Differential Returns to Academic Preparation for English Learners: Evidence from New York City

Abstract

This article reconsiders the role of prior academic preparation as it relates to English learners’ access to advanced coursework in high school. Using longitudinal data from New York City (n = 178,155), we test a set of interactions between middle school preparation—in the form of math achievement and advanced math and science coursework—and subgroups of English learners (ELs). We find that for all ELs, both components of prior preparation are significantly less predictive of college-preparatory STEM course-taking than for comparable non-ELs. We also find that ELs are less likely to benefit from enrollment in schools where the average prior preparation is higher. These findings problematize the common practice of controlling for prior achievement without permitting the effects to vary by subgroup—an approach that tends to overstate the role of prior academic preparation in explaining course-taking disparities in high school. This study thus provides evidence for a necessary analytic adjustment in a growing area of research, as well as the need for more attention on the high school transition for ELs’ access to STEM coursework.

Keywords

multilingual learners English learners academic tracking STEM courses middle school high school multilevel modeling

Efforts to improve postsecondary readiness in STEM have focused on establishing college-going math and science trajectories as early as middle school (e.g., Attewell & Domina, 2008), with the logic that early access facilitates later participation in these hierarchical subjects. Multilingual students classified as English learners (ELs)¹ appear to be especially at risk of exclusion in STEM, with limited access to algebra in Grade 8 and college-preparatory courses, such as chemistry and calculus, in high school (e.g., Johnson, 2019; Thompson, 2017). A number of recent studies have concluded that prior achievement explains some or all of the disparities in coursework for ELs in comparison with their non-EL peers (e.g., Hanson et al., 2016; Johnson, 2019; Umansky, 2016). However, these studies have largely assumed that prior preparation has the same relationships with future outcomes for ELs as it does for non-ELs. We formally test this assumption in the present study, investigating the extent to which math and science preparation in middle school differentially predict college-preparatory STEM coursework for different subgroups of ELs, in comparison with non-ELs.

Background

Evidence across states indicates that ELs are much less likely than their non-EL peers to take part in accelerated math sequences, with limited access to Algebra I in eighth grade (Hanson et al., 2016; Thompson, 2017; Umansky, 2016a) and notably lower participation in college-preparatory topics such as chemistry and calculus in high school (National Center for Education Statistics, 2022). These trends follow a broader pattern of documented barriers to core academic content for English learners, including concentration in schools with few advanced offerings, default school tracking practices, inappropriate assignment to academic intervention, poor instruction, and outright exclusion from particular content areas (see National Academies of Sciences, Engineering, and Medicine [NASEM], 2018 for a review; also Dabach, 2015; Iatarola et al., 2011; Kanno & Kangas, 2014; Umansky, 2016a). The consequences of early tracking may be particularly acute for ELs in math and science because the hierarchical content means that disparities tend to compound over time (Finkelstein et al., 2012; Riegle-Crumb, 2006). However, recent evidence on how ELs’ prior academic experiences actually translate into longer term coursework trajectories has been mixed.

A handful of recent studies have argued that prior academic achievement (i.e., middle school test scores) generally accounts for the large observed course-taking differences between ELs and non-ELs in high school (e.g., Hanson et al., 2016; Johnson, 2019). Hanson and colleagues (2016) found that although ELs in Washington State took fewer courses beyond Algebra II than their never-EL peers, these differences disappeared after controlling for achievement in eighth grade achievement and grade point average (GPA) in ninth grade. Johnson (2019) found that after controlling for eighth grade achievement, all subgroups of ELs were predicted to take more content area courses than similar non-ELs. Umansky (2016a) reports on corroborating evidence in middle schools; controlling for elementary exam scores attenuated differences between ELs and non-ELs in advanced course-taking in grades 6–8, though it did not explain outright exclusion from content areas. Such findings imply that ELs’ difficulty in accessing advanced portions of secondary curricula may have more to do with compounding effects of earlier exclusion than with problematic course placement policies later on. However, these studies have generally assumed that prior achievement has the same set of relationships with future outcomes for all students, as represented by a main effect of test scores in the analytic models. Given the variety of documented barriers to ELs’ coursework at both the student and school level (e.g., NASEM, 2018), there are reasons to be skeptical of this assumption.

Accounting for subgroup variation in the effects of prior preparation has been more common in work featuring course-taking as the predictor of interest. Callahan and Humphries (2016) found that when immigrant ELs participated in advanced coursework in high school, they received somewhat less benefit to their college enrollment outcomes than comparable immigrant students who were not ELs. Long and colleagues’ work (2009, 2012) on course-taking and college-going also found that advanced coursework had differential associations with outcomes by subgroup, though whether marginalized groups received more or less benefit depended on the specific outcome. Perhaps most relevant to the current study, Riegle-Crumb’s (2006) study of math trajectories by race/ethnicity and gender found that even when Black and Latino young men began their math coursework at the same level as other students, they tended not to advance as far through the curriculum. She argued that such differential returns to prior course-taking indicated ongoing disruptions in students’ coursework.

