Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

English Language Learners’ Pathways to Four-Year Colleges


by Yasuko Kanno & Jennifer G. Cromley - 2015

Background/Context: English language learners (ELLs) are the fastest growing segment of the K–12 student population in the United States, yet they encounter substantial problems entering higher education. The gap between ELLs and non-ELLs is particularly acute for four-year college access. Research has been largely silent on ELLs’ college advancement, and we know little about what inhibits ELLs’ college access.

Purpose: To examine the process of ELLs’ college planning in order to determine which stages of college planning present difficulties to ELLs and why. College planning is conceptualized as consisting of five milestones: (a) aspiring to college, (b) acquiring college qualifications, (c) graduating from high school, (d) applying to college, and (e) enrolling in college.

Research Design: Secondary data analysis of the Education Longitudinal Study of 2002. Only students who participated in all of the first three waves (2002, 2004, and 2006) of data collection were included (N = 12,450). Students were divided into three language background groups: (a) ELLs, (b) English-proficient linguistic minority students (EPs), and (c) native speakers of English (NSs). We first compared the college-access patterns of the three language groups. We then mapped out each group’s pathways through the milestones. Finally, we conducted multigroup analyses to examine whether and to what extent a different set of predictors shape the groups’ college pathways.

Findings/Results: It is the early stages of college planning (aspirations and college qualifications stages) that are particularly challenging to ELLs, such that the majority of ELLs never reach the later milestone of applying to a four-year college. Predictors known to matter significantly for the general population’s college access are not all significant for ELLs.

Conclusions/Recommendations: In order to enable more ELLs to reach four-year colleges, we should make a targeted effort to support them in the early stages of college planning. Racial/ethnic minority ELLs are particularly vulnerable and need more support. We also need to invest more effort into identifying the factors and conditions that specifically influence ELLs’ college planning.




INTRODUCTION


There appears to be a widely shared assumption in U.S. high schools that English language learners (ELLs), bilingual or multilingual students who are still in the process of acquiring grade-level English, are not candidates for higher education. As Callahan and Gándara (2004) write, “For recent immigrants, the goal is to teach them English; for long-term ELLs, the goal is drop-out prevention and—in the best-case scenario—high school graduation” (p. 110). If the best-case scenario for ELLs’ educational attainment is assumed to be high school graduation, the idea of ELLs enrolling in four-year colleges and obtaining a bachelor’s degree would seem like a fantasy.


A large body of research already exists regarding the college-access patterns of several underrepresented populations, including racial/ethnic minority, low-income, immigrant, and first-generation college students (e.g., Bowen, Kurzwell, & Tobin, 2005; Deil-Amen & Turley, 2007; Kim & Díaz, 2013; McDonough, 1997; Nuñez & Cuccaro-Alamin, 1998). However, higher education research has been largely silent on ELLs’ access to and success in college—despite the fact that ELLs constitute the fastest growing segment of the K–12 student population (Wolf, Herman, Bachman, Bailey, & Griffin, 2008). ELLs are predicted to represent 25% of the student body by 2025 (U.S. Department of Education, 2006). Language background may be examined as one of the factors related to the central topic of investigation in these studies, but it is rarely given center stage for inquiry in higher education scholarship.


Scholarship on ELLs’ college access as an area of inquiry has only recently begun to emerge, led by scholars who work in the intersection between language and education (e.g., Almon, 2010; Bunch & Endris, 2012; Callahan, 2005; Callahan, Wilkinson, & Muller, 2010; Kanno & Cromley, 2013; Kanno & Harklau, 2012; Kanno & Kangas, 2014; Kanno & Varghese, 2010, Mosqueda, 2011; Nuñez & Sparks, 2012; Ruecker, 2012). The scholarship in this area, however, is still at a nascent stage, and what we currently know about ELLs’ college access and success is outweighed by what we still do not know. National-level statistics that can inform policy are particularly scarce. In our effort to contribute to the development of the knowledge base in this area, in a previous study (Kanno & Cromley, 2013) we examined patterns of ELLs’ access to and attainment in postsecondary education by using the National Education Longitudinal Study of 1988 (NELS:88). We found that ELLs lagged far behind non-ELLs in both enrollment and degree attainment in postsecondary education (PSE). In particular, four-year college access and bachelor’s degree attainment were beyond the reach of many ELLs.


The present study, building on the findings of our NELS:88 study, focuses on the process of ELLs’ college planning. Our assumption is that if ELLs have a markedly lower rate of four-year college access than that of their non-ELL counterparts, there must be some stages of college planning that are particularly difficult for ELLs. Using the more recent Education Longitudinal Study of 2002 (ELS:2002), we conceptualize college planning as consisting of five critical milestones and examine which milestones present difficulties for ELLs and why.


ELLS’ COLLEGE ACCESS AND THEIR MULTIPLE DISADVANTAGES


It is important to note at the outset that ELLs in fact constitute a heterogeneous group of students (Wright, 2010). For example, although there is a pervasive image of ELLs being recently arrived immigrants, in fact only about 35% of ELLs are foreign-born students while the rest are born in the United States (Education Week, 2009).1 While some students develop their English proficiency and exit English as a second language (ESL) programs relatively quickly, others remain in ESL programs semipermanently (Menken & Kleyn, 2010; Umansky & Reardon, 2014). Although most ELLs are legal permanent residents or U.S. citizens, a small but important fraction of them are undocumented students (García, Kleifgen, & Falchi, 2008).


While fully recognizing such diversity within this population, it is also important to acknowledge that ELLs do share some characteristics when it comes to their college access and resources they bring to the process. It is those shared characteristics with respect to ELLs’ college planning, rather than diversity and different scenarios that certainly exist within their college planning, that we focus on in this study.


Beginning with the overall rates of ELLs’ college access and degree attainment, Kanno and Cromley’s (2013) analysis of NELS:88, mentioned above, found that only 18% of ELLs advanced to four-year colleges upon high school graduation, compared with 43% of monolingual English-speaking students and 38% of English-proficient linguistic minority (LM) students (i.e., students who speak a non-English language at home but who are also proficient in English). With regard to graduation, only 12% of ELLs attained a bachelor’s degree, compared with 32% of monolingual English speakers and 25% of English-proficient LM students within eight years after their high school graduation. Using another large-scale dataset, the Beginning Postsecondary Students Longitudinal Study 2004 (BPS: 2004), Nuñez and Sparks (2012) identified 11% of first-time and first-year students who began PSE in the 2003–2004 academic year to be LM students. Approximately the same proportions of LM and native-speaking students were enrolled in selective four-year institutions while a higher proportion (61%) of LM students than native-speaking students (56%) were enrolled in two-year institutions, suggesting a possible bifurcation among LM students in the type of PSE enrollment. Nuñez and Sparks speculate that English-proficient LM students are likely to be enrolled in selective four-year institutions while ELLs are more likely to attend two-year institutions.


Other studies are also beginning to identify factors that contribute to ELLs’ lower college-attendance rates, the most obvious of which is their low academic achievement. The results of the 2013 National Assessment of Educational Progress (NAEP) show a 55-point gap (test score range: 0–500) between 12th grade ELLs and non-ELLs in reading and a 44-point gap (test score range: 0–300) in mathematics (National Center for Education Statistics, n.d.). This is likely to be a major hindrance to ELLs’ college access and success since academic preparation in high school is a major predictor of college viability for the general population (Adelman, 2006; Cabrera, Burkum, & La Nasa, 2005).


There are structural barriers to ELLs’ academic achievement. Rodriguez and Cruz (2009) argue that educational institutions have consistently failed to invest enough resources in implementing researched-based best practices for fostering ELLs’ English language proficiency. The result is that many LM students remain in ESL programs on a semipermanent basis—and become long-term ELLs (Menken & Kleyn, 2010; Umansky & Reardon, 2014)—rather than quickly getting reclassified as English-proficient. At the same time, English language acquisition itself is treated as a “gatekeeping process for access to college preparatory content” (Rodriguez & Cruz, 2009 p. 2392), so that if students are not reclassified, their access to rigorous curricula is restricted. ELLs’ tracking into the noncollege streams starts as early as in middle school (Estrada, 2014; Wang & Goldschmidt, 1999). By high school, ELLs who qualify to take high-level college-preparatory courses such as honors and Advanced Placement (AP) are exceptions rather than the rule (Callahan, 2005; Callahan et al., 2010; Harklau, 1994; Kanno & Kangas, 2014). A comprehensive survey conducted in 2001 found that only 1% and 0.8 % of ELLs in high school were enrolled in AP math and AP science courses, respectively—much lower enrollment rates than the rate of 3.2% for all students (Hopstock & Stephenson, 2003).


Another impeding factor is ELLs’ low socioeconomic status (SES). Approximately 75% of ELLs come from low-income families (Zehler et al., 2003). Given the rising cost of college education, many ELLs may be priced out of higher education. Past research suggests that lack of finances affects both the choice of colleges that ELLs would consider (Kanno & Grosik, 2012) and their persistence in college once they are enrolled (Almon, 2010). A related issue to SES and college affordability is some ELLs’ legal status as undocumented immigrants. To our knowledge, no definitive information on the percentage of ELLs who are undocumented immigrants exists. However, among the undocumented immigrant youth (ages 15 to 30) who are eligible for the Obama administration’s Deferred Action for Childhood Arrivals (DACA)2, 31 % are limited English proficient (Batalova, Hooker, & Capps, 2013). We can surmise from this figure that a sizable proportion of first-generation-immigrant ELLs are undocumented immigrants and that their legal status in the United States is a major factor in their college planning. As of June 2014, only 17 states allow undocumented students to pay in-state resident tuition to attend public PSE institutions (National Immigration Law Center, 2014). Also, undocumented immigrant students are ineligible for federal financial aid and some state financial aid. The combination of steep tuition and lack of financial aid makes it extremely difficult for undocumented students to attend PSE (Kim & Díaz, 2013).


Lack of guidance from both their parents and schools is another potential barrier for ELLs’ college access. ELLs on the whole have less educated parents than non-ELLs (Kanno & Cromley, 2013; Zehler et al., 2003). For example, in Kanno and Cromley’s (2013) study, the mean educational level of ELLs’ parents was 2.19 as compared to 3.13 for the parents of monolingual English-speaking parents, with “2” indicating high school graduation and “3” indicating less than two years of college education. Such information suggests that ELLs’ parents may not be in the position to provide much guidance on PSE since most of them lack PSE experience. If the necessary guidance is not available from the parents, students need to receive it from somewhere else. However, ELLs are also more likely to attend resource-poor schools and are segregated from non-ELLs because they tend to be concentrated in a small number of schools. Furthermore, high-ELL-concentrated schools on average have less qualified teachers and principals and have lower academic achievement levels than low-ELL schools (De Cohen, Deterding, & Clewell, 2005; Fry, 2008). Thus, it is likely that many ELLs are not receiving necessary guidance from their schools either.


In summary, we are developing some understanding of ELLs’ access to and success in PSE and the factors that contribute to them. However, we still have not developed a coherent picture of how ELLs plan for college, which stages of planning they struggle with, and what factors facilitate or inhibit the reaching of each of the milestones. Just as importantly, one serious limitation of previous research is that it has largely assumed that the significant predictors for the general student population’s college access and degree attainment are also important for ELLs. But that may not be the case. Nuñez and Sparks (2012) found fewer significant predictors for the types of colleges that LM students enrolled in than for native-speaking students. Thus, in addition to mapping out ELLs’ pathways to four-year colleges, it is critical to investigate which factors specifically shape ELLs’ college planning. In this study, then, we ask the following research questions:


1.

What proportions of ELLs and non-ELLs enroll in four-year colleges after high school graduation?

2.

To what extent do ELLs’ pathways from college aspirations to enrollment differ from those of non-ELLs?

3.

Which of the critical milestones in the pathway to four-year colleges present difficulty to ELLs, and why?


