Reaching For The American Dream: Are Black immigrants more vulnerable to academic decline than other immigrants? A Thesis

Transcription

1 Reaching For The American Dream: Are Black immigrants more vulnerable to academic decline than other immigrants? A Thesis Presented in Partial Fulfillment of the Requirements for the Degree of Master of Arts in the Graduate School of The Ohio State University By Denise Obinna, B.A Graduate Program in Sociology The Ohio State University 2011 Thesis Committee: Claudia Buchmann, advisor Douglas Downey Reanne Frank

2 Copyright by Denise Obinna 2011

3 Abstract Ever since the successes of the Civil Rights Act of 1964 and the Hart-Cellar Immigration Act of 1965, black immigration to the United States has flourished. This growth in immigration has triggered a steady increase in black immigrant enrollment in American schools. Although research on the academic performance of black immigrants has been limited, few studies have identified that first and second generation black immigrants perform better than native blacks on standardized tests and college enrollment. Using data from the Educational Longitudinal Survey of 2002, I find that while black immigrants perform better than native blacks on standardized tests, they are not more likely to enroll in college. My analyses also illustrate that on standardized tests, second generation black immigrants earn scores that are comparable to other non-black second generation immigrants. Specifically, I find that while second generation Asian and Hispanic students outperform their native born counterparts on standardized tests; this does not apply to college enrollment. First and second generation Asian immigrants are not more likely to enroll in college than their native counterparts. In fact, first generation Hispanics have a significantly lower likelihood of enrolling in college than their native peers. ii

4 Dedication: To Mummy, Annie and Nke iii

5 Acknowledgements I would like to give sincerest thanks to my advisor and mentor Claudia Buchmann who has been instrumental in helping me refine my research question. Her constant support and feedback on preliminary drafts and statistical output helped to make this thesis possible. I am also grateful to Doug Downey and Reanne Frank for their service on my thesis committee and their continued support for my academic development. They have challenged me to be a better researcher and sociologist. In addition to my thesis committee, I would like to say a special thank you to Salvatore Restifo. His guidance on statistical and methodological issues was invaluable as I developed my regression models. I am truly grateful. iv

6 Vita January Born Newcastle, England May, B.A., Sociology, Claflin University September 2009 to August 2010 Graduate Fellow, The Ohio State University September 2010 to present Graduate Teaching Associate, The Ohio State University Fields of Study Major Field: Sociology v

7 Table of Contents Abstract ii Dedication... iii Acknowledgements.iv Vita....v Table of Contents... vi List of Tables... vii List of Figures... viii Introduction... 1 Literature Review... 3 Straight Line Assimilation... 3 Segmented Assimilation... 4 Bilingualism & English Fluency... 8 Family Structure & Socioeconomic Status... 9 Hypotheses Data & Methods Results Conclusion.. 29 References..30 Appendix: Tables & Figures..40 vi

8 List of Tables Table 1.Coding Scheme & Descriptive Statistics Table 2. Non-weighted Independent Variable Percentages for Each Racial Group Table 3. Bivariate Correlation Matrix Table 4. OLS Regression Models of Standrdized Tests Table 5. Odds Ratios for Logit Models on College Enrollment vii

9 List of Figures Figure 1. Distribution of English Fluency by Race & Generation viii

10 I. Introduction Within the last thirty years, immigrant children have become the fastest growing and most racially diverse group of students in the American school system (Passel 2011; Hernandez et al. 2008; Zhou 1997; Portes and Rumbaut 2001). Immigrant students currently make up a fourth of the nation s children and by the year 2050 it is projected that they will account for more than a third of all children in the United States (Suárez- Orozco et al. 2010; Tienda and Haskins 2011). As an important part of this demographic trend, the black immigrant population has been growing at a remarkable rate. Between 2001 and 2006, immigration accounted for 20 percent of the growth in the U.S black population and there are now over 1.5 million black immigrant children enrolled from Kindergarten to the 12 th grade in the United States (Shaw-Taylor 2009; McCabe et al. 2010). In a relatively short period of time, black immigrant students have become a significant part of the American population and their educational performance needs to be understood. Although scholarly interest in black immigrants has renewed in recent years, most of the literature on black education has referred to them as a single monolithic group. Few studies have separated the academic performance of black immigrants from native blacks (Thomas 2009; Bennett and Lutz 2009; Rong and Brown 2001). Yet, this distinction is necessary because black immigrants do perform differently from native 1