Current Study

Our paper is the first that we know of to investigate differences in returns to middle school preparation for English learners. Because we know that English learners face barriers both within schools as a function of tracking and between schools as a function of sorting, we also examine whether differential associations are due to individual or school-level factors. Our research questions are:

To what extent do math and science course-taking and achievement in middle school differentially predict college-preparatory math and science course-taking in high school for ELs in comparison to non-ELs?

To what extent do the high school averages of prior math and science course-taking and achievement differentially predict college-preparatory math and science course-taking for ELs in comparison to non-ELs?

For the sake of clarity, we refer to high school STEM coursework taken in middle school as advanced, and to advanced STEM subjects typically taken late in high school as college-preparatory.

Method

Policy Context

In New York State, advanced STEM course-taking in middle school is tantamount to early engagement in the high school Regents curriculum and may include taking Algebra I in eighth grade, and Living Environment and/or Earth Science (typical high school courses) in seventh and eighth grade. Although less common in middle school, students may take the Regents exit exams in these subjects even without the associated coursework, which would also permit them to move directly into later courses in the sequence. Also pertinent to the current study, New York City’s robust high school choice process means a high level of student sorting according to prior achievement and other background characteristics (e.g., Corcoran & Baker-Smith, 2018). Students with higher levels of achievement and preparation in middle school tend to enroll in high schools with similarly high-achieving classmates and, as a result, are likely to have wider access to college-preparatory courses in subsequent years. This kind of sorting may pose particular challenges for English learners, because the language of the achievement test in middle school may not allow them to demonstrate what they are capable of academically (e.g., Solórzano, 2008). There is also some evidence that English learners and their families do not participate in the school choice process in the same way as proficient English speakers (Mavrogordato & Harris, 2017), and that they are therefore less likely to enroll in high-performing schools. We attempt to address these school-level composition effects through our analytic approach.

Sample and Subgroups

This study uses New York City (NYC) Public Schools’ data sourced from the Research Alliance for New York City Schools. The work was approved by the ethics boards of the authors’ home institutions. We sampled three cohorts of first-time ninth graders enrolled in a NYC public school in the fall of 2013, 2014, or 2015 and who also had middle school records in NYC (n = 170,955).

In line with prior work (e.g., Johnson, 2019; Umansky, 2016a) and official NYC Public Schools (NYCPS) guidance (NYC Department of Education [NYCDOE], 2023), we examined six subgroups, defined in ninth grade (Table 1). The three subgroups of non-ELs were: English-only students (EOs; n = 97,342), who spoke only English at home; Initially Fluent English Proficient students (IFEPs; n = 24,819), who spoke another language at home but who were English proficient at school entry; and Reclassified Fluent English Proficient students (RFEPs; n = 38,711), who entered school as ELs but became English proficient prior to their ninth grade year. The three subgroups of ELs were: newcomer ELs (n = 8,929), with three or fewer years of language service prior to high school; developing ELs (n = 4,017), with four to six years of service; and long-term ELs (LTELs; n = 4,337), with seven or more years of service.

Table 1.

Selected Sample Characteristics by Student Subgroup.

	EO	IFEP	RFEP	New-EL	Dev-EL	LTEL
N	97,342	24,819	38,711	8,929	4,017	4,337
Female	50.6	51.9	49.7	45.3	41.0	38.5
Asian	8.4	32.2	28.8	26.1	26.0	7.8
Black, Non-Latine	46.3	4.7	4.5	8.4	7.8	4.8
Latine	27.1	47.3	56.6	55.1	59.4	83.9
White, Non-Latine	16.1	15.0	9.7	9.7	6.6	3.3
Median Household Income in Residential Census Tract	$60,610	$58,367	$51,188	$46,470	$43,532	$41,962
Missing Census Tract	2.1	2.2	1.8	0.5	1.3	2.2
Special Education in Grade 9	18.3	7.5	11.4	4.1	16.2	55.7
Std. MS Math Score	0.057	0.646	0.418	−0.556	−0.44	−0.799
Algebra I by Grade 8	26.4	41.2	28.9	3.9	6.6	5.7
Earth Science by Grade 8	9.0	16.9	9.7	0.9	0.7	0.6
Living Environment by Grade 8	14.6	18.3	13.6	1.2	2.7	4.0
Algebra II by Grade 12	70.4	84.2	81.0	62.9	59.9	48.4
Pre-calculus by Grade 12	22.9	40.5	31.9	12.4	11.0	6.1
Calculus by Grade 12	14.0	30.5	22.5	8.0	6.0	1.5
Chemistry by Grade 12	59.9	75.5	69.1	45.7	43.6	36.1
Physics by Grade 12	30.5	48.8	38.9	21.5	18.4	12.6
School Avg. MS Math Score	0.101	0.455	0.253	−0.235	−0.142	−0.175
School Avg. MS Regents Math Participation	28.3	37.4	29.9	15.0	17.9	18.9
School Avg. MS Regents Science Participation	23.7	31.0	24.9	12.3	15.0	15.7

