College Acceleration for All? Mapping Racial Gaps in Advanced Placement and Dual Enrollment Participation

Background About AP and DE

During the past six decades, there has been an increasing nationwide support for programs that allow high school students to earn college credit while in high school. The largest of these programs are Advanced Placement (AP) and Dual or Concurrent Enrollment (DE), which together enroll millions of high school students each year (College Board, 2017). AP is offered by the College Board and covers college-level curriculum content. It offers students the potential to earn college credits after students achieve a minimum score on a course-specific exam. Since its inception in 1955, AP has grown substantially: With more than 2.6 million exam takers in the academic year of 2015–2016, AP has become the largest mechanism through which high school students earn college credit in the United States. DE is the second largest, with roughly 1.4 million students participated in DE in 2010–2011. Different from AP programs, which are intended to be taken by high school students and are exclusively taught by high school teachers, DE is a broad category including many types of college course-taking arrangements, and are taught by either college instructors or college-approved high school teachers and through different modalities including at the college, at the high school, and online.

Researchers have noted several benefits of college acceleration programs on students’ postsecondary outcomes, particularly their potential to improve college attendance among underrepresented students (Berger et al., 2013; Klepfer & Hull, 2012). In addition to the policy and theoretical support for expanding programs that allow high school students to earn college credits, a number of studies have also provided empirical evidence for the benefits of AP and DE on student academic outcomes. Numerous studies of the AP program have compared the academic performance of non-AP and AP students, and generally found that AP students outperform their non-AP peers in a variety of academic achievement measures, such as ACT and SAT scores, college attendance rates, admission to selective colleges, college grade point averages (GPAs), college graduation rates, and time to degree (e.g., Ackerman et al., 2013; Flowers, 2008;; Klopfenstein, 2010). Interestingly, quotes from college administrators suggest that regardless of the score received on an AP exam, college admissions decisions may be favorably affected by a student’s AP participation alone (College Board, 2013). ¹

In a similar vein, a number of quasi-experimental studies also identified positive impacts of DE participation on a variety of academic outcomes, including high school graduation, college enrollment, college persistence, college GPA, and postsecondary degree completion (e.g., Allen & Dadgar, 2012; An & Taylor, 2019; Hemelt et al., 2019; Karp et al., 2007; Miller et al., 2018; Speroni, 2011). A handful of studies also examined whether the benefits of DE vary for subgroup populations of students (e.g., students from low socioeconomic backgrounds vs. those from more affluent backgrounds) and the results are mixed (e.g., An, 2013; Karp et al., 2007; Miller et al., 2018; Speroni, 2011).

Existing Evidence on Racial Gaps in AP and DE Participation

Given the likely benefits of AP and DE enrollment on college success for students, racial gaps in participation rates would serve as important indicators of educational inequality. Unfortunately, persistent racial disparities exist in AP enrollment and success rates, where Black students are most underrepresented: According to the 10th Annual AP report by College Board (2013), Black students represent only 9% of AP test takers in 2013 despite making up 15% of the 2013 graduating class. In a similar vein, disparities are also observed in dual-credit participation by race. Using the High School Longitudinal Study of 2009, National Center for Education Statistics (NCES, 2019) reported lower participation rates in DE programs among Hispanic students (30%) and Black students (27%) than White or Asian students (both 38%). These national patterns of racial gaps in DE participation are echoed in studies using data from particular states. For example, based on administrative data from Texas, Miller et al. (2017) found that while DE participation rates generally increased during 2000 to 2015 for all students, there was a persistent racial gap in participation rates, which seem to enlarge over time.

While the existing evidence on national and state-level patterns of racial gaps in AP and DE participation provides useful information about overall educational inequality in college acceleration opportunities, these aggregate statistics are less informative about whether these gaps are smaller or larger across smaller geographic units, such as school districts, therefore making it difficult to identify local contexts and factors that produce and sustain these gaps. In this article, we address this knowledge gap by providing detailed descriptive analyses of the patterns of racial gaps in AP and DE participation across thousands of school districts, and by identifying state-and district-level factors that are correlated with these gaps.

Correlates of Racial Gaps in AP and DE Participation

We draw on a rich body of literature and theories about racial disparities in educational choices and outcomes to provide a framework outlining the complex relationship between both school-related and nonschool factors that may be associated with racial gaps in AP and DE participation. Specifically, we focus on six broad categories of factors: (1) student academic preparation prior to high school, (2) family socioeconomic background, (3) racial composition in a district, (4) between-school segregation, (5) average characteristics of high schools in a district, and (6) state-level AP and DE policies. Below we discuss each category briefly.

