Abstract
This study assesses the excellence gap by examining those who enroll in advanced, honors, and advanced placement (AP) classes among a low-income and a majority-Latinx population. Prospective longitudinal data come from a diverse, urban sample (N = 32,885) where 82.2% of the students received free or reduced price lunch. We examined numerous predictors (i.e., demographics, school readiness skills, prior academic competence) for eventual enrollment in an advanced course (middle school advanced, honors in middle and high school, and AP courses in high school) via multivariate logistic regression analyses. Results suggest that demographic factors (socioeconomic status, ethnicity, English-language learner status) often played a smaller role in advanced course enrollment after controlling for school-entry skills and prior academic competence with the exception of AP courses, where demographic effects persisted. Implications include targeted early intervention to get qualified students in poverty enrolled in academically advanced courses.
Advanced coursework is an important way students can improve their odds of college acceptance and college success (Renbarger & Long, 2019). However, access to advanced coursework is uneven with disproportionately fewer students in poverty and students of color in these courses (College Board, 2014). This “excellence gap” is due to a variety of factors, including systemic differences in early opportunity such as disparities in housing; neighborhood resources, safety, and segregation; school resources; and time and resources available for in-school and out-of-school educational enrichment experiences (tutoring etc.; Carter & Welner, 2013). The first step in closing the excellence gap is identifying which students have access to spaces of excellence and which child circumstances and characteristics promote or impede this access. The current prospective, longitudinal study followed a large, predominantly low-income, Latinx, and Black sample of preschoolers through high school to identify which early child, family, and school characteristics predict advanced course taking in middle and high school.
“The Excellence Gap”
The excellence gap refers to three different aspects of disproportionate representation of students in spaces of academic excellence and achievement in terms of race, gender, wealth, and/or home language: access/enrollment, achievement, and retention/persistence (Plucker & Peters, 2016, 2018; Reardon, 2008; Wyner et al., 2007). Much of the work in this area focuses on achievement, for example, fewer students in poverty test at advanced levels on national and international tests (i.e., the National Assessment of Educational Progress [NAEP]), take advanced placement (AP) exams, and earn a graduate degree than their non-low-income counterparts (College Board, 2014; NAEP, 2015, Plucker & Peters, 2016, 2018). Fewer studies have centered their analysis of the excellence gap on access and enrollment, a topic that is critical because underrepresented students have to first be in such spaces (i.e., advanced classes) before one can examine performance differences across groups. Finally, the retention/persistence aspect of the excellence gap refers to the finding that students of color (Reardon, 2008) and low-income students (Wyner et al., 2007) are at higher risk of leaving such spaces after having entered, or losing their advanced or high achiever status once it has been obtained.
Advanced-level coursework includes advanced courses, honors courses, and AP courses. “Advanced” (ADV) courses are middle school courses intended to prepare students for honors-level coursework and as such include a more demanding and comprehensive curriculum than general courses. Honors courses are those which move at a faster pace, cover more topics, or go into greater detail than typical courses (Collegedata, 2020). AP courses are similarly fast paced and include more and deeper knowledge than offered in a traditional course; however, these courses are equivalent in difficulty to college coursework and passing the subject exam at the end of the course can grant the student college credit (College Board, 2016). Enrollment in these courses requires opting to take more difficult coursework early on and often teacher and/or school approval.
College Board, the institution behind AP courses, reports that in 2013, Black students accounted for 14.5% of high school graduates in the United States (College Board, 2014). However, Black students account for only 9.2% of those who take AP coursework. This is made more dramatic by the fact that Black students’ high school graduation rate is 69% compared with 86% for White students (“State high school graduation rates,” 2012). So the disparity between eligible students and those who actually enroll is even greater because College Board (2014) did not report on the large number of Black students who dropout before graduation. Somewhat more positively, but still underrepresented, Latinx students reportedly account for 18.8% of AP course takers in 2012 but they account for 23.5% of students (College Board, 2014; “Facts for features: Hispanic heritage month,” 2015).
The U.S. Department of Education Office for Civil Rights (2014) reports that although Black and Latinx students make up 37% of high school students, they make up only 27% of those who take at least one AP class. Another College Board (2013) report found that the discrepancy between the percent of those who graduate and the percent of those who enroll in AP was greatest among Black students. The discrepancy was smaller for Latinx students, however, meaning that for 24 states, the percent of Latinx students who enroll in AP is shrinking when held relative to the percent of Latinx students graduating from high school. Furthermore, Wakelyn and National Governors Association Center (2010) report that only 16% of low-income students have taken an AP or International Baccalaureate (IB) course, whereas 51% of their higher income counterparts have taken these courses (Moore & Slate, 2008). Of course, race and income are deeply intertwined in the United States due to historical and continuing systemic racism.
Despite the enrollment discrepancies for historically underrepresented populations, some improvement has been seen in the past two decades. Older studies report larger disparities such as only 33% of eligible students of color actually enrolling in an AP course (Burton et al., 2002). The reduced gaps seen in more current studies are at least partially due to major initiatives by the College Board, state education policy changes, and local efforts by individual school systems (including the one involved in the present study), that aim to increase access to these courses (Education Commission of the States, 2016). However, a reduction in enrollment gaps does not indicate that the problem is solved, and differences in overall enrollment rates of students of color and low-income students continue to exist.
Significance of Advanced Courses
Advanced courses play a major role in shaping the future of students. Students who take AP courses are more likely to get into college than those who do not, even controlling for other factors (Chajewski et al., 2011; Flowers, 2008; Taliaferro & DeCuir-Gunby, 2008; Warne et al., 2015). These students are also more likely to earn higher grades (Flowers, 2008; Patterson et al., 2011) and score better on standardized tests like the ACT than their nonadvanced course-taking counterparts, even controlling for ethnicity and socioeconomic status (SES; Warne et al., 2015). AP (and its counterpart—IB) coursework has become so important in the college admission process that for some universities, enrollment in these kinds of classes is all but required to achieve admittance (Conger et al., 2009). And its importance is only growing as weight is being shifted toward success in classes like these, rather than SAT and ACT scores as major markers for college acceptance (Conger et al., 2009).
Rigor in course work has an influence on college entry and college success as well. This means that even traditional honors courses with their increased difficulty but not the same weight as AP and IB classes can be beneficial. Horn and colleagues (2001) found that students who took more difficult course work in high school had a greater likelihood of graduating from college on time, even when controlling for SES. The importance of academic rigor is especially pronounced in low-income and ethnically diverse populations because children of parents who did not attend college increase their chances of attending and succeeding in college by taking advanced coursework. While not putting first-generation college students at the same rate of degree obtainment as their non-first-generation counterparts, high school academic rigor greatly increases the odds of obtaining a bachelor’s degree for first-generation students (U.S. Department of Education, 2006).