Note: Descriptive statistics for the full list of student- and school-level covariates are provided in the SOM. This table uses the following abbreviations: MS: middle school; EL: English learner; EO: English-Only; IFEP: Initially Fluent English Proficient; RFEP: Reclassified Fluent English Proficient; New-EL: Newcomer EL; Dev-EL: Developing EL; LTEL: Long-Term EL.

Analytic Approach and Measures

We used multilevel logistic regression, with students nested in high schools, to estimate the likelihood of taking college-preparatory math (Algebra II, Pre-Calculus, or Calculus) or college-preparatory science (Chemistry or Physics) before the end of expected Grade 12.

In order to address both individual- and school-level preparation, we separately estimated the associations of the outcomes with students’ prior preparation and with the prior preparation of their schools’ student body as a whole. Individual predictors were centered around the school-level means, and school means were centered around the city-wide grand means by cohort to facilitate their interpretation in the models (Raudenbush & Bryk, 2002). We focus our reporting on the interactions between student subgroup and each predictor, with English-only students serving as the reference.

Individual Middle School Math Achievement

We used the most recent available scaled score from the seventh and eighth grade New York State math achievement tests, standardized for all students who took the exam that year. Because middle schools may exempt students from taking the Grade 8 science exams—a practice that is often applied to students who are taking early Regents science coursework—we used the math achievement test for models with science outcomes, as well. The interaction terms that are the focus of our analysis indicate to what extent the prior achievement of the three EL subgroups predicts their future coursework, relative to the reference group of English-only speakers in the same school.

Individual Middle School Regents Participation

This predictor includes taking a course or exam associated with Algebra I, Living Environment, or Earth Science in either seventh or eighth grade. The individual indicators centered around the school-level means indicate not simply whether the student participated in the Regents in middle school but how unusual that experience was within the context of their high school. Whether ELs are able to leverage that prior coursework to the same extent as English-only speakers is indicated in the interaction terms.

Average Prior Achievement of Students Enrolled in the High School

School-level averages of incoming students’ middle school scores were calculated by cohort and centered at the city-wide grand mean. The measure is an indication of the average prior achievement of the student body in comparison with the overall city-wide average. In general, we would expect that schools with higher-achieving students would offer more college-preparatory coursework in response. The interaction terms indicate whether the three EL subgroups seem to benefit from the average achievement level to the same extent as English-only speakers in the same school.

Average Prior Advanced Course Participation of Students Enrolled in the High School

Similarly, the school-level component of the participation measure is the average rate of incoming students who participated in Regents math or science in middle school, centered at the district grand mean and unitized to 10 percentage points. The measure describes how much the student body as a whole might be prepared for college-preparatory coursework by advanced coursework in middle school. As with average achievement, we would anticipate that schools whose students have taken more advanced coursework in middle school would provide better access to college-preparatory coursework. Whether the EL subgroups are able to take advantage of this environment to the same extent as English-only speakers in their schools is the focus of our interaction analysis.

Other demographic and school-level covariates are described in the supplementary online materials (SOM), along with more details about the policy setting, sample, missing data, estimation procedure, and robustness checks using school fixed effects in place of multilevel modeling.

Results

As context for our findings, Figure 1 displays the observed incidence, by subgroup, of advanced and college-preparatory STEM coursework in middle school and high school, respectively. As elsewhere (e.g., Hanson et al., 2016; Johnson, 2019), all ELs were substantially less likely than non-ELs to have participated in advanced STEM courses in middle school. While English-only students took Algebra I in Grade 8 at rates close to 30%, ELs did so at rates of 5–8%. Similarly, about a quarter of English-only students took one or both of the advanced science subjects in middle school, while 5% or fewer of ELs did. Gaps in high school course-taking were particularly pronounced in the most advanced courses and for long-term ELs.

Figure 1.

Participation in Advanced Math and Science Curricula in Middle and High School, by Subgroup.