11 blacks on academic indicators such as grades and standardized tests (Kasinitz 2008; Waters 1996; Model 2008). A number of studies have found that black immigrant students are more likely to complete their high school education and attend college than their native born counterparts (Massey et al. 2007; Keller and Tillman 2008). Given these differences in academic outcomes, studies on immigrant generation and academic performance have indicated that black immigrants lose their competitive edge and experience a decline in educational performance with each successive generation (Portes and Zhou 1993; Perreira et al. 2006). For example, Rong and Brown (2001) report that the educational performance of black immigrant youths increases from the first generation (foreign-born children of foreign-born parents) to the second (U.S.- born children of foreign-born parents) and then declines sharply in the third or nativeborn generation (U.S.-born children of U.S.-born parents). Similarly, in their multivariate analyses, Kao and Tienda (1995) found that while first and second generation black immigrant students reported high reading and math test scores, the native-born generation had significantly lower scores. Thus, when compared to first and second generation blacks, researchers find that the native born generation underperforms on academic indicators such as standardized tests and college enrollment. Poor academic performance not withstanding, more research needs to understand the differences in academic performance between native and immigrant blacks. Also, research needs to examine whether black immigrants have academic outcomes which are comparable to non-blacks of the same immigrant generation. Using data from the Educational Longitudinal Survey (ELS: 2002) a nationally representative sample of 2

12 students in public and private schools in the United States, my study examines the academic performance of native, immigrant and non-immigrant blacks 1. With standardized test scores and college enrollment as measures of academic performance, I determine whether the performance of black immigrants is similar to native-born blacks or to non-blacks of the same immigrant generation. From these comparisons, my study provides a more complete understanding of the educational outcomes of black immigrants (relative to native blacks and other immigrants) within the American school system. My study also has theoretical implications because it builds on prior research on immigrant education and assimilation. By underscoring the crucial role of racial and minority status on black immigrant education, I expand on the straight line and segmented assimilation theories. Particular emphasis is placed on the role of English proficiency, parental education and socioeconomic status in accounting for academic disparities across racial and generational groups. Therefore, this study contributes to the literature on black immigration by answering three distinct research questions: 1) Net of socio-demographic characteristics, do native blacks under-perform when compared to immigrant blacks on standardized test scores and college enrollment? 2) Is the association between standardized tests and college enrollment similar or different for blacks and nonblacks of comparable immigrant generation? 3) Are differences in standardized test scores and college enrollment by immigrant generation due to socio-demographic factors such as parent s level of education, English fluency and socioeconomic status? 1 The non-black immigrant groups in the analysis are: Hispanics, Asians and Non-Hispanic whites. 3

13 II. Theoretical Explanations for Differential Academic Success Straight-line Assimilation Various theories have been proposed to explain the academic performance of immigrant and native-born youth in American schools. One of the most influential theories has been the straight line hypothesis model proposed by Milton Gordon (1964). This hypothesis states that as immigrant and native born youth interact, the former adopt an American identity which engenders success over a period of time. Other interpretations of this theory state that as length of stay in the United States increases, immigrant students become more proficient in English and are generally more comfortable with the educational curriculum than recently arrived immigrants (Rumbaut 1997; Shields and Behrman 2004; Rong and Fitchett 2008). According to the straight-line model, the academic performance of first-generation immigrants will be significantly lower than their native born peers because of the social and linguistic difficulties associated with immigration. However, with each passing generation, these difficulties will dissipate as they become more familiar with the American educational curriculum. Segmented Assimilation Despite the optimistic predictions of the straight-line model, it has often been criticized for its inability to explain why so many contemporary immigrants perform better than their native-born peers (Kao and Tienda 1995; Portes and Rumbaut 2001; 4

14 Alba and Nee 1997). Researchers have proposed a more selective or segmented assimilation model which emphasizes the multiple and often contradictory paths that immigrants take when they assimilate into American society (Grant and Rong 1999; Suárez-Orozco and Suárez-Orozco, 2001). Focusing on black adolescents, Ogbu (1987; 1991; 1994) concludes that the variability in performance among immigrant and native blacks is due to the fact that the latter were forcefully assimilated into American culture through slavery as involuntary (or caste-like) immigrants. As a result of this forceful assimilation, native blacks have developed an oppositional culture to deal with the social and economic subordination encountered from whites (Ogbu 1987; 1991). Within the school system, this culture translates into a behavioral attitude that devalues education and equates academic excellence as a white phenomenon. Therefore, black students who succeed academically are punished by their classmates because they are viewed as a threat to the collective black identity (Fordham and Ogbu 1987; Gibson and Ogbu 1991; Majors and Billson 1992). In contrast, voluntary black immigrants lack the historical scars of discrimination and are often more optimistic and amenable to the demands of schooling (Ogbu 1991; Hayes 1992; Ogbu1987). Thus, there is a clash of cultures when native and immigrant blacks interact with each other in school. Immigrant blacks who pursue academic excellence find themselves penalized by their native-black classmates as sellouts who are acting white (Portes and Zhou 1993; Gibson and Ogbu 1991). In order to prevent social ostracism, Portes and Zhou (1993) state that some black immigrant students adopt the oppositional culture of their native-born peers. However, 5

