CTE Mechanisms: The Effects of Career and Technical Education Center Admissions Offers in Baltimore

Abstract

While there is a growing body of literature related to the benefits of participation in career and technical education (CTE) in high school, there remains a dearth of causal work in this area. Relying on administrative data from Baltimore City Public Schools, which uses a ranking system with a strict cut score to assign students into CTE-focused high schools, we employ a regression discontinuity design to estimate the causal effects of attending one of these high-quality CTE schools on attendance, career pathway completion, high school GPA, and SAT scores. We find that students who are offered admission to a program within one of the CTE centers are significantly more likely to complete a CTE pathway, though they present significantly lower GPAs. Implications are discussed.

Keywords

high schools achievement regression discontinuity policy analysis quasi-experimental analysis career and technical education

Introduction

Across the country, there is a growing emphasis on the need for high schools to prepare students to be both college and career ready upon high school graduation (Every Student Succeeds Act, 2015; Strengthening Career and Technical Education for the 21st Century Act, 2018). In response, practitioners and policymakers have increasingly turned to career and technical education (CTE) as a potential means to promote both college and career readiness. Although initially branded as vocational education providing students with technical skills uniquely tailored to specific occupations (Barlow, 1976; Kantor & Tyack, 1982), contemporary CTE has been redesigned to better integrate academic and technical knowledge and skill development (Dougherty & Lombardi, 2016; Plasman & Gottfried, 2018).

In recognizing the need to better prepare students for life after high school, the two most recent authorizations of the Perkins Act—the federal policy governing and funding CTE in the United States—in 2006 and 2018 encouraged CTE programs to provide opportunities to align hands-on, job-related skills with academically challenging coursework to prepare students for whichever postsecondary options they choose to pursue (Brand et al., 2013; U.S. Department of Education, 2019). Through stronger alignment and integration between academic knowledge and technical skill attainment, students are meant to simultaneously better engage with school, develop new skills, and reinforce academic skills (Gottfried et al., 2014). As such, we evaluate how participation in CTE-focused high schools relates to key high school outcomes such as attendance, CTE pathway completion, and high school performance (as measured by GPA and SAT scores) using Baltimore as a test site.

Benefits of CTE Participation

With CTE programming now firmly entre-nched in districts around the country in both nonselective high schools and standalone CTE centers, career-related education can be viewed as a consistent means to encourage students to persist along the school-to-career pathway (Dougherty et al., 2019; Kemple & Willner, 2008). The re-search base on CTE participation is growing, with substantial causal work suggesting significant benefits. Specifically, Kemple and Snipes (2000) and Kemple and Willner (2008) conducted a longitudinal evaluation of participation in career-focused secondary schools using a random assignment strategy and found career ac-ademy students were more likely to graduate on-time and to experience a suite of labor market benefits including increased monthly earnings, higher average hourly wages, and more working hours per week, though many of these benefits were most pronounced for male students and those deemed to be at-risk. More current work taking advantage of similar random assignment into career academies came to similar conclusions (Hemelt et al., 2019), including improved odds of high school graduation and college enrollment, though benefits were again limited to male students.

In addition to reliance on random selection into career-focused secondary schools, a number of contemporary studies have used other causal estimation strategies such as regression discon-tinuity (RD) designs to estimate the causal effe-cts of attending such schools. In Massachusetts, Dougherty (2018) leveraged the use of a cutoff score used to determine admission into a career-focused secondary school and found CTE participation significantly predicts higher graduation rates, higher odds of earning an industry-recognized credential, and improved math scores. Using a similar admissions criteria for admission to CTE schools in Connecticut, Brunner et al. (2021) also employed an RD approach to establish a significant and positive impact of participation in CTE on later quarterly earnings. In California, Bonilla (2020) also employed an RD design based on the success of districts’ CTE programming funding application to estimate the effects of increasing spending for CTE on high school dropout and found districts that were awarded funding experienced significantly lower dropout rates in subsequent years.

CTE Mechanisms for Success

While the research base on the outcomes related to participation in contemporary CTE points to a wide range of benefits as identified above, less is understood about the intermediary outcomes that may lead to these benefits. To better understand how CTE participation may be achieving these lasting outcomes, Gottfried et al. (2014) proposed three mechanisms: new skill development, academic skill reinforcement, and engagement. The first linking CTE participation to beneficial outcomes is new skill development. Perkins V legislation requires CTE programs be designed to build skills to ensure students are college and career ready at graduation through the development of new skills relevant to the career pathway offer-ed (Oakes & Saunders, 2008; Stern & Stearns, 2006; Strengthening Career and Technical Edu-cation for the 21st Century Act, 2018). A developed interest in both general academic schoolwork and CTE coursework can help students develop new, field-specific technical skills that provide immediate and long-term employability benefits (Feldmann, 2016; Huq & Gilbert, 2013; Komariah, 2015). Secondary students who are able to internalize the intersection of coursework and employer requirements position themselves to develop skills and foster interests relevant to applicable college and career pathways. In addition, CTE coursework may help students develop skills in critical thinking, reasoning, logic, collaboration, research and development, and problem-solving (Brand et al., 2013; Schargel & Smink, 2001).

There is empirical evidence suggesting CTE participation is associated with new skill development. One direct measure of new skill development through CTE participation is by observing the acquisition of industry-recognized credentials, which signal to future employers that a student has successfully accumulated skills associated with competence in a given field (Dortch, 2014; Giani, 2022). In one of only a handful of causal analyses of CTE, Dougherty et al. (2018) found that students who participate in CTE have higher probabilities of earning industry-recognized credentials than nonparticipants. Theobald et al. (2019) and Walters et al. (2021) have come to similar conclusions through less causal analyses.

Second, participation in CTE coursework can reinforce and complement traditional academic skill growth through applied learning experiences (Bozick & Dalton, 2013; Plasman & Got-tfried, 2018; Shifrer & Callahan, 2010). This opportunity to complement theoretical knowledge through applied practices likely improves success in traditional academic settings (Stone et al., 2008). Overall, academic skill enhancement is also linked to greater college and career readiness via improved likelihood of progression along the education pipeline and career obtainment (Lessard et al., 2008). As students continue to develop both academic and technical skills through CTE programming, they are likely to experience greater feelings of self-efficacy (Kelly & Price, 2009), which in turn is a predictor of student success in school and improved high school completion (Finn, 1989; Stone & Aliaga, 2005).

Existing research suggests participation in STEM-related CTE programming early in high school is associated with improved math achievement, math self-efficacy, and a greater likelihood of enrolling in advanced STEM course-taking later in high school (Gottfried et al., 2014; Plasman & Gottfried, 2018; Sublett & Plasman, 2017). In addition, Michaels and Barone (2020) found that CTE participants outperformed their non-CTE counterparts on the ACT, while Gottfried et al. (2014) found that participation in STEM-related CTE was associated with higher standardized math scores. Using an RD design to estimate the causal effects of CTE, Brunner et al. (2021) find that students enrolled in CTE centers score higher on 10th-grade standardized tests.

The final mechanism, engagement, is a key predictor of high school completion (Bozick & Dalton, 2013). CTE coursework focuses on connecting traditional content with more specific career-based application and instruction, further promoting engagement and relevance for each student through applied contextual learning (Br-and et al., 2013; Elliott et al., 2002; Stone & Lewis, 2012). Empirical evidence suggests that participation in career interest activities and job readiness training increases school engagement as measured by attendance, dropout rates, and number of credits earned (National Dropout Prevention Center/Network, 2014). An increase in engagement and relevance should therefore be realized in the measure of student investment in their schools (Bozick & Dalton, 2013; Elliott et al., 2002).

Empirically, there is a growing body of res-earch suggesting there is a relationship between CTE and engagement. Although engagement is difficult to measure directly, it can be approximated through measures such as attendance and high school completion. Work by Plasman and Gottfried (2022) found that participation in CTE is associated with higher attendance and lower probability of chronic absence. Causal work by Brunner et al. (2021) came to a similar conclusion that CTE participation links to improved attendance. CTE participation is also associated with increased odds of high school completion, and this finding has remained consistent over time and across different locations as well as through both correlational and causal analyses (Bishop & Mane, 2004; Bonilla, 2020; Brunner et al., 2021; Dougherty, 2016, 2018; Gottfried & Plasman, 2018; Hemelt et al., 2019; Plank et al., 2008; Wonacott, 2002). Recent work by Kemple et al. (2023) identifies engagement as remaining on track to receive a New York State Regents diploma with students in CTE-dedicated schools more likely to be on track than their non-CTE counterparts.