Results of Logistic Regression

We focus our reporting of results on the subgroup interactions provided in the Model 2 columns of Tables 2 and 3. For all subgroups, advanced STEM course-taking and math achievement in middle school were positively related to participation in college-preparatory STEM in high school. However, all three groups of ELs appeared to receive less benefit from early preparation than non-ELs, as indicated by significant, negative interactions for ELs on achievement, and directionally negative (though sometimes nonsignificant) interactions for prior coursework (Tables 2 and 3, M2). Patterns for school-level achievement were similar, with strong associations between the prior achievement level of the student population and subsequent participation in college-preparatory STEM. For the math coursework, in particular, all three groups of ELs saw fewer returns to their enrollment in higher-achieving schools than did English-only students enrolled in the same school. School-level prior course-taking was less predictive of later course participation in both subject areas, and generally did not vary by subgroup.

Table 2.

Multilevel Logistic Regression Models for Taking Algebra II or Beyond by End of 12th Grade.

	Model 1			Model 2
	Est	(SE)	Odds	Est	(SE)	Odds
Intercept (EO)	1.61	(0.06)**	5.0	1.61	(0.06)**	5.0
New-EL	0.10	(0.03)*	5.5	−0.20	(0.06)**	2.7
Dev-EL	0.00	(0.04)	5.0	−0.30	(0.07)**	3.1
LTEL	−0.01	(0.04)	5.0	−0.56	(0.07)**	3.4
IFEP	0.25	(0.02)**	6.4	0.31	(0.03)**	7.4
RFEP	0.30	(0.02)**	6.7	0.34	(0.02)**	7.3
MS Math Achievement Scores (EO)	0.97	(0.01)**	2.6	0.99	(0.01)**	2.7
Math Score*New-EL				−0.31	(0.03)**	2.0
Math Score*Dev-EL				−0.27	(0.05)**	2.1
Math Score*LTEL				−0.42	(0.05)**	1.8
Math Score*IFEP				0.17	(0.04)**	3.2
Math Score*RFEP				0.13	(0.03)**	3.1
MS Advanced Math Course (EO)	0.88	(0.02)**	2.4	0.83	(0.03)**	2.3
MS Regents Math*New-EL				−0.30	(0.16)	1.7
MS Regents Math*Dev-EL				0.02	(0.19)	2.3
MS Regents Math*LTEL				−0.54	(0.16)*	1.3
MS Regents Math*IFEP				0.17	(0.07)+	2.7
MS Regents Math*RFEP				0.13	(0.06)+	2.6
School Average MS Math Scores (EO)	1.36	(0.08)**	3.9	1.40	(0.08)**	4.1
Avg. Math Score*New-EL				−0.28	(0.10)**	3.1
Avg. Math Score*Dev-EL				−0.35	(0.13)**	2.9
Avg. Math Score*LTEL				−0.50	(0.12)**	2.5
Avg. Math Score*IFEP				0.11	(0.08)	4.5
Avg. Math Score*RFEP				0.05	(0.06)	4.2
School Average MS Courses (EO)	0.16	(0.01)**	1.2	0.15	(0.01)**	1.2
Avg. Math Course in MS*New-EL				−0.01	(0.04)	1.2
Avg. Math Course in MS*Dev-EL				−0.07	(0.05)	1.1
Avg. Math Course in MS*LTEL				−0.05	(0.03)	1.1
Avg. Math Course in MS*IFEP				0.05	(0.02)*	1.2
Avg. Math Course in MS*RFEP				0.02	(0.02)	1.2
Observations	178,155			178,155

Note: n = 178,155. Models include all individual and school-level covariates, listed in the supplemental online materials. This table uses the following abbreviations: EL: English learner; EO: English-Only Student; IFEP: Initially Fluent English Proficient; RFEP: Reclassified English Proficient; New-EL: Newcomer EL; Dev-EL: Developing EL; LTEL: Long-Term EL.

p < 0.05; * p < 0.01; ** p < 0.001.

Table 3.

Multilevel Logistic Regression Models for Taking Physics or Chemistry by End of 12th Grade.