Data Sources

We linked multiple publicly available data sources to document geographic patterns of AP and DE racial/ethnic enrollment gaps among school districts in the United States. We describe each data source below:

Constructing Measures for AP and DE Participation

The 2015–2016 CRDC data provide new insight into the number of students participating in AP and DE. However, a major limitation of this dataset is that it only provides enrollments at the school level (instead of at the school-by-grade level). As a result, for high schools that also offer eighth grade and below, using total school enrollment as the denominator to calculate AP and DE participation rates is inappropriate, since CRDC specifically instructed districts and schools to only report AP and DE participants among students who are in Grades 9 through 12. In Supplementary Appendix B, we describe the selection criteria we used to identify eligible high schools to be included in our analytic sample, as well as our methodology for estimating the ninth- to 12th-grade enrollment counts for high schools that offer eighth grade or below. ⁶

Correlates of Racial Gaps in AP and DE Participation

Following the framework outlined in the section “Correlates of Racial Gaps in AP and DE Participation,” we estimate sources of variation in AP and DE racial participation gaps that fall within one of the following six broad categories: (1) student academic preparation prior to high school, (2) family socioeconomic background, (3) racial composition in a district, (4) between-school segregation, (5) average characteristics of high schools in a district, and (6) state-level AP and DE policies. Below, we briefly describe the variables included in each category. The full list of explanatory variables used and the data source for them are presented in Supplementary Appendix Table A1 (available online).

To capture academic preparation prior to high school, we include in our analysis a variable that measures the average pre–high school achievement in a district. This variable is available in the SEDA data and is based on standardized test scores taken by over 200 million students in Grades 3 through 8 for subjects English language arts and math. White-minority achievement gaps are calculated as the standardized difference in achievement between White and minority students.

We construct two variables to measure the average socioeconomic background of a district. The first measure is proportions of students eligible for free or reduced-price lunch, which was retrieved from CCD and indicates the percent of students in each school district eligible to receive a free or reduced-price lunch. The second measure is proportions of adults with a bachelor’s degree or higher in a district. This variable is available in the SEDA data set and was constructed using ACS-EDGE data for families with school-age children enrolled in public schools. White-minority gaps in both measures are also calculated and included in models that predict racial gaps in AP and DE participation.

We also construct three sets of variables to measure the local school context, including: (1) measures of racial and income segregation across schools within a school district constructed using the Thiel index, where higher values indicate greater levels of segregation; (2) indicators of racial/ethnic composition of a school district, measured as the proportion of Black or Hispanic students within a district; and (3) measures of average characteristics of high schools in a district, including proportions of students in a school district enrolled in urban schools, proportions enrolled in magnet schools, proportions enrolled in gifted and talented programs, average per-pupil instructional expenditures among high school students, average student-teacher ratios, and average student-counselor ratios. Given that DE opportunities rely on partnerships with local colleges, we also calculate the distance in kilometers to the closest 2– or 4–year public institution that offers dual enrollment or concurrent enrollment for each high school in our analytic sample and then take the average across districts. ⁷ Similarly, we also calculate the average number of AP courses offered at school to capture a district’s access to AP opportunities.

Last, we include state-level variables to reflect policies that either directly or indirectly influence AP and DE participation. ⁸ For AP enrollment, we identify nine relevant policies, grouped into three broad categories: (1) accountability and mandates surrounding access, (2) financial incentives and program support, and (3) accountability and mandates related to student outcomes. For DE enrollment, we identify 12 policies and group them into the same three broad categories listed above. Each category includes three values that indicate whether a state has strong, moderate, or weak AP (DE) policies for that category. A state is identified as having “strong” policies if it has at least half of the policies in place within a category, “moderate” if it has more than one but fewer than half of the policies in place, and “weak” if it has none or only one policy in place. ⁹ Details about the number and description of distinct policies included in each broad category are provided in Supplementary Appendix Table A2 (available online).

Empirical Model for Exploring Correlates of Racial Gaps

The outcome measures in our study, such as average AP participation rates in a district, follow a fractional response nature, which typically arises from averaged binary outcomes. Specifically, students either participated or did not participate in AP or DE programs. These binary responses were used to generate participation rates at the district level (see section “Constructing Measures for AP and DE Participation” and Supplementary Appendix B, for more detail). The AP and DE participation rate is therefore a continuous variable that is bounded between 0 and 1. Standard linear models are not appropriate in modeling fractional responses, as they can generate predictions that are greater than one and smaller than zero (Papke & Wooldridge, 2008). Building on the existing literature (Wooldridge, 2011), we instead use fractional logit regression with a logistic link that captures nonlinear relationships, especially when the value of the outcome measure is close to 0 or 1 to examine associations between district-level and state-level predictors and district participation rates.

It is important to note that our second set of outcome measures—racial gaps in participation rates—have a wider range than a normal fraction (i.e., they range from −1 to 1). We therefore perform a linear transformation to bound the racial gap variables between 0 and 1. Specifically, let W be the original outcome measure of racial gaps in AP or DE participation rates that is in [−1, 1]. Then, define Y = (W+ 1)/2, where Y would be in [0, 1]. This transformation allows us to use fractional logit regression to model the relationship between the transformed outcome measure Y and district-level predictors, which writes as follows:

Y i = b 0 + b 1 X i + v i

where

Y i = ( W i + 1 2 ) , ,

which could be further written as

( W i + 1 2 ) = b 0 + b 1 X i + v i

(1)

or,

W i = 2 b 0 + 2 b 1 X i − 1 + v i

In other words, the relationship between the racial gaps in DE participation W_i and district-level predictors X_i would be equal to two times b₁ from Equation 1. To make the coefficients easier to interpret, we report the average marginal effect. ¹⁰ As a result, the coefficient indicates the average changes in the predicted probabilities given a one unit change in an independent variable (in the case of a continuous independent variable), or as the independent variable changes from 0 to 1 (in the case of a binary independent variable).

