Migration Processes, Familial Characteristics, and Schooling Dropout Among Black Youths

Abstract

This study uses data from the 2000 U.S. census to examine whether the schooling advantage of black immigrants' children found in previous studies is robust. According to the results, the advantage associated with having migrant parents is not restricted to the children of immigrants. Black migrant parents, regardless of foreign-born status, have children with favorable schooling outcomes. Such parental-level influences, however, seem stronger among some immigrant groups than among native internal migrants. The study also suggests that the collective advantage of the children of immigrants is driven by positive migrant selectivity. Accordingly, comparisons between the children of native migrants and children in various immigrant groups reveal that the immigrant advantage is not robust. In fact, the results suggest that when immigrant ethnicity is considered, some children of immigrants may be disadvantaged relative to the children of native migrants. Among recent migrants, the children of native internal migrants also have more favorable outcomes than the children of immigrants, although these differences disappear after background factors are controlled. Further, internal-migrant and immigrant households are less likely to have characteristics that adversely affect schooling than nonmigrant households. Unsurprisingly, the children of non-migrants have the worst outcomes among black youths.

Introduction

The diversification of U.S. immigration flows has resulted in significant increases in the arrival of black foreign-born populations. Associated with these increases are analogous transformations in the ethnic composition of the black foreign-born population. Thus, while the traditional role of the Caribbean in black immigration is now well recognized (Johnson 2008; Rogers 2006), evidence points to an increasing significance of black immigration from Africa and other world regions (Arthur 2000; Djamba 1999; Hagopian et al. 2004; Kent 2007). Not surprisingly, the implications of these trends have received increased scholarly attention in recent studies (e.g., Gilkes 2007; Henke 2001; Logan and Deane 2003; Ogbaa 2003). As a result, despite their decades-old absence from the demographic literature, the children of black immigrants are gradually becoming more visible in contemporary research (e.g., Massey et al. 2007; Rong and Brown 2001; Stepick et al. 2001; Waters 1994).

One consequence of the increased focus on black immigrants' children is the expansion of studies on their educational outcomes. In general, most of these studies indicate that the children of black immigrants have higher levels of educational attainment than the children of U.S.-born blacks. For example, the children of immigrants have higher levels of postsecondary enrollment (Quigley 1996) and slightly higher GPAs (Kao 2004; Massey et al. 2007) than their counterparts with native black parents. Similarly, some studies on specific black-immigrant groups indicate that educational attainment is higher among second-generation African and Caribbean immigrant children than among the children of U.S.-born blacks (Rong and Brown 2001). On the whole, the black immigrant advantage in schooling is consistent with other studies showing that black immigrants outperform U.S.-born blacks in a number of other indicators (Djamba 1999; Dodoo 1997; Kalmijn 1996; Read and Emerson 2005). This, in the view of some scholars, indicates that black immigrants and their children have become America's new “model minority” group (Freeman 2002; Page 2007).

Despite progress in the study of black immigrants' children, a number of issues regarding their relative schooling success are insufficiently interrogated in the existing literature. The first concerns the role of migrant selectivity in explaining the comparative schooling advantage of black immigrants. Specifically, a number of commentators have used cultural frameworks, such as Ogbu's oppositional culture hypothesis (Ogbu 1992, 1993), to explain the advantage of black immigrants relative to U.S.-born blacks. One variant of this perspective suggests that black immigrants possess unique cultural attributes, which U.S.-born blacks need to cultivate, that allow them to succeed in U.S. schools (Rimer and Arenson 2004). Such perspectives, however, conflate the impacts of migration as a selectivity-inducing process and the apparent influence of immigrants' cultural attributes. Besides, the role of cultural factors in mediating schooling disparities among blacks has been systematically interrogated elsewhere (e.g., Butterfield 2006; Harris 2006; Lundy 2003; Lundy and Firebaugh 2005). Cultural explanations also have other implications for understanding nativity differences in schooling that are unresolved in the existing literature. One of these explanations is that the nativity differences imply that the immigrant advantage will persist even in comparisons between the children of black immigrants and the children of U.S.-born black internal migrants. However, the available evidence suggests that this may not be the case. In Model's (2008) study, for example, the black immigrant economic advantage is virtually eliminated in comparisons between black West Indian adults and their native black interstate migrant counterparts. Butcher (1994) also maintained that black West Indian immigrants and native-born black internal migrants have similar labor force outcomes. More generally, apart from a few older studies (e.g., Long 1975), research on the schooling outcomes of black internal migrants and their children is generally limited.

Another implication of the model-minority hypothesis, as applied to black immigrants, is that it suggests that the collective advantage of black immigrants will persist across all black immigrant groups. Thus, even though black immigrants vary in their human-capital endowments (Kent 2007), previous studies have not examined whether the immigrant advantage will disappear if parental schooling, for example, is lower among black immigrant families than among U.S.-born black internal-migrant families.

In this study, therefore, clarity is brought to the understanding of schooling differences among black youths by investigating the roles of immigration and internal migration in mediating differences in black schooling. A key objective of the study is to underscore the multidimensional significance of within-race heterogeneity for the analysis of educational attainment among black youths. Using schooling dropout as its outcome of interest, the study extends previous research in three specific ways. First, it expands our understanding of how migration processes mediate dropout disparities among black youths by comparing the outcomes of children in black immigrant and native black internal-migrant families. Second, given human-capital variations among black immigrant groups, and the association between parental schooling and children's educational attainment, it demonstrates that the black immigrant advantage is conditional on levels of parental education. The study, therefore, argues that by focusing on the collective advantage of black immigrants' children, the relative disadvantage of children from specific black immigrant groups is ignored while the advantage of native black internal migrants relative to selected black immigrant groups is discounted. A third contribution of the study is that it examines the role of migration, as a familial transformation process, on contextual influences associated with dropout differences among black youths. Drawing from research on the association between family contexts and schooling (Kohl et al. 2000; Tillman 2008a; Wojtkiewicz and Donato 1995) and Wilson's (2001) hypothesis that black migration leads to improvements in familial contexts, the implications of differences in the characteristics in migrant (i.e., immigrant and internal migrants) and non-migrant families for dropout disparities among black youths are also examined.