15 the adoption of this culture leads to a devaluation of education and the promotion of an adversarial stance towards whites (p. 81). Waters (1996) adds that the adoption of native-black culture causes immigrant students to perceive more racial discrimination and fewer rewards to their academic efforts than those who retain their ethnic identities (i.e. an identity associated with nationality such as Jamaican or Haitian). Students who identify only as black are likely to focus on the racial discrimination that limits their opportunities and tend not to perform well in school (Waters 1999; Vickerman 1999; Portes and Zhou 1993; Gibson and Ogbu 1991). The segmented assimilation model also states that immigrants who live in close proximity to low income blacks are at a greater risk of picking up norms that devalue education than immigrants who live close to whites (Fernandez-Kelly 1995; Portes and Rumbaut 2001; Portes and Zhou 1993; Zhou 1997; Zhou and Bankston 1999). While the model does not provide explicit comparisons of native and immigrant blacks educational outcomes, it suggests that the adversarial nature of black culture is negative and immigrant blacks who adopt it are likely to experience downward assimilation into a new underclass.at the bottom of society (Portes and Rumbaut 2001:59). In essence, black immigrants have a better chance for educational, economic and social advancement if they deliberately retain their immigrant culture and maintain ties to their ethnic communities (Waters 1996; Vickerman 1994; 1999; Gibson and Ogbu 1991). With regards to non-black racial groups, the segmented assimilation model also suggests that the retention of an immigrant identity is advantageous to academic performance. For instance, in their study of Vietnamese immigrants in New Orleans, 6

16 Zhou and Bankston (1998) found that despite the economic and educational disadvantages of their parents, Vietnamese children were more academically successful than their U.S born counterparts of similar socioeconomic status. Zhou and Bankston (1998) attribute the scholastic excellence of Vietnamese students to the dense multiplex system of family or kinship, religious, organizational and work ties (p. 222) which provides immigrant children with a support system that facilitates upward mobility, discipline and academic success. Thus, despite growing up in low income underprivileged neighborhoods, the ethnic enclave is a crucial source of social capital that provides a buffer against the adversarial youth subculture of the surrounding community. In a similar study espousing the benefits of segmented assimilation and the ethnic enclave, Gibson (1988) found that the academic success of Punjabi immigrants was due to the pressures of parents and family members who urged children to stay close to their immigrant identity and avoid over-exposure to American culture. Similarly in their study of indo-chinese immigrants, Caplan, Choy and Whitmore (1992) acknowledged that academic performance was directly influenced by the retention of ethnic values transmitted by parents. Among Hispanics, Portes and Zhou (1993) found that although first generation Mexicans often struggled with the American curriculum due to their lack of English fluency, they were described as courteous, respectful and serious about their schoolwork (p.88) by school administrators. Conversely, their native-born counterparts were often described as sullen apathetic and less motivated (p. 89). According to Portes 7

17 and Zhou (1993), Hispanic immigrants have come to the United States in search of economic security and they often view education as a catalyst into the middle class and to financial security. Therefore, their children are urged to do well in school. Building on the framework of Portes and Zhou (1993), Suárez-Orozco and Suárez-Orozco (1995) state that first generation Mexicans are motivated to do well in school despite economic hardships because of the struggles that their parents experienced in migrating to America. However, by the third generation there is a progressive deterioration of familial and academic motivation as students pick up the American culture of their peers. Although segmented assimilation has been influential in explaining educational disparities between immigrants and their native counterparts, it has received widespread criticism in the scholarly literature (Boyd 2002; Hirschman 2001; Farley and Alba 2002). The model has been criticized because it suggests that American culture is negative and that the immigrants who adopt it are bound to experience academic failure (Steinberg 2005; Alba and Nee 2003). Other critics state that the model stereotypes native blacks as possessing an oppositional culture which devalues education (Butcher 1994; Dodoo1997; Perlman and Waldinger 1997). While there has been some evidence that an oppositional culture exists among native-black students (Majors and Billson 1992; Farkas et al. 2002) a substantial amount of the literature contradicts the assumption that native-blacks are adversarial and devalue academic success (Carter 2005; Tyson et al. 2005; Ainsworth-Darnell and Downey 1998; Harris 2006). In particular, Alba and Nee (2003) strongly criticize the model for essentializing central-city black culture in the image of the underclass (p. 8). 8

18 It is therefore unfair to characterize assimilation into black culture as an inevitable assimilation into the underclass. Furthermore, Neckerman, Carter and Lee (1999) claim that segmented assimilation ignores the fact that black immigrants can also assimilate into the black middle or upper classes rather than the underclass. These criticisms notwithstanding, the segmented assimilation model acknowledges that racial discrimination makes it difficult for individuals classified as non-white to achieve upward mobility in America s racially stratified society (Waters 2010; Portes, Fernandez-Kelly and Haller 2009). Although these immigrants entered the country voluntarily, they are aware of the stigmas attached to those characterized as black or minority. For all immigrant groups, assimilating into American society means that they must develop an understanding of racism and all of its subtle nuances (Rumbaut 1994; Bonilla-Silva 2006). This is especially important for black immigrants who come from countries where they are the racial majority (Lindsey and Wilson 1994; Rong and Brown 2001). Most black immigrants come from Africa and the Caribbean where being black does not have the same negative meanings that it has in the United States (Rogers 2001; Waters 1990; 1996). Therefore, to dispel any negative stereotypes associated with being black and to gain acceptance from whites, black immigrants often find it advantageous to stress their ethnic identities rather than their racial ones (Rogers 2001; Vickerman 1994; 1999). III. Predictors of Academic Performance among Immigrant Blacks Bilingualism and English Fluency As a means of maintaining their ethnic distinction, black immigrants often use 9