Overall Patterns of AP and DE Participation

The top panel in Table 1 shows summary statistics for AP and DE participation among all the school districts in our sample. On average, districts had 11% participation rates in both AP and DE, with standard deviations of 12% and 14%, respectively, indicating that there are substantial variations across districts. Figure 1 further uses boxplot to show the distribution of district AP and DE participation rates visually by state (left and right panels, respectively), where states are ranked in descending order by the states’ median district participation rate. Overall, there is substantial variation in district AP and DE participation both within and across states. Interestingly, it seems that the states with the highest AP participation rates concentrate on the coastal areas (e.g., Maryland, California, Florida, etc.) while the states with the highest DE participation rates are in the middle of the country (e.g., Iowa, Wyoming, Kansas, etc.).

OPEN IN VIEWER

Table 1

Summary Statistics for Advancement Placement (AP) and Dual Enrollment (DE) Participation and Racial Gaps

	n	M	SD	25th percentile	Mdn	75th percentile
Overall AP and DE participation
AP participation	11,741	0.110	0.123	0.000	0.080	0.180
DE participation	11,741	0.113	0.137	0.004	0.070	0.170
Racial gaps in AP participation rates
White-Black gap	3,550	0.098	0.091	0.040	0.090	0.147
White-Hispanic gap	4,625	0.069	0.085	0.018	0.060	0.114
Racial gaps in DE participation gaps
White-Black gap	3,134	0.047	0.072	0.006	0.034	0.079
White-Hispanic gap	4,211	0.042	0.079	0.002	0.028	0.072

Note. The inclusion criteria used to construct the sample for overall AP (DE) participation is different from those used to construct the sample for racial enrollment gaps. Specifically, the sample for overall AP (DE) participation includes all school districts with at least 20 total student enrollments. The sample for racial enrollment gaps further restricts to districts with sufficient number of students for the two racial groups used to calculate the specified racial enrollment gap. For example, the sample for White-Black AP participation gap includes school districts with at least 20 White students and 20 Black students. Additionally, considering that it would be misleading to report a zero percentage point gap if the district had zero AP participation for both White and Black students, we further restrict the sample to districts that have a nonzero participation rate for at least one subgroup used in calculating the racial enrollment gap. Due to these additional sample inclusion criteria, the samples used for racial gaps are substantially smaller than the sample for the overall AP (DE) participation. Supplementary Appendix B describes these selection criteria in more detail. It should be noted that although there are substantial decreases in the number of districts as we restrict the analytic samples, these restrictions mainly exclude districts with very few students overall; as a result, districts that remain in the sample still cover at least two thirds of total students enrolled nationwide.

OPEN IN VIEWER

Figure 1.

Distribution of Advanced Placement (AP) and Dual Enrollment DE participation rate among school districts, by state.

To provide a more detailed overview of the distribution of AP and DE participation rates across all the districts in our sample, Figure 2 presents choropleth maps of district AP and DE participation rates. These maps show districts geographically in progressively darker shades of gray, based on quintiles of district AP and DE participation rates. ¹¹ Since we use the same coloring scheme for both the AP and DE choropleth maps, it allows us to compare participation rates between the two maps to gauge places of higher and lower AP or DE participation as well as regions that primarily offer AP or DE programs by comparing the color of a specific district or region between the AP and DE maps. For example, consistent with the patterns shown in Figure 1, districts in the middle of the country have higher participation rates in DE compared with AP, whereas districts in the coastal areas tend to have higher participation rates in AP.

OPEN IN VIEWER

Figure 2.

Maps of district Advanced Placement (AP) and Dual Enrollment (DE) participation rates.

Patterns of Racial Gaps in AP and DE Participation

The middle and bottom panels in Table 1 show summary statistics for racial gaps in AP and DE participation among school districts that met our inclusion criteria. ¹² There are larger racial gaps in AP participation (9.8 and 6.9 percentage points for the White-Black and White-Hispanic gaps, respectively) than in DE participation (4.7 and 4.2 percentage point gaps for the White-Black and White-Hispanic gaps, respectively). The standard deviation for these gaps ranges between 7.2 to 9.1 percentage points, suggesting that there is large variation across districts in the size of racial gaps.

Figure 3 illustrates the distribution of district racial gaps in AP and DE participation within each state. Separately for White-Black and White-Hispanic gaps for AP and DE participation, states are ranked in descending order by the states’ median district gap. The figure reveals three main patterns. First, the 25th percentile of racial gaps are higher than zero in most of the states, suggesting that nationwide, the majority of districts have racial equity gaps in both AP and DE participation. This is more clearly shown in Figure 4 which presents choropleth maps of district White-Black and White-Hispanic gaps in AP and DE participation rates. ¹³ Second, there is substantial variation between districts within a state. At one extreme, some school districts have White-Black and White-Hispanic gaps that are greater than 50%. At the other end, however, some school districts actually have larger minority student enrollment in AP and DE programs than White enrollment. ¹⁴ Finally, it seems that a number of states that have the highest AP and DE participation overall, as shown in Figure 1, also have relatively higher racial gaps in AP and DE enrollment. For example, among the top 10 states in overall AP enrollment, seven of them are among the top one third of states with the largest White-Black AP enrollment gaps.