Conceptual Issues

Migration theorists have long recognized the fact that major differences exist in the human-capital endowments of migrants and nonmigrants. Lee's (1966) work, for example, argued that given high levels of attraction to potential destination areas, individuals choosing to migrate will be positively selected on a number of human-capital attributes. In other words, migration processes attract the most able, motivated, and skilled individuals, while those less endowed with these attributes are less likely to migrate. Migrants are thus distinct from nonmigrants in terms of both observable (e.g., educational attainment) and unobservable (e.g., motivation) attributes (Bloom et al. 1994; Feliciano 2005a; Nakosteen et al. 2008). Chiswick (1999), however, suggested that in rare circumstances, migrants can be negatively selected or have worse socioeconomic indicators relative to nonmigrants.

In general, internal migration shares some similarities with international migration, although both processes sometimes differ in scale, spatial distance, and other dimensions (King et al. 2008). Internal migration, for example, involves changes in residence across subnational boundaries (e.g., state or regional boundaries), while international migration involves similar changes on an international scale. Earlier theorists, such as Ravenstein (1885) and Lee (1966), also developed frameworks on the determinants of migration, now utilized in international migration research, using insights on internal-migration processes. Yet, an essential feature of both internal- and international-migration processes is that they involve movements from less to more economically attractive destinations. Migrants in both processes, therefore, tend to be positively selected compared with nonmigrants (Borjas et al. 1992; Gabriel and Schmitz 1995).

Positive selection among international migrants may, therefore, undergird most schooling comparisons between immigrant and native populations. In fact, Chiswick and DebBurman (2004) argued that the children of immigrants have higher levels of schooling than children in U.S.-born families mainly due to the influence of positive migrant selectivity among immigrant parents. Immigrants' children, therefore, do well in school because their parents have higher average schooling levels than nonimmigrant parents (Feliciano 2005a) and because immigrant parents exert a significantly high level of influence on the education of their children (Kao 2004).

Among native-born populations, migrant selectivity also mediates disparities in human-capital indictors among internal migrants and nonmigrants. Accordingly, comparisons between U.S.-born internal migrants and nonmigrants have shown higher levels of socioeconomic attainment among the former relative to the latter. Research on the Great Migration, for example, indicates that U.S.-born black internal migrants were more highly educated than their nonmigrant counterparts (Margo 1988; Tolnay 1998, 2003). According to Vigdor (2002), this difference was due to the fact that educated blacks were more able to respond to the declining economic conditions in the South because they were better able to afford the financial costs of their journey. Analysis of more recent migration trends also shows that contemporary black internal migration is driven by the mobility patterns of the highly educated (Lee and Roseman 1997). Therefore, education disparities among native blacks, conditional on migration status, have two implications for the schooling outcomes of native black children that are largely discounted in previous studies. First, positive educational selectivity among adult black internal migrants is likely to result in disparities in parental schooling in internal migrant and nonmigrant families. Second, like immigrant parents, internal-migrant parents may also be positively selected for other parental-level influences that may mediate schooling disparities among the children of U.S.-born internal migrants and nonmigrants.

Few studies have accounted for selectivity-related issues in research comparing the outcomes of immigrant and U.S.-born blacks. When these issues have been accounted for by comparing adult black immigrants with their black internal-migrant counterparts, both groups have been found to have similar socioeconomic outcomes (Butcher 1994; Model 2008). Despite its heuristic appeal, however, this methodological perspective has not been incorporated in research on educational disparities among black youths. Still, migration theory would suggest that selectivity generally affects schooling comparisons among migrant (i.e., immigrants and internal migrants) and nonmigrant families in two important ways. First, because positive selectivity is greater among migrants traveling longer than shorter distances (Feliciano 2005b), parental schooling may be higher in immigrant families than in native internal-migrant families. Second, because nonmigrants are (by definition) not involved in the migration process, their respective outcomes would be worse than those found in either immigrant or native-born internal-migrant families.

Differences in parental schooling among black youths are, therefore, conceptualized as key mediators of disparities in their dropout status. Consistent with previous studies (e.g., Black et al. 2005; Driscoll 1999; Haveman et al. 1991), children are expected to have better schooling outcomes—that is, to be less likely to drop out of school—in families with higher parental schooling levels. Differential schooling attainment across black immigrant groups (Kent 2007), however, has additional implications for dropout disparities among black children. African and European black immigrants, for example, generally have higher levels schooling than other black immigrants (Kaba 2007; Kent 2007; Read and Emerson 2005), while the evidence suggests that the attainment levels of Hispanic black immigrants are among the lowest in the United States (Hirschman 2001). Conceptually, therefore, comparisons across black immigrant groups should reveal more favorable outcomes among children in African than, for example, West Indian Hispanic black immigrant families. However, despite higher dropout rates among Hispanic than non-Hispanic children in the United States (Fry 2003), little is known about relative dropout differences among children in Hispanic and non-Hispanic black immigrant families. For immigrant parents and their children, differences in another human-capital attribute—language proficiency—are also likely to mediate children's schooling disparities. In other words, low levels of English proficiency among Hispanic immigrants (Rumbaut 1995) may also mediate relative schooling differences among the children of Hispanic and non-Hispanic black immigrants.