	Model 1			Model 2
	Est	(SE)	Odds	Est	(SE)	Odds
Intercept (EO)	1.42	(0.08)**	4.1	1.42	(0.08)**	4.2
New-EL	−0.10	(0.03)*	3.7	−0.38	(0.05)**	2.9
Dev-EL	−0.26	(0.04)**	3.2	−0.60	(0.07)**	2.3
LTEL	−0.14	(0.04)**	3.6	−0.86	(0.07)**	1.8
IFEP	0.26	(0.02)**	5.4	0.27	(0.02)**	5.4
RFEP	0.21	(0.02)**	5.1	0.19	(0.02)**	5.0
MS Math Achievement Scores (EO)	0.96	(0.01)**	2.6	0.96	(0.01)**	2.6
Math Score*New-EL				−0.30	(0.03)**	1.9
Math Score*Dev-EL				−0.25	(0.05)**	2.0
Math Score*LTEL				−0.53	(0.05)**	1.5
Math Score*IFEP				0.25	(0.03)**	3.4
Math Score*RFEP				0.15	(0.03)**	3.0
MS Advanced Science Course (EO)	0.84	(0.02)**	2.3	0.80	(0.03)**	2.2
MS Regents Science*New-EL				−0.42	(0.18)+	1.5
MS Regents Science*Dev-EL				−0.43	(0.21)+	1.4
MS Regents Science*LTEL				−0.61	(0.18)*	1.2
MS Regents Science*IFEP				0.14	(0.06)+	2.5
MS Regents Science*RFEP				0.08	(0.05)	2.4
School Average MS Math Scores (EO)	1.18	(0.08)**	3.3	1.27	(0.08)**	3.6
Avg. Math Score*New-EL				0.02	(0.04)	3.6
Avg. Math Score*Dev-EL				−0.06	(0.05)	3.4
Avg. Math Score*LTEL				−0.11	(0.04)*	3.2
Avg. Math Score*IFEP				0.01	(0.02)	3.6
Avg. Math Score*RFEP				−0.01	(0.02)	3.5
School Average MS Courses (EO)	0.11	(0.01)**	1.1	0.11	(0.01)**	1.1
Avg. Sci. Course in MS*New-EL				0.02	(0.04)	1.1
Avg. Sci. Course in MS*Dev-EL				−0.06	(0.05)	1.1
Avg. Sci. Course in MS*LTEL				−0.11	(0.04)*	1.0
Avg. Sci. Course in MS*IFEP				0.01	(0.02)	1.1
Avg. Sci. Course in MS*RFEP				−0.01	(0.02)	1.1
Observations	178,155			178,155

p < 0.05; * p < 0.01; ** p < 0.001.

Math Outcomes

Differential returns associated with prior math achievement were large and meaningful, with ELs’ middle school math scores far less predictive of high school course-taking than those of English-only students (Table 2, M2). For EOs, an additional standard deviation in middle school math achievement was associated with 2.7 times higher odds of taking college-preparatory math in high school. For newcomer and developing ELs, higher math scores entailed only twice the odds of taking college-preparatory math or science, and for long-term ELs the corresponding odds were just 1.8. English-only students who participated in advanced math coursework in middle school math had odds of taking college-preparatory math in high school that were 2.3 times higher than those who did not; corresponding odds for long-term ELs were just 1.3. They were not significantly different for newcomer and developing ELs.

For all five subgroups, average prior achievement among the student body was a stronger predictor of access to college-preparatory math courses than even individual achievement. As with the individual predictors, ELs appeared to receive less benefit from their enrollment in higher-achieving schools than did English-only students. For EOs in math, enrollment in a school with higher prior achievement profile (+1 SD) was associated with a fourfold improvement in the odds of taking college-preparatory math, in comparison with 2.7 times higher odds for higher math achievement at an individual level. The corresponding improvement in odds was 3.1 for newcomers, 2.9 for developing ELs, and 2.4 for long-term ELs. For school-level average prior course-taking, which was a weaker predictor of math coursework for all students, there were no significant differences for ELs.

Science Outcomes

The patterns of association between prior achievement scores and college-preparatory coursework in science were almost identical to what we observed in math (Table 3, M2). An additional standard deviation in prior math achievement was associated with improved odds of taking chemistry or physics that were 2.6 for English-only students but only 1.9 for newcomers, 2.0 for developing ELs, and just 1.5 for long-term ELs. Similarly, EOs with advanced course-taking in middle school had odds of taking chemistry or physics that were 2.2 times greater than those who did not. The corresponding odds were just 1.5 for newcomers, 1.4 for developing ELs, and 1.2 for long-term ELs.

As in math, the school-level average prior achievement was more predictive of future course-taking in science for all subgroups, even over individual achievement. In science, however, there were fewer differences by subgroup. Only long-term ELs appeared to experience significantly lower returns than their peers from their enrollment in higher-achieving schools, and even these differences were relatively modest. English-only students enrolled in schools with higher average achievement (+1 SD) had 3.6 times higher odds of taking physics or chemistry, while the odds of long-term ELs were just 3.2. Also similar to the math results, school-level average prior course-taking was a weaker predictor of science coursework for all students. EOs enrolled in schools with 10% more students who had taken advanced science in middle school had significant but very small improved odds (1.1) of taking physics or chemistry. Long-term ELs were significantly different in that they experienced no returns to enrollment in schools where more of the students had taken advanced STEM in middle school.