Beyond just getting into college as a major reason for taking advanced classes, there are financial and time benefits as well, particularly with AP classes. Because AP classes allow one to take a test to receive college credit, students can use these as a way to finish college early. In fact, multiple studies show evidence that enrolling in AP or similar early college credit models (i.e., DE) is related to a reduction in the time to obtaining a degree and an increase in the odds of attaining a 4-year degree on time (Beard et al., 2019; Burns et al., 2019; Edmunds et al., 2020; Evans, 2019). Although not all colleges award college credit for such classes, there is still a time and a financial benefit to students. There is a fee associated with taking the AP test, but typically not the class, and this fee is sometimes covered by the high school itself. Despite the fee, AP classes often maintain financial appeal, particularly as the cost of the test is substantially lower than the cost of three credits of college tuition.
Mechanisms Behind Enrollment Discrepancies
A large portion of the existing work on enrollment discrepancy examines school-level factors such as course availability (Cisneros et al., 2014; Klopfenstein, 2004; Solórzano & Ornelas, 2004; Taliaferro & DeCuir-Gunby, 2008), with a smaller emphasis on student characteristics. However, Conger et al. (2009) proposed and tested three explanations for discrepancies in advanced course taking: pre-high school characteristic differences in students of color and low-income students, differences in the courses offered at the schools attended by these students, and differences in the characteristics of schools. They found that student pre-high school characteristics explained more of the variance in course-taking discrepancies than the other two explanations. Thus, we focused our examination on pre-high school student characteristics including student demographics, prior academic achievement, and course pathways.
Student Demographics
The deficit paradigm, or the notion that there is something inherent about being Black or Latinx that inhibits one’s ability to be equally successful to White students, is outdated and discredited (Tomlinson & Jarvis, 2014). Despite this, however, teachers often still express different expectations, recommendations, and attitudes toward students of different ethnic groups (Oates, 2009; Tenenbaum & Ruck, 2007). Potential consequences of this paradigm can be observed in the underrepresentation of students of color in gifted programs even after controlling for cognitive ability (Ford, 1998; Peters et al., 2019; Ricciardi et al., 2020). Taylor and colleagues (2001) examined differences in interpretation of behavior by ethnicity and found teachers labeled more of Black student’s behaviors as problematic than White students. Ferguson (2003) acknowledges how teacher’s knowledge of the Black–White achievement gap contributes to differential expectations and treatment in the classroom, specifically that knowledge of the Black–White achievement gap can inadvertently lead to teachers having lower expectations for Black students. Tenenbaum and Ruck (2007) found that teachers’ expectations varied by ethnicity where certain groups (Black and Latinx) garnered lower expectations, whereas others (Asians) garnered higher expectations than White students. This is important because it not only reflects differences among students of color but also specifically highlights distinct threats that Black and Latinx populations face. A lack of multicultural navigators and culturally sensitive and/or race-matched teachers compounds the systemic inequalities associated with teacher discipline, recommendation, and expectation (Carter, 2005). The unfounded belief that students and families of color are not as interested in education is another problem (Tyson et al., 2005).
While both Black and Latinx students face many similar barriers, it is important to note that their experiences are not the same (Becerra, 2010). Building on this, few studies assess the unique barriers faced by Latinx students, and those that do focus primarily on English language acquisition itself and not on how to boost Latinx students in other important areas like math (Gutierrez, 2008). This makes it exceedingly important that all students are given due attention in the research.
Early Academic Achievement
Conger et al. (2009) found that although demographic differences (free/reduced lunch status, gender, and ethnicity) and educational needs (English proficiency, receipt of special education services) in advanced course enrollment existed before controlling for cognitive ability, when controlling for eighth-grade test scores, there was a substantial reduction in poverty gaps and Black–White gaps in advanced course enrollment. As reported, Black students were 8.4 percentage points less likely to enroll in AP or IB math than their White counterparts, but after controlling for the aforementioned pre-high school ability characteristics, Black students actually had a 5.7 percentage point advantage in the likelihood of enrolling in AP or IB math. This means that when holding demographic and previous test scores constant, high-achieving Black students are actually more likely than their White counterparts to take an advanced course. This also suggests that Black students are receiving fewer school-based opportunities to succeed earlier in school, so while high-achieving Black students do get in, there are fewer of them. Importantly, the pre-high school measure of cognitive ability used in Conger et al. was eighth-grade standardized test scores. Because tracking is often already in effect before eighth grade, it is necessary to look earlier, and at multiple points in time, to more clearly understand the role of early cognitive ability and tracking in advanced course enrollment discrepancies.
Klopfenstein (2004) found that income was the most significant factor in differentiating who took AP courses and who did not. Hébert and Reis (1999) found that a network of high-achieving peers paired with a high level of familial support were helpful factors in promoting achievement in low-income students. Lack of information and guidance, as well as a sense of intimidation from currently advanced-tracked students, can also keep Black and Latinx populations from participating in higher level courses (Yonezawa et al., 2002).
However, AP is not the only kind of advanced course students encounter. Little research has examined honors courses. Much of the work related to traditional honors courses is about tracking. Some exceptions to this exist such as Corra et al. (2011) who conducted an important study that explored race and gender in relation to honors and AP enrollment. This study used SAT scores to control for cognitive ability, but unlike other similar studies, did not control for SES (Princeton Review, 2016). The authors combined the percentage of those who scored above the mean on the SAT for each demographic group with the percentage of the student body made up by that demographic group. This method was used to predict the “expected number” of students enrolling in a particular advanced course (by subject). Then, these expected numbers were compared with the actual number of enrollments by ethnicity.
What this showed is that Black students underenrolled in advanced-level courses and White students overenrolled based on their expected enrollment rates based on SAT scores. This might suggest a social component is involved in differentiating enrollment levels. This is somewhat different from what would be expected based on Conger et al. (2009) who found that enrollment differences were reversed when controlling for eighth-grade test scores. This could also, however, be due to differences in the tests used or the fact that Corra et al. did not control for SES. The SAT is an optional test used primarily to gain entrance to college, while the standardized testing used in Conger et al. is mandatory for all students, regardless of their postgraduation aspirations. Thus, Corra focused on an already high-aiming group, while Conger included a fuller range. In this study, we focus on predominately low-income students and use early statewide tests given to everyone.
Advanced Course Pathways
The opportunity to enroll in advanced classes is often predicated on being in an advanced academic pathway or sequence. Although the idea of courses having a scope and a sequence is reasonable, and ability grouping has been connected to positive outcomes for some (Steenbergen-Hu et al., 2016), the idea of courses stratified by ability has become intertwined with the concept of tracking. Tracking is defined as “the system of assigning high school students to different curricula according to their purported interests and abilities” (Gamoran & Mare, 1989, p. 1147). Although tracking is not practiced formally, distinct pathways still exist informally and have an effect on students of color and low-income groups’ progression through school (Archbald & Farley-Ripple, 2012; Gamoran & Mare, 1989; Jones et al., 1995). Thus, rather than courses serving students of all abilities, students can get stuck on the pathway they are assigned to early in their schooling. For example, students who take prealgebra in sixth grade may be funneled into algebra and geometry before leaving middle school. For those who do not take advanced math early on, everything that follows it becomes inaccessible or delayed.