19 their fluency in English and their native language as a way to differentiate from native blacks. For instance, many Haitian immigrants who are fluent in English continue to speak French and Creole with their children at home as an important marker of their social identity (Zephir 1997; Portes and Zhou 1993). This tendency is also found among black immigrants from Africa. Many first and second generation Nigerian immigrants maintain their indigenous tongue as well as their fluency in English 2 (Okome 2005; Kent 2007). Portes and Rumbaut (2001) argue that bilingualism is beneficial to immigrant children because it provides them with access to the supportive networks of their ethnic enclaves while English proficiency facilitates academic success in the United States. In a similar vein, Rumberger and Larson (1998) found that bilingualism had positive influences on a number of academic performance indicators. Their study of second generation Mexican immigrants indicated that bilingual students attained better grades and had lower dropout rates than monolingual students. They state that Spanish proficiency provides a cultural advantage while English proficiency provides an academic advantage. Studies by Zhou and Bankson (1994), Matute-Bianchi (1991) Stanton-Salazar and Dornbusch (1995) have all provided convincing evidence that bilingual students do indeed have access to ethnic and community networks which often translates into academic advantages. Although many studies have supported the positive findings on bilingualism, some researchers have been critical in accepting these conclusions. The strongest criticism has been that bilingual and monolingual students are not comparable and 10 2 Although almost all Nigerian immigrants are fluent in English, many still maintain their fluency in their indigenous tongues Yoruba, Igbo and Hausa. This language advantage is beneficial in segmented assimilation and enriches the language mosaic of their ethnic enclave (Okome 2005)

20 differences in academic performance can be explained by factors such as the student s individual motivation and socioeconomic status (Hakuta and Diaz 1985; Mouw and Xie 1999). Still, Portes and Hao (1998) and Portes (2002) contend that fluent bilinguals tend to have greater solidarity with their parents as well as higher levels of self esteem and ambition than monolingual students. Family Structure and Socioeconomic Status Researchers have indicated that disparities in the academic performance of immigrant and native-born children are due in part to differences in family structure and socioeconomic status (Buchmann and Parrado 2005; Amato 2005; Kao 2004). Although family structure comparisons between native and immigrant blacks are limited, researchers acknowledge that the children of native blacks are more likely to live in single parent households than their black immigrant counterparts (Thomas 2009; Landale, Thomas and Van Hook 2011; Brandon 2002). Family structure plays an important role in determining how immigrant children are integrated into American society. Existing research strongly suggests that children from two parent families fare better academically than children from single parent families (Thompson, Hanson and McLanahan 1994; Heiss 1996; Pong 1997). Also, immigrant children with both parents are more likely to report positive feelings about school and schoolwork than those in single parent households (Oropesa and Landale 1997; Amato 2005; Kao 2004). In addition to family structure differences, researchers have identified key disparities in the level of education earned by the parents of immigrant and native black 11

21 students (Thomas 2009; Bennett and Lutz 2009). In a study of students at selective universities, Massey et al. (2007) 3 found that relative to native blacks, a larger share of black immigrants had parents who had earned college and advanced graduate degrees. Furthermore, Hagy and Staniec (2002) observed that black immigrant high school graduates were more likely to attend college than their native black counterparts. Based on these differences in parental education and rates of college attendance, researchers have stated that immigrant blacks have an advantage over native born blacks because they are more likely to grow up in two parent households with parents who are highly educated (Bennett and Lutz 2009; Massey et al. 2007; Thomas 2009). Other studies identify that black immigration to the United States has been selective with a disproportionate share of immigrants coming from middle class educational and occupational backgrounds (Crowder and Tedrow 2001; Kasinitz et al. 2001; Logan and Deane 2003). Feliciano (2005) identifies that immigrants from Haiti, Jamaica and the Dominican Republic were more educated than their counterparts in their home countries. Also, Kent (2007) posits that more than a third of all immigrants from Kenya, Nigeria and Ghana have college degrees. In addition to being more educated, African and Caribbean immigrants on average have higher incomes than native blacks. According to U.S Census estimates in 2000, the median household incomes were: $ 43,650 for Africans, $ 42,900 for Caribbean immigrants and $33,790 for native blacks (McKinnon and Bennett 2005). Similarly, in a study of Caribbean immigrants in South Florida, Kasinitz et al. (2001) found that on average, Jamaican and Haitian immigrants in the 3 Massey et al. (2007) used data from the National Longitudinal Survey of Freshmen and found that 70 percent of black immigrants had parents with college degrees as opposed to 55.2 percent among native blacks. Furthermore, an additional 43.6 percent had earned advanced graduate degrees 12 while just 25.3 percent of native blacks had parents with advanced degrees.