This Study

In this study, we rely on the sharp cut score requirement for students to receive offers of admission to a program of study within one of the three CTE centers in the Baltimore City Public School District (BCPS), thereby allowing us to set up an RD analysis. Specifically, we look to directly examine how acceptance into a CTE program relates to a set of intermediary high outcomes—CTE pathway completion, high school GPA, SAT scores, attendance, and persistent high school enrollment—as suggested by three theorized mechanisms: new skill development, academic skill reinforcement, and engagement (Gottfried et al., 2014). Specifically, we ask the following questions:

What is the relationship between receiving an admission offer to a program in a CTE center and enrollment in a CTE center?

What is the effect of CTE participation on outcomes related to new skill development as measured by completion of a CTE program of study?

How does CTE participation impact the reinforcement of traditional academic skills as measured by GPA in core academic courses and SAT scores?

How does CTE participation effect outcomes related to engagement as measured by high school attendance and persistent high school enrollment?

While prior causal work has established a link between CTE participation and high school completion, this study explores the intermediary outcomes that may serve as key mechanisms that ultimately could mediate high school completion. Note that we do not specifically focus on how any CTE participation relates to each of our identified outcomes, as CTE programming is available to students in the non-selective high schools, and many students do choose to participate in such programs. Therefore, our focus is on how participation in high-quality CTE—as defined by participation in a CTE center—relates to these outcomes that align with the mechanisms of engagement, new skill development, and academic skill reinforcement using the city of Ba-ltimore as a test case.

High School in Baltimore

Estimating the causal impacts of CTE participation requires the use of more localized settings and must also rely on settings in which certain structures are in place to make these causal estimates. To create a more complete picture, therefore, it is necessary to rely on many different studies from many different locations to be able to better generalize findings. The city of Bal-timore offers a unique opportunity to evaluate outcomes related to CTE participation. BCPS serves approximately 80,000 students in a given year (Baltimore City Public Schools, 2023). A vast majority of these students are either African American (73%) or Latino (17%). Currently, the district has Community Eligibility Provision status, which allows all students to receive free meals. In 2016, the last year the district provided free/reduced-price lunch data, 76% of students were eligible for free or reduced-price meals. With respect to academic performance, BCPS has maintained a graduation rate of approximately 70% for the past 5 years, which is substantially lower than the state average of 87% over the same time period (Maryland State Board of Education, 2023). Similar to other large urban school districts, Baltimore offers a broad range of educational programming at the secondary level, including college preparatory as well as STEM and CTE-focused schools, through a district-wide high school choice system (Nathanson et al., 2013).

The BCPS high choice system is the specific mechanism that allows for a causal analysis CTE participation (Baltimore City Public Schools, 2019). Specifically, the high school choice model allows students to rank their top five high schools or CTE programs for high school attendance. Different high schools have different admissions criteria, thereby creating a tiered system. At the top exist a handful of highly selective high schools that emphasize strong academics and college preparation. The second tier of high schools are moderately selective and emphasize CTE programming designed to be integrated with a rigorous academic curriculum (henceforth referred to as CTE centers). The third tier of schools includes nonselective high schools (traditional, comprehensive high schools as well as other types of charter and contract high schools that have no entrance criteria) and attendance is determined by random assignment. All students are guaranteed a seat at minimum within one of the high schools in the third tier. While CTE programming is offered at many of the high schools across the city, CTE centers include a specific emphasis on CTE programs of study with the expectation that all students will complete both a CTE pathway as well as general academic diploma requirements.

School assignment takes place at the end of middle school based on students’ rank choices. Each student receives a composite score calculated as the weighted average of academic performance (GPA and test scores) and attendance rates in middle school. These composite scores are used to make high school enrollment assignments. To receive an offer of acceptance to one of the selective high schools (either the college-preparatory schools or CTE centers), students must clear two bars. First, students must have a higher composite score than the published minimum score for admission. Second, if more students clear this hurdle than there are seats in the school, the bar may move higher, and students compete with more qualified applicants. With respect to CTE centers, students do not apply to the school itself, but rather, they must apply to a specific CTE program (e.g., cosmetology, Project Lead the Way, etc.). Supplementary Table A1 in the online version of the journal provides a comprehensive list of programs within each of the CTE centers. Leveraging this selective admissions process, we are able to estimate the causal effect of an offer of admission to a CTE center to predict key high school outcomes.

Methodology

Data

To respond to our research questions exploring the link between CTE participation and intermediary high school outcomes, we use BCPS administrative data from the longitudinal archive housed at the Baltimore Education Research Consortium (BERC). The final dataset includes demographic characteristics, educational profiles of students (e.g., GPA, SAT scores, attendance, full course-taking history), and, central to this study, the middle school composite scores. The full dataset includes seven cohorts of students entering 9th grade for the first time between the autumn of 2011 and 2017. We have full demographic and educational profiles for each of these students.

Sample Description

Although we have access to the full population of students in BCPS high schools, we are only interested in focusing on a subset of this population. Given potential differences in motivation, ability, and other unobse-rvable factors, we are interested only in those students who clearly signal an interest in CTE by applying for admission to at least one program in one of the CTE centers in the choice ranking process. As such, any student who does not include at least one CTE center program in their ranked choices would not add to our analysis as there are clearly some important unobservable differences between these students and those who do indicate interest in CTE. This caveat immediately excludes any student who ranks only college-preparatory schools or only nonselective high schools (or some combination of these two) in the choice process. In addition to the requirements that a student is observed making rank choices in the high school selection process and that at least one of those choices must be a program in a CTE center, a student must also receive a placement through the process and must actually show up in the ninth-grade data.

For those students who indicate interest in CTE by ranking a CTE center program, there are three potential options: the application is offered (an offer of admission is made); the application is rejected (an offer of admission is not made); or the application is not considered (the student was offered admission at a higher-ranked school/program). This final set of applications is not considered in our analyses because the students are not causally linked within the sample, where causal linkage is defined as those who are above the threshold on the running variable are always winners (receive offer of admission), and those below the threshold are always losers (denied admission). This full process is discussed in more detail below and mirrors the work of Abdu-lkadiroğlu et al. (2014) in using ranked choice school selection based on composite scores to determine admission to identify effects of admission to a selective school on test scores and postsecondary decisions using an RD design. Based on these requirements, our final analytic sample in Baltimore included 14,910 students, with cohort sizes ranging from 1,820 in the 2011–2012 academic year to 2,330 in the 2017–2018 acade-mic year. We round all our sample sizes to the nearest 10 in accordance with general federal guidelines.

Outcomes

We base our selection of core outcomes on the three proposed mechanisms of new skill development, academic skill reinforcement, and engagement. In addition, we are interested in exploring CTE center participation as an outcome. As such, our first outcome is a binary indicator as to whether a student enrolled in (and attended) a CTE center as defined by attendance at the school on the first Friday of September. We define new skill development through a measure indicating whether a student can be identified as a CTE pathway completer. Here, we define a pathway completer as a student who passes at least three CTE courses within a single CTE cluster. Finally, we define academic skill reinforcement through two measures. First, we include a measure of cumulative high school GPA in core academic courses only (i.e., 4-year cumulative GPA across math, science, English language arts, and social studies courses). We also include a measure of the highest cumulative SAT score a student received as well as scores specifically in Math and Reading. For students who only took the ACT, we converted this score to its SAT equivalent. To measure engagement, we identify two different outcomes. First, we focus on the total number of absences during the freshman year of high school. This is a continuous measure that counts the number of days missed during high school. Second, we include a measure of persistent high school enrollment. Specifically, we can identify the high exact high school in which a student was enrolled in each year of their secondary school career. We define persistent high school enrollment as whether a student was observed as enrolled in the same high school for four consecutive years. These two measures allow us to gauge engagement immediately after the assignment process (attendance) as well as longer-term engagement (persistence).