The consequences of early course selection are compounded by potentially biased ways in which students are often classified into their paths, such as teachers being less likely to recommend Black students to gifted programs and advanced courses (Grissom et al., 2019; Grissom & Redding, 2016; McBee, 2006). When students are not put on an advanced track early on and school policy does not encourage upward mobility, students who were not successful early on may face additional difficulty accessing advanced-level coursework (Alexander et al., 1978; Oakes & Guiton, 1995; Promise, 2008).
The Current Study
Prior studies of the excellence gap are typically limited to data on middle or high school students and do not examine early school and child factors as predictors. This leaves a hole in understanding how pre-high school characteristics, like receipt of gifted status, readiness skills at school entry, pre-K experience, and elementary school achievement, affect later achievement at the highest levels. Furthermore, there is ample room for expanding the groups included when examining course-taking disparities, as Black–White comparisons have dominated at the relative expense of Latinx students, an understudied group in general who are also underrepresented in advanced coursework (Krogstad, 2016).
We used a cohort-sequential, 15-year, prospective longitudinal design to track a large (n = 32,885), ethnically diverse (59% Latinx, 34% Black) sample of students largely in poverty (82.2% free or reduced price lunch) from pre-K through high school to provide much-needed longitudinal information on early pathways and predictors of advanced course taking for underrepresented youth. We explore who among low-income, racially diverse students are taking advanced courses, what advanced courses they are taking and when, and what are the early predictors of taking advanced classes. Importantly, in addition to reporting on AP course taking (as others have done), we add crucial new data on honors course taking in middle and high school, and precursor advanced courses taken in sixth through eighth grade. Finally, to assist future researchers examining the effects of advanced course taking on students who must control for preexisting selection effects, or the ways in which advanced course takers are different from those who do not take such courses, we provide important information about early predictors of advanced course selection.
The following research questions were addressed:
Method
Participants
Participants for this study come from a large urban area in the Southeastern United States. The sample includes about 92% of students who received child care subsidies for low-income families to attend child care in the community or who attended public school pre-K at age 4 years in one large school district between 2002 and 2006. Our sample represents about 30% of the low-income population of the school system. The large-scale, university–community partnership utilized a cohort-sequential, longitudinal design with five cohorts of students who were assessed at age 4 years for school readiness. At the time of data analysis, all five cohorts were at least in seventh grade. One cohort had 11th-grade data, two had 10th-grade data, three had ninth-grade data, four had eighth-grade data, and all five cohorts had data from kindergarten through seventh grade. This sample included 32,885 students who had data for at least 1 year from sixth to 11th grade. The sample is 51.7% male, 82.2% on free or reduced lunch in sixth grade (73.3% free lunch, 8.9% reduced price lunch), 59.2% Latinx (N = 19,354), 34.2% Black (N = 11,198), 6% White/Other 1 (N = 1,965), and 0.6% Asian (N = 201). Attrition (leaving the school system) was roughly 3% a year.
Procedures
Students were assessed directly for school readiness at age 4 years. With the help of the school system, and consent from the families, the preschool sample was successfully matched with the school district database using a unique nonidentifiable ID code, and followed longitudinally throughout school (K–12), and deidentified administrative school record data were collected from the school system each year. Students were followed even when they changed schools as long as they remained within the same school district. All the middle and high schools in this district offered some sort of advanced courses. It is important to note that the district included in this study has won many awards, including AdvancED accreditation, College Board Advanced Placement Equity and Excellence District of the Year, and the Broad Prize for Urban Education, as well as garnered national attention for their dedication to improved access and inclusion in advanced academics.
Measures
Predictors
Demographics
Measure descriptive statistics are included in Table 1. Demographics were collected from sixth-grade administrative data unless otherwise specified both because this grade had the highest n’s across our included grades and because this allowed us to measure demographic and family data most proximally to the start of advanced course taking. Most demographic variables had little to no change over time. Poverty status was measured by receipt of free/reduced lunch in sixth grade. This service is available to those who receive food stamps, Temporary Assistance to Needy Families (TANF), or whose family income falls within a specified low range based on federal poverty lines. Although not a static measure of SES, free or reduced-price lunch (FRL) status varied little from year to year with the majority of those who changed lunch status moving in and out of the reduced-priced lunch category. Ethnicity was coded based on student/family-declared ethnicity according to school records and was grouped into White/Other, Latinx, Black, and Asian. All students in the sample were recruited through their preschool. As such, all children in the sample attended either community-based care on subsidies for low-income families (family or center-based child care) or public school preschool. This was tracked through original preschool agency records. English-language learner (ELL) status was assigned based on parent-reported home language. Upon kindergarten entry, if parents reported a language other than English was the primary language at home, participants were considered ELLs by the district.
Sample Description.
Note. ELL = English-language learner.
Disability status. Students who were identified by the district as having an intellectual disability, speech/language disorder, visual impairment, deafness, specific learning disability, autism, emotional disturbance, brain injury, or other health impairment were coded accordingly each year. An overall code indicating disability status in sixth grade where 0 equals none and 1 equals the presence of at least one of these codes (excluding gifted status) was used for disability status. These codes very rarely changed from year to year.
School readiness
Cognitive, language, and motor skills. Children’s cognitive, language, and motor skills were assessed individually and directly through the Learning Accomplishment Profile—Diagnostic (LAP-D; Nehring et al., 1992,). The LAP-D is a norm-referenced assessment that was given individually to pre-K students. Children were assessed both at the beginning (September/October) and again at the end of the school year (April/May) when they were 4 years old. This was administered by master’s level, trained, bilingual assessors (in whichever language was stronger for the child) for those children enrolled in community-based care. For those in public school pre-K, the child’s teacher administered the LAP-D after receiving the same 2-day training from the publisher. Alphas for internal consistency on the LAP-D subscales range from .76 to .92 (Nehring et al., 1992; Winsler et al., 2008). Content and construct validity ranged from .64 to .86 with the Batelle Developmental Inventory (Winsler et al., 2008).
Social skills and behavior problems. The Devereux Early Childhood Assessment (DECA; LeBuffe & Naglieri, 1999) is a 37-item measure that has an identical teacher and parent report form. It was used to evaluate the socioemotional competence of participants at school entry and is appropriate for ages 2 through 5 years (LeBuffe & Naglieri, 1999). Both teacher- and parent-reported DECA scores were included as predictors. Within the DECA, the subscales are initiative, self-control, attachment, and overall behavioral concerns (BCs). Initiative, self-control, and attachment are combined into an overall measure of “total protective factors” (TPFs). The DECA is a reliable and frequently used measure of socioemotional skills, and it is known to maintain its integrity when evaluating ethnically diverse and low-income children; particularly, alphas for internal consistency ranged from .71 to .94 in this sample (Crane et al., 2011).