OPEN IN VIEWER

Figure 3.

Distribution of racial equity gaps in Advanced Placement (AP) and Dual Enrollment (DE) participation among school districts, by state.

OPEN IN VIEWER

Figure 4.

Quintile maps of district racial equity gaps in Advanced Placement (AP) and Dual Enrollment (DE) participation.

In light of the results presented above, we then examine whether there are districts where minority students have both high AP and DE participation rates and where White-minority gaps are small. In identifying these “star districts,” we first narrow down to districts where the AP (DE) participation rates among the specified minority group reach above the national median. We then further zoom in on districts with a White-minority gap below 1 percentage point in AP (DE) enrollment rates to identify districts where minority students do well in both absolute levels and relative to their White peers. Among the thousands of districts examined, we identify 318 star districts in term of AP enrollment among Black students (i.e., districts that have above-median AP enrollment among Black students and also below 1 percentage point White-Black AP participation gap), 649 districts in AP enrollment among Hispanic students, 595 districts in DE enrollment among Black students, and 968 in DE enrollment among Hispanic students. The overlap among the four categories of districts is fairly small, where only 15 districts are “stars” in all four categories.

To better understand the characteristics of districts with high minority enrollment rates and smaller racial gaps, Table 2 provides the summary statistics for these “star districts” using both the district- and state-level variables grouped into six broad categories. To ease interpretation, we standardize all the continuous variables (such as number of AP courses offered). Thus, a positive value for a continuous variable indicates that the districts have an average value that is above the national average. In a similar vein, for the binary variables (such as whether the district is in a state with strong financial incentives for AP/DE participation), we report the difference between the average value and national average.

OPEN IN VIEWER

Table 2

Districts With Above-Median Enrollment Rates Among Black/Hispanic Students and <1 Percentage Point Racial Enrollment Gap

	White-Black gap		White-Hispanic gap
District-/state-level variable	AP	DE	AP	DE
Academic preparation (pre–high school)
Average district achievement: Grades 3–8	−0.097	0.040	0.054	0.054
White-minority achievement gap	−0.889	−0.226	−0.607	−0.274
Family socioeconomic background
Proportion receiving FRPL in public high schools	0.034	−0.005	−0.022	−0.027
Proportion of adults with a BA+ degree or higher	−0.010	0.196	0.004	0.043
White-minority gap: proportion receiving FRPL	0.259	0.348	0.127	0.175
White-minority gap: proportion of adults with BA+	−0.176	−0.138	−0.057	0.016
Between-high-school segregation in district
Between-school free lunch/not free lunch segregation	0.006	0.149	0.016	0.014
Between-school segregation, White-minority	−0.176	−0.007	−0.104	−0.051
Racial composition among high school students in district
Proportion Black in district	0.153	0.258	0.037	0.047
Proportion Hispanic in district	0.291	0.181	0.171	0.185
Average characteristics of high schools in a district
Proportion of enrollment in urban schools	0.519	0.589	0.355	0.326
Proportion of enrollment in magnet schools	0.034	0.068	−0.005	0.035
Proportion of enrollment in gifted/talented programs	0.057	0.140	0.147	0.103
Average student-counselor ratio	0.006	−0.014	0.041	0.053
Average student-teacher ratio	0.119	0.152	0.106	0.115
Per-pupil instructional expenditures	−0.088	−0.024	−0.168	−0.112
Distance to nearest public institution	−0.237	−0.279	−0.151	−0.140
Average number of AP courses offered per school	0.559	0.612	0.538	0.410
State-level policies
Moderate accountability for access	0.000	0.039	0.023	−0.004
Strong accountability for access	0.094	0.012	0.133	0.083
Moderate financial support	0.001	0.049	−0.037	0.007
Strong financial support	0.041	0.020	0.080	0.040
Moderate accountability for student outcomes	—	0.055	—	0.063
Moderate accountability for student outcomes	0.052	−0.012	−0.010	0.023

Note. This table includes districts where the AP (DE) participation rates among the specified minority group is above the national median on average and the White-minority gap in AP (DE) is below 1 percentage point. Each cell provides the summary statistic for the relevant explanatory variable. All continuous variables are standardized. Therefore, a positive value indicates that the districts have an average value that is above the national average. For the binary variables (i.e., state-level policies), we report the difference between the average value and the national average. FRPL=free or reduced-price lunch; AP = Advancement Placement; DE = Dual Enrollment; FRPL = free or reduced-price lunch.