Previous research on black migrant families is also silent on the comparative impacts of parental dropout status on subsequent dropout among the children of migrants and nonmigrants. However, a number of studies have shown that children with dropout parents are more likely than those without to drop out of school (Perreira et al. 2006; Wojtkiewicz and Donato 1995). Because higher parental education in migrant than nonmigrant families implies that the children of migrants are less likely than those of nonmigrants to have parents who dropped out of school, the negative impact of parental dropout is likely to be greater among the former than the latter.

Within families, a variety of sibling characteristics are also known to significantly affect children's schooling outcomes. Living with stepsiblings or half-siblings, for example, negatively affects schooling performance (Tillman 2008a), and increasing coresidence with such siblings accentuates this negative effect across time (Tillman 2008b). Other studies, such as Peraita and Pastor (2000), indicated that schooling dropout is negatively associated with having younger than older siblings. More generally, prior research (e.g., Slomkowski et al. 2001; Whiteman et al. 2007; Whiteman and Christiansen 2008) showed that children's sociodemographic outcomes are directly affected by sibling modeling influences that can have positive effects on schooling (Smith 1990). Thus, higher schooling attainment among the children of immigrants relative to the children of the U.S.-born suggests that the former are more likely to have dropout siblings than the latter.

Assimilation theory provides a final conceptual perspective useful for understanding the schooling outcomes of immigrant's children. Although conventional assimilation suggests that immigrants' schooling will improve as they assimilate, segmented assimilation theory suggests differential outcomes among the children of immigrants as they assimilate, conditional on (among other things) their class or their modes of incorporation (Bankston and Zhou 1997; Portes and Zhou 1993; Zhou 1997). Accordingly, with increasing generational status, middle- and upper-class immigrants—with high levels of human capital—are expected to have more favorable outcomes than their lower-class counterparts as they assimilate. Segmented assimilation theory, therefore, implies that any collective immigrant advantage in schooling dropout will be driven by the outcomes of groups with higher than lower levels of human capital.

Hypotheses

Four hypotheses are, therefore, examined in the study's empirical analysis. The first is that the children of black immigrants will collectively be less likely to drop out of school than the children of U.S.-born blacks. However, because immigrant parents are positively selected on various unobservable characteristics (e.g., motivation), the collective advantage of their children will persist even in seemingly disadvantaged familial contexts—for example, in single-parent families. Second, the study hypothesizes that among U.S.-born blacks, children with internal-migrant parents will be less likely to drop out of school than children with nonmigrant parents. In other words, having migrant instead of nonmigrant parents will result in a lower probability of schooling drop out, regardless of parental foreign-born status. Third, because immigration is associated with higher levels of migrant selectivity than internal migration, the children of immigrants are hypothesized to have a collective dropout advantage over the children of native black internal migrants. However, the fourth hypothesis posits that this collective black immigrant advantage relative to native internal migrants will be driven by the outcomes of immigrants from highly selective immigrant groups. Specifically, the black immigrant advantage is expected to driven by the outcomes of the children of African immigrants, who, the evidence suggests, are the most highly selective black immigrant group (Dodoo 1997). Africans are expected to be followed by European blacks because studies indicate that the latter may also be a highly selective group, given their generally high levels of schooling (Read and Emerson 2005). The children of West Indian non-Hispanics are expected to outperform the children of native-black migrants and nonmigrants because although West Indians are less selected for schooling, they are generally highly selected for unobservable factors, such as parental ambition and determination (Model 2008). Finally, given low schooling levels of West Indian Hispanic blacks, their children are expected to do less well than the children of other black immigrants and the children of U.S.-born black internal migrants.

Data and Methods

Data used in the study are taken from the noninstitutionalized population found in the 5% sample of 2000 U.S. census. These data are available in the Integrated Public Use Microdata Samples database of the Minnesota Population Center (Ruggles et al. 2009) and include individual-level information on basic demographic indicators (e.g., age and sex); socioeconomic characteristics (e.g., educational attainment); and migration-related attributes. In addition, unique household identification numbers make it possible to identify all individuals living within the same household. Using these numbers, along with information on individuals' relationship with their household heads, data on parental-level attributes (e.g., parent's education and migration status) are linked with individual-level data for all children living within a household.

Children of black immigrants are identified as black children with a black immigrant household head or spouse.¹ Immigrant household heads and spouses are identified by using information on the respondent's foreign-born status (i.e., whether born in a country other than the United States) collected for all individuals during the census. This information is also used to assess the validity of the selectivity hypothesis across black immigrant groups by distinguishing between the children of black immigrants from six mutually exclusive black immigrant family types: African, West Indian non-Hispanic, West Indian Hispanic (i.e., West Indian immigrants of Hispanic origin), South American, European, and a residual category referred to as “Other” black immigrants. Because Guyana and Belize are usually classified among the countries of the West Indies (see, e.g., Lewis and Knight 2004), they are not classified as South Americans but as West Indians. In addition, children with both a household head and a spouse of a household head from different pan-national groups are classified according to pan-national origins of the former. In general, despite linguistic differences among these groups, more-selective groups (such as Africans) are still expected to have the lowest probability of schooling dropout after language differences are controlled for. These differences are accounted for using a measure of language proficiency (i.e., whether children and parents are English proficient) constructed by using census indicators of whether individuals speak English very well or speak only English.

For all children of immigrants, assimilation effects are inferred from the observed dropout disparities between first- and second-generation children. These disparities represent tentative summaries of the expected changes in dropout status with increasing assimilation because the census data do not capture differences in, for example, modes of incorporation. Generational status is thus determined by using child-level information on place of birth. Accordingly, first-generation children are defined as the foreign-born children of immigrants, while second-generation children are immigrants' U.S.-born children.