CONCEPTUAL FRAMEWORK


Our inquiry into understanding ELLs’ college pathways is informed by scholarship that has conceptualized college enrollment as the culmination of a multistep process that develops over several years (Cabrera & La Nasa, 2000, 2001; Hossler & Gallagher, 1987; Roderick, Nagaoka, Coca, & Moeller, 2008). Hossler and Gallagher’s (1987) well-known model of college choice conceptualizes the process up to enrolling in college in terms of (a) predisposition to attend college, (b) searches for potential colleges to apply to, and (c) making the choice about which college or university to attend. Focusing specifically on the search and choice phases of Hossler and Gallagher’s model, Cabrera and La Nasa (2000, 2001) argue that four-year college enrollment requires successful completion of three critical steps: (a) acquiring necessary college qualifications, (b) graduating from high school, and (c) applying to college. To this model, Roderick et al. (2008) added an initial step: aspiring to attend a four-year college. We believe that this step is a critical first step to the road to a four-year college because such aspirations are likely to motivate students to take the subsequent steps for college planning.


In all then, we conceptualize the college planning process for four-year college enrollment as consisting of five critical milestones: (a) aspiring to college in early high school, (b) acquiring the necessary college qualifications, (c) graduating from high school, (d) applying to college, and finally (e) enrolling in college. We describe students who meet all five milestones as being on the four-year college pathway.


Further, in order to understand what sets of variables predict students reaching each milestone, we bring to bear theories of economic, cultural, and social capital as originally conceptualized by Bourdieu (e.g., 1977, 1986, 1991). Bourdieu (1987) conceptualized capital as resources that are “capable of conferring strength, power and consequently profit on their holder” (p. 3). One of Bourdieu’s critical insights is that what renders power to individuals is not simply their economic capital (money, property) but also cultural capital (familiarity with the dominant culture) and social capital (access to social networks that yield useful resources and information) (Bourdieu, 1977, 1986). Previous higher education studies have found that in addition to economic capital (McDonough, 1997; Walpole, 2007), information and guidance available from parents (Perna & Titus, 2005), students’ own academic preparation in high school (Adelman, 2006; Cabrera & La Nasa, 2001), information available from college-bound friends (Cherng, Calarco, & Kao, 2013; McDonough, 1997), and information and resources available from school (McDonough, 1997; Nuñez & Kim, 2012; Perna & Thomas, 2008) shape their chances for college access and success. These variables also constitute our starting point. However, as we noted above, it is important to keep in mind that the variables that contribute to non-ELL students’ college access may not be relevant to ELLs’ college access. Thus, we need to investigate which variables are significant predictors specifically for ELLs’ college access.


Here, it is also important to explain why we focus specifically on ELLs’ access to four-year colleges and universities in this study. Clearly, four-year institutions are not the only PSE option available; nor is advancing directly from high school to four-year college the only path to a bachelor’s degree. Nonetheless, we chose to focus on ELLs’ four-year college access for the following reasons. First, our analysis of NELS:88 demonstrated that going to a four-year college and earning a bachelor’s degree were particularly elusive goals for ELLs. There was a 25-percentage-point difference in four-year college access and a 20-percentage-point difference in bachelor’s degree attainment between ELLs and monolingual English-speaking students (Kanno & Cromley, 2013).


Second, a bachelor’s degree is the surest ticket out of poverty into the middle-class that our current education system offers to an underprivileged person (Swail, 2000). This function of college education is particularly important for ELLs, who are more likely to come from low-income families than non-ELL students (Kanno & Cromley, 2013; Zehler et al., 2003). Over a worker’s lifetime a bachelor’s degree is worth over $1 million more than a high school diploma and over $0.5 million more than an associate’s degree (Carnevale, Rose, & Cheah, 2011).


Finally, we focus on the direct pathway from high school to four-year college because it offers the highest chance of resulting in a bachelor’s degree. Cabrera et al. (2005) showed that even among highly academically prepared students, only 30% who first went to two-year colleges completed a bachelor’s degree, compared to 78% who went straight to four-year colleges. Moreover, studies that examined ELLs’ retention in community colleges have found that ELLs are even less likely than non-ELL students to transfer to four-year colleges (Almon, 2010; Razfar & Simon, 2011). Based on these findings, we believe that if we are serious about helping more ELLs obtain a bachelor’s degree, we must identify and eliminate systemic barriers that prevent ELLs from advancing directly from high school to a four-year college.


DATASET


ELS:2002, sponsored by the National Center for Education Statistics (NCES), began in 2002 by collecting data from a nationally representative stratified random sample of 15,360 tenth graders from 750 randomly selected schools.3 After the base-year (BY) survey in 2002, students were resurveyed three times (2004, 2006, and 2012). Students were followed from the time they were 15–16 years old in 2002 until they were 25–26 years old in 2012, approximately eight years after their high school graduation. Therefore, ELS:2002 constitutes the ideal, most up-to-date dataset to use to analyze students’ transition from high school to college and from college to workforce.


We used the ELS:2002 restricted-use dataset from the first three waves (2002, 2004, and 2006; from 10th grade to two years after high school graduation), which includes surveys from students, parents, and schools; reading and mathematics achievement tests; and high school transcripts. NCES intentionally oversampled Asian and Hispanic students in order to include sufficient numbers of these students in the sample to permit statistical analyses by racial/ethnic groups and also included individual weights to correct for the oversampling. Since Hispanic and Asian students represent large proportions of ELLs, their overrepresentation in the data makes ELS:2002 a particularly suitable dataset for analyzing ELLs’ educational trajectories because there are enough ELLs sampled to allow for statistically stable estimates.


We included in our analyses only those students who participated in all first three waves of data collection. Of the 15,360 students in the original sample, 12,590 students participated in all three waves of data collection. Of these 12,590 students, we excluded the small number of Native American students due to problems with model convergence. We also excluded students whose language background and/or postsecondary destination were missing and approximately 10 additional students whose postsecondary destination appeared contradictory with other data (e.g., students who were reported be enrolled in a four-year college even though other information indicated that they failed to graduate from high school; students who did not apply to any four-year college and yet showed up as enrolling in one). This resulted in an analytical sample of 12,450 students.


Distilling an analytical sample this way has important implications for what kinds of students are included in our analysis. Preliminary examination of the descriptive statistics of those students who participated in all three waves (“three wavers”) as compared to all students who were included in the BY data collection (“BY participants”) showed that the three wavers were on the whole somewhat better off in terms of the cultural, economic, and academic forms of capital they possessed. This is unsurprising since despite NCES’s efforts to follow BY participants through subsequent data collection, in reality it is easier to track students who stay in school. Notably, the differences between three wavers and BY participants were in fact smaller for ELLs than for non-ELLs. This is most likely because those ELLs who were included in the BY data collection were already a selective group. The ELS:2002 sampling was based on a random selection of high schools; thus, those who were not enrolled in high school in 10th grade were automatically excluded. Since ELLs are more likely to drop out of school than non-ELLs (Kim, 2011), more ELLs are likely to have been excluded from this sampling method than non-ELLs. Moreover, in the BY survey, students with severely limited English proficiency (i.e., not able to read the surveys in English) were excluded. Consequently, it is safe to assume that our analytical sample includes those ELLs who are somewhat higher academic performers with better English proficiency than the ELL population at large. It is important for the reader to keep these characteristics of ELLs in our sample in mind when interpreting our findings.


DATA ANALYSIS


Our analyses proceeded in four steps: (a) categorizing ELS:2002 students into three language background groups; (b) identifying and analyzing the groups’ college enrollment patterns; (c) mapping out the groups’ trajectories through the milestones; and (d) analyzing a model that includes a set of predictors that account for students’ achieving each milestone. See the Appendix for the list of variables used in our analyses. For all analyses, we used the appropriate panel weight (F2BYWT).


THREE LANGUAGE BACKGROUND GROUPS


For all subsequent analyses, we divided students into three language background groups, following methods from previous studies of NELS:88 and ELS:2002 (Bennici & Strang, 1995; Callahan et al., 2010; Kanno & Cromley, 2013).


English language learners (ELLs): students whose first language is not English and who exhibit signs of limited English proficiency (n = 490)

English proficient linguistic minority students (EPs): students who are nonnative speakers of English but who currently exhibit no sign of difficulty using English (n = 1,580)

Native speakers (NSs): students who are native speakers of English (n = 10,380)


We first categorized students into native and nonnative speakers of English based on the BY student survey. Students who answered “yes” to the question: “Is English your native language?” were categorized as NSs and those who answered “no” were categorized as nonnative speakers (NNSs). NNSs were further divided into English proficient (EP) students and ELLs. We used multiple sources of data to identify the two groups of NNSs. First, those who self-rated at least one of their four skills (listening, speaking, reading, and writing) in English in the bottom two categories of the four-point scale in the BY survey were identified as ELLs.4 Also, ELS:2002 surveyed two teachers for each student in the BY survey. If at least one of the teachers identified a NNS student as an ELL, the student was also identified as an ELL.5 However, relying solely on students’ self-report would eliminate those ELLs who did not answer key language background questions on the survey. We therefore also used high school transcript data to identify additional ELLs by adopting Callahan et al.’s (2010) method of categorizing any students whose high school transcripts indicated that they took at least one ESL, bilingual education,6 or sheltered English content class as ELLs.


Finally, we excluded ELLs from the whole NNS group to identify NNSs who are not ELLs: i.e., EPs. Altogether, we identified 10,380 NSs (86%), 1,580 EPs (10%), and 490 ELLs (4%) for a total of 12,450 students.


Table 1 summarizes the relevant background characteristics of the three groups (all analyses are weighted). Sixty percent of ELLs and 56% of EPs are Hispanic, and much larger proportions of ELLs and EPs are racial/ethnic minority students than NSs. In terms of family SES, ELLs’ parents are at lower income levels and have less education than the parents of EPs and NSs. Parents’ educational aspirations for their children are not much different among the three groups: In fact, EP parents’ aspirations are higher than those of NS parents, perhaps reflecting immigrant optimism (Kao & Tienda, 1995). In contrast, ELLs’ own aspirations are markedly lower than those of EPs and NSs. ELLs’ academic preparation and performance are both much lower than those of EPs and NSs. Moreover, ELLs and EPs attend schools with much larger percentages of minority students and low-income students than do NSs. Further, ELLs on the whole attend high schools that send less than a quarter of their graduates to four-year colleges. Many of these characteristics have strong bearings on ELLs’ college planning, as we will see in our results section.