22 Miami-Dade and Broward counties of Florida had median household incomes which were significantly higher than their native black counterparts. Considering that black immigrants often come from families that are more educated and economically advantaged, it is not surprising that they are more likely to attend college than native blacks. Despite these academic distinctions, black immigrants have received little research attention. While they constitute a relatively small percentage of students in American schools, black immigrants are a growing part of the U.S. black population whose academic performance needs to be understood. Given the close correlation between academic success and economic well being, a more complete understanding of how black immigrants differ from native born blacks can be useful in reducing academic disparities between the two groups. Furthermore, by comparing black immigrants to non-black immigrants, researchers can learn how closely the educational performance of black immigrants resembles non-blacks of the same immigrant generation. IV. Hypotheses In this study, I evaluate academic performance by using one measure of academic achievement (i.e., standardized test scores) and one measure of academic attainment (i.e., college enrollment). I control for the impact of family structure, bilingualism, socioeconomic status and parental education which have all been said to influence academic performance (Landale, Thomas and Van Hook 2011; Kasinitz et al. 2001; Buchmann and Parrado 2005). Based on the evaluation of prior literature, eight hypotheses emerge and they are tested empirically in the analyses that follow. The first 13

23 four hypotheses are related to standardized test scores and the last four are related to college enrollment. Hypothesis 1: First and second generation black, Asian and Hispanic students will have higher standardized test scores than their native born counterparts of the same race. This hypothesis draws from the segmented assimilation model which suggests that first and second generation students will fare better than their native peers because of the deliberate retention of an immigrant identity which facilitates academic success (Waters 1996; Vickerman 1994; 1999; Gibson and Ogbu 1991; Portes and Zhou 1993). Hypothesis 2: When differences in parental education, family structure and socioeconomic status are controlled, the achievement gaps between black, Asian and Hispanic immigrants and their native born counterparts will be greatly reduced or eliminated completely. This hypothesis is based on research which suggests that immigrant students have an advantage over their native born peers due to the fact that the former come from families that are more educated and economically more advantaged than their native born counterparts (McKinnon and Bennett 2005; Crowder and Tedrow 2001; Kasinitz et al. 2001; Logan and Deane 2003) Hypothesis 3: Standardized test score gaps are due to differences in family structure, socioeconomic status and fluency in the language of the host country. Black, Asian and Hispanic students who are not fluent in English will produce lower test scores than students who are native English speakers. After controlling for family background factors, these achievement gaps will be greatly reduced or eliminated. Hypothesis 4: Unlike students who are not fluent in English, black, Asian and Hispanic 14

24 students who are bilingual will have test scores that are higher than (or not significantly different from) native English speakers net of family background characteristics. Hypotheses 3 and 4 are supported by the segmented assimilation model and work by Portes and Rumbaut (2001) who argue that immigrant students who maintain fluency in their native language and fluency in English will perform better than students who are not fluent in English (Portes and Hao 1998; Portes 2002; Mouw and Xie 1999; Zhou and Bankston 1994). Hypotheses 3 and 4 are also supported by the straight line assimilation model which suggests that immigrants who face social and linguistic barriers will exhibit poor academic performance (Rumbaut 1997; Shields and Behrman 2004; Rong and Fitchett 2008). Hypothesis 5: First and second generation blacks, Hispanics and Asians will have higher rates of college enrollment than their native born counterparts of the same race. Like Hypothesis 1, this hypothesis draws from the segmented assimilation model which suggests that first and second generation immigrants will have higher rates of college enrollment due to their retention of an immigrant identity rather than a racial one (Waters 1996; Vickerman 1994; 1999). Hypothesis 6: Gaps in college enrollment among black, Asian and Hispanic immigrants and their native counterparts are due to differences in parental education, family structure and socioeconomic status. When these effects are controlled, gaps in college enrollment will disappear or reduce greatly. This hypothesis is supported by previous research that has identified that immigrant students often come from families with parents that have earned college degrees and more economically advantaged than their native born peers 15