Although the summary statistics vary widely across the four groups of “star districts,” shared patterns still emerge for a number of district-level and state-level variables. Overall, it seems that these districts tend to have substantially smaller White-minority gaps in both pre–high school achievement and income (as measured by eligibility to free or reduced-price lunch at school). They also tend to be more racially diverse than the national average, more likely to be located in urban areas, closer to local postsecondary institutions, and have a greater number of AP courses offered. Interestingly, these districts do not seem to be necessarily better resourced in general than the national average. For example, districts in all four categories have a student-teacher ratio that is above the national average, and a per-pupil instructional expenditures below the national average. Finally, these districts seem to concentrate in states with stronger accountability regarding access and financial support for AP and DE participation.

Taken together, these raw descriptive patterns seem to suggest that star districts tend to have lower racial baseline gaps, greater access to college acceleration opportunities, and have stronger financial support. In the next section, we use regression adjusted models to identify the relationship between specific variables and AP (DE) gaps holding other factors constant.

Correlates of AP and DE Participation

We begin by estimating the correlations between these factors and overall AP and DE participation in a district using fractional logit regression models explained in the section “Empirical Model for Exploring Correlates of Racial Gaps.” Since all the continuous variables are standardized, the coefficient of a continuous variable indicates the changes in AP or DE participation rates given a 1–standard deviation change in that variable. The results are presented in Table 3 and reveal three general patterns. First, a handful of local-level factors are correlated with greater levels of participation in both AP and DE programs. Specifically, districts that have greater levels of between-school income segregation are related to lower levels of participation in AP and DE programs. In contrast, per-pupil instructional expenditures and academic acceleration opportunities before high school, such as the proportion of students enrolled in gifted and talented programs, are associated with higher levels of AP and DE participation.

OPEN IN VIEWER

Table 3

Regression Estimates Predicting Advancement Placement (AP) and Dual Enrollment (DE)

Variable	AP	DE
Academic preparation (pre–high school)
Average district achievement: Grades 3–8	0.004 (0.002)	0.022*** (0.003)
Family socioeconomic background
Proportion receiving FRPL in public high schools	−0.001 (0.002)	−0.008*** (0.003)
Proportion of adults with a BA+ degree or higher	0.016*** (0.002)	−0.018*** (0.003)
Between-high-school segregation in district
Between-school free lunch/not free lunch segregation	−0.002** (0.001)	−0.003** (0.001)
Racial composition among high school students in district
Proportion Black in district	−0.001 (0.002)	−0.007*** (0.002)
Proportion Hispanic in district	0.001 (0.002)	−0.002 (0.002)
Average characteristics of high schools in a district
Proportion of enrollment in urban schools	0.008*** (0.002)	−0.000 (0.002)
Proportion of enrollment in magnet schools	0.000 (0.001)	−0.006*** (0.002)
Proportion of enrollment in gifted/talented programs	0.004*** (0.001)	0.005*** (0.002)
Average student-counselor ratio	−0.004*** (0.002)	−0.000 (0.002)
Average student-teacher ratio	0.021*** (0.004)	−0.030*** (0.006)
Per-pupil instructional expenditures	0.007*** (0.002)	0.006** (0.003)
Distance to nearest public institution	−0.013*** (0.004)	0.001 (0.002)
Average number of AP courses offered per school	0.047*** (0.004)	−0.021*** (0.002)
State-level policies
Moderate accountability for access	−0.001 (0.002)	0.021*** (0.004)
Strong accountability for access	0.032*** (0.004)	0.059*** (0.005)
Moderate financial support	0.040*** (0.005)	−0.018*** (0.005)
Strong financial support	0.045*** (0.005)	−0.016*** (0.005)
Moderate accountability for student outcomes	—	0.030*** (0.005)
Strong accountability for student outcomes	0.004 (0.003)	0.000 (0.004)
N	8,746	8,746

Note. The sample includes school districts with nonmissing values for the explanatory variables. All the continuous explanatory variables are standardized; thus, the coefficient indicates the changes in AP or DE participation rates given a one standard deviation change in that variable. Average high school characteristics in a district include high schools only. All the state-level policies are dummy variables and use weak accountability as the reference group. See Supplementary Appendix Table A2 for details about the coding scheme for state-level policies. FRPL = free or reduced-price lunch.

*p < .10. ^**p < .05. ^***p < .01.

Second, there are a number of cases where the local-level variables are associated with both AP and DE enrollment, but in opposite directions. For example, the proportion of educated adults with school children in a district is associated with an increase in AP participation rates but a decrease in DE participation rates. Similarly, the average number of AP courses offered at a school is associated with higher participation rates in AP programs but are negatively associated with DE enrollment. These patterns suggest that AP and DE programs may serve as substitutes to each other when schools are allocating resources among different college acceleration programs. These patterns also suggest that students with more educated parents may interpret the value these programs have to offer differently.

Last, among the state-level variables, having strong accountability mandates seems to be an important predictor for both AP and DE participation— districts in states with strong accountability and mandates for access to AP (DE) programs are associated with higher AP (DE) participation rates than states without or with weak accountability. Additionally, strong financial incentives are important for AP participation. Specifically, states that offer moderate to strong financial incentives for participating in AP programs, such as reducing or waiving exam fees for low-income students, have AP enrollment rates that are five percentage points larger than states without or with weak financial incentives.