Changes in Subgroup Main Effects

Also noteworthy in our results were changes in the subgroup main effects from Model 1 to Model 2 in both subjects. In math, where observed differences in college-preparatory courses were largely explained by the main effects of prior preparation, gaps between ELs and non-ELs reappeared after including interactions by subgroup. In science, where the main effects of advanced preparation did not erase differences between ELs and non-ELs, the negative magnitudes for subgroup effects were substantially larger after including interactions. In both cases, differences across the models indicate that disparities in high school course-taking for ELs are partially or fully obscured by controlling for prior preparation without allowing its effects to vary by group. By contrast, early preparation was more predictive of future course-taking for the other two groups of multilingual students, IFEPs and RFEPs; only those students who continued to be classified as ELs in high school experienced lower returns to prior preparation.

Robustness Checks

As an additional check on the robustness of our findings, we ran a set of models using the same individual covariates with fixed effects for high schools in order to account for any unobserved differences between schools that might not have been fully accounted for in our multilevel models. Although the grand mean centering produced a different baseline likelihood for the intercept (with English-only students as the reference group), the direction, significance, and magnitude of interaction effects were virtually identical with the multilevel models we use in our main reporting (see SOM for details).

Discussion

This study examined differential returns to STEM preparation for English learners. We found that although middle school course-taking and achievement were significant predictors of later course-taking for all students, they were far less predictive for ELs than for non-ELs. ELs were also less likely to benefit from enrollment in higher-achieving high schools than were otherwise similar peers enrolled in the same schools.

These findings echo earlier work arguing that ELs face both early and ongoing barriers to course access at the secondary level. Participation in advanced course-taking in middle school was low enough to suggest systematic exclusion of ELs from the Regents curriculum prior to high school. Furthermore, substantially lower returns to prior preparation for ELs indicate that even when ELs are academically prepared, they tend not to be able to leverage those early experiences to the same extent as non-ELs. This appeared to be particularly the case for long-term ELs in our sample, more than half of whom also received special education services during high school. Their notably lower odds of the outcomes, even after controlling for prior academics and background, may point to the scheduling constraints associated with receiving both language and special education services through high school (Kangas, 2014).

Although our analysis was not designed to uncover the causal mechanisms for these differences, the broader literature on tracking for ELs points to a confluence of factors that might be in play. Teachers who make course recommendations may underestimate English learners’ academic skills, either because of the challenge of assessing content knowledge in their second language (e.g., Solórzano, 2008) or because of misplaced assumptions about English learners’ need to prepare for college (Blanchard & Muller, 2015). Existing scheduling practices may simply default to placing English learners into lower-level trajectories of STEM coursework, and families of English learners may not have the same cultural capital to advocate for their students’ postsecondary preparation (Kanno & Kangas, 2014).

Our finding of strong associations between school-level preparation and individual course-taking echoes other work on differences in course availability across schools (e.g., Iatarola et al., 2011). If ELs are both less likely to be enrolled in schools with college-preparatory coursework and less able to take advantage of better course access even when they are, the interaction of individual and school-level disparities is likely amplifying the course-taking disparities we observe across the city as a whole.

These findings suggest that efforts to expand curricular rigor in middle school and high school (e.g., Attewell & Domina, 2008) may not be working for ELs—both because so few ELs participate in advanced course-taking and because those who do will not experience the same returns, even when their achievement is comparable. Therefore, policies that push for more rigorous middle school math may not, without adaptation, work for ELs. Instead, schools may need to attend more directly to discontinuities in programming across the high school transition, with better differentiation and course assignment on the basis of both language and academic assessment. Middle schools may also need to do more to ensure that ELs and their families have the information they need to take advantage of the high school choice process.

Our findings also problematize one of the fundamental analytic decisions that previous well-designed quantitative studies have made—namely, controlling for prior achievement without allowing it to vary by subgroup (e.g., Johnson, 2019; Umansky, 2016a). Our findings suggest that this approach likely overestimates of the role of prior academic achievement in explaining disparities in course-taking. In fact, it may be that the differential returns we observe in our sample offer an efficient way to characterize the magnitude of ongoing challenges to ELs’ course-taking, in comparison with otherwise similar non-ELs. Regardless, the presence of differential returns to prior preparation in the largest and most diverse school district in the country suggests that future quantitative work in this area needs to test for subgroup interactions more consistently than it has done to date.