Early achievement
Standardized test scores. Students in the district/state take a standardized, high-stakes assessment of reading and math aligned with state education benchmarks each year. We used math and reading scores in fifth grade as a measure of prior academic performance. Scores on this test are highly correlated across grades (r’s > .70) so only fifth-grade scores were included to reflect the most proximal measure of academic achievement before students entered middle school. In multivariate analyses, only the reading test was included because the highly correlated (r = .73) nature of these tests could lead to collinearity problems. Reading was chosen because it had stronger relationships to advanced course taking univariately and because it is the major test used for high-stakes decisions in the state. Internal consistency reliability coefficients range from .88 to .92. Criterion-related validity obtained by correlating test scores to Stanford Achievement Test 10 scores is .79 and .71 for math and reading, respectively. The test changed versions and corresponding scoring metrics during this longitudinal study (in 2011–2012). However, the state maintained a consistent ordinal scoring system which ranged from 1 to 5. In this metric, scores of 1 or 2 were considered failing to meet statewide standards. A score of 3 equated proficiency, and scores of 4 and 5 indicated the standards had been exceeded. Thus, this ordinal score was utilized and treated as continuous to maintain power and reduce potential bias due to variation over time in the test.
GPA. Another authentic measure of student academic performance are the grades received at the end of the year by teachers in all their subjects. These grades were converted to a 5-point scale where A = 5, B = 4, C = 3, D = 2, and F = 1. The grades were then averaged across all subjects to create a composite measure of academic performance for the fifth-grade school year.
Gifted status. Gifted status was coded as a “1” if the student ever had “gifted” as their primary exceptionality in elementary school (K-G5). This means that their school has marked the student as gifted/talented and they were likely, but not necessarily, receiving some type of services for gifted students. According to the Exceptional Student Education Policies and Procedures Handbook set by the school district, a student is eligible for participation in gifted programs if the student meets the following criteria: The student demonstrates the need for a special program, possesses a majority of characteristics of gifted students according to a standard scale or checklist, or has superior intellectual development as displayed by an intelligence quotient of at least two standard deviations above the mean (a score of 130 or higher) on an individually administered standardized IQ test. Alternatively, for students in poverty and ELL students, an alternative identification matrix can be used to increase access to gifted programs for these underrepresented group which includes a creativity measure, teacher report, student grades, and test scores, along with the traditional identification criteria.
Retention, suspension, and grade skipping. Prior retention, school suspension, and grade skipping were also used as additional indicators of academic achievement/success/progression. If a child completed a grade, repeated that grade, and had final, end-of-year grades for that grade a second time, the child was coded as having been retained that year. If this occurred in elementary school (K-G5), they were considered as having been previously retained. If a child ever received in-school or out-of-school suspension during elementary school according to school records, they were coded as having been previously suspended. Finally, if a student had data in a grade in a particular year and then appeared in a grade 1 year higher than they normally should the following year, they were flagged as having skipped a grade. This was exceptionally rare in our sample (see Table 1).
Outcome variables
Enrollment in advanced courses
Advanced-level coursework includes “advanced” courses, “honors” courses, and “Advanced Placement” (AP) courses. Each course title with “honors,” “AP,” or “advanced” specified was carefully coded as an advanced type of course, and children were categorized each year in school as to whether they had one of these classes on their transcript. Each child was coded yes/no for each type of advanced course in each Grade 6 through 11 as well as for which subject area that advanced course was in (i.e., AP Calculus was classified as an “AP Math”). Subject areas included math, science, language, and social studies. The participant received a 1 if they were enrolled in the specified type of advanced course in the specified year (e.g., a student would be coded 1 for the honors sixth grade if they were enrolled in an honors class in sixth grade). If a student was not enrolled in the specified type of course in the specified year, then that student received a code of 0 for that variable. This information was used to create the dichotomous dependent variables included in the study.
Ever enrolled
This variable flags those who took any type of advanced course at any point in Grades 6 through 11. To measure this, students are coded 1 if they received a 1 in any advanced course type in any Grade 6 through 11 and assigned a 0 for if they did not have a 1 in any advanced course flag for Grades 6 through 11. This variable breaks down into three additional variables: ever enrolled in each grade (i.e., enrolled in any type of advanced in G6 yes/no), ever enrolled in each type (i.e., ever enrolled in honors in any grade yes/no), and ever enrolled by year and type (i.e., enrolled in AP in G8).
Results
Research Question 1
Research Question 1 asks how many students are taking which advanced classes and when. Descriptive statistics were run to calculate overall frequencies. Table 2 summarizes the percent of students who took any type of advanced course in Grades 6 through 11 among those eligible. Eligibility here is defined as having reached the grade for which that type of advanced course became available. This is sixth grade for ADV and honors, and ninth grade for AP. As Table 2 also breaks down the taking of each advanced course in each Grade 6 through 11, students had to reach that grade to be included in the “Students Eligible” column. For instance, the percent column indicates that of the students who had been to seventh grade, 48.78% took some kind of advanced course.
Enrollment in Advanced Courses by Type and Grade.
Note. Students eligible includes all those who had progressed far enough through middle/high school for enrollment in the noted course type to have been an option. AP = advanced placement; ADV = advanced.
The table reveals that 65% of students (who had data for at least one Grade 6 through 11) had taken at least one advanced class of some kind (AP, honors, or ADV). The overall percentage of students who took an advanced class in a particular grade ranged from 48% in sixth grade to 67% in 11th grade. This makes sense in terms of the availability of advanced courses (low in sixth grade but higher by 11th grade). The table shows decreasing raw numbers of course takers as grade increases because the cohort-sequential nature of the data means fewer students had reached the later years than had reached the earlier years. As such, the percentage of students is a more useful measure (compared with N’s) of student enrollment.
Overall, there was a relatively high level of participation in some sort of advanced course work (more than 65%). Breaking advanced course work into its subtypes reveals further information on advanced course taking. ADV courses had the highest level of participation (59%), which is in line with such courses being the first step in the accelerated pathway. The next highest level of participation was honors classes (43%), followed by AP classes (22%).
Research Question 2
Research Question 2 asks whether demographic factors, school readiness skills, and prior academic competence predict later advanced course enrollment for our predominantly low-income, ethnically diverse pre-K sample. First, bivariate statistics were used to determine the role of each of these factors independently and for comparison to similar studies which do not include all the same predictors. Ever enrolled in any type of advanced course was examined first, followed by enrollment in ADV, honors, and AP courses. Results for the bivariate analyses are found in Online Supplemental Table A1 and A2. Then, multivariate analyses examining how multiple variables interact to predict later course taking are described.
Bivariate analysis summary
Demographics
In bivariate analyses, Black and Latinx students were less likely to enroll in advanced courses than White and Asian students, girls were more likely to enroll than boys, and ELLs were more likely to enroll than non-ELLs. The unadjusted relationship between poverty and advanced course taking varied by advanced course subtype, but overall those not on reduced lunch or on reduced lunch enrolled at higher proportions than those receiving free lunch. Finally, a higher percentage of public school pre-K attendees took an advanced course compared with those who attended child care in the community at age 4 years.
School readiness and prior achievement
Bivariate analysis of the continuous school readiness and prior achievement predictors can be seen in Online Supplemental Table A2. Analyses reveal that the cognitive, language, and fine motor subscales of the LAP-D were significantly related to later advanced course enrollment for all types of advanced courses. In each case, the mean score on the assessment at age 4 years for those who enrolled in the advanced courses at least 7 years later was significantly higher than the mean for those who did not enroll. The gross motor subscale was significant for ADV, honors, and AP courses. Cohen’s d suggests that cognitive and fine motor skills at school entry had the largest effect sizes for later course taking.