25 (McKinnon and Bennett 2005; Crowder and Tedrow 2001; Kasinitz et al. 2001). Therefore, this hypothesis suggests that when differences in socioeconomic status and parent s level of education are controlled, there should be little variation in college enrollment rates between immigrant and native born students. Hypothesis 7: Disparities in college enrollment are due to family structure, socioeconomic status and English proficiency. Black, Asian and Hispanic students who are not fluent in English will have lower rates of college enrollment than students who are fluent in English. After controlling for family background factors, college enrollment gaps will be greatly reduced or eliminated. Hypothesis 8: Black, Asian and Hispanic students who are bilingual will have rates of college enrollment that are higher than (or not significantly different from) their native born counterparts who are native English speakers net of family background characteristics. Hypotheses 7 and 8 are based on research by Portes and Rumbaut (2001) as well as the segmented assimilation theory which suggests that immigrant students who maintain their fluency in their native tongue as well as English will report better academic performance than students who are not fluent in English. These hypotheses also support the straight line assimilation model which suggests that immigrant students who face linguistic challenges will also face challenges academically (Rumbaut 1997; Shields and Behrman 2004; Rong and Fitchett 2008). Hypotheses 7 and 8 suggest that poor academic performance will be illustrated by low rates of college enrollment among students who are not fluent in English. 16

26 V. Data and Methods Data Source This study uses data from the Educational Longitudinal Study (ELS), a survey of 2002 high school sophomores conducted by the National Center for Education Statistics (NCES). Data collection began in 2002 with a nationally representative sample of 16,197 sophomore level students from 752 public and private schools across the United States. Students were tested at two year intervals and transcript data were collected from their high schools in 2005, one year after their expected graduation. I use data from the baseline survey and the second follow-up in 2006 when most students had graduated from high school. The ELS data are ideal for addressing the research questions in my study because they include an extensive amount of information on the family and academic backgrounds of immigrant and native-born students. The longitudinal nature of the data allows for the comparison of academic performance at the sophomore year as well as the rates of college enrollment following student graduation. The non-weighted analytic sample for this paper includes a total of 13,150 students. I have restricted the sample to those students who reported a racial/ethnic group and who had complete information on the mother and child s country of birth. I also excluded Native American students (n=130) because of the emphasis on immigrant generation. The main immigrant group of focus in the study is black immigrants who 17

27 comprise a total sample of 1,669 students (87 first-generation, 101 second-generation, and 1,481 native born). Although I would have preferred a larger sample of first and second generation blacks, my data are weighted using the second follow-up, base year panel weight 4 which ensures that the sample of respondents is representative of students who were in the tenth grade in Thus, I am confident about any statistically significant differences between black immigrants and other immigrant groups. The final weighted sample for all respondents is 11,408. Dependent Variables: In this study, academic performance is the key dependent variable. It is measured by student performance on standardized tests and the rate of college enrollment after high school. For standardized test scores, I use the ELS composite on math and reading from the base year. This composite score is the average math and reading scores taken from the ELS: 2002 Cognitive Tests. These scores were re-standardized to a national mean of 50.0 and a standard deviation of Some students only had scores for the math and reading tests, but not both. For the students who did not have both sets of scores, the composite is based on the scores which were available. To find out the rates of college enrollment after high school, I examined how many students were enrolled in college by January Given that most students who were sophomores in the base year of 2002 should have graduated by 2004, I use this as a robust measure in estimating how many immigrant and native born students had graduated and enrolled in college. This measure of enrollment was coded as a dummy variable (1=Student enrolled in college, 0=Student did not enroll in college). 4 (F2BYWT) See National Center for Educational Statistics (NCES) for details on the calculation of weights. 18

28 Independent Variables: Race was self-reported by students at the base year, with response categories including: Black, Asian, Hispanic and white. Generational status was ascertained by the responses that mothers gave when they were asked about the country of birth of their child: Was your tenth grader born in the United States (that is, any of the fifty states or the District of Columbia), in Puerto Rico 5, or in another country or area? I use this information to determine the generation status of the student. Students are considered first-generation if they were not born in the United States and students born in the United States with foreign-born mothers are considered second-generation. When students and mothers are born in the United States, they are designated as native-born generation 6. The multivariate analyses also include a number of individual and family level variables. First, to distinguish if the respondent was a male or female, I created a dummy variable named female (1=respondent is female 0=respondent is male). English fluency was self reported by students who stated their fluency in English when it was not their native language. It is measured by three dummy variables (1=not fluent 2=bilingual 3= native English speakers). Parent s level of education is the report of the parent s educational level. It is measured by a series of dummy variables: high school diploma, some college, graduated from college and advanced degree (less than high school is the reference category). Family composition reports the number of people who live with the respondent at home. It is divided into two categories: single parent if the student lives with either their mother or father or a single guardian and two parents (the reference 5 Due to a lack of complete information in the ELS data, I am unable to determine country of birth. Therefore, students born in Puerto Rico are considered to be first-generation I used mother s and not father s nativity to determine student s generation status because the ELS data includes more information on mother s birthplace than father s birthplace.