Table 1. Descriptive Statistics by Linguistic Background, Weighted

 

Linguistic background

 

ELL (unweighted n ~ 490)

EP (unweighted n ~ 1,580)

NS (unweighted n ~ 10,380)

 

M or % (SD)

Missing value %

M or % (SD)

Missing value %

M or % (SD)

Missing value %

Predictor

Sex

 

0%

 

0%

 

0%

  Female

50%

 

51%

 

50%

 

  Male

50%

 

49%

 

50%

 

Race/ethnicity

 

0%

 

0%

 

0%

  Asian

16%

 

20%

 

2%

 

  Hispanic

60%

 

56%

 

8%

 

  Black

7%

 

6%

 

15%

 

  White

15%

 

13%

 

70%

 

  Mixed race

2%

 

3%

 

5%

 

Family income

7.38 (2.68)

0%

7.98 (2.42)

0%

9.23 (2.27)

0%

Parental education

3.41 (2.22)

0%

3.69 (2.27)

0%

4.53 (1.99)

0%

Parental education aspirations

5.24 (1.37)

0%

5.56 (1.32)

0%

5.33 (1.25)

0%

Student education aspirations

4.70 (1.66)

15%

5.16 (1.49)

11%

5.19 (1.40)

9%

Highest math taken

4.61 (1.49)

5%

5.42 (1.65)

12%

5.48 (1.60)

7%

10th grade GPA

1.94 (1.66)

0%

1.75 (2.38)

0%

2.17 (2.12)

0%

Math test score (theta)

-.992 (.825)

0%

-.548 (.857)

0%

-.283 (.787)

0%

Reading test score (theta)

-1.26 (.707)

0%

-.723 (.792)

0%

-.355 (.785)

0%

Financial aid receiveda

27%

47%

42%

28%

45%

27%

How many friends plan to attend   

  4-yr college

2.89 (1.17)

4%

3.11 (1.13)

3%

3.33 (1.07)

2%

Students at school receiving

  free/reduced-price lunch

4.11 (1.86)

6%

4.23 (1.98)

10%

3.23 (1.79)

9%

% of minority students at school

58.88 (29.78)

0%

58.78 (31.14)

0%

28.83 (28.72)

0%

Graduates at school went to 4-yr

  college

3.98 (1.08)

37%

4.14 (1.12)

27%

4.47 (1.08)

23%

 

Note. Weighted by F2BYWT. Percentages may not add up to 100% because of rounding. Family income units correspond most closely to increments of $10,000 per year, where 7 represents <$20K to $25K, 8 represents <$25K to $35K, 9 represents <$35K to $50K, and 10 represents <$50K to $75K. Parental education is measured on a 8-point scale: 3 corresponds to some college education, 4 to 2-year degree, and 5 to more than 2-year degree but less than 4-year degree. Both parental and student educational aspirations are measured on a 7-point scale: 4 corresponds to more than 2-year degree but less than 4-year degree, 5 to obtaining a bachelor’s degree, and 6 to a Master’s degree. Math and reading test scores are IRT-scaled scores. “How many friends plan to attend 4-year” is measured on a 5-point scale: 2 corresponds to “a few,” 3 to “some,” and 4 to “most.” “Students at school receiving free/reduced-priced lunch” refers to the degree of the representation of low-income students in the high school that each participant attended and is measured on a 7-point scale: 3 corresponds to “11-20%,” and 4 to “21-30%.” “Graduates at school went to 4-year college” refers to the % of graduates at the high school that each participant attended who advanced to four-year colleges or universities upon graduation, and is measured on a 5-point scale: 3 corresponds to “11-24%,” 4 to “25-49%” and 5 to “50-74%.”

a The missing values for financial aid are large because they include cases of legitimate skip (i.e., students who did not advance to postsecondary education and therefore did not apply for financial aid).




COLLEGE ENROLLMENT PATTERNS

In order to answer our first research question, “What proportions of ELLs and non-ELLs enroll in four-year colleges after high school graduation?” we first cross-tabulated students’ language background (ELL, EP, and NS) with their first PSE institution types, as reported in the 2006 wave (two years after high school graduation): (a) enrolled in four-year college, (b) enrolled in two-year college, (c) enrolled in less than two-year college, (d) high school diploma/GED but no PSE education, or (e) some high school.


COLLEGE PATHWAYS

In order to address the second research question, “To what extent do ELLs’ pathways from college aspirations to enrollment differ from those of non-ELLs?” we charted the college pathways through the five milestones for each language background group. Our objective in these analyses was to identify the percentage of students who persisted from one milestone to the next, as well as the percentage of students who achieved all five milestones: (a) four-year college aspirations, (b) college qualifications, (c) high school graduation, (d) four-year college application, and (e) four-year college enrollment. Each of these five milestones was conceived of as a dichotomous variable: Either a student reached the milestone (1) or did not (0).


Aspire 4-year indicates whether students, in 10th grade, aspired to earn a bachelor’s degree or higher. Students were asked in the BY survey, “How far do you think you will get in school?” Students were assigned a 1 if they responded that they thought they would attain at least a bachelor’s degree; they were assigned a 0 if they reported aspirations that were anything lower than a bachelor’s degree.


College qualifications refer to “academic qualification for four-year college work” (Berkner & Chavez, 1997, p. 21). In our analysis, we used the highest level math course completed in high school as a proxy for this variable. NELS:88 has an NCES-derived composite variable college qualifications, which incorporates high school GPA, class rank, NELS test scores, and college examination (SAT and ACT) scores (Berkner & Chavez, 1997). Unfortunately, the equivalent composite variable has not been created for ELS:2002. However, a report by an advisory committee for the U.S. Department of Education (U.S. Department of Education Advisory Committee on Student Financial Assistance, 2006) noted that approximately the same proportion of high school graduates took at least algebra II in ELS:2002 as the proportion of the students who were at least “minimally college-qualified” (i.e., the top 75% of those students who entered four-year colleges and universities) in NELS:88, and therefore proposed the use of “at least Algebra II” as an index of college qualifications for ELS:2002. The close match between college qualifications and mathematics course-taking is unsurprising because mathematics, more than any other subjects, plays gatekeeping functions in higher education and in society (Battey, 2013; Martin, 2009). We adopted this approach and recoded an ELS:2002 transcript-based composite variable, F1RMAPIP, for the highest math course taken. College qualified is a dichotomous variable for which students were assigned a 1 if the highest level math course they took was Algebra II or above; otherwise they were assigned a 0.


High school graduation was derived from a composite variable for students’ high school graduation status in 2006. Those who held a high school diploma or a GED were coded as 1; those who had not achieved a diploma or a GED were coded as 0.


Apply 4 year and enroll 4 year indicate whether students had applied to, and enrolled in, a four-year institution, respectively, by 2006. For apply 4 year we coded as 1 those students who applied to any type of four-year college; those who did not apply to any four-year college were coded as 0. For enroll 4 year, we recoded the same variable we used for the college enrollment patterns above into a simple dichotomous variable: Those students were enrolled in a four-year institution in 2006 were assigned a 1; those who were enrolled in less-than-four-year institutions or did not attend any postsecondary school were assigned a 0.


MULTIGROUP ANALYSES OF FIVE MILESTONES

In order to address the third research question, “Which of the critical milestones in the pathway to four-year colleges present difficulty to ELLs, and why?” we ran a multigroup analysis for each milestone, using a set of demographic, familial, academic, and school predictors. The idea of examining the predictors for each language background group was drawn from Cabrera and La Nasa (2001), who ran similar analyses for students of different SES levels. However, instead of running a separate regression for each of the groups, as Cabrera and La Nasa did, we employed a single multigroup analysis for each dependent variable. This enabled us to statistically test for differences across groups in the effect of each predictor. For example, we were able to test whether 10th grade GPA was as strong a predictor of four-year college aspirations for NSs, EPs, and ELLs (i.e., whether the coefficients for each group were significantly different from each other).


For the analysis of reaching Milestone #1, aspire 4 yr, we used multigroup logistic regression in Mplus Version 7 (Muthén & Muthén, 1998–2012) to analyze the effects of: (a) students’ demographic characteristics, (b) social and cultural capital available from the parents (familial capital), (c) students’ academic preparation in high school (academic capital), and (d) school resources and guidance (school capital). Models were considered to show good fit to the data if they met Hu and Bentler’s (1999) criteria: RMSEA < .06 and CFI > .95. One advantage of Mplus is that it uses Full Information Maximum Likelihood (FIML) to handle any missing data. Briefly, FIML uses all available responses for the numerator of each test statistic (using all available responses rather than excluding anyone who missed even a single measure) but uses only respondents with complete data to calculate standard errors.


For students’ demographic characteristics, we entered three variables: gender, race/ethnicity, and family income. We used five categories of race/ethnicity: Asian, Black, Hispanic, mixed race, and White (the reference group). The family income, the index of students’ economic capital, is based on the parents’ responses in the BY survey and is divided into 13 categories, from no income (1) to $200,001 or more (13). Although it is measured in categories, normality statistics suggested this variable could be treated as a continuous variable in all analyses.7


For the measures of familial capital, we included both parental education and parents’ aspirations for their children’s highest educational attainment. Parental education refers to the highest level of education achieved by either of the parents and is measured on an 8-point scale: did not complete high school (1) to completed Ph.D., M.D., or other advanced degree (8). The scale for parental aspirations similarly ranges from less than high school education (1) to obtain Ph.D., M.D., or other advanced degree (7).


As discussed previously, there is by now solid evidence that high school academic preparation has a large bearing on students’ college access and viability (e.g., Adelman, 2006; Cabrera & La Nasa, 2001)—at least for the general student population. Three variables serve as indices of academic capital: GPA in 10th grade, and math and reading test scores. In Adelman’s (2006) study of college persistence, high school GPA was the strongest predictor of students’ persistence in college—more so than test scores. We also included math and reading test scores, IRT-scaled theta scores8 from 10th grade, in order to examine the effect of students’ academic abilities above and beyond the GPA, since GPA may reflect other factors such as attendance and studiousness.


Finally, for the school capital factors, we wanted to know the extent to which the characteristics of the school a student attends influences his or her college choice and enrollment. We included percentage of 10th graders receiving free or reduced-priced lunch at the school as an index of the SES composition of the school. This is a 7-point measure ranging from 0%–5% (1) to 76%–100% (7). Percentage of minority students at school was used to capture the ethnic composition of the school. Finally, percentage of graduates who went to four-year colleges upon graduation was used as an index of the college-bound orientation of the school. Schools that routinely send many of their graduating seniors to four-year colleges are likely to know how to help students make the transition to four-year colleges and have the necessary resources to achieve this goal (McDonough, 1997). This predictor is a 6-point measure, ranging from 0% (1) to 75%–100% (6).


To this set of predictors, for the analysis of reaching Milestone #2, college qualifications, and further models we added student aspirations as a predictor because students’ own aspirations early in the college planning process are known to have an impact on their subsequent planning (Schneider & Stevenson, 1999). When we used student aspirations as a dependent variable, we dichotomized the variable. However, when we used student aspirations as an independent variable, we used the full 7-point response scale, ranging from less than high school education (1) to obtain Ph.D., M.D., or other advanced degree (7).


Similarly, for the analysis of reaching Milestone #3, high school graduation, and further models, we added college qualifications as a predictor. Given that college qualifications are known to be an important predictor for college access (Adelman, 2006; Cabrera & La Nasa, 2001), we hypothesized that it is likely to affect students’ ability to reach some of the milestones along the way as well. As noted before, the highest math course taken was used as a proxy for college qualifications, and as an independent variable, we used an 8-point measure ranging from no math at all (1) to calculus (8).


For the analysis of reaching Milestones #4, applying to four-year college, and #5, enrolling in four-year college, we also added as a predictor the number of friends that the participant had who were going to a four-year college. This variable ranges from none (1), to some (3), or all (5). Recent research suggests that the kinds of friends that one has affect one’s academic achievement as well as college enrollment (Cherng et al., 2013). Finally, it is well documented that the availability of financial aid has a large impact on students’ decision to enroll in college (National Center for Public Policy and Higher Education, 2002). We therefore wanted to examine the extent to which the availability of financial aid (including grant, scholarship, work-study, loan, tuition waiver/discount) affects students’ enrollment in four-year colleges, and how the impact of financial aid differs among different language background groups. The problem with this variable, however, is that it is pertinent only to those who applied to PSE. Therefore we conducted two multigroup regression analyses of reaching Milestone #5, enroll 4 year: (a) one that includes all students but without financial aid as a predictor, and (b) one that includes only those students who applied to four-year colleges and with financial aid as a predictor.


Since the dependent variables in these multigroup analyses are dichotomous, we used Snijders and Bosker’s (1999) formula for Intraclass Correlation Coefficient (ICC) to determine whether analyses needed to account for students being nested within schools. The ICC was extremely small (ranging from .003 to .015, far below the usual cutoff of .05), indicating that multilevel analyses were not warranted.


RESULTS


COLLEGE ENROLLMENT PATTERNS

The cross-tabulation clearly showed that ELLs lagged behind NSs and EPs in four-year college enrollment (Table 2). Only 19.0% of ELLs were enrolled in four-year colleges two years after their scheduled high school graduation date whereas 34.6% of EPs and 44.8% of NSs were enrolled in four-year colleges. At the other end of the spectrum, 46% of ELLs either had not graduated from high school or had not advanced to PSE compared to 27.6% of EPs and 26.1% of NSs.