Parent and teacher ratings of student social skills (TPF) and BCs in preschool were also significantly related to later course taking for all types of courses. Those who enrolled in advanced courses had higher social skills and better behavior in preschool, on average, than those who did not. Effect sizes for teacher ratings were generally higher than for parent ratings; however, parent BCs were particularly salient for ADV course taking (d = .61).
Math scores, reading scores, and GPA in fifth grade were all significantly related to later advanced course enrollment for all subtypes, such that higher GPAs and test scores were found for those who later enrolled in each type of advanced course. Cohen’s d suggests that the largest effects, in order, occur for GPA, reading, then math. Having skipped a grade in elementary school was only significantly related to honors course enrollment where skippers were more likely to enroll in an honors course later compared with those who did not skip. A consistently lower percentage of students who were retained or suspended in elementary school later enrolled in advanced courses. Gifted status was strongly related to advanced course enrollment where nearly 100% of students identified as gifted in elementary school enrolled in some sort of advanced course in Grades 6 through 11.
Multivariate analyses
Multivariate analysis design
Although bivariate analyses reveal significant relationships between nearly all our variables of interest and later enrollment in advanced coursework, they do not account for the possible confounding influence of related variables. The bivariate results reported above are useful to compare with other studies because often research on disparities in advanced course taking cannot control for all these factors. Multivariate analyses were conducted and reported below to assess each predictor accounting for the others. Multivariate analyses were conducted via three-step, logistic regression using SPSS. Each regression followed a standard structure. In the first step, only demographic factors were included (ELL status, preschool type, gender, special education status, ethnicity, and poverty level). For ethnicity, Latinx was chosen as the reference group because it is our largest group and our desire to avoid default comparisons to White students as being the “norm.” A supplemental Black/White comparison was also included in the tables to facilitate comparison with other studies. This step illustrates how demographic factors like ethnicity, and early life and schooling factors like preschool type relate to later advanced course enrollment. This provides a picture of group differences on enrollment before children enter school and before taking elementary academic skills into account.
The second step incorporates school readiness skills as measured at age 4 years. This provides a picture of how school-entry skills relate to advanced course enrollment later on. Finally, the third step adds prior elementary school competence variables (gifted status, fifth-grade GPA and test scores). This provides a picture of how demographic variables do and do not still matter when prior competence is included, as well as how different early competence markers relate to later enrollment. Performing the regression in three steps not only provides information on how the role of each factor changes as other factors are added but also creates an approximate impact timeline. Moving from the earliest factors related to advanced course taking to factors like grades in the preceding year highlights not only where but also when variables are related to access.
Multivariate results—Ever enrolled
Ever enrolled—Any type of advanced course: Step 1. The first of these regressions (found in Table 3) examined ever enrolling in any type of advanced course as the outcome. The overall model significantly predicted enrollment in an advanced course of any type, χ2(23, N = 13,723) = 5,555.24, p < .001. Step 1 reveals that when just assessing demographic factors, poverty, ethnicity, gender, preschool type, and special education status were each significantly related to enrollment, even accounting for the other variables. Attending public school pre-K, as opposed to community child care, was related to a 48% increase in likelihood of enrolling in some sort of advanced course (p < .001). Those who received reduced price lunch were 82% more likely than those who received free lunch to enroll in an advanced course, while those who did not receive lunch aid were 45% more likely to enroll in an advanced course than those who received free lunch. This suggests that even slight income differences produce meaningful differences in advanced course access.
Three-Step Logistic Regression Predicting Enrollment in Any Type of Advanced Course by Demographics in Step 1, School Readiness in Step 2, and Prior Competence in Step 3 (N = 13,723).
Note. TPF = total protective factor; BC = behavioral concern; ELL = English-language learner; GPA = grade point average.
p < .05. **p < .01. ***p < .001.
Furthermore, there were significant ethnicity effects. Both Latinx and Black students were less likely than White students to enroll in advanced courses (by about 60%). Asian students were significantly more likely to enroll in an advanced course than Latinx students (by more than 4 times), while Black students were less likely than Latinx students to enroll in an advanced course. Being male was associated with about a 30% decrease in odds of enrolling in an advanced course, while receiving special education services in sixth grade was associated with about a 75% decrease in the odds of enrolling in an advanced course. The only demographic factor not significantly associated with advanced course enrollment when all were entered together was ELL status.
Step 2. Adding in school readiness skills changed the results notably. All subscales of the LAP-D (cognitive, language, fine motor, and gross motor), as well as both parent- and teacher-rated reports of student social skills at school entry, were significantly related to advanced course enrollment later. Scoring higher on the cognitive, language, and fine motor skills at school entry was associated with an increase in the odds of enrolling in an advanced course years later. The largest of these effects was for the cognitive subscale, where a 1-point increase on the cognitive subtest was associated with a 0.01 increase in the odds of enrolling in an advanced course years later. This means that for a 50 percentile point discrepancy on preschool cognitive skills, the odds of enrolling in an advanced course would be increased by 50% (0.01 × 50 = 0.50). The smallest effect size among the LAP-D subscales was gross motor skills, where an increase of 1 point on the gross motor subscale was associated with a slight decrease of 0.002 in the odds of later advanced course enrollment. Both parent and teacher ratings of students’ social skills at school entry were associated with a 0.3% increase in the odds of enrolling in an advanced course. Parent and teacher perceptions of child BCs were associated with a similar-sized decrease in the odds of later advanced course enrollment years later.
Of note, after adding in readiness at school entry, many demographic variables lost their significance. Attending public school pre-K was no longer associated with an increase in the odds of taking an advanced course, which suggests that pre-K programs increase school readiness and it is school-entry skills that matter for later access to advanced courses. Latinx and Black students remained less likely to enroll in advanced courses compared with White students, but the degree to which they differed was mitigated by school readiness. Black students were still less likely to take an advanced course than Latinx students after controlling for skills at school entry. Thus, enrollment differences between Black and Latinx students and White students were not fully explained by differing readiness at school entry. Being a male now resulted in only a 15% decrease in the odds of advanced course enrollment suggesting that boys were less ready for school than girls according to the assessments. Receiving special education services remained significant where the odds of enrolling in an advanced course for those with disabilities reduced to 65% less than for those who did not receive services. Notably, being an ELL was now significantly associated with a 19% increase in the odds of advanced course enrollment controlling for school-entry skills.