Correlates of Racial Gaps in AP and DE participation

Table 4 presents results that further use the local- and state-level variables to predict racial gaps in AP and DE participation. In addition to all the variables included in Table 3 that examines general AP and DE participation, we further add four additional variables into the model to better capture racial gaps in pre–high school academic achievement and in their socioeconomic status. ¹⁵ Columns 1 and 2 present the results for White-Black enrollment gaps and columns 3 and 4 present the results for White-Hispanic gaps. Following our conceptual framework presented in the section “Correlates of Racial Gaps in AP and DE Participation,” below, we discuss the key findings for each of the six categories of variables examined.

OPEN IN VIEWER

Table 4

Regression Estimates Predicting White-Black and White-Hispanic Participation Gap

Local-/sate-level variable	White-Black gap		White-Hispanic gap
	AP	DE	AP	DE
	1	2	3	4
Academic preparation (pre–high school)
Average district achievement: Grades 3–8	0.005 (0.003)	0.020*** (0.004)	0.002 (0.001)	0.004** (0.002)
White-minority achievement gap	0.038*** (0.002)	0.011*** (0.002)	0.016*** (0.001)	0.008*** (0.001)
Family socioeconomic background
Proportion receiving FRPL in public high schools	0.008** (0.003)	0.006* (0.003)	0.004*** (0.001)	−0.001 (0.002)
Proportion of adults with a BA+ degree or higher	0.004 (0.003)	−0.015*** (0.003)	0.004*** (0.001)	−0.004** (0.002)
White-minority gap: proportion receiving FRPL	0.010* (0.006)	−0.003 (0.005)	0.007*** (0.002)	−0.003* (0.002)
White-minority gap: proportion of adults with BA+	0.005** (0.002)	0.004* (0.002)	0.003** (0.001)	0.006*** (0.001)
Between-high-school segregation in district
Between-school free lunch/not free lunch segregation	−0.003 (0.002)	0.000 (0.002)	−0.003** (0.001)	0.001 (0.001)
Between-school segregation, White minority	−0.001 (0.002)	−0.004** (0.002)	−0.001 (0.001)	−0.002** (0.001)
Racial composition among high school students in district
Proportion Black in district	0.011*** (0.002)	0.007*** (0.002)	0.006*** (0.001)	0.003** (0.001)
Proportion Hispanic in district	0.002 (0.002)	−0.001 (0.002)	0.001 (0.001)	0.001 (0.001)
Average characteristics of high schools in a district
Proportion of enrollment in urban schools	0.007*** (0.002)	−0.001 (0.002)	0.002** (0.001)	−0.001 (0.001)
Proportion of enrollment in magnet schools	0.001 (0.001)	−0.002*** (0.001)	−0.000 (0.000)	−0.000 (0.000)
Proportion of enrollment in gifted/talented programs	0.005*** (0.002)	−0.002 (0.001)	0.001 (0.001)	−0.001 (0.001)
Average student-counselor ratio	−0.004** (0.002)	0.006** (0.002)	−0.001 (0.001)	−0.001 (0.001)
Average student-teacher ratio	−0.005 (0.003)	−0.011** (0.005)	−0.003 (0.002)	−0.005*** (0.002)
Per-pupil instructional expenditures	0.007*** (0.002)	0.000 (0.003)	0.004*** (0.001)	0.002 (0.001)
Distance to nearest public institution	−0.012* (0.007)	−0.005 (0.007)	0.002 (0.002)	0.003 (0.003)
Average number of AP courses offered per school	0.023*** (0.002)	−0.008*** (0.002)	0.007*** (0.001)	−0.004*** (0.001)
State-level policies
Moderate accountability for access	0.009** (0.004)	0.012** (0.006)	0.001 (0.002)	0.006** (0.003)
Strong accountability for access	0.008* (0.004)	0.019*** (0.006)	−0.002 (0.002)	0.010*** (0.003)
Moderate financial support	0.008 (0.007)	−0.019*** (0.006)	0.009** (0.004)	−0.005** (0.002)
Strong financial support	0.000 (0.008)	−0.037*** (0.006)	0.007 (0.005)	−0.014*** (0.003)
Moderate accountability for student outcomes		0.001 (0.005)		−0.003 (0.002)
Strong accountability for student outcomes	0.018*** (0.004)	0.006 (0.005)	0.007*** (0.002)	0.000 (0.002)
N	2,117	1,885	2,554	2,188

Note. The sample includes school districts with nonmissing values for the explanatory variables. All the continuous explanatory variables are standardized; thus, the coefficient indicates the changes in AP or DE participation rates given a one standard deviation change in that variable. Average high school characteristics in a district include high schools only. All the state-level policies are dummy variables and use weak accountability as the reference group. See Supplementary Appendix Table A2 for details about the coding scheme for state-level policies. AP = Advanced Placement; DE = Dual Enrollment; FRPL = free or reduced-price lunch.

*p < .10. ^**p < .05. ^***p < .01.