Table 2. Access to Postsecondary Education, by Language Group, Weighted


 

Language Status

 
 

ELL

EP

NS

Total

4-Year Institution

    

N

20,568

101,599

1,105,820

1,227,987

%

19.0%

34.6%

44.8%

42.7%

Contribution to chi square

-25,775.4

-23,915.0

49,690.4

 


2-Year College

    

N

33,925

101,578

665,485

800,988

%

31.3%

34.6%

26.9%

27.9%

Contribution to chi square

3,696.3

19,708.0

-23,404.3

 


Less than 2-Year College

    

N

4,064

9,430

53,784

67,278

%

3.7%

3.2%

2.2%

2.3%

Contribution to chi square

1,525.0

2,553.4

-4,078.4

 


High School Diploma or GED

    

N

31,784

57,999

522,301

612,084

%

29.3%

19.8%

21.1%

21.3%

Contribution to chi square

8,684.4

-4,562.8

-4,121.5

 


Some High School

    

N

18,070

23,009

123,214

164,293

%

16.7%

7.8%

5.0%

5.7%

Contribution to chi square

11,869.7

6,216.4

-18,086.1

 


Total


108,411


293,615


2,470,604


2,872,630




Chi square tests showed that these differences were significant (χ2 [4, Nweighted = 2,872,630] = 60342.5, p < .001, Φ = .145). The biggest contribution to chi square was NSs’ overrepresentation in 4-year institution, followed by ELLs’ and EPs’ underrepresentation in the same category. In contrast, ELLs were overrepresented in some high school, suggesting that NSs’ and ELLs’ college access patterns are almost a mirror image of each other (see Table 2). EPs were underrepresented in 4-year institution but were instead overrepresented in 2-year college, which indicates that many EPs choose two-year colleges as their first PSE destination. Overall, EPs’ college access patterns fell between those of ELLs and NSs but resembled NSs’ patterns more than ELLs’, which was also the case in our NELS:88 analysis (Kanno & Cromley, 2013).


COLLEGE PATHWAYS

While 75% of NSs and 71% EPs aspired to graduate from four-year colleges in 10th grade, only 58% of ELLs had the same aspirations, suggesting that from the beginning of the college planning, ELLs lag behind NSs and EPs (Table 3). Obtaining college qualifications was another major hurdle for ELLs, where approximately half of those with college aspirations dropped off course while 76% of NSs and 66% of EPs with college aspirations went on to become college-qualified. In contrast, high school graduation hardly posed a hurdle to those who were college-qualified: Across the language background categories, more than 99% of those with college aspirations who were also college-qualified graduated from high school. Applying to four-year colleges was another milestone at which a substantial portion of ELLs fell out of the pathway: Only 62% of high school graduate ELLs who originally had four-year college aspirations, who were college qualified, and who graduated from high school, applied to four-year colleges, compared to 80% of NSs and 76% of EPs. On the other hand, those ELLs who had stayed on the four-year-college pathway thus far seemed to go on to enroll in four-year colleges just as well as EPs: Seventy-one percent of ELLs who had passed through the previous four milestones ultimately enrolled in four-year colleges. However, since ELLs had dropped off the four-year-college pathway at previous milestones (except for high school graduation) at disproportionately higher rates than NSs and EPs, the cumulative effect is that only 13% of ELLs managed to stay on the four-year college pathway all the way through to enroll in four-year colleges whereas 36% of NSs and 25% of EPs did so.


Table 3. Students’ Trajectories through Five Milestones from Aspirations to Enrollment, Weighted


 

Full Sample

Aspire 4Yr

College Qualified

HS Graduation

Apply 4Yr

Enroll 4Yr

NS

2,470,604

1,840,707

1,392,627

1,385,854

1,108,702

889,098

100%

75%

56%

(76% of Aspire4yr)

56%

(100% of ColQual)

45%

(80% of HG Grad)

36%

(80% of Apply4Yr)

EP

293,615

208,320

138,230

137,861

105,129

73,214

100%

71%

47%

(66% of Aspire4Yr)

47%

(100% of ColQual)

36%

(76% of HG Grad)

25%

(70% of Apply4Yr)

ELL

108,411

62,878

32,314

32,253

20,087

14,219

100%

58%

30%

(51% of Aspire4Yr)

30%

(100% of ColQual)

19%

(62% of HG Grad)

13%

(71% of Apply4Yr)

















One could make an argument that as long as students in the end manage to enroll in four-year colleges, it does not matter how they get there: that is, it is of no importance in the end whether they reached all the previous four milestones. On the surface this argument sounds reasonable. However, our analysis shows that staying on the four-year college pathway from the beginning and achieving all the four milestones is the surest way for a student to reach a four-year college. This is true regardless of students’ language background. In addition to the four-year college pathway (Pathway [a] in Table 4), there are three other possible pathways for students to eventually enroll in four-year colleges (Pathways [b–d]). The last three milestones are the same for all four pathways because logically if one does not graduate from high school and apply to a four-year college, one cannot enroll in a four-year college. In other words, the variation is in the values of the first two milestones. If we examine the “survival rate” of each of these four pathways to four-year college enrollment, it is clear that a much larger percentage of students go on to four-year colleges if they follow the four-year college pathway than any other alternative pathway (see Table 4). Forty-four percent of ELLs who started with four-year college aspirations and then achieved college qualifications enrolled in four-year colleges (64% NSs and 53% of EPs). In contrast, only 7% of those ELLs who started with four-year college aspirations but failed to fulfill college qualifications subsequently went on to enroll in four-year colleges (15% of NSs and 9% of EPs). Similarly, only 17% of ELLs who did not start out with four-year college aspirations but who did become college-qualified enrolled in four-year colleges (22% of NSs and 18% of EPs) while 0% of ELLs who did not aspire to four-year colleges in the first place and did not become college-qualified enrolled in four-year colleges (4% of NSs and 5% of EPs).


Table 4. Rates of Different Pathways to Reach Four-Year College Enrollment, Weighted

 

College Pathways

NS (%)

EP (%)

ELL (%)

a.

Aspire 4yr à College qualified à High school grad à Apply 4yr à Enroll 4yr

64

53

44

b.

Aspire 4yr à NOT College qualified à High school grad à Apply 4yr à Enroll 4yr

15

9

7

c.

NOT Aspire 4yr à College qualified à High school grad à Apply 4yr à Enroll 4yr

22

18

17

d.

NOT Aspire 4yr à NOT College qualified à High school grad à Apply 4yr à Enroll 4yr

4

5

0



At the same time, it is interesting to observe that a small portion of students began without four-year college aspirations but eventually enrolled in four-year colleges. A cross-tabulation (weighted) between aspire 4yr and enroll 4 yr shows that 4% of NSs, 6% of EPs, and 7% of ELLs who did not have four-year college aspirations in 10th grade nonetheless went on to enroll in four-year colleges. These students would be the ones who changed their minds during the last two years of high school, perhaps discovering that they were academically capable and/or being encouraged by their teachers, counselors, and parents.  


MULTIGROUP ANALYSES

Multigroup logistic regressions showed that for some of the milestones the best fitting model was one in which all predictors were constrained across groups: That is, for some milestones there was no variation by language background group regarding how strongly the predictors affected students’ chances for reaching a particular milestone. For other milestones, however, significant predictors and their effect sizes varied considerably across the groups. Generally, we found that more predictors were significant for NSs and EPs than for ELLs, suggesting that the predictors that have traditionally been considered important for students’ college access account for native or near-native speakers’ college access better than ELLs’ college access. In all multigroup analyses below (Tables 5–10), coefficients are expressed as odds ratios. For example in Table 5, the coefficients in the first row shows that the odds of female students espousing four-year college aspirations are 36% higher (eb = 1.36) than the odds for male students across language groups when controlling for all other variables.   


Although much information can be gleaned from the multigroup analyses, the following discussion focuses on the patterns that were salient for ELLs.


Four-year college aspirations. What is immediately noticeable in this set of analyses is the significance of race/ethnicity for ELLs’ college aspirations, on the one hand, and a much smaller impact of academic performance on their aspirations, on the other. Being Asian, Black, or Hispanic (vs. White) hindered ELLs’ aspirations substantially. Controlling for other predictors, Hispanic ELLs had roughly one-quarter the odds, Asian ELLs one-fifth the odds, and Black ELLs only one-tenth the odds relative to White ELLs, respectively (Table 5).


Students’ academic performance was strongly predictive of their aspirations for NSs and, to a lesser extent, EPs, but not for ELLs. Whereas GPA, math test scores, and reading test scores were all significant predictors of NSs’ aspirations, the only significant predictor of ELLs’ aspirations was the math test score. These contrasting patterns between ELLs and non-ELLs suggest that academic performance, traditionally considered a powerful predictor of college access and success (Adelman, 2006; Cabrera & La Nasa, 2000, 2001), may not be such a clear predictor of ELLs’ eventual college access even at the aspirations stage.


Table 5. Multigroup Analysis for Four-Year College Aspirations in 10th Grade, Weighted


  

NS

EP

ELL

Predictor

 

eb

eb

eb

Female   

 

1.36***

1.36***

1.36***

Asian

 

1.06

1.37

0.21***

Black

 

1.22*

1.32

0.11**

Hispanic

 

1.02

1.07

0.25**

Mixed Race

 

1.01

1.01

1.01

Family Income

 

1.02

1.02

1.02

Parental Education

 

1.05**

1.00

1.10

Educational Aspirations of Parents

 

1.32***

1.33***

1.17

10th Grade GPA

 

1.35***

1.38**

1.04

10th Grade Math Test Score

 

1.45***

1.30

1.56*

10th Grade Reading Test Score

 

1.26***

1.29

1.17

%10th Graders Receiving FRPL

 

0.97

0.97

0.97

% Minority Students in School

 

1.01***

1.00

1.01

% Graduates Going to 4-Yr College

 

1.09**

1.10

1.29

     

N

7,723

6,521

949

253

χ2(df)

9.60 (8)

   

RMSEA

(90% CI)

.009 (< .001, .026)

   

CFI

.995

   

Note. The predictors in boldface are the ones for which in the best-fitting model we allowed coefficients to vary by language background groups.

*p < .05 ** p < .01 *** p < .001


College qualifications. The best-fitting model for college qualifications was the one in which all the predictors were constrained across groups (Table 6). Among the demographic variables, being Black was a positive predictor, when controlling for other variables, boosting the chances of becoming college qualified by 33% compared to being White. However, this result needs caution in interpreting because in reality all the other factors are rarely equal for Black students given the history of institutional discrimination they face (e.g., Hirschman & Lee, 2005; Kao & Thompson, 2003; Yonezawa, Wells, & Serna, 2002).


By far, the strongest predictors for college qualifications were academic capital predictors, with all the predictors in this category being significant positive predictors. Both GPA (eb = 2.01) and math test score (eb = 1.95) had a large impact on college qualifications, each boosting odds by almost 100%. A high correlation between college qualifications and math test score was expected since the highest math course taken was used as the proxy for college qualifications. Also students who attended schools that sent a large percentage of their students to four-year colleges were more likely to become college-qualified (eb = 1.24). This suggests that the schools that have a strong four-year-college orientation offer more advanced college-preparatory courses and push students to take them.


Table 6. Multigroup Analysis for College Qualifications, Weighted


  

NS

EP

ELL

Predictor

 

eb

eb

eb

Female

 

1.09

1.09

1.09

Asian

 

1.02

1.02

1.02

Black

 

1.33**

1.33**

1.33**

Hispanic

 

0.94

0.94

0.94

Mixed Race

 

0.89

0.89

0.89

Family Income

 

1.05***

1.05***

1.05***

Parental Education

 

1.02

1.02

1.02

Educational Aspirations of Parents

 

1.05*

1.05*

1.05*

10th Grade Student Aspirations

 

1.13***

1.13***

1.13***

10th Grade GPA

 

2.01***

2.01***

2.01***

10th Grade Math test score

 

1.95***

1.95***

1.95***

10th Grade Reading test score

 

1.16**

1.16**

1.16**

%10th Graders receiving Free/Reduced Lunch

 

1.00

1.00

1.00

% Minority Students in School

 

1.01***

1.01***

1.01***

% Graduates Going to 4-Yr College

 

1.24***

1.24***

1.24***

     

N

7,723

6,521

949

253

χ2(df)

27.80 (30)

   

RMSEA

(90% CI)

< .001 (< .001, .013)

   

CFI

1.00

   

*p < .05 ** p < .01 *** p < .001


High school graduation. High school graduation was another milestone for which predictors behaved the same way in all three groups (Table 7). Whereas for college qualifications, being Black was a positive predictor, for high school graduation, it was a negative predictor: When controlling for other variables, Black students (regardless of their language backgrounds) had only half the odds (eb = 0. 51) of White students of graduating from high school. Academic capital variables once again were important predictors of graduation: The strongest predictors were college qualifications, 10th grade GPA, and reading test score. Although math test score was not a significant predictor, we believe that this is because college qualifications and math test score were capturing a similar construct, with college qualifications already accounting for high performance in math in this analysis. However, it is important to note that it was college qualifications rather than math test score that had an impact on high school graduation. A subsequent analysis showed that taking Algebra II was the dividing line: The vast majority of the students who took at least Algebra II in high school went on to graduate whereas the chances for graduation were diminished for those who took less than Algebra II (Figure 1).