Control Variables

Using only administrative data, we are limited as to the range of control variables we are able to include. As such, we predominantly rely on a set of demographic and academic history variables in our analyses. With respect to demographic information, we include indicators for gender, race/ethnicity (Black, Hispanic, White, and Other race or ethnicity), English language learner status, special education status, and overage for grade. Our academic variable is the composite score that includes a weighted average of middle school test scores, attendance, and GPA. Table 1 presents the descriptive statistics for our control variables as well as outcomes of interest.

Table 1

Descriptive Statistics

Variable	M	SD
Eligible for CTE Center	0.37	—
CTE Center Enrollment	0.36	—
CTE Concentrator	0.35	—
Cumulative HS GPA (core academic classes)	1.55	(0.78)
Highest SAT (Cumulative)	801.77	(103.58)
Highest SAT (Math)	395.32	(60.61)
Highest SAT (reading)	406.34	(57.24)
9th grade absences	25.65	(29.56)
Persistent high school enrollment	0.55	—
Male	0.53	—
Black	0.94	—
White	0.03	—
Hispanic	0.03	—
Other race/ethnicity	0.005	—
English language learner	0.01	—
Special education status	0.19	—
Overage for grade	0.28	—
Composite score	458.10	(71.78)

Note. All variables binary unless noted here (SAT cum. 400–1370; SAT math 200–680; SAT reading 200–690; ninth grade absences 0–180; Composite score 0–659.27).

Analytic Approach

OLS/Linear Probability

We respond to each of our research questions using a similar methodology. In the case of binary outcomes (i.e., CTE enrollment and CTE concentrator), we employ a linear probability model, while we use a basic ordinary least squares (OLS) model for outcomes with continuous variables (i.e., absence, GPA, and SAT scores). The following equation presents our estimation model:

Y_{i c j p} = α_{0} + α_{1} C T E_{i} + {X^{'}}_{i} θ + φ_{c} + γ_{j} + π_{p} + ε_{i c j p} .

Here, Y represents a placeholder for the outcome of interest for student i in cohort c who applied to school j and was in the lottery for program p. On the right side of the model, CTE is an indicator for whether a student enrolled in a CTE center, and the associated coefficient α₁ is the primary estimand of interest. The vector X′ contains the full set of control measures as defined above and presented in Table 1. The terms φ, γ, and π represent cohort-fixed effects, school-fixed effects, and program lottery–fixed effects, respectively—which are ultimately incorporated as program by school by year-fixed effects. Finally, the error term ε represents a robust estimate of the standard error using a Huber–White adjustment.

RD—CTE Center Eligibility and Participation

At the heart of our study is the question as to whether students enrolled in CTE centers exhibit better outcomes than their peers who do not att-end CTE centers. As such, enrollment in a CTE center serves as the main malleable intervention on which we focus. As mentioned previously, admission to one of the CTE centers is based on a deferred acceptance algorithm that takes into account both the rank order of the chosen school (or program) as well as clearance of the lower bound of the threshold on the running variable (the middle school composite score). Selective schools and programs proceed to rank students in order of priority for admission based on their composite scores from highest to lowest, where the composite score is based on middle school GPA, test scores, and attendance. Following the logic that schools and programs have only a set number of admissions slots in any given year, offers for admission are assigned using the following criteria:

In the first round, student applications to first choice school/program is considered. Schools/programs provisionally admit the highest ranked students up to seat capacity.

In the second and subsequent rounds, students who were not provisionally admitted to their preferred school/program in previous rounds are considered for their next most preferred school/program. These students, in conjunction with students provisionally admitted in previous rounds, are then considered with respect to their composite score and are again provisionally admitted up to the school/program seat capacity.

The process ends when all students have been assigned to one of their chosen schools/programs or are not assigned to any school/program on their choice list, in which case they are randomly assigned to a nonselective high school.

Given the opportunity for students themselves to choose which school they want to attend, there is clearly some level of unobserved bias for which we are unable to account in our naïve estimates identified above. Based on the admissions criteria, it is necessary to describe how this process results in a final causally linked sample on which we base our analyses and which serves to allow for a truly causal estimate. Specifically, students are represented in the causally linked sample if their positive or negative offer of admission to a given program can be deterministically linked to the cutoff for that program (Abdulkadiroğlu et al., 2014). For example, all students who apply to and are granted admission to a given cosmetology CTE program are in the causally linked sample. Those who clear the cutoff to that cosmetology program but are admitted to a higher-ranked choice are not in the causally linked sample as their application was not officially considered in that lottery. Students who do not clear the cutoff but are admitted to a different CTE program—for example, HVAC—are not considered in the cosmetology program causally linked sample but would appear in the HVAC sample. Those students who rank, but are not offered admission to any, CTE programs are in the causally linked sample for all CTE programs they ranked. This final subset is an important point to consider—students who lose all CTE center lotteries can and do appear in a more than one causally linked sample, while those who win a CTE lottery are only in the causally linked sample for the lottery they won.

The deferred acceptance process creates a number of discontinuities at the cutoff between admission to a school/program or not. Given the composite score serves as the determinant of admission, it can be leveraged as a running variable in an RD design as students who have composite scores above the cutoff are offered admission while those who have composite scores below the cutoff are not offered admission. As such, we can assume a causal relationship between composite score and admission to an identified CTE center program.

As noted in Abdulkadiroğlu et al. (2014), though these cutoffs do create an ideal opportunity to evaluate schooling choices using RD designs, the concept of deferred acceptance can loosen the direct link between the running variable and the offer of admission as students’ preferences for other schools may ultimately infl-uence the final determination of admissions cutoffs. This loosening can be recovered through a process of centering the rank ordering of students who apply to a given CTE program on the admissions cutoff and rescaling the ranks for each CTE program. To illustrate, in a lottery in which 100 applicants enter and 20 are granted admission, the top 20 individuals would be ranked from 1 to 20, with the 20th individual making the cut by a single unit. In a lottery of only 10 applicants, of which two are granted admission, the top two individuals would be ranked from one to two. The second ranked individual would make the cut by a single unit. This second ranked individual is equivalent to the 20th ranked individual from the larger lottery analytically with respect to distance from the cutoff threshold. Due to the small sizes of many of the lotteries, we pool all individual program discontinuities for our main analyses. This if further warranted given a common rating variable is used by all CTE programs and schools in each year and given our centering and rescaling of ranks.

Importantly, there is a causal relationship between the composite score and the offer of admission; however, there is the potential that not all students enroll in the school to which they were initially assigned. For example, so-me students who are initially offered admission to a CTE center may choose to enroll in a noncompetitive high school or may be moved off a waitlist into a more competitive, college-preparatory school. Conversely, some students who did not initially receive an offer of admission to a CTE center may end up enrolling in one. As such, we are presented with a fuzzy RD. In this instance, we are essentially estimating how an offer of admission to a CTE center relates to our outcomes of interest. To better approach our true interest, CTE center enrollment, we use the offer of admission as an instrument in a two stage least squares app-roach to predict actual enrollment while usi-ng the discontinuity to predict the offer of ad-mission.

We implement a standard fuzzy RD design (Dougherty, 2018; Imbens & Lemieux, 2008) using a uniform kernel due to the recentering of each lottery. We specify the first-stage linear probability model predicting CTE enrollment for student i in cohort c in school j in program lottery p as follows:

\begin{matrix} P {(C T E = 1)}_{i c j p} = α_{0} + α_{1} {O F F E R}_{i c j p} \\ + α_{2} C S C O R E_{i c j p} \\ + α_{3} C S C O R E x O F F E R_{i c j p} \\ + {X^{'}}_{i} θ + φ_{c} + γ_{j} + π_{p} + ε_{i c j p}, \end{matrix}

where OFFER indicates whether a student re-ceived an offer of admission to a CTE center based on the observed composite score (CSC-ORE). The interaction term, CSCORExOFFER, indicates whether the student was actually gra-nted an offer of admission given their cutscore. The remaining indicators are identical to those identified in the OLS regression model above.