Step 3. Step 3, our main model of interest, includes demographic factors and school readiness skills, but adds measures of elementary school competence. Prior academic competence, as expected, was strongly related to the likelihood of later advanced course enrollment. Being assigned the designation gifted, having a higher GPA in Grade 5, higher standardized test scores in Grade 5, and not being retained in elementary school were all significantly related to increased odds of enrolling in an advanced course later. The odds of gifted students enrolling in an advanced class were more than 5 times higher than those of students without a gifted designation. This is a massive effect, especially when considering that this controls for GPA and test scores which are used by school systems to measure academic success. Similarly, an increase of 1 GPA point in fifth grade (e.g., moving from a “C” to a “B” average) increased the odds of enrolling in an advanced course by more than 3 times. In addition, moving up one ordinal standardized test score classification (i.e., moving from a 3 to a 4) more than doubled the odds of enrolling in an advanced course. In contrast, being retained in elementary school decreased the odds of ever enrolling in an advanced course by roughly 60%. Finally, skipping a grade (a rare event) was not significantly associated with a change in odds of advanced course enrollment.
Even accounting for the very powerful effects of prior academic competence, certain school readiness measures were still significantly related to advanced course enrollment years later. Higher cognitive skills at age 4 years still increased the odds of enrollment by 0.2% per percentage point increase (odds ratio [OR] = 1.002, p < .01). In other words, moving from the 25th percentile on the cognitive test to the 75th percentile increased the odds of advanced course enrollment by 10%, even controlling for fifth-grade GPA and test scores. The same was true for child gross motor skills at school entry. Preschool teacher perceived behavior problems were negatively related to the odds of enrollment. All other school readiness skills at age 4 years were no longer significantly related to advanced course enrollment after taking later elementary school academic competence into account.
Finally, even when accounting for all other variables in the model, disability status, preschool type, and poverty level were still significantly associated with advanced course access. Similar to the previous models, having some sort of exceptionality is related to a substantial decrease in the likelihood of advanced course enrollment (by about 70%, p < .001), even after controlling for academic competence. Interestingly, regarding poverty, when holding all other variables in the model constant, those who did not receive any lunch subsidies (the most affluent group in the sample) were actually about 40% less likely to gain admittance to some sort of advanced course than those who received free lunch (those with highest poverty in the sample). Those who attended public school pre-K were about 12% more likely to enroll in an advanced course than those who attended center-based child care. None of the other demographic variables (including ethnicity) remained significant in the final model. This suggests that prior academic competence is explaining more of the variance in advanced course taking than ethnicity, such that after controlling for those who are achieving similarly in elementary school, there is no statistically meaningful difference in the likelihood of enrolling in advanced courses by ethnicity.
Ever enrolled—ADV. Next, the same analysis was run separately for each of the subtypes of advanced course (ADV, honors, AP). Tables for subtype analyses are contained in the Online Supplemental Materials—Tables A3–A5), where bolded terms indicate deviations from the “ever enrolled any” advanced course model just discussed. Table 4 contains an overview of the Step 3 OR patterns across each model for comparison. For simplicity, only the final models with all variables included will be discussed below.
A Comparison of Step 3 Findings Across Each Type of Advanced Class.
Note. Only findings from Step 3 of each model are included; ○ = nonsignificant finding, + = significant finding with an OR > 1, − = significant finding with an OR < 1. AP = advanced placement; ADV = advanced; TPF = total protective factor; BC = behavioral concern; ELL = English-language learner; GPA = grade point average; OR = odds ratio.
Enrollment in ADV courses specifically (Grades 6–8; Table A3) yielded very similar results to that of the advanced courses at large, and the final model similarly significantly predicts ADV enrollment, χ2(23, N = 13,723) = 5,919.89, p < .001. The most notable differences from any-type model results discussed above were seen in the role of school readiness skills, gender, and suspension. In the any advanced course model, only cognitive skills, gross motor skills, and teacher-perceived behavior concerns were significant in Step 3, but for ADV courses in middle school, increased cognitive skills, teacher-rated social skills, and teacher-rated BCs in pre-K were associated with increased odds of ADV enrollment. While not significant in the any type of advanced model, gender was significant when predicting enrollment in ADV courses such that makes were more likely to enroll by about 12%. Being suspended in elementary school was related to about a 20% decrease in the odds of enrolling in an ADV course.
Ever enrolled—honors. The next step in the traditional advanced course-taking pathway is honors coursework. Again, the full model significantly predicted honors enrollment, χ2(23, N = 13,722) = 5,777.85, p < .001. Predicting enrollment in honors courses (Table A4, Online Supplemental Materials) looks very similar to predicting any advanced course. In the final model, most relationships continued to follow the pattern seen in the any-type model. However, ELL status and gender remained significant in the final model such that those who were ELLs and males were more likely to enroll in an honors course years later than those who were not ELLs or females. In contrast to the any-type model, those who received reduced price lunch were about 25% more likely to enroll in an honors course than those on free lunch. Many school readiness variables lost their significant association with enrollment in the honors model. Although still following the pattern observed in the any-type model, the effect size of gifted status on honors course taking was substantially smaller than the effect gifted status had on any type of advanced course taking.
Ever enrolled—AP. The final step in the traditional advanced course pathway is AP (Table A5, online). Predictors of enrollment in at least one AP class were notably different from those of enrollment in ADV or honors classes; however, the full model did still significantly predict enrollment, χ2(23, N = 8,258) = 2,696.0, p < .001. Similar to the final any-type model, all prior competence variables, except skipping a grade, were significantly associated with later AP enrollment. Gifted status was still a significant predictor of AP enrollment, as receiving gifted status more than doubled the odds of taking an AP course, but to a smaller degree than for any-type or ADV models. Of school readiness skills, only the language subscale of the LAP-D was still significant in the final model. This is different from the any-type model where cognitive skills and teacher-rated BCs were significantly related to enrollment, not language skills.
Also unlike in the any-type model, even after controlling for skills at school entry and prior competence at the end of elementary school, Black students were still nearly 40% less likely to enroll in an AP class than their similarly skilled White peers. Asian students were significantly more likely than Latinx students to enroll in AP courses, whereas Black students were significantly less likely than Latinx students to enroll in AP courses. In the full (Step 3) model, being Asian increased the odds of enrollment in an AP course by almost 3 times. In addition, for AP course taking, males remain less likely to enroll than females in the final model. The effect of poverty on course taking is also different when looking at AP classes specifically. First, the distinction between the free and reduced lunch was significant for predicting AP course enrollment, unlike for any type of advanced course. Second, and more strikingly, unlike in any of the previous models, those not in poverty were significantly more likely (OR = 1.51) to take an AP course than those who received free lunch. Furthermore, for the first time, attending public school pre-K decreased the odds of enrollment (OR = 0.85, p < .05) and qualification for special education services was not significantly related to AP enrollment.
Discussion
We explored how frequently students largely in poverty from diverse backgrounds who participated in different kinds of pre-K programs at age 4 years participate in advanced courses later in middle and high school, and what factors serve to boost or constrain the likelihood of enrollment for these populations. This is in response to the large amount of research looking at achievement gaps (Burney & Beilke, 2008; Ford & Harmon, 2001; Hallett & Venegas, 2011; Jimenez-Castellanos, 2008; Klugman, 2013; E. Taylor, 2006) and the substantially smaller amount of research that focuses on the excellence gap (Plucker et al., 2010). Unlike previous studies, we had data stretching from preschool through high school enabling assessment of the predictive validity of multiple measures across childhood. In addition, we had a large, diverse sample that permitted a more nuanced look at how ethnicity is related to advanced course selection overall, and we examined all three types of advanced course taking—ADV, honors, and AP.