Variation Across Districts

First, our results suggest that the majority of districts have nontrivial racial gaps in both AP and DE programs, with more pronounced gaps in AP than DE and with wider White-Black gaps than White-Hispanic gaps in both programs. Yet, even for White-Black gaps in AP participation, the largest gap on average among the four, our descriptive findings identify several hundred districts where Black students both have high participation rates—defined as reaching above the national median—and White-Black participation gaps are near-zero or even negative. The variation we display presents an opportunity to examine and scale innovations to both expand access to college acceleration opportunities and close gaps in participation. A sensible next step would be for researchers, practitioners, and policymakers to collaborate in identifying and documenting the policies and practices among districts (and their partner colleges in the case of DE coursework) that are highly effective in serving students through college acceleration programs.

Choice Between AP and DE Programs

Second, we find suggestive evidence that AP and DE programs may serve as substitutes to each other for schools—when they are allocating resources among different college acceleration programs, and for parents—when they are considering which program to enroll their children in. While there have been a growing number of studies that assess the benefits of AP and DE separately, there is less information about what factors families and students consider when they choose between different types of college acceleration programs. To help students and schools make informative decisions, future research may wish to examine whether the relative benefits of different college acceleration programs vary depending on specific student characteristics, such as academic ability, family socioeconomic background, and type of college intended.

Predictors of Equity Gaps

Third, among all the factors examined, the White-minority achievement gap prior to high school is the strongest predictor of racial gaps in AP and DE participation. Once we take into account the White-minority achievement gap, White and minority students are predicted to be equally likely to participate in AP and DE classes. This finding is consistent with a number of other studies that identified substantially smaller or even reversed racial gaps in models that condition on measures of academic achievement (e.g., Conger et al., 2009). We also find that conditional on the achievement gap, differences in district- and state-level contexts are only able to explain a small amount of variation in equity gaps. However, it is not entirely clear how to unpack this finding.

One possible explanation is that the pre–high school achievement gap has “caused” minority students to have a smaller chance to be admitted into AP and DE programs. For example, the emphasis on performance criteria in determining a student’s eligibility could serve as a strong obstacle for many students to take advantage of AP and DE opportunities and influence minority students disproportionately. A potential problem with this approach is that students’ school grades only have low to moderate correlations with their performance on AP examinations, ¹⁶ suggesting that a student’s probability of success in AP programs may vary along dimensions other than school grades (College Board, 1998). Thus, incorporating additional measures of students’ likelihood of success in AP and DE programs and multiple eligibility criteria for AP and DE participation would enable districts to improve the accuracy of the screening process and may also have implications for the racial gaps in AP and DE participation (Richardson et al., 2016).

An alternative explanation, however, is that the factors underlying racial achievement gaps, such as school resources and policies, largely overlap with the factors that explain racial gaps in AP and DE participation. This type of scenario suggests that a single policy or practice may be insufficient to eliminate or narrow the current racial gaps in AP and DE participation, and the goal of integrating minority students into college acceleration programs can hardly be achieved alone without comprehensive and persistent efforts to reform the social and school structure where racial achievement gaps are rooted.

Resources and Access Alone May Be Insufficient

Fourth, our results indicate that a handful of local factors that are associated with higher overall program participation, such as a larger number of AP courses offered and higher per-student instructional expenditures, are associated with wider racial gaps in AP enrollment. This is consistent with our descriptive results that districts with higher AP and DE participation overall also tend to have larger racial gaps in AP and DE enrollment. Along a similar vein, districts in states with stronger mandates in access to AP and DE programs have larger White-minority gaps than districts in states with weak mandates, indicating that White students, as compared with minority students, might be in a better position to take advantage of these types of incentives.

An important implication of these findings is that districts with greater resources surrounding college acceleration programs, while achieving strong AP and DE participation overall, may also engender racial disparity if there are inadequate efforts to prioritize equitable access to and success in such programs. For example, racially minoritized students—particularly those from lower-income families—may, as a result of accumulated educational disadvantage, experience other barriers in accessing college acceleration programs even when they meet the performance criteria, such as limited guidance about how to take advantage of AP and DE programs. In addition, research indicates that better-resourced schools are often more likely to use academic tracking and other forms of homogeneous ability grouping, such as gifted programs (Loveless, 2009). While these programs may provide targeted instruction early on to prepare students for more advanced coursework such as AP, they may also serve as a driver of segregation within schools along lines of race and family socioeconomic background (Roda, 2015; Rui, 2009). To the extent that curricular rigor influences students’ educational aspirations and college choices (Cabrera & La Nasa, 2000; Hauser & Anderson, 1991), the underrepresentation of minority students in gifted programs and high ability tracks could engender racial disparities in their probability of participating in college acceleration programs even when they are eligible to.