Future research will also need to attend to a group that was necessarily excluded from our study given our research questions: newly arrived English learners who do not have achievement or course-taking records in middle school. Given their absence from our sample, our estimates likely underestimate the magnitude of differences in high school course-taking that exist between ELs and non-ELs.

This limitation notwithstanding, our findings suggest that general efforts to widen access to college-preparatory STEM coursework may not benefit ELs without targeted efforts to do so. They also suggest that our analytic methods must account for the differential role that prior preparation plays, given the many challenges ELs face to their content course access.

Supplemental Material

sj-docx-1-tcz-10.1177_01614681251341047 – Supplemental material for Differential Returns to Academic Preparation for English Learners: Evidence from New York City

Supplemental material, sj-docx-1-tcz-10.1177_01614681251341047 for Differential Returns to Academic Preparation for English Learners: Evidence from New York City by Kristin E. Black, Ben Le, Lindsay Romano, Coleen D. Carlson, Jeremy Miciak, David J. Francis and Michael J. Kieffer in Teachers College Record

Footnotes

Acknowledgements

This study is a part of a larger project with the Center for the Success of English Learners (). We are grateful to the Research Alliance for New York City Schools at New York University for their data and technical assistance.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research reported here was supported by the Institute of Education Sciences, U.S. Department of Education, through Grant R305C200016 to the University of Houston. The opinions expressed are those of the authors and do not represent views of the Institute or the U.S. Department of Education.

ORCID iDs

Kristin E. Black

Ben Le

Lindsay Romano

Jeremy Miciak

Michael J. Kieffer

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Kristin E. Black is a research scientist at the Institute for Human Development and Social Change at New York University. Her work focuses on postsecondary literacy and on policies and practices that support racially, economically, and linguistically diverse students in the transition to college. She also has 15 years of experience teaching reading, writing, and literacy education in colleges and high schools. Her work has appeared in Educational Researcher, Journal for Adolescent and Adult Literacy, Written Communication, and Assessing Writing.

Ben Le is a doctoral candidate in the Sociology of Education program at New York University. His research examines how the intersectional identities of racialized students classified as English learners shape their K–12 education experiences and outcomes. He was first author on a paper titled “Ever English Learner 4-Year Graduation: Toward an Intersectional Approach” in Educational Researcher, which demonstrated the importance of intersectional quantitative analyses when studying multiply marginalized populations. Before his PhD program, Ben was a Fulbright English Teaching Assistant in Madrid, Spain.

Lindsay Romano is a PhD candidate in special education in the department of Teaching and Learning at New York University. Her research examines how systemic oppressions, such as racism, ableism, and linguicism, impact the educational trajectories and outcomes of secondary students experiencing multiple marginalizations. She is particularly interested in how teachers play a role in perpetuating/disrupting inequities in their instruction and how critical and contemplative pedagogies may be used as tools for social justice in the classroom. Lindsay is a former high school special education teacher in New York City.

Coleen D. Carlson is an assistant research professor at the University of Houston and the associate director of the Texas Institute for Measurement, Evaluation, and Statistics (TIMES). Dr. Carlson served as principal investigator of the Texas Reading First Evaluation, the Tejas LEE instrument development, and multiple other large-scale projects. She is a co-investigator for the recently funded Institute of Education Sciences National Research and Development Center on English Learners in Middle and High School, under which this research is being conducted.

Jeremy Miciak is an associate research professor in the Department of Psychology at the University of Houston, and affiliated faculty with the Texas Institute for Measurement, Evaluation, and Statistics (TIMES). He conducts research around learning disabilities and learning risk, with a special emphasis on issues of identification and intervention, particularly in culturally and linguistically diverse settings. Dr. Miciak earned a doctorate in special education from the University of Texas at Austin. He has served as principal investigator and/or co-investigator on multiple grants funded by the National Institutes of Health and the Institute of Education Sciences.

David J. Francis is the Hugh Roy and Lillie Cranz Cullen Distinguished Chair of Quantitative Methods; director of the Texas Institute for Measurement, Evaluation, and Statistics; and director of the Center for the Success of English Learners. Dr. Francis is a fellow of APA, APS, and AERA, and recipient of the Hedges Lecture Award from the Society for Research in Educational Effectiveness. He has served on numerous governmental advisory panels, including serving as chair of the 2018 National Academy of Sciences, Engineering and Medicine consensus report on English learners in STEM subjects.

Michael J. Kieffer is professor of Literacy Education at New York University. A former middle school teacher, he focuses on the language and literacy development of students from multilingual backgrounds. He has authored over 50 peer-reviewed articles and received funding and awards from the Institute of Education Sciences, Spencer Foundation, American Educational Research Association, and International Reading Association. He holds an Ed.D. in Language and Literacy from Harvard Graduate School of Education.