Ethnicity
Race/Ethnicity is one of the primary constructs discussed in the excellence gap literature (College Board, 2014; Plucker et al., 2010). Studies that attempt to clarify why and by how much ethnicity-based enrollment discrepancies can be narrowed show somewhat dissonant findings. Some found support for continued ethnic and racial gaps in enrollment even after controlling for academic competence (Corra et al., 2011), while other studies found support for students of color actually out-enrolling White students after controlling for prior academic proficiency (Conger et al., 2009).
Our study was uniquely situated to provide some clarity on this, with the ability to not only break down different types of advanced courses but also use many different possible predictors across the early lifespan. Our findings were more in line with the findings of Corra et al. as, particularly with AP taking, ethnicity continued to play a role even after accounting for prior academic competence and early skills. This was not always the case (e.g., with honors course taking), but our results never showed the full reversal of enrollment patterns by ethnicity that was seen in Conger et al. (2009). However, these other studies did not specifically deconstruct each advanced course-taking subtype (e.g., honors and AP) like was done in this study. In addition, much of the previous research uses more nationally representative samples, while this one focuses on predominately low-income children who are ethnically diverse and largely Latinx. What our results suggest is that bivariately, ethnicity is strongly related to all types of advanced course enrollment. This effect persists when controlling for poverty, ELL status, and other highly related constructs suggesting that ethnicity itself is a factor in enrollment. However, when controlling for elementary school competence, the effects of ethnicity become much more complicated.
Once elementary school competence was controlled for, excellence gaps between students narrowed, and in many cases, lost statistical difference. For example, in the full model (any type of advanced), the Latinx/White comparison stops being significant after controlling for elementary school performance (i.e., the odds of advanced enrollment are statistically the same for Latinx and White students). This parallels the findings of Grissom and Redding (2016) regarding gifted identification after controlling for math and reading assessments. However, Black and Latinx students were significantly less likely than White students to enroll in AP classes specifically, even after controlling for elementary school outcomes.
The transition from Steps 1 to 3 in our analyses suggests that ethnicity is related to elementary school competence, opportunity, and academic skill development, and it is this that is truly pushing enrollment differences in advanced courses later (Peters & Engerrand, 2016; Plucker & Peters, 2018). Thus, to decrease racial disparities present in advanced classrooms, steps need to be taken prior to a student reaching middle school. Of course, the persistence of ethnicity effects in the case of AP course taking suggests that even after earlier intervention is used to reduce opportunity gaps in elementary school, further support is needed for students of color to take advanced courses. This is a place where culturally competent and diverse teachers and guidance counselors could make a huge impact on encouraging all qualified kids to seek this rigorous coursework (Grissom & Redding, 2016; Ohrt et al., 2009).
Poverty
Students who grow up in poverty often have reduced access to physical resources, like books, computers, and educational toys. Students in poverty are also less likely to have access to opportunities like educational enrichment programs and diverse settings and materials for learning (Olszewski-Kubilius & Corwith, 2017). All these things together, especially when experienced early in life, relate to a reduced opportunity to learn and thrive in educational settings (Burney & Beilke, 2008; Plucker & Peters, 2018). Students in poverty arrive to school less ready than their high-income peers (Reardon & Portilla, 2016; Ryan et al., 2006; Winsler et al., 2008) and once in school are less likely to gain access to programs like gifted which could support their progress (Plucker et al., 2010). Although a great deal of previous research (with more financially advantaged samples) finds strong advanced course enrollment disparities along SES lines (Wakelyn & National Governors Association Center, 2010), our results suggest something slightly different.
In the first step of the models, those in poverty enrolled less frequently than those not in poverty. However, after controlling for fifth-grade competence, those in poverty were actually more likely to enroll in an advanced class than those from non-low-income backgrounds. This was not true in the case of AP enrollment, where those not in poverty were always more likely to enroll, no matter what other factors were controlled. This could be a result of restricted range with our predominantly low-income sample, or it could be that some competent students do not enroll for free or reduced lunch even though they are eligible due to social stigma reasons (Poppendieck, 2011), or it could indicate that what is barring poor students from accessing advanced courses is actually that the students in poverty are doing worse in school early on so that by the time they reach middle and high school, their track is set (Hemphill & Vanneman, 2011). The latter is supported by the directional change in the OR for poverty in Steps 1 versus 3. In essence, our findings suggest that the primary way poverty is linked to advanced course enrollment is through early measures of academic achievement like standardized test scores. The story then is that low-income students are less likely to enroll overall (as suggested by the bivariate results and previous work by Klopfenstein, 2004), but when they are doing well in elementary school, they are just as likely or even more likely to enroll (Young et al., 2017). This could be reflective of a high degree of resilience in the face of numerous obstacles for high-achieving students in poverty.
Retention and Suspension
A finding that was novel, though not surprising, was that suspension and retention in elementary school were both consistently related to decreased odds of advanced course taking even after controlling for elementary school performance. Regarding ever enrolling, retention consistently more than halved the odds of enrollment. This is an important, potential negative consequence of retention not previously considered in the literature, and with high-stakes testing policies (like the one in the included state), it could be introducing an unforeseen barrier to achievement. Previous research has found evidence that retention is linked to a decrease in later academic performance and behavior (Pagani et al., 2001). It is also possible that the social ramifications of retention (Jimerson & Ferguson, 2007) are related to a reluctance to select into advanced academic spaces.
A meta-analysis by Noltemeyer and colleagues (2015) found a consistent negative relationship between school suspension and academic achievement. This was mirrored in our findings where suspension had an overall smaller effect than retention, but still played a significant role in preventing advanced course enrollment. This could suggest that behavioral factors play an important role in deciding who makes it to advanced courses. This relationship could also be related to the punishment gap in education in which students of color are disproportionately singled out for harsh and exclusionary disciplinary action. Morris and Perry (2016) found that one fifth of Black–White achievement differences in school performance for their sample could be attributed to school suspensions. This suggests that even after controlling for race differences in course enrollment, suspension could still be negatively affecting enrollment rates for all students but especially students of color.
School Readiness
Many studies have explored the long-term impact of school readiness, but the definition of “long-term effects” for most of these studies does not extend past late elementary/early middle school (Duncan et al., 2007; Sabol & Pianta, 2012). This project took perhaps one of the longest looks at school readiness yet. This is partially what makes it so notable that slight differences at school entry are still related to advanced course taking many years later in middle and even high school. This also helps explain why school readiness skills were more strongly related to ADV taking than to AP courses. ADV courses are middle school courses and are met earlier in a child’s life, whereas AP and honors are taken in high school, further down the road from pre-K. It makes sense that the further a student gets from school entry, the more course-taking differences are subsumed by the more proximal GPA and test score differences. Of course, studies show that school readiness is related to both GPA and standardized test scores in elementary school (Davies et al., 2016), so school readiness is playing a role even if it is being mediated by other academic factors. This is likely part of a developmental cascade in which students who attend public school pre-K are attaining better school readiness (Winsler et al., 2008), and those higher in readiness skills are performing better in early academics and are more likely to get into ADV courses, promoting a trajectory of excellence beginning even before elementary school.