Thus, in addition to increasing educational resources and offering abundant college acceleration opportunities, it is also critical for districts and schools to be committed to integration and make intentional efforts to alleviate racial gaps in utilizing college acceleration resources. A number of studies have documented several promising efforts in this regard. Among these discussions, one promising avenue that has received great emphasis is more effective advising processes from teachers and school counselors, such as targeting students for advising, providing advisors with professional development opportunities to learn about college acceleration programs, and making information easily available to both parents and students (e.g., Flores & Gomez, 2011; Goldhaber et al., 2015; Kerr, 2014; Kettler & Hurst, 2017; Quintero, 2019; Whiting & Ford, 2009). For example, Kerr (2014) documented the efforts made by teachers in a high school to close the non-White/White AP enrollment gap. Specifically, the teachers created a space for middle and high school teachers to collaborate which, first and foremost, made middle school teachers aware of the racial disparities that existed in advanced courses and their role in closing them.

In addition, given the critical role teachers’ recommendation plays in determining a student’s eligibility to AP and DE programs, it is important that school districts provide professional development opportunities around explicit and implicit bias training to ensure that those who advise students for college acceleration programs do not hold biases that disadvantage underrepresented minorities. Prior literature has documented the way in which implicit biases can foster negative attitudes and lead to stereotypical beliefs about individuals from different backgrounds (Carnes et al., 2012; Gershenson & Dee, 2017). A number of studies focused on teaching and advising practices surrounding AP and DE and identified mixed findings regarding the existence of these biases and how they affected who was targeted and encouraged to pursue college acceleration opportunities. For example, based on interview data with teachers and counselors in Texas, Miller et al. (2018) found little evidence to support the existence of implicit biases or discrimination in advising practices. On the other hand, using an adapted audit study where the authors asked a sample of school counselors to make AP recommendations based on transcripts that had identical information but were randomly assigned student names suggestive of gender and race, Francis et al. (2019) found that Black female students were less likely to be recommended for AP Calculus and were rated as being the least prepared. Other research has also shown the extent to which teacher biases and expectations are systematically related to student racial/ethnic background and how this affects the course recommendation process (Archer-Banks & Behar-Horenstein, 2012; Campbell, 2012; Francis et al., 2019; Grissom & Redding, 2016; Ho & Cherng, 2018; Oakes, 2005). Implicit bias training has the potential to make teachers and counselors aware of their role as gatekeeper and learn about ways to move past biases so that all eligible and potentially successful students are encouraged to enroll in college acceleration courses.

Policies to Remove Financial Barriers

Finally, districts in states with stronger financial support to DE participation are associated with smaller White-minority DE participation gaps. This is not surprising given that the financial hurdles in regard to DE participation faced by students from less affluent families have been well documented (e.g., Karp et al., 2004; Museus et al., 2007; Roach et al., 2015). These barriers include both direct costs of participation, such as the tuition associated with college courses, and indirect costs, such as textbooks and transportation to local colleges. Results from our analysis further highlights the importance for policymakers to take into consideration the financial constraints of potential participations, especially those from less affluent backgrounds, in the allocation of funding and targeted support.

Limitations

Our study is subject to a number of limitations and caveats. First, dual enrollment participation is new for the 2015–2016 CRDC data collection. Dual enrollment programs and courses are offered in different modalities and taught by both high school and college instructors, compared with AP coursework, which is more standardized. Although the CRDC has provided detailed definitions regarding what counts as DE and requires that schools and districts certify the accuracy of their data on submission, there may still be measurement errors (beyond what may be expected in this type of national administrative data) on the reporting of DE participation.

Additionally, the broad definition of DE used by the CRDC also prevents us from differentiating between different types of DE programs in this study. Since DE programs vary substantially in multiple dimensions, the type of students enrolled in different DE programs may vary considerably. As a result, the way various local- and state-level factors predict overall participation and racial gaps in DE participation may largely depend on the specific characteristics of particular DE programs. Therefore, it is critical for future data collection to include detailed information regarding the specific attributes of a DE course or program.

Furthermore, since the primary focus of our article is on racial gaps in AP and DE participation, we had to restrict our sample to districts with at least one racial group having a nonzero participation rate. As a result, our findings may not speak to the factors that are associated with the availability of AP (DE) programs in general, especially local factors that lead to complete absence of AP or DE opportunities in a district.

Finally, though the focus on this study was on access to college acceleration programs, educational leaders and policymakers should also be focused on student success in AP and DE courses, tracking student progression and momentum into and through college. For example, previous research, including documentation from the College Board, indicates that there remain racial and socioeconomic gaps in AP exam pass rates, even among AP course participants (College Board, 2017). Our study uses a broader definition of access to AP, counting students as participating if they ever took an AP course. As a result, the racial gaps reported in our study are likely to be larger if we were able to also take into account disparities in the number of students who take and pass the AP exam.

Despite these limitations and caveats mentioned above, our analysis provides several important policy implications regarding college acceleration programs. Although we cannot make any causal claims based on the observed correlations between racial gaps in AP and DE participation and district-level and state-level factors, they point to potential channels that may affect minority students’ participation in college acceleration opportunities. In light of the positive relationship between these opportunities and important student educational outcomes, such as subsequent college enrollment and performance, channels identified in this study could contribute to racial gaps in education attainment and labor market opportunities. Further analysis of these associations through longitudinal research designs that enable researchers to draw causal conclusions would be a valuable direction for future work and has the potential to inform broader policy discussions around college access and completion.