Table 7. Multigroup Analysis for High School Graduation, Weighted


  

NS

EP

ELL

Predictor

 

eb

eb

eb

Female

 

0.96

0.96

0.96

Asian

 

0.95

0.95

0.95

Black

 

0.51**

0.51**

0.51**

Hispanic

 

1.02

1.02

1.02

Mixed Race

 

0.60

0.60

0.60

Family Income

 

0.99

0.99

0.99

Parental Education

 

1.03

1.03

1.03

Educational Aspirations of Parents

 

1.01

1.01

1.01

10th Grade Student Aspirations

 

1.04

1.04

1.04

College Qualifications

 

1.26***

1.26***

1.26***

10th Grade GPA

 

1.56***

1.56***

1.56***

10th Grade Math Test Score

 

0.89

0.89

0.89

10th Grade Reading Test Score

 

1.48**

1.48**

1.48**

%10th Graders Receiving Free/Reduced Lunch

 

1.03

1.03

1.03

% Minority Students in School

 

1.00

1.00

1.00

% Graduates Going to 4-Yr College

 

1.03

1.03

1.03

 

 




N

 7,721

 6,520

 948

 253

χ2(df)

 19.39 (32)




RMSEA

(90% CI)

 < .001 (< .001, .000)




CFI

 1.00




*p < .05 ** p < .01 *** p < .001


Figure 1. High school graduation rates by highest math course completed, weighted

 

[39_18155.htm_g/00002.jpg]

Note. Non-Academic = basic math; Low Academic = pre-algebra or Algebra I; Middle Academic = Algebra I + geometry; Middle Academic II = Algebra II; Advanced I = Algebra II + Trigonometry or Statistics; Advanced II = pre-calculus; Advanced III = calculus. The black vertical line in the middle of the figure indicates the dividing line between less than Algebra II and Algebra II or above.





Four-year college application. Significant predictors for four-year college application varied across the three language background groups, with similar patterns shared by NPs and EPs but a distinct pattern emerging for ELLs (Table 8). Being Black was a strong positive predictor of four-year college application across the groups (eb = 2.03), after accounting for all other predictors in the model. Being Hispanic, in contrast, was a major negative predictor of ELLs’ college application. Family income was a significant predictor for ELLs’ college application but not for NSs and EPs. This suggests that the issue of college affordability enters the equation for ELLs even at the stage of college application. College qualifications are another predictor that was significant across the groups: For each higher-level math course taken, the odds of applying to a four-year college increased by 23%. Importantly, none of the other academic factors influenced ELLs’ college application whereas for both NSs and EPs, GPA was a major predictor of college application (eb = 1.45 for NSs, and eb = 1.79 for EPs) and the math test score was another significant predictor for EPs (eb  = 1.45). The number of college-bound friends influenced NSs’ college application but not EPs’ and ELLs’. The percentage of the school’s graduates going to four-year colleges mattered for NSs and EPs, but it was less of an influence for ELLs, suggesting that even when many students are going to four-year colleges and the resources exist at the school for going to four-year colleges, ELLs may not be benefitting from them as fully as NSs and EPs.


Table 8. Multigroup Analysis for Four-Year College Application, Weighted


 

 

NS

EP

ELL

Predictor

 

eb

eb

eb

Female

 

0.93

0.93

0.93

Asian

 

1.06

1.06

1.06

Black

 

2.03***

2.03***

2.03***

Hispanic

 

1.14

0.94

0.18*

Mixed Race

 

1.20

4.26**

0.40

Family Income

 

1.01

0.98

1.13*

Parental Education

 

1.06***

1.09*

1.07

Educational Aspirations of Parents

 

1.07*

1.01

1.13

10th Grade Student Aspirations

 

1.15***

1.02

1.06

College Qualifications

 

1.23***

1.23***

1.23***

10th Grade GPA

 

1.45***

1.79***

1.15

10th Grade Math Test Score

 

0.99

1.45*

1.44

10th Grade Reading Test Score

 

1.03

1.03

1.03

# of Friends Going to 4-Yr College

 

1.33***

1.14

1.09

%10th Graders Receiving Free/Reduced Lunch

 

0.99

0.99

0.99

% Minority Students in School

 

1.00

1.01*

1.01

% Graduates Going to 4-Yr College

 

1.21***

1.22*

1.17

     

N

6,709

5,702

824

183

χ2(df)

11.92 (12)

   

RMSEA

(90% CI)

< .001 (< .001, .021)

   

CFI

1.00

   

Note. The predictors in boldface are the ones for which in the best-fitting model we allowed coefficients to vary by language background groups.

*p < .05 ** p < .01 *** p < .001


Four-year college enrollment. For the multigroup analyses of four-year college enrollment, we first used data from all of the available students and ran the multigroup analysis without entering financial aid as a predictor. In the second analysis, we then entered financial aid into the analysis and analyzed data for only those students who applied to four-year colleges.  


In the first analysis, ethnicity played divergent roles for ELLs’ college enrollment: Black ELLs had almost 50% higher odds of enrolling in four-year colleges than White ELLs after accounting for all predictors in the model, but Asian and Hispanic ELLs had only about a third of White ELLs’ odds (Table 9). Students’ aspirations in 10th grade had no effect on the ELLs’ actual four-year college enrollment, suggesting that many circumstances and steps happen along the way that mitigate the impact of ELLs’ initial aspirations. College qualifications and reading test scores were important positive predictors across the groups. But while GPA and the school’s college orientation were significant predictors for NSs and EPs, they were not significant for ELLs.



Table 9. Multigroup Analysis for Four-Year College Enrollment (Without Financial Aid), Weighted


 

 

NS

EP

ELL

Predictor

 

eb

eb

eb

Female

 

1.02

1.02

1.02

Asian

 

1.20

1.33

0.31*

Black

 

1.48***

1.48***

1.48***

Hispanic

 

0.89

1.02

0.34*

Mixed Race

 

1.17

1.17

1.17

Family Income

 

1.03**

1.03**

1.03**

Parental Education

 

1.07***

1.07***

1.07***

Educational Aspirations of Parents

 

1.05*

0.91

1.21

10th Grade Student Aspirations

 

1.11***

1.18*

1.08

College Qualifications

 

1.27***

1.27***

1.27***

10th Grade GPA

 

1.60***

2.00***

1.37

10th Grade Math Test Score

 

1.04

1.04

1.04

10th Grade Reading Test Score

 

1.20***

1.20***

1.20***

# of Friends Going to 4-Yr College

 

1.43***

1.05

1.37

%10th Graders Receiving Free/Reduced Lunch

 

0.98

0.98

0.98

% Minority Students in School

 

1.00

1.00

1.01

% Graduates Going to 4-Yr College

 

1.25***

1.30***

1.25

     

N

7,608

6,437

933

238

χ2(df)

15.69 (18)

   

RMSEA

(90% CI)

< .001 (< .001, .015)

   

CFI

1.00

   

Note. The predictors in boldface are the ones for which in the best-fitting model we allowed coefficients to vary by language background groups.

*p < .05 ** p < .01 *** p < .001


When we considered only those students who had applied to four-year colleges and entered financial aid as a predictor, the picture changed somewhat (Table 10). First of all, when we limited the sample to only those who applied to four-year colleges, the three groups seemed to resemble each other more, with shared patterns. Thus, being an ethnic minority had neither a positive nor negative effect whereas academic capital factors as well as the number of college-bound friends were equally important predictors for all three groups.


At the same time, what was different even with this select group of students is telling. Financial aid mattered to all three groups, boosting the odds of four-year college enrollments by more than 200% at least. However, the impact of financial aid was less on ELLs’ four-year college enrollment than for NSs’ and EPs’. We believe that this is most likely because some ELLs who decide that they cannot afford a four-year college education do not apply in the first place without fully taking into account the possibility of financial aid (see the analysis of four-year college application above). Also, the college-orientation of the school had a smaller influence on ELLs’ decisions to enroll in four-year colleges, once again indicating a smaller impact of the school culture on ELLs’ college access compared to NSs’ and EPs’.



Table 10. Multigroup Analysis for Four-Year College Enrollment (With Financial Aid), Weighted


  

NS

EP

ELL

Predictor

 

eb

eb

eb

Female

 

0.96

0.96

0.96

Asian

 

1.06

1.06

1.06

Black

 

1.11

1.11

1.11

Hispanic

 

0.82

0.82

0.82

Mixed Race

 

1.07

1.07

1.07

Family Income

 

1.06**

1.06**

1.06**

Parental Education

 

1.04*

1.04*

1.04*

Educational Aspirations of Parents

 

1.01

0.82**

1.09

10th Grade Student Aspirations

 

1.04

1.04

1.04

College Qualifications

 

1.22***

1.22***

1.22***

10th Grade GPA

 

1.26***

1.26***

1.26***

10th Grade Math Test Score

 

1.12

1.12

1.12

10th Grade Reading Test Score

 

1.14*

1.14*

1.14*

# of Friends Going to 4-Yr College

 

1.23***

1.23***

1.23***

Financial Aid Offered

 

2.37***

3.92***

2.22

%10th Graders Receiving Free/Reduced Lunch

 

0.93**

1.03

1.00

% Minority Students in School

 

1.00

1.00

1.00

% Graduates Going to 4-Yr College

 

1.16***

1.25*

1.13

     

N

5,095

4,373

627

95

χ2(df)

22.55 (28)

   

RMSEA

(90% CI)

< .001 (< .001, .014)

   

CFI

1.00

   

Note. The predictors in boldface are the ones for which in the best-fitting model we allowed coefficients to vary by language background groups.

*p < .05 ** p < .01 *** p < .001



DISCUSSION


Our analysis of the most recent national representative sample of 10th graders making a transition to PSE indicates that ELLs’ four-year college access is severely limited largely due to their difficulty reaching the critical milestones on the way to four-year institutions.


In answering our first research question, “What proportions of ELLs and non-ELLs enroll in four-year colleges after high school graduation?” we found a large gap between ELLs’ four-year college enrollment and those of non-ELLs. Whereas almost half (45%) of NSs and a third (35%) of EPs advance to four-year colleges directly from high school, only a fifth of ELLs (19%) do so. This result echoes our finding of ELLs’ access rate in NELS:88 (Kanno & Cromley, 2013) and shows that even after a whole new generation of students—there is a 14-year distance between NELS:88 and ELS:2002—access to a four-year college immediately after high school graduation remains beyond reach for most ELLs. In contrast, the difference among the three groups in two-year college access is much smaller, indicating that it is four-year college enrollment that remains an elusive goal for ELLs.


The question then becomes, why is access to a four-year college so difficult for ELLs? Our answers to the second research question, “To what extent do ELLs’ pathways from college aspirations to enrollment differ from those of non-ELLs?” begin to provide some insights into the question. Our college pathway analyses show that it is in the earlier stages of college planning, in particular the aspirations and college qualifications stages, that ELLs drop off from the four-year college pathway at disproportionately high rates. Schneider and Stevenson (1999) in the 1990s found that approximately 70% of students expected to attend four-year colleges and therefore called them “the ambitious generation.” We have shown in our research that NS and EP high school youths of the 2000s are even more ambitious, with more than 70% of them espousing four-year college aspirations. However, this scenario does not apply to ELLs. From the outset, ELLs begin with much more modest aspirations: In 10th grade only 58% of them expect to obtain a bachelor’s degree.