Skipping a Grade
One predictor that was surprisingly unrelated to advanced course taking was skipping a grade in elementary school. As grade skipping is offered as an option for the extremely accelerated student, it would have been expected for these students to be taking coursework designed to challenge advanced students. In a meta-analysis including 38 studies on grade acceleration, findings suggest that grade acceleration was positively related to academic achievement (Steenbergen-Hu & Moon, 2011). However, skipping was only related to honors course taking when other variables were controlled for. This suggests that by this measure of academic success, skipping a grade is not providing added benefit to accelerated students in terms of advanced course taking down the road.
Access Versus Selection
The question of access versus selection seeks to understand whether evidenced disparities in advanced course taking are due to a lack of access (students are not allowed to or are not referred, or they attend schools that do not offer the courses) or a choice to not select into advanced courses when they are available. In essence, are disparities in course taking attributable to unequal opportunity in course offerings in schools and restrictive admittance criteria, or to a lack of belongingness, desire, or knowledge about and within advanced course spaces? The schools studied here provide an excellent place to explore this concept, as they have an open enrollment policy for advanced course taking, and all schools offered advanced courses.
This means that anyone who would like to can sign up for these advanced classes without the need for teacher recommendation/approval or qualifying scores. This, to some extent, removes the barrier of teacher bias and testing bias from the equation. That being said however, students are given a recommended course schedule based on teacher recommendation and test scores. To select into something not recommended for the student by the teachers, parents must reach out to the school. According to the school system, only about 10% of parents do this. Thus, in the included school district, barriers to accessing advanced courses are significantly reduced but not entirely gone. Similarly, as discussed in the introduction, tracking, deficit thinking, and unequal early opportunity conflate what access truly means for underrepresented populations. Although the relative equal availability of advanced courses in the current school system would suggest that enrollment differences by ethnicity observed were a result of voluntary selection differences, the previous sections make clear that early education experiences and circumstances are related to enrollment, and that these early experiences often vary by ethnicity. Our results suggest that selection differences in open enrollment exist, but that selection differences may be driven by access differences and skill development in earlier schooling.
Implications
Discrepancies in advanced course enrollment clearly exist nationwide. However, a thorough understanding of whether these disparities persist with the inclusion of many relevant factors, and how these discrepancies vary across type of advanced course was previously unavailable. The uniquely situated sample allowed a novel exploration of these concepts which sheds light on many aspects of academic placement that need attention. As such, conclusions here shed light on how performance (even as early as elementary school), school readiness skills from before a child even enters school, and demographic background factors contribute to the problem of advanced course enrollment discrepancies. Particularly, this study looked at multiple popular forms of advanced courses (ADV, honors, and AP) which allows for more specific policy change and suggestions tailored to the factor of interest (i.e., poverty) or the advanced coursework of interest (i.e., AP). This is particularly important because, as the results show, AP courses are influenced by different factors than other types of advanced course enrollment. As such, the support students need to enter an ADV class is different from what they need to enter an AP class.
The cumulative, tracked nature of education means that the skills a student has when they enter kindergarten can relate to courses they take in late high school. Our findings support this notion empirically and suggest that intervention for advanced academics needs to occur far earlier than middle school. In addition, it is clear that there needs to be a concerted effort to increase enrollment, particularly for Black students, early on and in gateway advanced courses like ADV. Frontloading, or preparing early for later advanced work, is considered a key strategy for closing excellence gaps (Plucker & Peters, 2018). One possible way to encourage early enrollment is an increase in guidance counselors all together, and specifically an increase in culturally competent counselors and teachers. Studies suggest that diverse teachers promote the inclusion of diverse students in advanced academics (Grissom & Redding, 2016). This suggests that trainings to improve cultural competence for teachers could be a promising way to disrupt systemic racism and promote enrollment in advanced coursework for qualified students, regardless of racial or economic background. Furthermore, policies like universal screening for gifted programs, which have been shown to enroll more high-achieving students of color in gifted programs (Card & Giuliano, 2016), could increase enrollment in advanced courses.
Limitations
There were three main limitations in this project. The first is the sample. Although this sample provided major advantages, such as having sufficient diversity to look at ethnic differences beyond just Black/White, having large numbers overall, and a high level of income and ethnic diversity, these same things threaten broad generalizability. This is particularly true because the region that makes these sample characteristics possible is relatively unique within the United States. As such, it is not safe to assume that certain findings, particularly those relating to Latinx students, generalize to situations in which Latinx students are not the numerical majority. That being said, in 2015, 266 counties in the United States containing 31% of the country’s population were “majority minority” counties—78 of which gained this designation between 2000 and 2013 (Misra, 2015). In addition, schools in this sample were open enrollment for their advanced courses. This means that some of these findings may not be generalizable to school districts in which there are more administrative barriers to enrollment in advanced courses. Furthermore, this district is noted for their innovative policies and efforts dedicated to improving access to advanced academics for all, meaning that findings here may not be mirrored in districts not making equivalent strides. Finally, given our large sample size and the number of analyses run, there is the potential for Type 1 error.
Future Directions
This study took a broad look at the access facet of the excellence gap through the lens of advanced course enrollment. In that way, it provided a foundation upon which future studies can be conducted delving deeper into particular topics, using more advanced statistical methods to understand different aspects of the data and the world the data reflect, and upon which related concepts can be explored. Specific future studies that should be conducted include an investigation of the accumulation of and persistence in advanced courses over time as well as an examination of advanced course trajectories where differing trajectories can be tracked in detail to account for the full spectrum of course-taking differences. Future research should expand upon school-based factors in advanced course selection (Klopfenstein, 2004) to examine opportunity and resource differences. In addition, more work should be done on those who accelerate through grade skipping specifically and why they are not more likely to enroll in advanced courses than their peers. Finally, future work should build on these findings to understand how self-regulation, motivation, and personal goals affect advanced course-taking decisions and success.
Supplemental Material
sj-pdf-1-joa-10.1177_1932202X21990096 – Supplemental material for Selection Into Advanced Courses in Middle and High School Among Low-Income, Ethnically Diverse Youth
Supplemental material, sj-pdf-1-joa-10.1177_1932202X21990096 for Selection Into Advanced Courses in Middle and High School Among Low-Income, Ethnically Diverse Youth by Courtney Ricciardi and Adam Winsler in Journal of Advanced Academics
Footnotes
Acknowledgements
We thank the participating school district, children, agencies, and staff.
Authors’ Note
The views expressed are those of the authors and do not necessarily reflect the opinions of Foundation staff members or Board of Directors.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by a grant from the Jack Kent Cooke Foundation (JKCF). Research projects funded by the Cooke Foundation examine issues that affect high-achieving students with financial need.
Supplemental Material
Supplemental material for this article is available online.
Notes
About the Authors
References
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.