Unlike the children of immigrants, the children of U.S.-born blacks have only black parents who were born in the United States. Among this group, the children of internal migrants are defined as children living in households with either an internal-migrant household head or an internal-migrant spouse of the household head. However, in the census data, internal migrants can be identified in two ways. Lifetime internal migrants can be identified as individuals whose current state of residence differs from their state of birth; the census also provides specific information on individuals who migrated between states within the past five years. Information from these two sources is, therefore, exploited to generate two categories of internal migrants: (1) recent internal migrants, or internal migrants who migrated within the past five years; and (2) long-term internal migrants, or lifetime internal migrants who did not migrate during this same period. These two categories are mutually exclusive because lifetime migrants who migrated in the past five years are captured by the indicator of recent internal-migrant status. The recent and long-term migrant statuses of internal-migrant families are then determined according to the timing of internal migration among household heads and their spouses. Families with mixed duration-specific migration statuses among heads and spouses are classified by using the duration-specific migration status of the household head. Distinguishing between these two groups also makes it possible to compare internal migrants with their recent and long-term immigrant counterparts (i.e., immigrant families with parents who arrived in the United States in the past five years, and six or more years ago, respectively). In doing so, the analysis is able to capture whether differences in schooling between the children of internal migrants and immigrants are also driven by differences in parental post-migration duration of residence.

Given the growing interest in the adjustment of immigrant adolescents (Fry 2003; Fuligni and Witkow 2004; Gordon-Larsen et al. 2003), the analysis focuses on dropout status among all black youths between ages 13 and 19.² Among this group, children who dropped out of school are defined as children who completed at least one year of schooling but did not graduate from high school and are not currently enrolled in school (i.e., between the first and twelfth grades). Similar measures of status dropout, or schooling dropout based on current-status data, have been employed in a number of studies using cross-sectional data (e.g., Fry 2007; Lillard and DeCicca 2001; Marschall 2006).

Table A1 in Online Resource 1 presents percentage distributions of the highest completed grades for dropouts and non-dropouts in the sample. Overall, the dropout rate in the sample is about 6.2%. Other sources, such as Laird et al. (2006), reported slightly higher dropout rates for black youths. This difference is the result of two factors. The first is that because this study focuses on the noninstitutional population, black teenagers who are institutional inmates are excluded from the sample. According to the 2000 census data, for example, the dropout rate among children in this population (a little more than 2% of all black teenagers), is upwards of 35%.³ Secondly, most studies reporting higher black dropout rates focus on individuals in much older ages (e.g., between 18 and 25).⁴

The study's analytical strategy also addresses a key methodological issue essential for understanding the dynamics of immigrant schooling disparities. In other words, schooling dropout among first-generation children may be influenced by transitions between foreign and U.S. educational systems (Fry 2003). In this study, this issue is addressed using a proxy indicator—arrival in the United States at age 7 and older—to account for dropout disparities between the children most likely to have started their schooling careers abroad, (i.e., those who arrived at age 7 and older) and those who started schooling in the United States.

Empirically, the analysis employs logistic regression models to examine the determinants of schooling dropout among black youth. The basic form of the estimation equation is as follows:

$$D=-\mathrm{\alpha }+{\mathrm{\beta }}_a\mathit{\text{FM}}+{\mathrm{\beta }}_b\mathit{\text{SA}}+{\mathrm{\beta }}_c\mathrm{C}+{\mathrm{\beta }}_{d}H+\varepsilon$$

D, the dependent variable, is an indicator of whether a child dropped out of school. This measure is a function of family migration status (FM); whether the child had first school enrollment abroad (SA); a vector of other child-level covariates (C); and a number of familial-level influences (H), including sibling characteristics,⁵ family structure, and parental schooling. Because children are clustered within households, however, standard error estimates for the right-side variables are likely to be underestimated. Thus, the regression models are estimated by using robust standard errors that adjust for clustering.

In the empirical analysis, the process of interrogating the black immigrant advantage proceeds in four stages. In the first stage, all children of U.S.-born blacks are compared with the children of black immigrants. To examine whether the immigrant advantage is robust within seemingly adverse familial contexts (i.e., single-parent families, or families with dropout parents), interaction terms are used to examine whether factors such as family structure and parental dropout status have differential influences among children in black U.S.-born and immigrant families. In the second stage, children with U.S.-born black internal-migrant parents are compared with those with nonmigrant parents to determine whether parental internal-migration status has a similar association with children's schooling as parental immigration status. In short, we examine whether the schooling dropout “advantage” associated with having migrant parents is restricted to the children of immigrants. The robustness of the collective immigrant advantage to comparisons with U.S.-born internal migrants and nonmigrants is examined in the third stage. The final stage of the analysis then examines whether the presumed black immigrant advantage is robust to differences in black immigrant panethnic origins.

Results

Descriptive Findings

Summary characteristics of the children in the sample are presented in Table 1. Revealed in these findings are significant variations in familial characteristics that are conditional on parental migration status. Consistent with Wilson's (2001) hypothesis, Table 1 indicates that black children with migrant (i.e., immigrant or internal-migrant) parents are generally less likely than those with nonmigrant parents to live in families with characteristics known to adversely affect schooling outcomes. Thus, while factors such as living in single-parent families and having dropout siblings are usually associated with worse schooling outcomes (Astone and McLanahan 1994; Kohl et al. 2000), Table 1 suggests that they may be a lesser constraint to schooling among children with migrant than among those with non-migrant parents. In addition, the children of immigrants are the least likely to have older siblings or any sibling who dropped out of school. Among native blacks, similar patterns are observed in comparisons between children in internal-migrant and nonmigrant households. Parental migration status, however, has no consistent association with whether children live with nonbiological siblings. As such, although the children of immigrants are collectively less likely to have nonbiological siblings than the children of U.S.-born blacks, the children of native internal migrants appear more likely to live with such siblings than the children of nonmigrants.