Acquiring minimum college qualifications during high school (e.g., taking Algebra II) is another major hurdle for ELLs. Cabrera and La Nasa (2001) argued that “the defining characteristic of the college enrollee is the acquisition of college qualifications” (p. 23). We would argue, rather, that it is the combination of aspiring to a four-year college and getting college-qualified that is the defining characteristic of the four-year college enrollee. Across the language background groups, those who had four-year college aspirations and became college qualified were far more likely to enroll in four-year institutions than those who failed to reach one or both of these milestones. At the same time, as Table 3 shows, only about half of the ELLs who had four-year college aspirations become college qualified as opposed to two-thirds of EPs and three-quarters of NSs. This underscores past research findings documenting the difficulty ELLs have in accessing college-preparatory courses (Callahan et al., 2010; Harklau, 1994; Hopstock & Stephenson, 1999; Kanno & Kangas, 2014).


College application is another challenging milestone for ELLs. It is remarkable that only 62% of ELLs who originally espoused four-year college aspirations, became college-qualified, and graduated from high school applied to four-year colleges whereas 80% of NSs and 76% of EPs with the same qualifications did. In their study of Chicago Public School (CPS) students, Roderick et al. (2008) observed,


Acceptance is less of a barrier than might be expected; only 8 percent of students who planned to enroll applied to a four-year college and were not accepted. Rather, a larger issue is that many CPS students never face a college acceptance decision because they do not apply to four-year colleges. (p. 33, added emphasis)


This statement applies equally well to ELLs in our study. Those ELLs who have reached all the previous milestones and then applied to four-year colleges go on to enroll in four-year colleges at a comparable rate as EPs. But what is different is that a much smaller group of ELLs apply to four-year colleges in the first place than EPs and NSs. In sum, these patterns strongly suggest that only a small fraction of ELLs reach four-year colleges right after high school graduation because many of them do not apply to four-year colleges in the first place. Furthermore, many of them do not reach the four-year college application milestone because they fail to reach the earlier milestones that lead to college application.


With our college pathway analysis, we have already started to address our third research question, “Which of the critical steps in the pathway to four-year colleges present difficulty to ELLs, and why?” Aspiring to obtain a four-year degree, becoming college-qualified, and applying to four-year colleges are the milestones that are particularly difficult for ELLs to reach.


On the whole, we learned that we know much less about what factors contribute to ELLs’ college access than the factors for NSs and EPs. The predictors we have considered were chosen based on what previous studies had identified as critical predictors. Those predictors accounted for NSs’ college access—and to a lesser extent, EPs’—much better than ELLs’ college access. This is unsurprising since the majority of previous studies in higher education are based on samples that are comprised largely of native speakers of English. Our results suggest that we cannot assume that the same set of variables predict ELLs’ college access.


One consistent pattern we found across many multigroup analyses is that race/ethnicity mattered more for ELLs whereas academic predictors mattered less for them, compared to NSs and EPs. Being non-White was a major negative predictor of ELLs’ college aspirations. This reminds us of Pimentel’s (2010) assertion that “language is color-coded” (p. 26). That is, racial/ethnic minority ELLs are particularly vulnerable to negative stereotyping and low expectations from teachers and counselors. Interestingly, we saw no evidence of the pervasive stereotype of Asian immigrants as the “model minority” (Lee, 1996). Being Asian, when it was a significant predictor, was always a negative predictor for ELLs. Perhaps, when Asian students have limited English proficiency, they are just likely to be deemed “non-college material” as any other racial/ethnic minority ELLs.


In later stages of the college pathway, Black students tend to have an advantage when all other variables are held constant. Across language background groups, when a Black student has the same set of conditions as a White student, it is the Black student who is more likely to be college-qualified, apply to four-year colleges, and enroll in a four-year college. Perhaps when a Black student has the same scores and qualifications as a White student, the prevalent emphasis on diversity across U.S. higher education makes the Black student a more attractive candidate than a White student with the same qualifications. At the same time, as we noted above, it is important to keep in mind that few Black students in reality have the same resources and educational opportunities as White students (e.g., Hirschman & Lee, 2005; Kao & Thompson, 2003). Opportunity to be exposed to high-level courses in high school, for example, is much more limited for Black students than for White students (College Board, 2014; Yonezawa, Wells, & Serna, 2002). Given that the impact of academic capital outweighs the impact of being Black in reaching these milestones, few Black students are in fact in a position to enjoy the “advantage” that our analyses indicated when we controlled for all other variables.


In contrast, even when all other variables are held constant, being a Hispanic ELL remains a negative predictor for later milestones (i.e., college application and enrollment). In other words, there is something about being a Hispanic ELL, as compared to being a White ELL, hindering Hispanic ELLs’ college access that our set of predictors did not capture. Notably, it is the combination of being Hispanic and being an ELL that puts a student at a particularly high risk; among NSs and EPs, being Hispanic was not a significant predictor. Although this is obviously subject to future research, we believe that a few factors are contributing to Hispanic ELLs’ disadvantage. One is the persistent negative stereotypes about Hispanic immigrant students as nonacademic, unwilling to learn English, and lacking in high aspirations, which can lead to low teacher expectations (Gándara & Contreras, 2009; Ruecker, 2013; Wortham, Mortimer, & Allard, 2009). Another factor may be negative perceptions surrounding Hispanic ELLs’ bilingualism. Although in K–12 public school education in the United States, ELLs in general are far more likely to be characterized as limited English proficient than “emergent bilinguals” (García, 2009), hostility towards Hispanic students’ use of Spanish and the denial of their linguistic capital seem particularly strong, which could lead to ELLs’ own sense of self-doubt and lack of confidence in their linguistic identity (Holmes, Fanning, Morales, Espinoza, & Herrera, 2012). Finally, we believe that this is one area in which ELLs’ immigration status plays a distinct role. Eighty-seven percent of undocumented immigrants under 18 years old are Hispanic (Passel & Cohn, 2011). Given that the vast majority of the ELLs in ELS:2002 are first-generation-immigrant students, it is reasonable to assume that undocumented students, if there are any in the sample, are disproportionately concentrated among Hispanic ELLs. As such, the immigration status may be functioning as a hidden factor inhibiting Hispanic ELLs’ college access.


Another critical finding is that academic capital factors play a lesser role for ELLs’ college access. We are not claiming that academic preparation is irrelevant for ELLs’ college access since college qualifications matter to ELLs just as much as to NPs and EPs. However, the influence of other academic capital indicators, especially high school GPA, is less deterministic for ELLs than for the other two groups. This means that ELLs who do well in class are not necessarily those who advance to four-year colleges. The lack of a direct relationship between ELLs’ four-year college access and their high school academic achievement suggests that other mitigating factors, not captured in the current analyses, are at play.


One of such factors may be ELLs’ lack of college knowledge (Conley, 2005; Vargas, 2004): the practical “knowledge about how to prepare for and apply to college” (Vargas, 2004, p. 3). Even when a student is a high achiever, if she does not know how to translate her qualifications into viable academic capital by taking necessary steps for college application, her academic achievement will not necessarily lead to four-year college enrollment. Other relevant factors may include heavy family obligations such as taking care of young siblings and serving as translators for non-English-speaking parents and relatives (Almon, 2010), the preconceived notion that four-year colleges are only for native speakers (Kanno & Varghese, 2010), and the desire to become financially independent fast (Harklau, 2013). These are not factors that have been included in the analysis of college access for the general population but may be highly relevant for ELLs’ college access.


The impact of financial aid for four-year college enrollment is less strong for ELLs than for NSs and EPs, which seems contradictory at a first glance. Given that ELLs are on average poorer than non-ELLs (Table 1), one would think that the availability of financial aid is particularly critical for ELLs’ ability to enroll in four-year colleges. We believe that the explanation for this apparent contradiction lies in ELLs’ and their families’ underdeveloped understanding of the financial aid system and their failure to factor financial aid into their college planning. Low-income ELLs who assume that they cannot afford a four-year college education may simply opt out at the application stage whereas NSs and EPs may wait until they receive the offers of financial aid packages to decide whether or not to attend a four-year college. The fact that family income was a significant predictor for ELLs’ college application but not for NSs and EPs supports this explanation.


Finally, we noted in relation to Table 1 that ELLs tend to attend schools that send smaller proportions of their graduates to four-year colleges than EPs and NSs. And yet, apart from college qualifications, this variable seems to have a less direct impact on ELLs than on NSs and EPs. We believe that this is an important finding that needs further investigation. First of all, ELLs tend to be segregated in schools where four-year college attendance is not the norm, which limits their opportunity to be college-qualified (Adelman, 2006). However, even when they do attend schools where many students routinely advance to four-year schools, ELLs may not be included as part of the expectation of college enrollment. As a consequence, ELLs may not be as much encouraged as non-ELLs to take advanced-level courses or given as much guidance on four-year college enrollment.


LIMITATIONS AND IMPLICATIONS


LIMITATIONS


This study utilized one of the newest nationally representative datasets and employed statistically rigorous methods to examine ELLs’ pathways to four-year colleges. However, there are some limitations to the study. First, as discussed at the beginning of the study, this study used only the data of students who participated in all of the first three waves of data collection. Our analytical sample therefore is biased towards those students who are on the whole better off in terms of the various forms of capital they possess. The tendency to capture more resourceful and higher-achieving students is particularly pronounced in our ELL sample for the reasons we already discussed. Consequently, the picture of high school seniors going to college that we painted in this study is likely to be somewhat more optimistic than if a truly nationally representative sample had completed all three waves. Furthermore, although the gap between ELLs’ college access rate and those of NSs and EPs is staggering, the ELLs represented in this study are likely to be more advantaged than the ELL population as a whole.


Second, given the heterogeneity of the ELL population, it might have been revealing to disaggregate the ELL sample in terms of some additional characteristics, such as their legal status in the U.S., the length of enrollment in ESL programs, and their immigrant generation status. However, the already small sample size of ELLs (n = 490) did not allow such analyses; further disaggregation of this group would have caused convergence problems. Moreover, even if the sample size had been adequate, the information available in ELS:2002 would not have allowed reliable categorization of students into subgroups. For example, NCES did not (most likely intentionally) collect any information on students’ legal status in the United States. Similarly, although some information was available to identify students’ immigrant generation status, a large portion of ELLs were missing data on these variables.


IMPLICATIONS


There are several policy, pedagogical, and ethical implications we can draw from the findings of this study. One clear implication is that in order to enable more ELLs to reach four-year colleges, we should make a targeted effort to support them in the early stages of the process, especially at the aspirations and college qualifications stages. Espousing four-year college aspirations early on is important because if students have clear four-year college aspirations as they enter high school, they are far more likely to seek relevant information and take the necessary steps (Cabrera & La Nasa, 2001). This suggests that efforts to raise awareness about college should begin before high school, in middle school and even in elementary school, so that by the time students enter high school, they have clear PSE goals.


Enabling ELLs to acquire necessary college qualifications is another area that needs more attention. There are multiple reasons why ELLs have limited access to high-level courses in high school: For example, the need to secure time for ESL instruction may not allow enough time on their roster to take high-level courses that are desirable for college application but are not required for high school graduation (Callahan et al., 2010). However, we believe that the root cause of ELLs’ limited access to high-level courses is that high school educators do not see ELLs as serious contenders for four-year colleges (Callahan, 2005). In the area of mathematics education, Martin (2009) argues that research on racial achievement gaps in mathematics tends to characterize African American, Latino, and Native American students as mathematically illiterate while masking the “long-standing inequitable patterns of access to mathematical opportunities” (p. 314) that privilege White students. Similarly, we would argue that ELLs’ limited access to a four-year college has generally been understood as a natural consequence of their low academic achievement and limited English proficiency; yet, reduced access to college-preparatory courses and the knowledge of the importance of such coursework, which have prevented ELLs from building meaningful academic credentials, have been largely ignored (Kanno & Kangas, 2014). Given the critical importance of college qualifications for college access, we would argue that preventing ELLs from taking advanced-level college preparatory courses the way NSs and EPs routinely do amounts to a form of discrimination against a particular group of people based on their linguistic background.