References

Attewell

Domina

(2008). Raising the bar: Curricular intensity and academic performance. Educational Evaluation and Policy Analysis, 30(1), 51–71.

Blanchard

Muller

(2015). Gatekeepers of the American dream: How teachers’ perceptions shape the academic outcomes of immigrant and language minority students. Social Science Research, 51, 262–275.

Callahan

R. M.

Humphries

M. H.

(2016). Undermatched? School-based linguistic status, college going, and the immigrant advantage. American Educational Research Journal, 53(2), 263–295.

Corcoran

S. P.

Baker-Smith

(2018). Pathways to an elite education: Application, admission, and matriculation to New York City’s specialized high schools. Education Finance and Policy, 13(2), 256–279. https://doi.org/10.1162/edfp_a_00220

Dabach

D. B.

(2015). Teacher placement into immigrant English learner classrooms: Limiting access in comprehensive high schools. American Educational Research Journal, 52(2), 243–274.

Estrada

Wang

Farkas

(2020). Elementary English learner classroom composition and academic achievement: The role of classroom-level segregation, number of English proficiency levels, and opportunity to learn. American Educational Research Journal, 57(4), 1791–1836. https://doi.org/10.3102/0002831219887137

Finkelstein

Fong

Tiffany-Morales

Shields

Huang

(2012). College bound in middle school and high school? How math course sequences matter. Center for the Future of Teaching and Learning at WestEd.

Hanson

Bisht

Greenberg Motamedi

(2016). Advanced course enrollment and performance among English learner students in Washington state (REL 2017–187). U.S. Regional Educational Laboratory Northwest. https://files.eric.ed.gov/fulltext/ED570326.pdf

Iatarola

Conger

Long

M. C.

(2011). Determinants of high schools’ advanced course offerings. Educational Evaluation and Policy Analysis, 33(3), 340–359.

10.

Johnson

(2019). A matter of time: Variations in high school course-taking by years-as-EL subgroup. Educational Evaluation and Policy Analysis, 41(4), 461–482.

11.

Kangas

S. E.

(2014). When special education trumps ESL: An investigation of service delivery for ELLs with disabilities. Critical Inquiry in Language Studies, 11(4), 273–306.

12.

Kanno

Kangas

S. E.

(2014). “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, 51(5), 848–878.

13.

Long

M. C.

Conger

Iatarola

(2012). Effects of high school course-taking on secondary and postsecondary success. American Educational Research Journal, 49(2), 285–322.

14.

Long

M. C.

Iatarola

Conger

(2009). Explaining gaps in readiness for college-level math: The role of high school courses. Education Finance and Policy, 4(1), 1–33.

15.

Mavrogordato

Harris

(2017). Eligiendo escuelas: English learners and access to school choice. Educational Policy, 31(6), 801–829.

16.

National Academies of Sciences, Engineering, and Medicine. (2018). English learners in STEM subjects: Transforming classrooms, schools, and lives. National Academies Press. https://doi.org/10.17226/25182

17.

National Center for Education Statistics. (2022). High school mathematics and science course completion. Condition of education. U.S. Department of Education, Institute of Education Sciences. https://nces.ed.gov/programs/coe/indicator/sod

18.

New York City Department of Education. (2023). Policy and reference guide for multilingual learners/English language learners, 2023–2024 school year. New York City Department of Education. https://infohub.nyced.org/docs/default-source/default-document-library/ell-policy-and-reference-guide.pdf

19.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Applications and data analysis methods. Sage.

20.

Riegle-Crumb

(2006). The path through math: Course sequences and academic performance at the intersection of race/ethnicity and gender. American Journal of Education, 113(1), 101–122.

21.

Solórzano

R. W.

(2008). High stakes testing: Issues, implications, and remedies for English language learners. Review of Educational Research, 78(2), 260–329. doi:10.3102/0034654308317845.

22.

Thompson

K. D.

(2015). Questioning the long-term English learner label: How classification and categorization can blind us to students’ abilities. Teachers College Record, 117(2), 1–50.

23.

Thompson

K. D.

(2017). What blocks the gate? Exploring current and former English learners’ math course-taking in secondary school. American Educational Research Journal, 54(4), 757–798.

24.

Umansky

I. M.

(2016a). Leveled and exclusionary tracking: English learners’ access to academic content in middle school. American Educational Research Journal, 53(6), 1792–1833.

25.

Umansky

I. M.

(2016b). To be or not be EL: An examination of the impact of classifying students as English learners. Educational Evaluation and Policy Analysis, 38, 714–737.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.21 MB

0.00 MB