29 category) if the student lives with both parents. Lastly, Family Income is measured by the total family income earned from all sources in the year Four dummy variables were created to measure income (measured in thousands): $20,000 or less, $ 20,001-$ 35,000, $35,001-$75000 and more than $75,000. These income categories were created by collapsing the 13 income groups of the ELS data into 4 income categories to even out the distribution of respondents between the groups and for ease of analysis 7. These groups roughly divide the data into quartiles with the exception of students from families who earn less than $20,000 a year who make up just 14 percent of the sample. Detailed definitions and coding scheme for all variables used are presented in Table 1. ***Insert Table 1 here*** Analytic Strategy To begin, in Table 2, I present percentages for the key variables by race and generational status. Since these differences may be due to family background characteristics in addition to English fluency, Figure 1 shows the distribution of English fluency by race and generational status. For the second stage of the data analyses, I present a correlation matrix (see Table 3) for the dependent variables (standardized test scores and college enrollment) which indicate that no issues of collinearity are present. Next, I perform an ordinary least squares regression (OLS) to examine the relationship between race and generational status on standardized test scores net of other factors (see Table 4). In this table, Model 1 is the baseline model which incorporates generational status, race and sex for each student. In Model 2, I include interaction terms for race and 20 7 Respondents who earn less than $ 20,000 a year are considered to be below the poverty line according to 2001 Census estimates (US. Census Bureau of Poverty in the United States 2001)

30 generational status for first and second generation blacks, Asians and Hispanics. For each of the interaction terms, the native-born generation is the reference category. In Model 3, I add controls for family structure, parent s level of education and family income. In addition to the controls in Model 3, Model 4 incorporates controls for English fluency. In the next stage of the data analyses, I use binary logistic regression models to estimate the odds of college enrollment. Model 1 in Table 5 is a baseline model which examines the impact of generational status, race and sex on college enrollment. In model 2, I include interactions for generational status and race. The native born generation serves as the reference category for each racial group. In Model 3, I include controls for family structure, parent s level of education, as well as family income. Lastly, in the final model (Model 4, Table 5) I include controls for English fluency into the analyses. To ensure that all the dependent variables were regressed on the same sample, I conducted a listwise deletion on the independent variables used for each regression. I also used regression-based imputation to account for missing data. All data in the models are weighted to account for sampling design. VI. Results ***Insert Table 2 Here*** Descriptive Analysis Table 2 presents percentages by race and immigrant generation for all variables. The analyses show that first and second generation immigrants are more likely to be of Hispanic or Asian origin than white or black students. All of the racial groups in the study come from two parent families although 45.5 percent of blacks are from single 21

31 parent households. Majority of the respondents have parents who have earned at least a high school diploma or attended some college. However, Asian and white students have the highest proportions of parents with graduate/advanced degrees at 18.5 and 13.3 percent respectively. Also, with the exception of Asians, female students make up a slightly higher proportion of the respondents than males. In terms of family income, the majority of whites and Asians are from families that earn $35,001-$75,001 and more than $ 75,000 a year. While majority of the Hispanics and blacks come from families that earn $20,001-$35,000 and less than $ 20,000 a year. ***Insert Figure 1 here*** The second stage of the descriptive analysis illustrates the distribution of English fluency by race and generational status. As expected, Figure 1 shows that an overwhelming majority of the native born generation are native English speakers although this proportion appears to be strongest among whites and blacks. With regards to bilinguals, Figure 1 shows that almost 70 percent of first generation Hispanics is bilingual and approximately 57 percent of the second generation is bilingual. A similar pattern exists for Asians as approximately 61 percent of the first generation is bilingual and 60 percent of the second generation is bilingual. Among blacks and whites, the largest bilingual proportions are within the first generation with 28 and 51 percent respectively. Within the sample, a relatively small proportion of respondents are not fluent in English and this is largely among first and second generation Hispanic and Asian students with an estimated 20 percent and 70 percent respectively. From the results above, the descriptive data produces some interesting findings 22

32 that will be informative in the regression analyses that follow. First, since most of the blacks (regardless of generational status) are fluent in English with a smaller percentage that is bilingual, this should not have a negative impact on their academic performance. English fluency might have more of an impact on Hispanic and Asian students since they have the largest amount of non-fluent and bilingual students. The descriptive analyses also indicate that parental education and family income will have an effect on the academic performance of certain groups. For instance, Asians and whites collectively have the most educated parents and they come from more affluent families than their black and Hispanic counterparts. Based on these descriptive results, the regression models (see Tables 4 and 5) further explore how race and generational status (in addition to family background factors and English fluency) influence academic performance. ***Insert Table 3 here*** Standardized Test Score Performance Table 4 shows the results of OLS regression models on standardized test scores. The baseline model (Model 1) shows that first generation students earned scores that are 1.09 points lower than their native born counterparts. Similarly among the racial groups, black and Hispanic students earned scores that are significantly lower than whites. For instance, blacks and Hispanics scored 8.95 and 7.32 points lower than whites on standardized tests. In Model 2, when racial and generational interactions are included, blacks and Hispanics still report scores that are significantly lower than whites. However, unlike the baseline model, blacks scored 9.27 points lower than whites and Hispanics scored 6.51 points lower than whites. The decreased magnitude in standardized test 23