Providing better financial aid information to ELLs and their parents would also go a long way in increasing ELLs’ likelihood of applying to four-year colleges, and hence their likelihood of four-year college enrollment. Decoding the financial aid system is challenging even for native speakers (Roderick et al., 2008); for families consisting of limited English proficient speakers, the challenge is magnified (Kanno & Grosik, 2012). Therefore, providing information on financial aid early on in high school and stressing that the gap between what the family can afford and the sticker price for a college does not mean the end of a four-year college dream would be important components of college access support for ELLs.


We also learned from our findings that the combination of racial/ethnic minority and ELL status puts students at a particular disadvantage in college planning. This particular finding raises questions about our deep-seated prejudice involving the relationship between race/ethnicity and language proficiency, that is, what Pimental (2010) calls our “racially informed language expectations” (p. 25). When we see a racial/ethnic minority student who is also an ELL, her race/ethnicity may highlight her limited English proficiency while her limited English proficiency may also accentuate her racial/ethnic minority profile, both of which tend to bring out a deficit orientation in educators (e.g., Crosnoe, 2006; Gándara & Contreras, 2009). Pedagogically, our finding clearly points to the need to focus our intervention on racial/ethnic minority ELLs’ college planning. However, above all, we need to challenge our own racially informed language expectations and assumptions about academic capability that accompany these expectations.  


Lastly, for future research, our findings strongly point to the importance of not assuming that the same predictors influence all language background groups’ college planning the same way and the need to identify factors and conditions that influence ELLs’ college planning. Based on the emerging research, we believe that ELLs’ lack of college knowledge is one such factor. However, there may be a host of other factors yet to be identified. Qualitative studies that closely examine ELLs’ college planning would be particularly effective in identifying such factors. Once new factors are identified in qualitative studies, we then will be able to test them using large-scale datasets such as ELS:2002 and BPS:2004.


Acknowledgment


We are grateful to Dr. Judith Stull for facilitating our access to ELS:2002.

Notes


1. However, it is important to note that in in our analytical sample, 80% (weighted) of ELLs are first-generation, foreign-born students. This is because the majority of U.S.-born students who might have started their K–12 education as ELLs are reclassified as English-proficient by the time they reach high school (Kim, 2011). At the high school level, therefore, ELLs tend to be foreign-born students who have arrived in the United States relatively recently.

2. The DACA offers “eligible immigrant youth work authorization and temporary relief from deportation” (Batalova et al., 2013, p. 1).

3. Following the policy required by the National Center for Education Statistics (NCES), which collected the ELS:2002 data, we are rounding the numbers to the nearest 10 when reporting unweighted sample sizes in order to avoid the identification of individual students.

4. The four choices in the BY survey for the self-rating of four English skills (listening, speaking, reading, and writing; BYS70a-d) were “Very well,” “Well,” “Not well,” and “Not at all.” The bottom two categories (“Not well,” and “Not at all”) were used as markers of limited English proficiency.

5. Using teacher identification alone seems to overidentify ELLs (i.e., there were many students who had no signs of being an ELL other than the teacher identification). Therefore, we used teacher identification only for students who self-identified as NNS.

6. Taking bilingual education classes is counted as an index of limited English proficiency because the vast majority of bilingual education programs at the high school level are transitional bilingual education programs serving ELLs. Developmental bilingual programs and two-way immersion programs, which include English-proficient bilingual students, largely happen at the elementary level (Center for Applied Linguistics, 2011).

7. Similarly, initial screening suggested that all the other categorical variables we used as independent variables in multigroup analyses, except for race/ethnicity, could be treated as continuous variables. Thus, the categorical variables that we treated as continuous variables in regression analyses were: family income, parental education, parental educational aspirations for the child, students’ own educational aspirations, highest math course taken, number of friends going to four-year college, percent of 10th graders at school receiving free/reduced lunch, and percent of graduates from the school who went to four-year colleges upon graduation.

8. Item Response Theory, or IRT, is a modern approach to validating measures, where student ability and item difficulty are estimated simultaneously. The theta score for each item represents the ability level a student must have in order to have a 50% chance of answering the item correctly. Student ability can be calculated for each student from the theta values of the questions that student answered correctly.


References


Adelman, C. (2006). The toolbox revisited: Paths to degree completion from high school through college. Washington, DC: U.S. Department of Education, Office of Educational Research and Involvement.


Almon, P. C. (2010). English language learner engagement and retention in a community college setting. (Doctoral dissertation). Retrieved from ProQuest Dissertations and Thesis database. (UMI No. 3408684).


Batalova, J., Hooker, S., & Capps, R. with Bachmeir, D., & Cox, E. (2013). Deferred action for childhood arrivals at the one-year mark: A profile of current eligible youth and applicants. Retrieved from http://www.migrationpolicy.org/pubs/CIRbrief-DACAatOneYear.pdf


Battey, D. (2013). Access to mathematics: “A possessive investment in Whiteness.” Curriculum Inquiry, 43(3), 332–359. doi:10.1111%2Fcuri.12015


Bennici, F. J., & Strang, W. E. (1995). An analysis of language minority and limited English proficient students from NELS:88. Washington, DC: Office of Bilingual Education and Minority Language Affairs, U.S. Department of Education.


Berkner, L., & Chavez, L. (1997). Access to postsecondary education for the 1992 high school graduates (NCES 98-105). Washington, DC: National Center for Educational Statistics, U.S. Department of Education.


Bourdieu, P. (1977). Cultural reproduction and social reproduction. In J. Karabel & A. H. Halsey (Eds.), Power and ideology in education (pp. 487–511). New York, NY: Oxford University Press.


Bourdieu, P. (1986). The forms of capital. In J. G. Richardson (Ed.), Handbook of theory and research for the sociology of education (pp. 241–258). Westport, CT: Greenwood Press.


Bourdieu, P. (1987). What makes a social class? On the theoretical and practical existence of groups. Berkeley Journal of Sociology, 32, 1–8.


Bourdieu, P. (1991). Language and symbolic power. J. B. Thompson (Ed.), (G. Raymond & M. Adamson, Trans.). Cambridge, MA: Harvard University Press.


Bowen, W. G., Kurzwell, M. A., & Tobin, E. M. (2005). Equity and excellence in American higher education. Charlottesville, VA: University of Virginia Press.


Bunch, G. C., & Endris, A. K. (2012). Navigating “open access” community colleges: Matriculation policies and practices for US-educated linguistic minority students. In Y. Kanno & L. Harklau (Eds.), Linguistic minority students go to college: Preparation, access, and persistence (pp. 165–183). New York, NY: Routledge.


Cabrera, A. F., Burkum, K. R., & La Nasa, S., M. (2005). Pathways to a four-year degree: Determinants of degree completion among socioeconomically disadvantaged students. In A. Seidman (Ed.), College student retention: Formula for student success: Formula for student success (pp. 155–214). Westport, CT: Praeger.


Cabrera, A. F., & La Nasa, S. M. (2000). Three critical tasks America's disadvantaged face on their path to college. New Directions for Institutional Research, 2000(107), 23–29. Doi:10.1002/ir.10702


Cabrera, A. F., & La Nasa, S. M. (2001). On the path to college: Three critical tasks facing America's disadvantaged. Research in Higher Education, 42(2), 119–149. doi:10.1023/A:1026520002362


Callahan, R. M. (2005). Tracking and high school English learners: Limiting opportunity to learn. American Educational Research Journal, 42(2), 305–328. doi:10.3102/00028312042002305


Callahan, R. M., & Gándara, P. (2004). On nobody's agenda: Improving English-language learners' access to higher education. In S. Michael (Ed.), Teaching immigrant and second-language students: Strategies for success (pp. 107–127). Cambridge, MA: Harvard Education Press.


Callahan, R. M., Wilkinson, L., & Muller, C. (2010). Academic achievement and course taking among language minority youth in U.S. schools: Effects of ESL placement. Educational Evaluation and Policy Analysis, 32(1), 84–117. doi:10.3102/0162373709359805


Carnevale, A. P., Rose, S. J., & Cheah, B. (2011). The college payoff: Education, occupations, lifetime earnings. Retrieved from https://cew.georgetown.edu/report/the-college-payoff/


Center for Applied Linguistics. (2011). Grade levels served in two-way immersion programs, by school sites. Retrieved from http://www.cal.org/twi/directory/grades.htm


Cherng, H.-Y. S., Calarco, J. M., & Kao, G. (2013). Along for the ride: Best friends' resources and adolescents' college completion. American Educational Research Journal, 50(1), 76–106. doi:10.3102/0002831212466689


College Board. (2014). The 10th annual AP report to the nation. Retrieved from http://media.collegeboard.com/digitalServices/pdf/ap/rtn/10th-annual/10th-annual-ap-report-to-the-nation-

single-page.pdf


Conley, D. T. (2005). College knowledge: What it really takes for students to succeed and what we can do to get them ready. San Francisco, CA: Jossey-Bass.


Crosnoe, R. (2006). Health and the education of children from race/ethnic minority and immigrant families. Journal of Health and Social Behavior, 47, 77–79. doi:10.1177/002214650604700106


De Cohen, C. C., Deterding, N., & Clewell, B. C. (2005). Who’s left behind?: Immigrant children in high and low-LEP schools. Retrieved from http://www.urban.org/UploadedPDF/411231_whos_left_behind.pdf


Deil-Amen, R., & Turley, R. L. (2007). A review of the transition to college literature in sociology. Teachers College Record, 109(10), 2324–2366.


Education Week. (2009, January). A distinctive population. Quality Counts 2009: Portraits of a Population, How English Language-Learners Are Putting Schools to the Test, 15.


Estrada, P. (2014). English learner curricular streams in four middle schools: Triage in the trenches. The Urban Review. Advance online publication. doi:10.1007/s11256-014-0276-7.


Fry, R. (2008). The role of schools in the English language learner achievement gap. Washington, DC: Pew Hispanic Center. Retrieved from http://pewhispanic.org/files/reports/89.pdf


Gándara, P., & Contreras, F. (2009). The Latino education crisis: The consequences of failed social policies. Cambridge, MA: Harvard University Press.


García, O. (2009). Emergent bilinguals and TESOL: What’s in a name? TESOL Quarterly, 43(2), 322–326. doi:10.1002/j.1545-7249.2009.tb00172.x


García, O., Kleifgen, J. A., & Falchi, L. (2008). From English language learners to emergent bilinguals. Equity Matters: Research Review No. 1. Retrieved from http://files.eric.ed.gov/fulltext/ED524002.pdf


Harklau, L. (1994). Tracking and linguistic minority students: Consequences of ability grouping for second language learners. Linguistics and Education, 6, 217–244. doi:10.1016/0898-5898(94)90012-4


Harklau, L. (2013). Why Izzie didn’t go to college: Choosing work over college as Latina feminism. Teachers College Record, 115(1), 1–32.


Hirschman, C., & Lee, J. (2005). Race and ethnic inequality in educational attainment in the United States. In M. Rutter & M. Tienda (Eds.), Ethnicity and causal mechanisms (pp. 107–138). Cambridge, UK: Cambridge University Press.


Holmes, M., Fanning, C., Morales, A., Espinoza, P., & Herrera, S. (2012). Contextualizing the path to academic success: Culturally and linguistically diverse students gaining voice and agency in higher education. In Y. Kanno & L. Harklau (Eds.), Linguistic minority students go to college: Preparation, access, and persistence (pp. 201–219). New York, NY: Routledge.


Hopstock, P. J., & Stephenson, T. G. (2003). Descriptive study of services to LEP students and LEP students with disabilities. Special topic report #2: Issues in studying learning outcomes for LEP students. Washington, DC: Office of English Language Acquisition, U.S. Department of Education.