The second-stage model takes the following form:

\begin{matrix} Y_{i c j p} = β_{0} + β_{1} C T E_{i c j p} + β_{2} C S C O R E_{i c j p} \\ + β_{3} C S C O R E x O F F E R_{i c j p} \\ + {X^{'}}_{i} θ + φ_{c} + γ_{j} + π_{p} + ε_{i c j p} . \end{matrix}

Y_icjp again represents the outcome of interest. Here, the parameter of interest is β₁, which represents the causal effect of participation in CTE center program on later outcomes at the margins of having been admitted. We estima-te the models using a uniform kernel and the empirically defined bandwidth as described by Calonico et al. (2014), referred to as the CCT bandwidth. Note that in the instance of CTE offer predicting CTE enrollment, we do not employ the fuzzy discontinuity design, but ra-ther a sharp discontinuity given that all students above the cutoff are initially offered ad-mission while those below the cutoff are not.

Sensitivity Tests

To provide further evidence as to the accuracy of our estimates, we include a number of different specifications. First, we estimate each of our models using different bandwidths. In addition to the CCT bandwidth for each identified outcome, we estimate the outcome at a bandwidth of 15 and 4. In an additional sensitivity test, we estimate each outcome without inclusion of our set of control variables.

Results

Relationship Between Offer of Admission and Enrollment

Recall our first research question asked about the relationship between receiving an offer of admission to a CTE center/program and eventually choosing to enroll in that program. This estimate is central to our overall evaluation of CTE center participation, as a positive, significant finding establishes that the offer of admission to the program creates an observable discontinuity in eventual CTE center enrollment and provides evidence as to the credibility of our instrumental variables approach. Table 2, panel A presents the linear probability estimates linking the offer of CTE admission to eventual CTE enrollment. Panel B presents the RD results. In panel A, we see a strong, positive, and significant relationship (0.65, p < .001) between receiving an offer of admission and ultimate enrollment in a CTE center. Note that there are differences related to race/ethnicity, such that White, Hispanic, and students from other backgrounds are significantly less likely to enroll in a program at a CTE centers than are Black students. Panel B presents similar findings, with our RD estimate suggesting that students at the margins of eligibility who do receive an offer of admission are approximately 59% more likely to eventually enroll in a CTE center.

Table 2

CTE Center Enrollment

Panel A: Linear probability estimates^a
	Enrolled in CTE Center
	Coef.	SE
Eligible for CTE Center	0.72	(0.01)
Male	0.01	(0.01)
White	−0.10***	(0.02)
Hispanic	−0.10***	(0.02)
Other race/ethnicity	−0.08	(0.05)
English language learner status	−0.01	(0.03)
Special education status	−0.01	(0.01)
Constant	0.12***	(0.02)
Program by school by year-fixed effects	Y
N		14,910
Panel B: RD estimated treatment effect^b
	Enrolled in CTE Center
	Coef.	SE
Estimate	0.66***	(0.02)
CCT bandwidth		11
Control variables	Y
Program by school by year-fixed effects	Y
N		5,400

Note. Standard errors in parentheses.

Coefficients estimated using linear probability models. ^bCoefficients presented represent 2SLS results.

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

Figures 1 and 2 help to establish the internal validity of our estimation strategy. Figure 1 shows that there is no observable manipulation at the discontinuity and also presents the overall distribution of the forcing variable, with a relatively normal distribution centered around the cut. Figure 2 illustrates a clear presence of a discontinuity at the cutoff with respect to receiving an offer of admission and ultimately enrolling in a program of study at a CTE Center. The vertical line in the figure indicates the cut point between admission and nonadmission, with observations to the right of the vertical line indicating the point at which students first received an offer (CSCORE > 0). There is a clear jump in probability of attending a CTE center at this cutoff. These findings serve to support that the first-stage estimates in our later fuzzy RD estimates are viable. However, it is worth noting there is not a perfect relationship, providing further evidence as to the need to use a fuzzy RDD. Further supporting the credibility of this instrumental variable approach, the F-statistic in all our first-stage estimates substantially exceeds the minimum threshold of 10 as suggested in prior work (Stock & Yogo, 2002).

Impact of CTE Center Admission on High School Outcomes

Our subsequent research questions inquired as to the effect of receiving an offer of admission to a program in a CTE center on a range of outcomes related to new skill development (con-centration in a career field), academic skill reinforcement (GPA and SAT score), and engagement (attendance and persistence). Table 3 pr-esents the OLS/linear probability model estimates. After accounting for demographic characteristics and prior academic achievement, our results indicate a significant relationship between CTE center enrollment and earning high school concentrator status. Specifically, students in CTE centers were approximately 28% more likely to be CTE concentrators. With respect to academic skill reinforcement, CTE center participants are expected to earn significantly lower GPAs than their peers at nonselective high schools by just more than a tenth of a point. There is no relationship between CTE center enrollment and overall SAT score. However, when breaking SAT score out by math and reading, CTE center students do score significantly higher on math but not on reading. In considering engagement, CTE center enrollment is associated with fewer absences (−1.30, p < .05) during the first year of high school. Further suggesting an overall engagement relationship, students in CTE centers were 14% more likely to exhibit persistent high school enrollment. Note that for each outcome, there is a significant relationship between rank score and a change in a beneficial direction.

Table 3

Linear Probability and OLS Estimated Program Effects

	(1)		(2)		(3)		(4)
	CTE concentrator		GPA		Total SAT		Math SAT
Variable	Coef	SE	Coef	SE	Coef	SE	Coef	SE
Enrolled in CTE Center	0.28***	(0.01)	−0.11***	(0.02)	2.68	(3.03)	3.79*	(1.79)
Control variables
Male	−0.04***	(0.01)	−0.21***	(0.02)	7.22*	(3.38)	8.28***	(1.97)
White	−0.10***	(0.02)	0.07	(0.05)	31.97**	(10.64)	13.35*	(5.41)
Hispanic	−0.12***	(0.03)	−0.00	(0.07)	17.07	(9.96)	11.98*	(5.81)
Other race/ethnicity	−0.09	(0.05)	0.26*	(0.12)	52.14*	(16.43)	29.60**	(11.14)
ELL	0.15***	(0.04)	0.57***	(0.09)	−9.28	(10.27)	−3.90	(5.87)
Special education status	−0.01	(0.01)	−0.01	(0.02)	−47.51***	(3.10)	−20.01***	(1.95)
Overage	−0.07***	(0.01)	−0.17***	(0.02)	−25.95***	(2.84)	−12.89***	(1.80)
Rank score	0.00***	(0.00)	0.01***	(0.00)	1.36***	(0.06)	0.71***	(0.03)
Program by school by year FE	Y		Y		Y		Y
N		14,910		11,230		5,617		5,617
	(5)			(6)			(7)
	Reading SAT			9th grade absences			HS persistence
	Coef	SE		Coef	SE		Coef	SE
Enrolled in CTE Center	−1.28	(1.76)		−1.30*	(0.53)		0.14***	(0.01)
Control variables
Male	−1.05		(1.96)	−1.44*		(0.60)	−0.04***	(0.01)
White	19.38**		(6.76)	8.37***		(1.96)	−0.03	(0.03)
Hispanic	4.80		(5.62)	10.03***		(2.18)	−0.11**	(0.03)
Other race/ethnicity	23.11**		(7.87)	0.35		(3.56)	−0.12	(0.06)
ELL	−4.37		(6.47)	−11.84***		(2.49)	0.07	(0.04)
Special education status	−27.49***		(1.79)	−2.86***		(0.63)	0.01	(0.01)
Overage	−13.09***		(1.60)	10.38***		(0.60)	−0.06***	(0.01)
Rank score	0.66***		(0.03)	−0.18***		(0.01)	0.00***	(0.00)
Program by school by year FE	Y			Y		Y
N			5,617		14,910			14,910

Note. Standard errors in parentheses. Coefficients estimated using linear probability models. Sample sizes vary based on non-missing values on identified outcomes.

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

In addition to establishing a strong relationship between admission to a program in a CTE center and eventual enrollment in a CTE center as evidence of acceptability for our instrument, it is also necessary to establish a measure of equivalence across our observable variables based on our selected bandwidth to further support our use of an RD. Table 4 presents multiple approaches to estimating balance. Panel A presents the results of a manipulation test based on density discontinuity (Cattaneo et al., 2018). The results suggest there is no evidence of manipulation at either a bandwidth of 15 or a more restrictive bandwidth of 4. We discuss the selection of these bandwidths in more detail below. This is an unsurprising finding, given the cutoff is unknowable to the student applicants as the minimum admission threshold was unpublished and based entirely on the number of applicants exceeding the threshold. Panel B shows differences in observable characteristics between individuals who received offers of admission and those who did not. We present differences for the full sample, at a bandwidth of 15, and finally at a bandwidth of 4. Notably, there are significant differences across many of our variables in the full sample. Within the bandwidth of 15, we see that the differences are decreased in every instance (except for ELL, which remains the same). Under our most re-strictive bandwidth, we observe no significant differences except among students identified as receiving special education services. In other words, even when we observe remaining nonzero estimates in our balance tests, the magnitudes are substantially smaller. However, we do continue to include our demographic variables as covariates in our RD models.

Table 4

Test of Balance

Panel A: Manipulation testing
		T	p value
No offer = offer
Bandwidth = 15		0.44	.66
Bandwidth = 4		−0.30	.76
Effective n
Bandwidth = 15		3,610	3,290
Bandwidth = 4		1,080	1,080
Panel B: Estimated differences between admittees and nonadmittees
	Male	Black	White	Hispanic	Other	ELL	Sp. Ed.	Overage
Full sample	0.05***	0.00	0.00	0.00	0.00	0.01***	0.15***	0.14***
	(0.01)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.01)
N	14,910	14,910	14,910	14,910	14,910	14,910	14,910	14,910
Bandwidth = 15	0.03*	−0.01	0.01	0.00	0.00	0.01*	0.11***	0.08***
	(0.01)	(0.01)	(0.00)	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)
N	6,900	6,900	6,900	6,900	6,900	6,900	6,900	6,900
Bandwidth = 4	0.00	0.00	0.00	0.00	0.00	0.01	0.06***	0.03
	(0.02)	(0.01)	(0.01)	(0.01)	(0.00)	(0.00)	(0.02)	(0.02)
N	2,160	2,160	2,160	2,160	2,160	2,160	2,160	2,160

Note. Standard errors in parentheses. Panel A results estimated using Cattaneo et al. (2018) local density estimator. Panel B results presented as difference in means.

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

Table 5 presents an additional interpretation of potential imbalance between those receiving offers of admission and those who do not. As expected, there are significant differences with respect to nearly every covariate when estimating an imbalance in the full model. Furthermore, the F-statistic reveals a clear difference. At a reduced bandwidth of 15, there remains a significant full model imbalance, though the differences are reduced. However, at a bandwidth of 4, there is no longer a significant full-model difference between students offered admission and students not offered admission. As such, we present multiple specifications at different bandwidths in our RD estimates (see Table 6 below). We also include estimates in which we exclude the covariates (see Supplementary Table A2 in the online version of the journal). Furthermore, given the findings by Brunner et al. (2021) that students with special needs may have been treated differently in the selection process and our findings here that there is a significant difference in admissions for students in special education, we include supplemental analyses in Supplementary Table A3 in the online version of the journal estimated without this population of students. Note we ran similar robustness checks with respect to English learner students, but there were no observable differences given there were fewer than 100 such students within the largest identified bandwidth of 16.

Table 5

Omnibus Differences Between Admittees and Nonadmittees

	(1)		(2)		(3)
Variable	Coef	SE	Coef	SE	Coef	SE
Male	−0.01	(0.01)	0.00	(0.02)	0.01	(0.03)
White	0.09***	(0.02)	−0.01	(0.03)	0.01	(0.07)
Hispanic	0.14***	(0.03)	0.07	(0.04)	0.10	(0.09)
Other	0.09**	(0.05)	0.01	(0.08)	0.21	(0.18)
English language learner status	−0.30***	(0.04)	−0.21***	(0.06)	−0.20	(0.10)
Special education status	−0.20***	(0.01)	−0.16***	(0.02)	−0.09*	(0.03)
Overage	−0.14***	(0.01)	−0.08***	(0.01)	−0.03	(0.02)
F		33.53***		6.16***		0.65
Program by school by year FE		Y		Y		Y
Bandwidth		Full sample		15		4
N		14,910		7,290		2,180

Note. Standard errors in parentheses. Coefficients estimated using linear probability models.

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

Table 6

Regression Discontinuity Estimates

Panel A: First stage estimates
	(1)		(2)		(3)		(4)
	CTE concentrator		GPA		Total SAT		Math SAT
	Coef	SE	Coef	SE	Coef	SE	Coef	SE
Offer of admission	0.67***	(0.02)	0.67***	(0.02)	0.65***	(0.03)	0.64***	(0.03)
Panel B: Estimated treatment effects
	CTE Concentrator		GPA		Total SAT		Math SAT
	Coef	SE	Coef	SE	Coef	SE	Coef	SE
Estimate	0.21***	(0.03)	−0.21***	(0.06)	7.91	(10.51)	8.59	(6.40)
CCT bandwidth		16		12		12		14
N		7,240		4,430		2,310		2,590
Alternate specifications
Bandwidth = 4	0.15*	(0.08)	−0.32*	(0.14)	−13.60	(24.36)	−4.03	(15.80)
N		2,160		2,160		870		870
Bandwidth = 15	0.22***	(0.03)	−0.17**	(0.06)	16.85	(9.70)	10.09	(6.15)
N		6,895		5,275		2,725		2,725
Control variables	X		X		X		X
Program by school by year FE	X		X
School by year FE					X		X
Panel A: First stage estimates
	(5)			(6)			(7)
	Reading SAT			9th grade absences			HS persistence
	Coef	SE		Coef	SE		Coef	SE
Offer of admission	0.65***		(0.03)	0.67***		(0.02)	0.67***	(0.02)
Panel B: Estimated treatment effects
	Reading SAT			9th grade absences			HS persistence
	Coef	SE		Coef	SE		Coef	SE
Estimate	3.42		(6.12)	−1.27		(2.06)	0.17***	(0.04)
CCT bandwidth			12			12		14
N			2,310			5,790		6,540
Alternate specifications
Bandwidth = 4	−8.07		(14.27)	−3.07		(4.37)	0.15*	(0.08)
N			870			2,160		2,160
Bandwidth = 15	7.16		(5.59)	−3.14		(1.87)	0.19***	(0.04)
N			2,725			6,895		6,895
Control variables		X		X			X
Program by school by year FE				X		X
School by year FE		X

Note. Standard errors in parentheses. Coefficients represent 2SLS results. Sample sizes vary based on bandwidth and nonmissing values on identified outcome. Estimates related to SAT scores include only school by year FE due to a more limited sample size, while CTE concentrator, GPA, absences, and persistence include program by school by year FE.

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

Under our RD estimates, we attempt to acc-ount for many of the unobserved biases that are present in our baseline OLS/linear probability models. Figure 3 provides visual evidence of the discontinuities for five outcomes (we do not include figures for SAT subtests). There are clear discontinuities in probability of CTE concentration (Figure 3A), high school cumulative GPA (Figure 3B), and persistence (Figure 3E). As in our OLS estimate, GPA appears to be lower for students at the margins of admission. There does not appear to be an obvious discontinuity related to ninth-grade absences or SAT score.

These visual conclusions are supported by the estimates presented in Table 6. Panel A presents the first-stage results of our instrumental variables approach. The results for each model align with our initial findings that there is a strong, positive, significant relationship between an offer of admission and eventual CTE enrollment. However, given we are working within empirically defined bandwidths for each outcome, it is important to include the results here as the different estimations suggest slightly different bandwidths, resulting in different analytic sample sizes.

Panel B presents the results of our RD analyses. As shown in model 1, receiving an admission offer to a CTE center clearly impacts the probability of ultimately concentrating in a career field. Specifically, students at the margins of admission who do receive the offer of admission are 21% more likely to concentrate. Another clear discontinuity relates to GPA (see model 2). Specifically, students at the margins of admission who receive the offer ultimately have a GPA more than two tenths lower than students who do not receive an offer. Models 3 to 5 provide evidence that there is no significant relationship between CTE center offer of admission and SAT score. However, it is worth noting that in the figure, the further from the discontinuity a student falls in the distribution, the higher they tend to score on the SAT. There is very little evidence, visually, of a clear discontinuity with respect to absences, and this is supported by our estimates in model 6. Although the estimate is negative (which would suggest students at the margins who are offered admission are absent less often), it is not significant, indicating no true difference. However, as with SAT scores, there does seem to be a benefit of CTE center participation the further a student is from the cutoff threshold, as supported by our OLS estimates. This is likely to be at least as much a function of prior ability, however, as CTE center participation. Finally, there is a clear relationship between participating in a program of study at a CTE center and persistent high school enrollment. Specifically, students in CTE centers are approximately 17% more likely to remain in the same high school for four consecutive years than non-CTE center students.

To support these overall findings, we also present the results of our estimates using alternate bandwidths. While there are slight differences, the general conclusions remain the same. Specifically, students in CTE centers are more likely to be CTE concentrators (0.15, p < .05) and more likely to persist in the same high scho-ol (0.17, p < .001) under our most restrictive bandwidth. CTE center students under this model are also still expected to perform lower than non-CTE center students with respect to GPA. It is also worth noting that the direction of the estimate is reversed with respect to each of the SAT outcome measures, though all remain nonsignificant. Under our estimates with a bandwidth of 15, our estimates are quite similar in both magnitude and direction to our main specifications.

Discussion

Based on three proposed mechanisms—eng-agement and relevance, new skill development, and academic skill reinforcement (Gottfried et al., 2014)—as to how participation in CTE relates to end-of-high school and postsecondary outcomes, we relied on the unique case of the Baltimore City Public School District to evaluate high-quality CTE offerings in the City’s CTE centers. Under our causal RD estimates, there is strong evidence that students who attended CTE centers were more likely to develop new skills as evidenced by completion of a CTE pathway than those students who received admissions to nonselective high schools. However, there was no evidence that students in CTE centers benefited with respect to improved academic performance. In fact, students who attended CTE centers presented significantly lower GPAs at the cutoff. Finally, though we did not find evidence that participation in high-quality CTE was related to improved engagement as evidenced by reduced absences in ninth grade, there is evidence that CTE center participation impacts engagement through persistent high school enrollment.

At first glance, the results from our analyses showing that CTE center participants exhibit worse outcomes with respect to GPA may seem to run contrary to much of the existing research on CTE participation showing significant benefits related to improved engagement and relevance and academic achievement (Bozick & Dalton, 2013; Dougherty, 2018; Fletcher et al., 2020; Plasman et al., 2021; Plasman & Gottfried, 2018). By design, programs of study at the BCPS CTE centers are very rigid in structure. Students are required to complete key courses on time and in sequence. Should a student fail one of the introductory CTE courses within the program of study, they could have to wait until the following year to attempt that course again, thereby setting the student back in the program. This could prove to be a severe detriment toward engagement with the school, which may then be exhibited through increased absences as we see in our OLS model. A similar process may explain the negative impact on GPA.

There are two important caveats in the Baltimore context that must be considered in interpreting the findings related to GPA. First, this is not truly an evaluation of CTE, but rat-her an evaluation of CTE centers—which focus on small learning communities and integrati-ng career and academic curricula (Fletcher et al., 2020). Students who are not granted admission to a CTE center still have access to CTE in nonselective high schools. Many of the prior studies mentioned above compared individuals who participated in CTE to those who do not, whereas here we are comparing CTE participants in selective schools to all students in nonselective schools, which is the second caveat—the CTE centers are selective. This is essentially more comparable to the Abdulkadiroğlu et al. (2014) evaluation in which the authors find that students at the margins of admission into elite schools are nearly indistinguishable on academic achievement outcomes. In our study, students who just miss out on admission are now situated at the higher end of the academic achievement distribution in the schools without entrance criteria, while students who just clear the threshold for admission are now at the lower end of the academic achievement distribution in the selective CTE centers.

Although we did not observe benefits with respect to GPA, there is some suggestion that students in CTE centers did see other benefits. First, for the average student in a CTE centers as estimated in our OLS models, there did appear to be some benefit with respect to higher SAT scores in math. In addition, CTE center students under our naïve OLS models were also expected to be absent less often. While these benefits with respect to SAT performance and attendance were not visible at the margins as estimated in our RD models, the direction of the estimates were in the desired direction (though not for SAT performance under our most restrictive bandwidth). In addition, students who received admissions to CTE centers had significantly higher probability of concentrating in a CTE field under our RD estimates. Note that a substantial portion of students in nonselective high schools did still concentrate in CTE as shown in Figure 3A. This finding aligns with prior research suggesting that participation in CTE relates to new skill development (Dougherty et al., 2019; Kemple et al., 2023; Walsh et al., 2019). Typically, this has been measured by receipt of an industry-recognized credential, but our measure of CTE concentration approximates a similar signal. Furthermore, there is strong evidence that students enrolled in CTE centers are significantly more likely to remain in those schools throughout their high school careers than students who do not enroll. This persistence indicator is an important indicator of engagement as it suggests that students are less likely to transfer schools, which is a strong predictor of eventual high school dropout (Bedard & Do, 2005; Schwerdt & West, 2013). Ultimately, then, this is an encouraging finding given the design of the CTE centers to focus on career education and career preparation and further supports evidence that CTE links to improved school engagement.

Implications

Policymakers and practitioners in the United States continue to push for expanding educational opportunities that promote college and career readiness while maintaining high standards of rigor. One method is through offering CTE programming. The process by which BCPS assigned students to high schools allowed us to perform an evaluation of one aspect of CTE that is often theorized but rarely empirically explored, namely how participation in a CTE within a small learning community of like-minded individuals may be beneficial. Given that students who enroll in CTE centers actually enroll in specific programs of study, they are surrounded by individuals with similar goals and motivations. This shared vision within an intimate group of learners may encourage students to develop a more close-knit environment and positive school climate (Fletcher et al., 2019; Stern et al., 2010). While we do not observe improvement in engagement related to improved attendance that is often associated with school climate, this improved school environment may ultimately influence students’ decisions to complete a career pathway and to remain in the same school throughout high school.

As such, our findings present a number of CTE-relevant implications that apply to academic researchers and policymakers, and students and practitioners. We began our exploration of CTE center participation with a hypothesis guided by three theorized mechanisms: new skill development, academic skill reinforcement, and engagement. Although, as mentioned above, we are not truly evaluating CTE, we expected that high-quality CTE as delivered in more selective CTE programs would produce strong results related to these three mechanisms. Although prior research has shown some empirical evidence supporting the mechanisms, our work shows, perhaps, that it may not necessarily be the location of CTE delivery that matters, but rather that a student participates at all. It is difficult to disaggregate this consideration empirically in our work, given the opportunity to participate in CTE in nonselective high schools as well as CTE centers.

With respect to policy, the design of the BCPS high school choice system during the years of our study is a clear policy choice. Under the new Perkins V legislation, there is a strong emphasis on equity of access to CTE. A choice program that relies on academic achievement to determine admissions may serve to limit equitable access, which we observe through the underrepresentation of English language learners and students with special needs in CTE centers, even at the margins. However, there is clearly still an important place for such CTE centers as they do produce strong outcomes with respect to concentration in CTE and persistent high school enrollment. This is perhaps one of the central findings of this study. CTE centers are doing, at their core, exactly what they are expected to do and what we would expect of them—increasing the amount of CTE in which a student participates.

Finally, our findings are relevant for educational policy beyond CTE considerations. Policies that establish thresholds for admissions into certain selective programs or schools should look to establish an additional set of supports for those students who just clear the threshold. These students may actually be at a greater risk of failure than their peers who just miss out on admission, even though those gaining admission may have previously exhibited slightly higher prior academic achievement. Students who just miss the cut can now be considered as the big fish in a little pond, and there is empirical evidence that there is an observable big-fish-little-pond effect (Loyalka et al., 2018). In other words, students who just clear the threshold are now comparing themselves to peers with higher academic achievement, thereby negatively impacting their own perceptions of ability, which in turn negatively impacts their performance. This threshold of admission may serve as an empirical indicator for students who could be at-risk of falling under the influence of peer comparison and better help identify students who will need the most support immediately upon entry into the school or program. Notably in the visual representations of our analyses, this negative effect—significant with respect to GPA, and observable though nonsignificant for attendance—appears strongest very near the cutoff and quickly diminishes.

Limitations and Future Research

As with any research study, there are always limitations worth noting. First, though we attempted to deal with issues related to potential unobserved biases that may be present in our OLS models through the use of a RD approach, there remain limitations to our conclusions. Chi-ef among these limitations is that the results are not necessarily generalizable beyond the population of students immediately surrounding the threshold of admission. A second limitation in our RD design has to do with the sheer number of discontinuities that exist in the selection process. Since students must apply to a specific program within a CTE center, there are more than 300 total individual lotteries across all our cohorts. Each program lottery has a different threshold for admission, meaning that the composite score is not a clear designation for acceptance to CTE as it also depends on the lottery to which a student applied. We have attempted to deal with this issue as has been done in prior work through recentering the various lotteries and ranking students on order of acceptance (Abdulkadiroğlu et al., 2014; Dougherty, 2018) as well as including program-, school-, and cohort-fixed effects and an instrumental variable predicting enrollment. However, there is the possibility that we are still unable to account for key differences. Given the need for more causal evidence on CTE, however, we deem this limitation worth the cost. Although not possible with this dataset given the small sample sizes for each lottery, future work could address explore similar outcomes for a single lottery.

A second limitation as we noted above is that this is not truly an evaluation of CTE since students who do not attend CTE centers can still participate in CTE programs of study at other high schools. If we attempted to include CTE participants at other schools in our analysis, our causal mechanism would no longer be valid, and we would only be able to explore any outcomes in a descriptive manner. However, as mentioned, in this study, we are still able to examine how CTE offered in small learning communities and delivered through rigorous, integrated curricula remains an important standalone focal point. To truly evaluate CTE through a comparable methodological design, we would need to observe a setting in which students who did not clear the threshold were denied the opportunity to participate in CTE. Again, our causal mechanism relied on the idea that receiving an offer of admission was significantly predictive of enrollment.

A final limitation has to do with our sample selection. Given our research interests focused on outcomes measured at the end of high school (e.g., CTE concentration, cumulative GPA, SAT performance, total absences, and persistent enrolment) within the city of Baltimore, we did not include those students who may have left BCPS. Note that inclusion in our sample was not contingent on high school graduation, just that they were present in the BCPS system. Future work could expand the sample of students to all those students who were observed in BCPS in their ninth-grade year to explore additional outcomes related to transfer out of the district. In addition, we purposefully chose intermediary mechanisms as our outcomes as a way to empirically test the theorized mechanisms, but a logical expansion on this work could observe the mediating influence of these factors on later outcomes such as high school completion, postsecondary enrollment, and labor market participation.

With the continued emphasis on college and career readiness as evidenced through federal policies such as ESSA and the Perkins Act as well as countless state initiatives toward similar ends, our evaluation of the BCPS CTE centers adds to the growing research on CTE. That we are able to make causal claims in this regard should add further weight to the results we present. Given the growing work linking CTE concentration to im-proved labor market returns (Dougherty, 2016; Dougherty et al., 2019; Kemple & Willner, 2008; Plasman, 2019), our findings that CTE center participation increases the odds of CTE concentration may ultimately bode well with respect to labor market outcomes for these students. Continued policy discussions around how best to support students in various transitional periods should inevitably include CTE considerations as a means to open opportunities for all.

Supplemental Material

sj-pdf-1-epa-10.3102_01623737241239299 – Supplemental material for CTE Mechanisms: The Effects of Career and Technical Education Center Admissions Offers in Baltimore

Supplemental material, sj-pdf-1-epa-10.3102_01623737241239299 for CTE Mechanisms: The Effects of Career and Technical Education Center Admissions Offers in Baltimore by Jay Plasman, Marc L. Stein, Rachel E. Durham and Zyrashae Smith-Onyewu in Educational Evaluation and Policy Analysis

Footnotes

Correction (August 2024):

Article updated to correct the funding statement.

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 R305A210049 to Johns Hopkins University. 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

Jay Plasman

Marc L. Stein

Rachel Durham

Zyrashae Smith-Onyewu

Supplemental Material

Supplemental material for this article is available online.

Authors

JAY PLASMAN, PhD, is an assistant professor in the Department of Educational Studies at The Ohio State University. His research focuses on education policy as it relates to college and career readiness for high school students.

MARC L. STEIN, PhD, is the executive director of the Improvement Colaboratory at Improving Education. His current research focuses on developing practical tools and methods to conduct educational continuous quality improvement projects.

RACHEL E. DURHAM, PhD, is an associate professor at Notre Dame of Maryland University. Her research focuses on high school and postsecondary transitions and research-practice partnerships.

ZYRASHAE SMITH-ONYEWU, PhD, is a postdoctoral scholar in the College of Education at the University of Florida. Her research focuses on college access for low-income and minoritized students.

References

Abdulkadiroğlu

Angrist

Pathak

(2014). The Elite illusion: Achievement effects at Boston and New York exam schools. Econometrica, 82(1), 137–196.

Baltimore City Public Schools. (2019). Middle and high school choice guide for 2019-20.

Baltimore City Public Schools. (2023). City schools at a glance.

Barlow

M. L.

(1976). 200 years of vocational education, 1776-1976: The vocational education age emerges, 1876-1926. American Vocational Journal, 51(5), 45–58.

Bedard

(2005). Are middle schools more effective? The impact of school structure on student outcomes. Journal of Human Resources, 40, 660–682.

Bishop

J. H.

Mane

(2004). The impacts of career-technical education on high school completion and labor market success. Economics of Education Review, 23(4), 381–402.

Bonilla

(2020). The dropout effects of career pathways: Evidence from California. Economics of Education Review, 75, 101972. https://doi.org/10.1016/j.econedurev.2020.101972

Bozick

Dalton

(2013). Career and technical education and academic progress at the end of high school: Evidence from the Education Longitudinal Study of 2002. RTI International.

Brand

Valent

Browning

(2013). How career and technical education can help students be college and career ready: A primer. American Institutes for Research.

10.

Brunner

Dougherty

Ross

S. L.

(2021). The effects of career and technical education: Evidence from the Connecticut technical high school system. National Bureau of Economic Research.

11.

Calonico

Cattaneo

M. D.

Titiunik

(2014). Robust non-parametric confidence intervals for regression-discontinuity designs. Econometrica, 82, 2295–2326.

12.

Cattaneo

M. D.

Jansson

(2018). Manipulation testing based on density discontinuity. StataJournal, 18, 234–261.

13.

Dortch

(2014). Career and technical education (CTE): A primer. Congressional Research Service.

14.

Dougherty

S. M.

(2016). Career and technical education in high school: Does it improve student outcomes? Thomas B. Fordham Institute.

15.

Dougherty

S. M.

(2018). The effect of career and technical education on human capital accumulation: Causal evidence from Massachusetts. Education Finance and Policy, 13(2), 119–148. https://doi.org/10.1162/edfp_a_00224

16.

Dougherty

S. M.

Gottfried

M. A.

Sublett

(2019). Does increasing career and technical education coursework in high school boost educational attainment and labor market outcomes? Journal of Education Finance, 44(4), 423–447.

17.

Dougherty

S. M.

Grindal

Hehir

(2018). The impact of career and technical education on students with disabilities. Journal of Disability Policy Studies, 29(2), 108–118.

18.

Dougherty

S. M.

Lombardi

A. R.

(2016). From vocational education to career readiness: The ongoing work of linking education and the labor market. Review of Research in Education, 40, 326–355.

19.

Elliott

M. N.

Hanser

L. M.

Gilroy

C. L.

(2002). Career academies: Additional evidence of positive student outcomes. Journal of Education for Students Placed at Risk, 7(1), 71–90.

20.

Every Student Succeeds Act, 20 U.S.C. § 6301. (2015).

21.

Feldmann

(2016). Considerations in the design of WBL settings to enhance students’ employability. Higher Education, Skills and Work-Based Learning, 6(2), 131–145.

22.

Finn

J. D.

(1989). Withdrawing from school. Review of Educational Research, 59(2), 117–142.

23.

Fletcher

E. C.

Jr. Dumford

A. D.

Hernandez-Gantes

V. M.

Minar

(2020). Examining the engagement of career academy and comprehensive high school students in the United States. The Journal of Educational Research, 113(4), 247–261.

24.

Fletcher

E. C.

Jr. Tan

T. X.

Hernandez-Gantes

V. M.

(2019). A comparative analysis of student engagement in career academies and a comprehensive high school. Career and Technical Education Research, 44(2), 144–176.

25.

Giani

M. S.

(2022). How attaining industry-recognized credentials in high school shapes education and employment outcomes. Thomas B. Fordham Institute.

26.

Gottfried

M. A.

Bozick

Srinivasan

S. V.

(2014). Beyond academic math: The role of applied STEM coursetaking in high school. Teachers College Record, 116(7), 1–35.

27.

Gottfried

M. A.

Plasman

J. S.

(2018). Linking the timing of career and technical education courset-aking on high school dropout and college-going behavior. American Educational Research Jour-nal, 55(1), 325–361. https://doi.org/10.3102/000-2831217734805

28.

Hemelt

S. W.

Lenard

M. A.

Paeplow

C. G.

(2019). Building bridges to life after high school: Contemporary career academies and student outcomes. Economics of Education Review, 68, 161–178.

29.

Huq

Gilbert

D. H.

(2013). Enhancing graduate employability through work-based learning in social entrepreneurship. Education + Training, 55(6), 550–572.

30.

Imbens

Lemieux

(2008). Regression discontinuity designs: A guide to practice. Journal of Econometrics, 142(2), 615–635.

31.

Kantor

Tyack

D. B.

(1982). Work, youth, and schooling: Historical perspectives on vocationalism in American education. Stanford University Press.

32.

Kelly

Price

(2009). Vocational education: A clear slate for disengaged students? Social Science Research, 38(4), 810–825.

33.

Kemple

J. J.

Snipes

J. C.

(2000). Career academies: Impacts on students’ engagement and performance in high school. Manpower Demonstration Research Corporation.

34.

Kemple

J. J.

Unterman

Dougherty

Sludden

Kamin

S. J.

(2023). NYC as a laboratory for learning about career and technical education: Lessons from CTE-dedicated high schools. The Research Alliance for New York City Schools.

35.

Kemple

J. J.

Willner

(2008). Career academies: Long-term impacts on labor market outcomes, educational attainment, and transitions to adulthood. Manpower Demonstration Research Corporation.

36.

Komariah

(2015, February). The role of work-based learning in building employability skills of vocational education students [Conference session]. 3rd UPI International Conference on Technical and Vocational Education and Training, Bandung, Indonesia.

37.

Lessard

Butler-Kisber

Fortin

Marcotte

Potvin

Royer

. (2008). Shades of disengagement: High school dropouts speak out. Social Psychology of Education, 11(1), 25–42. https://doi.org/10.1007/s11218-007-9033-z

38.

Loyalka

Zakharov

Kuzmina

(2018). Catching the big fish in the little pond effect: Evidence from 33 countries and regions. Comp-arative Education Review, 62(4), 542–564.

39.

Maryland State Board of Education. (2023). Cohort graduation and dropout rates. Division of Assessment, Accountability and Performance Reporting.

40.

Michaels

Barone

(2020). Career and technical education: Academic achievement as measured by National Testing. Career and Technical Education Research, 45(3), 3–20.

41.

Nathanson

Corcoran

Baker-Smith

(2013). High school choice in NYC: A report on the school choices and placements of low-achieving students. The Research Alliance for New York City Schools, NYU Steinhardt.

42.

National Dropout Prevention Center/Network. (2014). Career and technical education.

43.

Oakes

Saunders

(2008). Beyond tracking: Multiple pathways to college, career and civic participation. Harvard Education Press.

44.

Plank

S. B.

DeLuca

Estacion

(2008). High school dropout and the role of career and technical education: A survival analysis of surviving high school. Sociology of Education, 81(4), 345–370. https://doi.org/10.1177/003804070808100402

45.

Plasman

J. S.

(2019). Linking occupational concentration to hourly wages for non-college going individuals. Journal of Career and Technical Education, 34(1), 29–51.

46.

Plasman

J. S.

Gottfried

M. A.

(2018). Applied STEM coursework, high school dropout rates,and students with learning disabilities. Educa-tional Policy, 32(5), 664–696. https://doi.org/10.1177/0895904816673738

47.

Plasman

J. S.

Gottfried

M. A.

(2022). School absence in the United States: Understanding the role of STEM-related vocational education and training in encouraging attendance. Journal of Vocational Education and Training, 74(4), 531–553. https://doi.org/10.1080/13636820.2020.1765841

48.

Plasman

J. S.

Gottfried

M. A.

Klasik

(2021). Do career-engaging courses engage low-income students? AERA Open, 7(1), 1–17. https://doi.org/10.1177/23328584211053324

49.

Schargel

F. P.

Smink

(2001). Strategies to help solve our school dropout problem. Eye On Education.

50.

Schwerdt

West

M. R.

(2013). The impact of alternative grade configurations on student outcomes through middle and high school. Journal of Public Economics, 97, 308–326.

51.

Shifrer

Callahan

(2010). Technology and communications coursework: Facilitating the progression of students with learning disabilities though high school science and math coursework. Journal of Special Education Technology, 25(3), 65–76.

52.

Stern

Dayton

Raby

(2010). Career academies: A proven strategy to prepare high school students for college and careers. Career Academy Support Network.

53.

Stern

Stearns

(2006). Combining academic and career-technical courses to make college an option for more students: Evidence and challenges. College & Career Alliance Support Network, UC Berkeley.

54.

Stock

J. H.

Yogo

(2002). Testing for weak instruments in linear IV regression. National Bureau of Economic Research.

55.

Stone

J. R.

Alfeld

Pearson

(2008). Rigor and relevance: Enhancing high school students’ math skills through career and technical education. American Educational Research Jour-nal, 45(3), 767–795. https://doi.org/10.3102/000-2831208317460

56.

Stone

J. R.

Aliaga

O. A.

(2005). Career & technical education and school-to-work at the end of the 20th century: Participation and outcomes. Career and Technical Education Research, 30(2), 125–144.

57.

Stone

J. R.

Lewis

M. V.

(2012). College and career ready in the 21st century: Making high school matter. Teachers College Press.

58.

Strengthening Career and Technical Education for the 21st Century Act. (2016). 114 U.S.C.

59.

Sublett

Plasman

J. S.

(2017). How does applied STEM coursework relate to mathematics and science self-efficacy among high school students? Evidence from a national sample. Journal of Career and Technical Education, 32(1), 29–50. https://doi.org/10.21061/jcte.v32i1.1589

60.

Theobald

R. J.

Goldhaber

D. D.

Gratz

T. M.

Hol-den

K. L.

(2019). Career and technical education, inclusion, and postsecondary outcomes for students with learning disabilities. Journal of Learning Dis-abilities, 52(2), 109–119.

61.

U.S. Department of Education. (2019). Bridging the skills gap: Career and technical education in high school.

62.

Walsh

O’Kane

Noronha

Taska

(2019). Where credentials meet the market: State case studies on the effect of high school industry credentials on educational and labor market outcomes. Burning Glass Technologies.

63.

Walters

Stinnett

Plotner

(2021). Industry recognized credentials in inclusive postsecondary education programs for students with intellectual disability (Insight: A Think College Brief on Policy, Research & Practice). Think College.

64.

Wonacott

M. E.

(2002). Dropouts and career and technical education. Myths and realities. Office of Educational Research and Improvement.

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.

0.00 MB

0.14 MB