Table 1. Percentage distribution of children in the sample.

	Children of Immigrants	Children of U.S.-born Blacks
	All	All	Internal Migrants	Nonmigrants
Single-Parent Families	42.7*	61.2	50.5*	67.2*
Dropped Out of School	4.1*	6.4	5.3*	7.1*
Lives With Nonsiblings	34.7*	40.7	43.1*	39.4
Oldest Sibling Dropped Out of School	3.4*	4.7	3.9*	5.2*
Any Sibling Dropped Out of School	5.3*	6.7	5.7*	7.3*
Highest Parental Schooling
Less than completed high school	22.3	23.0	18.1*	25.8*
High school graduate	23.2*	31.4	26.2*	34.4*
Some college education	29.7*	32.8	37.8*	30.0*
College graduate	24.7*	12.7	17.9*	9.8*
Parental Dropout Status
One dropout parent	23.4*	25.3	22.0*	27.2*
Two dropout parents	6.5*	3.6	3.6	3.6
N	17,841	160,274	57,856	102,418

Source: 5% sample of the 2000 U.S. census.

^*Significantly different from all children of U.S.-born blacks at the p < .05 level

Other differences in Table 1 highlight the association between family migration status and selected parental human-capital indicators. In terms of highest level of parental schooling, or the highest schooling level found among parents within households, Table 1 shows that the children of internal migrants have parental post-secondary schooling levels that exceed the averages for native blacks. Immigrant households, however, have the highest percentage of children with parents who graduated from four-year colleges. In fact, the prevalence of college-graduate parents among immigrants exceeds that among all native black families by a factor of about 2.

These disparities, however, mask considerable variation in parental schooling across black immigrant groups. These variations are presented in Table A2 (see Online Resource 1), which shows, for example, that the highest levels of parental schooling are found among African immigrants and the lowest levels are found among West Indian Hispanic blacks. Strikingly, Table A2 also indicates that when compared with children from selected black immigrant groups, the children of U.S.-born internal migrants are more likely to have parents with postsecondary education qualifications than other children of immigrants, except those in African and European immigrant households. Furthermore, as expected, the least favorable parental-schooling profiles are found among the children of black native nonmigrants.

There are also significant variations in schooling dropout among children in the sample. For example, Table 1 indicates that about 6.4% of all children with U.S.-born parents had dropped out of school; the respective figure for the children of immigrants is 4.1%. Furthermore, in line with the selectivity hypothesis, children in native black nonmigrant households have higher dropout rates than their counterparts in internal-migrant households. Variations in dropout rates are also evident among the children of immigrants from various pan-national groups. Among these groups, Table A2 (Online Resource 1) shows that dropout rates are lowest among Africans (3.0%) and West Indian non-Hispanics (3.8%), but are considerably higher among “other” blacks (7.8%) and West Indian Hispanics (6.9%). Moreover, only children in African and West Indian non-Hispanic households have lower rates of dropout than the children of native black internal migrants that are statistically significant.

Surprisingly, however, parental schooling dropout is generally higher among children with immigrant than U.S.-born black parents. In addition, across family type, percentages of children with one dropout parent exceed the respective percentages of children whose highest parental schooling level is less than completed high school. However, these differences are not in themselves contradictory.⁶ On the whole, these summary findings are consistent with the notion that the greater positive selection in international than in internal-migration process may also be associated with greater levels of familial transformation among immigrants than internal migrants.

Multiple Regression Results

Regression results from the analysis of dropout disparities, conditional on parental immigration status, are presented in Table 2. Model 1 provides baseline estimates that reveal a significant collective advantage among all children of black immigrants relative to children in U.S.-born black families even after demographic factors are controlled. Further clarity on this collective advantage of immigrants is provided in Model 2 by accounting for whether children started their schooling careers abroad. When this difference is accounted for, the analysis suggest that the size of the immigrant advantage shown in Model 1 may be constricted by higher levels of schooling dropout among children with a foreign rather than a U.S. schooling debut. As a result, the immigrant advantage is accentuated further after this factor is controlled.

Table 2. Logistic regression coefficients estimating the likelihood of dropping out of school among the children of black immigrants and U.S.-born blacks.

	Model 1	Model 2	Model 3	Model 4
All Children of Black Immigrants	−0.53***	−0.86***	−0.71***	−0.47**
All Children of U.S.-born Blacks (ref.)	(0.00)	(0.00)	(0.00)	(0.00)
Age	0.70***	0.70***	0.71***	0.70***
Male	0.25***	0.25***	0.31***	−0.31***
Started Schooling Abroad (yes 0 1, no 0 0)		0.38**	0.30*	0.31*
Single-Parent Household			0.47***	0.50***
Immigrant × Single-parent household				−0.23*
Family Size			−0.03*	−0.03*
Household Income (ln)			−0.06***	−0.06***
Highest Parental Schooling Level
Less than completed high school			0.70***	0.66***
High school graduate			0.58***	0.59***
Some college education			0.33***	0.35***
College graduate (ref.)			(0.00)	(0.00)
One Parent Dropped Out of School			0.80***	0.85***
Immigrant × One parent dropped out				−0.48***
Both Parents Dropped Out of School			0.91***	0.97***
Immigrant × Both parents dropped out				−0.43**
Household Head is English Proficient			−0.25**	−0.27***
Sibling Characteristics
Total nonbiological siblings			0.07***	0.07
Immigrant × Nonbiological sibling				0.04
Oldest sibling dropped out of school			−0.26***	−0.25***
Immigrant × Oldest sibling dropout				0.00
Total dropout siblings			1.06***	1.03***
Immigrant × Total dropout siblings				0.40*
Constant	−14.0***	−14.0***	−15.5***	−14.5***
N	178,115	178,115	178,115	178,115
Log-Likelihood	−34,496	−34,492	−31,539	−31,513

Source: 5% sample of the 2000 U.S. census.

^*p < .05;

^**p < .01;

^***p < .001

Model 3 shows that the black immigrant advantage persists even after familial characteristics are controlled, effectively discounting the possibility that the immigrant advantage is entirely explained by the more favorable structural characteristics of their families. Model 3 also reveals other noteworthy associations between familial characteristics and schooling dropout. A number of these associations, such as the positive association between income and schooling dropout, are consistent with findings reported in previous studies. Still, Model 3 also shows a positive association between schooling dropout and the number of dropout parents living within households. This association seems explained neither by background economic factors nor by the impacts of parental dropout status on parental schooling achievement. As such, the positive association number of dropout parents and dropout among their children persists even after income and parental schooling are controlled. As expected, the likelihood of schooling dropout also increases as the number of dropout siblings increases. However, after other factors are controlled, dropout risks are generally lower among children whose oldest siblings dropped out of school than among other children.⁷

Interaction terms in Model 4 reveal findings that are consistent with the hypothesis that the relative advantage of immigrants persists within disadvantaged familial contexts. Living in single-parent families, for example, is associated with lower dropout risks among children with immigrant than U.S.-born parents. Likewise, although the children of immigrants are more likely than the children of U.S.-born blacks to have two dropout parents (Table 1), having such parents is still associated with lower dropout risks among the former than the latter.⁸ Surprisingly, however, having more dropout siblings is associated with greater increases in the likelihood of schooling dropout among children with immigrant parents than among those with native-born parents. One implication of this finding is that sibling and parental dropout status may have contrasting influences on the likelihood of schooling dropout among children living in immigrant families.

In Table 3, the analysis examines whether the advantage associated with having migrant than nonmigrant parents can also be found among U.S.-born black families. Accordingly, the results show a clear schooling dropout advantage among children with internal-migrant parents compared with children of nonmigrant parents. Significantly, however, the disparities observed in comparisons between children in internal-migrant and nonmigrant families are closely similar to those found in comparisons between children with immigrant and U.S.-born parents. For example, in Model 1, the advantage of the children of internal migrants is not explained by differences in child-level demographic factors. Similarly, in the full model (Model 2), the internal-migrant advantage remains robust to the influence of other background factors. Yet, the decrease the internal-migrant advantage in the full model implies that differences in parental schooling and other family-level characteristics explain most, although not all, of their schooling advantage. Interaction terms in Model 3 further indicate that disparities between children with internal-migrant parents and those with nonmigrant parents in seemingly disadvantaged familial contexts are broadly analogous to those shown in Table 2. For example, in families with at least one dropout parent, schooling dropout risks are lower among children with internal-migrant parents than among those with nonmigrant parents. In addition, as observed among the children of immigrants, having an increasing number of dropout siblings is also associated with higher dropout risks among native internal migrants' children than native nonmigrants' children.

Table 3. Logistic regression coefficients estimating the likelihood of dropping out of school among the children of U.S.-born black internal migrants and nonmigrants.

	Model 1	Model 2	Model 3
Children of U.S.-born Black Internal Migrants	−0.29***	−0.09*	−0.02
Children of U.S.-born Black Nonmigrants (ref.)	(0.00)	(0.00)	(0.00)
Age	0.70***	0.72***	0.71***
Males	0.25***	0.32***	0.32***
Single-Parent Household		0.50***	0.52***
Migrant × Single-parent household			−0.04
Family Size		−0.03***	−0.03***
Household Income (LN)		−0.06***	−0.06***
Highest Parental Schooling Level
Less than completed high school		0.71***	0.70***
High school graduate		0.61***	0.62***
Some college education		0.37***	0.38***
College graduate (ref.)		(0.00)	(0.00)
One Parent Dropped Out of School		0.84***	0.92***
Migrant × One parent dropped out			−0.23***
Both Parents Dropped Out of School		0.96***	0.92***
Migrant × Both parents dropped out			0.11
Sibling Characteristics
Total nonbiological siblings		0.07***	0.07***
Migrant × Nonbiological sibling			−0.02
Oldest sibling dropped out of school		−0.25***	−0.22**
Migrant × Oldest sibling dropout			−0.09
Total dropout siblings		1.02***	0.95***
Migrant × Total dropout siblings			0.22*
Constant	−14.6***	−15.6***	−15.7***
N	160,274	160,274	160,274
Log-Likelihood	−31,708	−29,034	−29,015

^*p < .05;

^**p < .01;

^***p < .001

Comparisons between the children of native internal migrants and their counterparts with black immigrant parents are presented in Table 4. Regardless of whether parental migration occurred within the past five years, having either immigrant or internal-migrant parents is associated with lower dropout risks than having nonmigrant parents. Model 1, however, provides further illumination on the role of parental duration of residence in mediating schooling disparities among migrant families. For example, it shows that gross differences between children with immigrant and internal-migrant parents are conditional on the timing of parental migration. Thus, among the children of recent migrants, dropout risks are lower, not higher, among those with native internal-migrant parents rather than among those with immigrant parents (p < .05). Increasing parental duration of residence is, however, associated with more improvements among the children of immigrants than among the children of native internal migrants. Accordingly, among the children of long-term migrants, the likelihood of schooling dropout is much lower among those with immigrant parents than among those with internal-migrant parents. Model 2 includes additional controls accounting for age and sex, as well as for whether children started their schooling abroad. After accounting for these factors, the advantage of children of recent internal migrants over their counterparts with immigrant parents is eliminated, while the schooling advantage of the children of long-term immigrants over other children is accentuated. Additionally, in Model 3, the advantage of the children of long-term migrants persists even after controlling for the possible impacts of other household and parental-level factors (not shown).

Table 4. Logistic regression models estimating the likelihood of dropping out of school among the children of black immigrants, U.S.-born black internal migrants, and U.S.-born black nonmigrants.

	Model 1	Model 2	Model 3	Model 4	Model 5
Children of Black Immigrants
Recent	0.10	−0.40*	−0.42*
Long-term	−0.68***	−0.97***	−0.76***
First generation				−0.23*	−0.74***
Second generation				−0.90***	−0.88***
Children of U.S.-born Black Internal Migrants
Recent	−0.25***	−0.20**	0.06
Long-term	−0.32***	−0.31***	−0.12***
All				−0.31***	−0.10***
Children of U.S.-Black Nonmigrants (ref.)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Age		0.69***	0.70***		0.70***
Male		0.26***	0.31***		0.31***
Started Schooling Abroad (yes = 1, no = 0)		0.30*	0.25		0.36*
Constant	−2.58***	14.4***	−15.1***	−2.81*** −15.1***
N	178,115	178,115	178,115	178,115	178,858
Log-Likelihood	−41,112	−34,392	−31,522	−41,102	−31,621

^*p < .05;

^**p < .01;

^***p < .001

Subsequent models examine the question of whether generational status mediates dropout differences between the children of immigrants and the children of internal migrants. Model 4 focuses on gross differences (i.e., without accounting for other background factors) and suggests that the children of internal migrants are collectively less likely to drop out of school than first-generation children of immigrants. However, Model 4 also implies that the children of immigrants outperform the children of internal migrants as generational status increases, with second-generation children being comparatively less likely to drop out of school than all children of native internal migrants. Model 5, which controls for other background characteristics, indicates that differences between the children of internal migrants and first- and second-generation children are further mediated by other background factors. When these factors are controlled, both first- and second-generation children have comparatively lower dropout risks than the children of U.S.-born black internal migrants.

Disparities between the children of internal migrants and their counterparts from specific black pan-national groups are examined in Table 5. As expected, the results indicate that the dropout advantage associated with having black immigrant parents is not robust across pan-national groups. After differences in child age and sex are controlled (not shown) in Model 1, the results suggest that the collective immigrant advantage is driven by the children of African and West Indian non-Hispanic immigrants. Moreover, the coefficients also suggest that the children of native black internal migrants have lower dropout rates than the children of West Indian black Hispanics, and significantly outperform the children of “other” black immigrants (as defined earlier in this article). Additionally, although both Africans and Europeans are considered to be the more educationally selective black immigrant groups, the children of West Indian non-Hispanics appear to be less likely to drop out of school than the children of black Europeans. This suggests that although these West Indian parents may be less selected on schooling than black European parents (e.g., as in Table A2, Online Resource 1), they may be more highly selected on unobservable factors that positively influence their children's schooling outcomes.

Table 5. Logistic regression models estimating the likelihood of dropping out of school among the children of black immigrants from various pan-national groups and the children of U.S.-born blacks.

	Model 1	Model 2	Model 3
Children of Black Immigrants
African	−0.65***	−0.77***	−0.64***
European	−0.16	−0.18	−0.24
West Indian (non-Hispanic)	−0.42***	−0.52***	−0.76***
West Indian (Hispanic)	0.22	−0.10	−0.41
South American	−0.02	−0.29	−0.52
Other black	0.34**	0.17	−0.20
Children of U.S.-born Blacks
Nonmigrants	0.29***	0.30***	0.10***
al migrants (ref.)	(0.00)	(0.00)	(0.00)
Constant	−14.72***	−14.23***	−15.2***
N	178,115	178,115	178,115
Log-Likelihood	−34,457	−34,434	−31,518

^*p < .05;

^**p < .01;

^***p < .001

In Model 2, however, the analysis includes an additional control for whether heads of children's households are English proficient. This addition results in a sharp decline in the comparative disadvantage of the children of West Indian Hispanics relative to the children of native internal migrants and nonmigrants. This suggests that the West Indian Hispanic dropout disadvantage is unlikely to be driven by negative selectivity among black Hispanic immigrants but reflect their limited English-language proficiency. Language issues also appear to explain the slight under-performance of “other” black immigrants relative to native nonmigrants and migrants (Model 1). As such, the suggested disadvantage of the “other” black immigrants, relative to black natives, is eliminated after controlling for the language proficiency of household heads. Instructively, however, controlling for household-level factors (Model 3) does not eliminate the comparatively lower dropout risks of children in African and West Indian non-Hispanic households, nor does it explain the disadvantage of native nonmigrants relative to native internal migrants.⁹ Seemingly, therefore, having African or West Indian non-Hispanic parents, or having native internal-migrant parents, provides a schooling dropout buffer that is explained neither by the schooling levels of these parents nor by the background characteristics of their respective families. Rather, the buffering influences of these migrant parents appear to be driven by unobserved parental characteristics that positively influence the likelihood of schooling dropout, more so among highly selected immigrants than among native internal-migrant parents.

Discussion and Conclusions

Increased scholarly focus on the children of black immigrants has resulted in a growing body of work showing that their educational attainment exceeds that of the children of U.S.-born blacks. In this study, however, the robustness of this immigrant advantage has been interrogated in four significant ways. First, the study investigated whether the collective immigrant advantage persists after familial characteristics are controlled. Second, it examined whether the advantage associated with having migrant parents is restricted only to the children of immigrants. Third, it explored whether the advantage of black immigrants' children persists when they are compared with the children of black internal migrants. Finally, the study examined whether the immigrant advantage is robust across all black immigrant panethnic groups. Although the analysis focuses on children living in migrant and nonmigrant households, sensitivity tests in Table A3 (Online Resource 1) indicate that the study's findings generally remain robust among both biological and nonbiological children. In general, therefore, the analysis provides several insights into the ways in which migration processes mediate dropout differences among black youths in the United States.