Hossler, D., & Gallagher, K. S. (1987). Studying college choice: A three-phase model and the implication for policy makers. College and University, 63(2), 207–221.


Hu, L.-T., & Bentler, P. M. (1999). Cutoff criteria for fit idexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55. doi:10.1080/10705519909540118


Kanno, Y., & Cromley, J. (2013). English language learners’ access to and attainment in postsecondary education. TESOL Quarterly, 47(1), 89–121. doi:10.1002/tesq.49


Kanno, Y., & Grosik, S. A. (2012). Immigrant English learners’ transitions to university: Student challenges and institutional policies. In Y. Kanno & L. Harklau (Eds.), Linguistic minority students go to college: Preparation, access, and persistence (pp. 130–147). New York, NY: Routledge.


Kanno, Y., & Harklau, L. (Eds.) (2012). Linguistic minority students go to college: Preparation, access, and persistence. New York, NY: Routledge.


Kanno, Y., & Kangas, S. E. N. (2014). “I’m not going to be, like, for the AP’: English language learners access to advanced college-preparatory courses in high school. American Education Research Journal, 51(5), 848-878. doi:10.3102/0002831214544716


Kanno, Y., & Varghese, M. (2010). Immigrant English language learners’ challenges to accessing four year college education: From language policy to educational policy. Journal of Language, Identity, and Education, 9(5), 1–19. doi:10.1080/15348458.2010.517693


Kao, G., & Thompson, J. S. (2003). Racial and ethnic stratification in educational achievement and attainment. Annual Review of Sociology, 29, 417–442. doi:10.1146/annurev.soc.29.010202.100019


Kao, G., & Tienda, M. (1995). Optimism and achievement: The educational performance of immigrant youth. Social Science Quarterly, 76(1), 1–19.


Kim, E., & Díaz, J. (2013). Immigrant students and higher education. ASHE Higher Education Report, 38(6), 1–170.


Kim, J. (2011). Relationships among and between ELL status, demographic characteristics, enrollment history, and school persistence (CRESST Report 810). Los Angeles, CA: University of California, National Center for Research on Evaluation, Standards, and Student Testing (CRESST).


Lee, S. J. (1996). Unraveling the “model minority” stereotype: Listening to Asian American youth. New York, NY: Teachers College Press.


Martin, D. B. (2009). Researching race in mathematics education. Teachers College Record, 11(2), 295–338.


McDonough, P. M. (1997). Choosing colleges: How social class and schools structure opportunity. Albany, NY: State University of New York Press.


Menken, K., & Kleyn, T. (2010). The long-term impact of subtractive schooling in the educational experiences of secondary English language learners. International Journal of Bilingual Education and Bilingualism, 13(4), 399–417. doi:10.1080/13670050903370143


Mosqueda, E. (2011). Teacher quality, academic tracking and the mathematics performance of Latino English learners. In K. Téllez, J. N. Moschkovich, & M. Civil (Eds.), Latina/os and mathematics education: Research on learning and teaching in classrooms and communities (pp. 315–339). Charlotte, NC: Information Age Publishing.


Muthén, L. K., & Muthén, B. O. (1998–2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Muthén & Muthén.


National Center for Education Statistics. (n.d.). NAEP data explorer. Retrieved from http://nces.ed.gov/nationsreportcard/naepdata/


National Center for Public Policy and Higher Education. (2002). Losing ground: A national status report on the affordability of American higher education. Retrieved from http://www.highereducation.org/reports/losing_ground/ar.shtml


National Immigration Law Center. (2014). Basic facts about in-state tuition for undocumented immigrant students. Retrieved from http://www.nilc.org/basic-facts-instate.html


Nuñez, A.-M., & Cuccaro-Alamin, S. (1998). First-generation students: Undergraduates whose parents never enrolled in postsecondary education. Retrieved from http://nces.ed.gov/pubs98/98082.pdf


Nuñez, A.-M., & Kim, D. (2012). Building a multicontextual model of Latino college enrollment: Student, school, and state-level effects. Review of Higher Education, 35(2), 237–263. doi: 10.1353/rhe.2012.0004


Nuñez, A.-M., & Sparks, P. J. (2012). Who are linguistic minority students in higher education; An analysis of the Beginning Postsecondary Students Study 2004. In Y. Kanno & L. Harklau (Eds.), Linguistic minority students go to college: Preparation, access, and persistence (pp. 110–129). New York, NY: Routledge.


Passel, J. S., & Cohn, D. (2011). Unauthorized immigrant population: National and state trends, 2010. Retrieved from http://www.pewhispanic.org/files/reports/133.pdf


Perna, L. W., & Thomas, S. L. (2008). Theoretical perspectives on student success: Understanding the contributions of the disciplines. ASHE Higher Education Report, 34(1), 1–87. doi:10.1002/aehe.3401


Perna, L. W., & Titus, M. A. (2005). The relationship between parental involvement as social capital and college enrollment: An examination of racial/ethnic group differences. Journal of Higher Education, 76(5), 485–518. doi:10.1353/jhe.2005.0036


Pimental, C. (2010). Color-coded bilingualism. In L. D. Soto & H. Kharem (Eds.), Teaching bilingual/bicultural children: Teachers talk about language and learning (pp. 19–27). New York, NY: Peter Lang.


Razfar, A., & Simon, J. (2011). Course-taking patterns of Latino ESL students: Mobility and mainstreaming in urban community colleges in the United States. TESOL Quarterly, 45(4), 595–627. doi:10.5054/tq.2011.268060


Roderick, M., Nagaoka, J., Coca, B., & Moeller, E. (2008). From high school to the future: Potholes on the road to college. Retrieved from http://ccsr.uchicago.edu/downloads/1835ccsr_potholes_summary.pdf


Rodriguez, G. M., & Cruz, L. (2009). The transition to college of English learner and undocumented immigrant students: Resource and policy implications. Teachers College Record, 111(10), 2385–2418.


Ruecker, T. (2013). High-stakes testing and Latino/a students: Creating a hierarchy of college readiness. Journal of Hispanic Higher Education, 12(4), 303–320. doi:10.1177/1538192713493011


Ruecker, T. (2012). Writing across institutions: Studhing the curricular and extracurricular journeys of Latino/a students transitioning from high school to college. (Doctoral dissertation). Retrieved from ProQuest Dissertations and Thesis database. (UMI No. 3512021).


Schneider, B., & Stevenson, D. (1999). The ambitious generation: America’s teenagers, motivated but directionless. New Haven, CT: Yale University Press.


Snijders, T. A. B., & Bosker, R. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London, UK: Sage.


Swail, W. S. (2000). Preparing America's disadvantaged for college: Programs that increase college opportunity. New Directions for Institutional Research, 107, 85–101. doi:10.1002/ir.10706


Umansky, I. M., & Reardon, S. F. (2014). Reclassificaton patterns among Latino English learner students in bilingual, dual immersion, and English immersion classrooms. American Education Research Journal, 51(5) 879-912. doi:10.3102/0002831214545110


U.S. Department of Education. (2006). Building partnerships to help English language learners. Retrieved from http://www2.ed.gov/nclb/methods/english/lepfactsheet.pdf


U.S. Department of Education Advisory Committee on Student Financial Assistance. (2006). Mortgaging out future: How financial barriers to college undercut America's global competitiveness. Retrieved from http://www2.ed.gov/about/bdscomm/list/acsfa/mof.pdf


Vargas, J. H. (2004). College knowledge: Addressing infomation barriers to college. Retrieved from http://www.teri.org/pdf/research-studies/CollegeKnowledge.pdf


Walpole, M. (2007). Economically and educationally challenged students in higher education: Access to outcomes. ASHE Higher Education Report, 33(3), 1–113. doi:10.1002/aehe.3303


Wang, J., & Goldschmidt , P. (1999). Opportunity to learn, language proficiency, and immigrant status effects on mathematics achievement. The Journal of Educatioanal Research, 93(2), 101–111. doi:10.2307/27542253


Wolf, M. K., Herman, J. L., Bachman, L. F., Bailey, A. L., & Griffin, N. (2008). Recommendations for assessing English language learners: English language proficiency measures and accommodation uses (CRESST Report 737). Retrieved from http://www.cse.ucla.edu/products/reports/R737.pdf


Wortham, S., Mortimer, K., & Allard, E. (2009). Mexicans as model minorities in the new Latino diaspora. Anthropology and Education Quarterly, 40(4), 388–404. doi:10.1111/j.1548-1492.2009.01058.x


Wright, W. (2010). Foundations for teaching English language learners: Research, theory, policy, and practice. Philadelphia, PA: Caslon Publishing.


Yonezawa, S., Wells, A. S., & Serna, I. (2002). Choosing tracks: “Freedom of choice” in detracking schools. American Educational Research Journal, 39(1), 37–67. doi:10.3102/00028312039001037


Zehler, A., Fleischman, H. L., Hostock, P. J., Stephenson, T. D., Pendzick, M. L., & Sapru, S. (2003). Descriptive study of services to LEP students in and ELP students with disabilities (Volume 1, Research Report). Retrieved from http://ncela.net/files/rcd/BE021199/special_ed4.pdf



APPENDIX

List of Variables


Variable

      Type

Source

Dependent variable

  

College enrollment as of 2006

Categorical (1-5)

Derived from F2PS1LVL and F2EDLEVL

Aspire 4 yr

Dichotomous

Derived from BYSTEXP

College qualifications

Dichotomous

Derived from F1RMAPIP

High school graduation

Dichotomous

Derived from F2HSSTAT

Apply 4 yr

Dichotomous

Derived from F2PSAPSL

Enroll 4 yr

Dichotomous

Derived from F2PS1LVL

Independent variable

  

ELL

Dichotomous

Derived from:

BYSLANG, BYS70A-D, BYTE12B, BYTM12B, and high school transcripts

EP

Dichotomous

Derived from:

BYSLANG, BYS70A-D, BYTE12B, BYTM12B, and high school transcripts

NS

Dichotomous

Derived from BYSLANG

Sex

Dichotomous

BYSEX

Race

Categorical (1-5)

Derived from BYRACE

Family income

Categorical (1-13)

BYINCOME

Parental education

Categorical (1-8)

F1PARED

Parents’ educational aspirations for child

Categorical (1-7)

BYPARASP

Students’ aspirations

Categorical (1-7)

BYSTEXP

10th grade GPA

Continuous

F1RGP10

Math course-taking

Categorical (1-8)

F1RMAPIP

Math test score

Continuous

BYTXMTH

Reading test score

Continuous

BYTXRTH

Number of friends going to 4 yr colleges

Categorical (1-5)

F1S65D

Financial aid offered

Dichotomous

Derived from F2PS1AID

School % free/reduced price lunch

Categorical (1-7)

BY10FLP

School % minority

Continuous

CPO02PMIN

School % graduates going to 4-yr colleges

Continuous

F1A19A

Panel weight

 

F2BYWT





Cite This Article as: Teachers College Record Volume 117 Number 12, 2015, p. 1-44
https://www.tcrecord.org ID Number: 18155, Date Accessed: 10/16/2021 7:36:08 AM

Purchase Reprint Rights for this article or review
 
Article Tools
Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Yasuko Kanno
    Temple University
    E-mail Author
    YASUKO KANNO is an associate professor of TESOL in the College of Education at Temple University. Her research focuses on linguistic minority students’ educational opportunities, especially their access to and degree attainment in postsecondary education. Her recent publications include “’I’m not going to be, like, for the AP’: English language learners’ limited access to advanced college-preparatory courses in high school” (American Educational Research Journal, 2014), and Linguistic Minority Students Go to College (Routledge, 2012).
  • Jennifer Cromley
    University of Illinois at Urbana Champaign
    E-mail Author
    JENNIFER G. CROMLEY is an associate professor of Educational Psychology at the College of Education at the University of Illinois at Urbana Champaign. Her research focuses on two areas: achievement and retention of STEM students in higher education and reading comprehension of illustrated scientific text. Some of her recent work on these topics has been published in Learning and Instruction, the Journal of Educational Psychology, and Contemporary Educational Psychology.
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS