what their socioeconomic attainments are. The greater diversity of this era of immigration, and the greater prominence of change for women alongside men, calls for more refined methods for measuring mobility. We hope that the insights yielded from the proposed immigrant generation cohort method will contribute to continued evolution of the literature on immigrant assimilation.