First, both immigration and internal-migration processes appear to influence dropout status by their respective associations with transformations in children's familial contexts. Thus, in line with Wilson's (2001) hypothesis, children in immigrant and internal-migrant families are less likely to have familial characteristics known to adversely affect schooling (e.g., single-parent families, and families with dropout siblings) than children in nonmigrant families. These differences partly explain the advantage of immigrants relative to natives (Table 2) and that of native internal migrants relative to native nonmigrants; in both cases, the magnitude of the migrant advantage is reduced after familial factors are controlled. Within families, however, the impacts of seemingly adverse characteristics on schooling dropout appear to be moderated by the influence of migrant selectivity on the parental characteristics of black children. As such, consistent with the first hypothesis, the study finds that having immigrant parents moderates the influence of adverse familial contexts on the likelihood of dropping out of school. In other words, part of the explanation for the collective advantage of children of immigrants relative to the children of U.S. natives may lie in the fact that these buffering influences allow the former to be more able to succeed in less favorable familial contexts relative to the latter.

Second, and more importantly, the schooling advantage associated with having migrant parents is not restricted to black immigrants' children. As such, the study's findings are consistent with the hypothesis that the children of U.S.-born internal migrants also have schooling advantage over the children of U.S.-born nonmigrants. Furthermore, as observed among the children of immigrants, having native migrant instead of nonmigrant parents is also associated with favorable dropout outcomes in less-favorable native household contexts. Consequently, the study argues that black migrant parents, regardless of foreign-born status, generally have children who do well in school. Thus, the same factors that drive the advantage of children in immigrant households, relative to their counterparts in native households, are also likely to explain why the children of native internal migrants outperform the children of native nonmigrants.

A third finding presented in the analysis is that children of black immigrants are collectively less likely to drop out of school than children of native black migrants. While this finding is consistent with the study's third hypothesis, the results indicate that disparities between the children of immigrants and native internal migrants are also conditional on three factors: parental duration of residence, children's generational status, and parental human capital. For example, foreign-born (i.e., first-generation) children and the children of recent immigrants have worse schooling outcomes than the children of U.S.-born internal migrants. However, with increasing parental duration of residence and children's generational status, the children of immigrants outperform the children of native internal migrants.

Also confirmed by the analysis is the hypothesis that the black immigrant advantage is driven by the outcomes of the children of highly selected black immigrants. Specifically, the immigrant advantage is driven by the outcomes of the children of African and West Indian non-Hispanic immigrants. In fact, no evidence is found to suggest that children from additional black immigrant groups, such as those from South America, outperform the children of native black migrants. Rather, the results suggest that some children of black immigrants may be disadvantaged when compared with the children of native black migrants. This disadvantage is partly driven by limited levels of English proficiency in immigrant families, especially among black Hispanics. Instructively, however, native black migrant parents also have higher levels of schooling than the parents of more disadvantaged children of immigrants. Thus, the mediating role of disparities in parental-level influences between these immigrant families and native migrant families should not be discounted. Evidence from some studies also appears to suggest that a native-migrant schooling advantage may be further driven by strong systems of familial support among native migrants (Chavin 1991). However, the role of such systems in mediating schooling differences among children has not been examined in this study.

For black immigrant groups driving the collective immigrant advantage, the influence of parental immigrant selectivity also appears to operate through its effects on both parental schooling and unobserved parental influences. Accordingly, low dropout rates are found among the children of African and West Indian non-Hispanics who, along with the children of Europeans, are among the most likely to have parents with postsecondary schooling (Table A2, Online Resource 1). Yet, after parental background factors are controlled, only the children of Africans and the children of West Indian non-Hispanics outperform the children of native migrants. For these two immigrant groups, therefore, dropout status is seemingly determined by factors other than parental education and more favorable household contexts. Similarly, as with the children of immigrants, the relative advantage of the children of native-born internal migrants over the children of nonmigrants generally persists after parental- and household-level factors are controlled. Beyond the influence of parental schooling, therefore, the dropout outcomes of the children of both immigrants and internal migrants appear to be driven by the unobservable characteristics of these migrant parents. These similarities further underscore the study's argument that the same types of selectivity influences associated with unobserved parental-level attributes that positively influence the schooling outcomes of the children of immigrants may also operate among the children of U.S.-born internal migrants.

Instructively, therefore, the study's findings challenge the notion that the collective schooling advantage of black immigrants' children, relative to the children of native blacks, is driven by the unique cultural attributes of the former. Instead, the empirical evidence is consistent with the notion that black immigrant parents are positively selected on attributes, such as parental aspirations, that explain the schooling successes of their children. Previous studies on African and Caribbean immigrants indicate that these parents strongly emphasize the importance of schooling to their children (Arthur 2000; Kasinitz et al. 2001). In addition, prior studies (e.g., Rumbaut 1997) have argued that such aspirations are generally higher among immigrant parents than among U.S.-born parents. According to this analysis, positive parental influences on schooling may also exist among native U.S. migrants. More importantly, although these findings deal with the possible role of such influences among blacks, such selectivity-related factors may affect broader comparison between immigrants and natives. Systematic studies on parental influences on schooling among internal migrants in the United States are, however, absent from the literature. The expansion of research on native internal-migrant families may, therefore, provide useful insights into the determinants of their well-being and potentially shape future discourses on how broader migration processes mediate socioeconomic disparities among children in the United States.