33 scores among Hispanics and the increased (negative) magnitude among black students indicates that some of the differences between Hispanics and blacks relative to whites can be explained by generational factors. Consistent with prior literature, the interactions show that black immigrants perform better than native blacks on standardized tests (Kao and Tienda 1995). This is most evident among second generation blacks who reported scores that are 4.91 points higher than native blacks. Conversely, first generation Hispanics reported scores that are 3.12 points lower than their native born counterparts. The results of Model 2 confirm some of the projections of Hypothesis 1 because although black immigrants have higher test scores than native blacks, this trend does not extend to Asian or Hispanic students. In fact, first generation Hispanics reported lower test scores than native born Hispanics and the second generation did not perform significantly better than the native born generation. The hypothesis is further disproved among Asian students as neither the first nor second generation reported test scores that were higher than the native born generation. The effects of Hypothesis 1 are only confirmed with regards to second generation black immigrants as they are the only group of students to out perform the native born generation. These findings for second generation blacks lend support to the segmented assimilation model which argues that an immigrant status is beneficial to academic performance although this immigrant status is more beneficial to second generation black immigrants. In Model 3, controls for parent s level of education, family structure and income are included into the analysis. The results of this model indicate that native blacks and Hispanics still earned 7.0 and 4.7 points lower than whites even when a host of 24

34 demographic characteristics are controlled. Among the interactions, the second generation reports test scores which are significantly higher than the native born generation for all racial groups. This association is strongest among black immigrants who earned scores that are 3.15 points higher than native blacks. Furthermore, second generation Asian immigrants earned 3.04 points higher than the native-born generation. Also in this model, Hispanics scored 1.28 points higher than native Hispanics. The performance of the second generation in Model 3 is noteworthy because it is consistent among all racial groups in the study. When parents level of education, income, and family structure are held constant, second generation students appear to have an advantage over their native born counterparts that results in higher standardized test score performance. In line with existing research, Model 3 shows that students who have a parent who has earned a high school degree, a college degree or a more advanced degree have higher test scores than students with parents who have less than a high school degree. Similarly, students from households that earn less than $ 20,000 per year have test scores that are significantly lower than students from households that earn between $20, ,000 per year. The results of Model 3 disprove Hypothesis 2 completely because even when differences in parental education, family structure and socioeconomic status are controlled, second generation blacks, Hispanics and Asians still report higher test scores than their native born counterparts. Interestingly, this trend only exists for second generation students as first generation students do not report significant differences from the native born generation. Thus despite the projections of Hypothesis 2, that achievement gaps between immigrant and native born students would disappear 25

35 when family background characteristics were taken into account, second generation students still have an advantage over their native born peers. Building on the results of Model 3, Model 4 incorporates controls for English fluency. The results of this model indicate that when controlling for language (in addition to other generational factors) the first generation performs significantly better than the native born generation. Put another way, first generation students scored 1.67 points higher than the native born generation after English fluency, and family background characteristics were controlled. In line with prior research, students who are not fluent in English scored 8.18 points lower than students who are fluent English speakers. However, students who are bilingual also earned scores that were 2.19 points lower than native English speakers. Among the different racial groups, English fluency appears to have little effect on the performance of native blacks and Hispanics as they still report scores which are significantly lower than whites. From an analysis of the interactions, second generation blacks and Asians have the highest test scores with 3.15 and 3.04 points higher than their native born counterparts. For Hispanics, a similar trend persists with second generation Hispanic students earning 2.12 points higher than native Hispanics. The results of this model confirm the importance of English fluency on immigrant academic performance. Second generation immigrants (regardless of race) outperform their native-born counterparts on standardized tests when language fluency is held constant (in addition to other socio-demographic factors). As expected, students with parents who have earned more than a high school degree and come from households that have incomes of more than $20,000 a year score higher than students who have parents 26

36 with less than a high school degree. The results of Model 4 lend validity to Hypothesis 3 because students who are not fluent in English scored significantly lower than their native born peers on standardized tests net of family background characteristics. These results give support to the segmented and straight-line assimilation theories because they both underscore the salience of English fluency on academic success. Despite the credibility which Model 4 gives to Hypothesis 3, Model 4 disproves the predictions of Hypothesis 4 because students who are bilingual did not outperform native English speakers; they actually underperformed when they were compared to them. These findings do not reinforce the assumptions of Portes and Rumbaut (2001) or Portes and Hao (1998) since the performance of bilingual students does not exceed the performance of native English speakers. Therefore, with regards to standardized tests, there are clear differences in the performance of immigrants relative to native born students. These effects appear to be strongest among the second generation and it exists for blacks, Hispanics and Asians alike. The findings become more pronounced when English fluency is controlled (net of other family characteristics) indicating that fluency has a strong influence on the performance of all second generation immigrants (regardless of race). In the regression analyses that follow, I evaluate if the results from the regression analyses on standardized tests are present for college enrollment estimates as well. ***Insert Table 4 Here*** College Enrollment Estimates Table 5 presents the results of binary logistic regression models predicting college 27