Impacts of School Racial Composition on the Mathematics and Reading Achievement Gap in Post Unitary Charlotte-Mecklenburg Schools

Abstract

This is a longitudinal study of the change in the academic achievement gap between African American and European American students from elementary to high schools with large administrative data from a school district in the United States. Analysis of variance between eight tracks of students defined by the school environment of isolated schools or diverse schools indicated that middle school is a critical period for closing the achievement gap and that students who stayed in diverse schools from elementary to high schools benefited the most in both reading and mathematics standardized test scores. Multilevel linear growth models show that staying in isolated elementary and middle schools has a negative impact on the students’ reading achievement and their annual growth rate in mathematics for all students regardless of race.

Keywords

educational policy multicultural education school improvement social justice

The long-standing and intractable difference between many African American students and their European American peers in grades, standardized test scores, course selection, dropout rates, and college-completion rates is not new (Barton & Coley, 2009). The “gap” in academic performance between groups of students, however, is one of the most pressing education-policy challenges that schools currently face (National Governors Association Center for Best Practices, 2009). It has been a focal point of education reform efforts at many levels for many years. For example, yearly National Assessment of Educational Progress (NAEP) report cards provide updated information on the Black–White achievement gap of elementary and secondary schools in the United States and the differences are of great interest to educational researchers and policy makers. Braun, Chapman, and Vezzu (2010) compared the pre and post No Child Left Behind (NCLB) data from NAEP and concluded that NCLB only had a very modest impact on the rates of improvement for African American students and the change of Black–White achievement gap. In another recent study, Barton and Coley (2010) reviewed achievement gap studies on three periods of American history trying to identify possible factors that might have an impact on the large reductions in the achievement gaps of the 1970s and 1980s and the halt of the closing of the gaps of the 1990s and the 21st century. Their findings were inconclusive with possible reasons including but not limited to family and demographic changes, early education and nutrition, course taking and tracking, desegregation, class size, minimum competence testing, social capital, and disappearing African American fathers. They urged “the research and policy communities to put this question high on the list of priorities, and encourage funders to make the resources available” (Barton & Coley, 2010, p. 4). This study examined the possible impact of school-level characteristics on the Black–White academic achievement gap from elementary through high schools with longitudinal data.

School Segregation

School segregation and integration have been a political debate over the last half-century. The Brown v. Board of Education (1954) held that separate but equal was unconstitutional in education because separate could not be equal. In 1954, the U.S. Supreme court called for the end of racial segregation in schools. Wake county in North Carolina puts substantial efforts at racial desegregation in 1965 after the Elementary Education Act made the district choose between receiving federal funds or staying segregated. Wake county implemented a race-based reassignment plan to make sure that each school reflect the racial composition of the county. In 1994, however, the U.S. Supreme court ruled that the practice of desegregation based solely on race is not allowed. Starting from the 2000–2001 academic year, Wake county switched to reassigning students on the basis of family income rather than race. To investigate how the reassignment of students to schools impact the students achievement, Hoxby and Weingarth (2005) used data (total reading and mathematics End-of-Grade [EOG] scale score) on third through eighth graders in Wake county in North Carolina from the 1994–1995 through 2002–2003 school years. They found strong evidence that peer’s ethnicity has only slight effects on students’ academic achievement. A student’ score is unaffected by the mean of his/her class’s previous year test scores when other fixed effects are controlled. Hoxby and Weingarth’s (2005) results support that a student will have higher achievement when he or she is surrounded by peer with similar characteristics. Hoxby and Weingarth (2005) thought that this is probably because the environment is made to cater to these students’ needs and teachers might be able to organize lessons and materials around the learning style of a student if there is a critical mass of his or her type. They conclude that “fears of racial, ethnic, and economic desegregation are overblown, . . . policy makers who pin all their hopes for achievement on such desegregation are unduly optimistic” (p. 30).

Charlotte-Mecklenburg Schools (CMS) was sued by a parent because the child was denied entrance to a magnet program based on race. This case forced CMS to adopt a neighborhood-based school choice plan in December 2001, and new school boundaries were drawn for the academic year 2002. Using administrative records from CMS that span Kindergarten through 12th grade and the academic years from 1995 to 2010, Billings, Deming, and Rockoff (2012) examined the impacts of re-zoning on student academic achievement. They found that both White and African American students scored lower on high school exams when they were assigned to schools with more minority students and concluded that the re-zoning widened racial inequality. Specifically, Billings et al. (2012) estimated that re-zoning in CMS widened the racial gap in mathematics scores by 0.04 standard deviations and larger gaps for students in segregated schools. Recently, Mickelson, Bottia, and Lambert (2013) conducted a meta-analysis of the effect size of school racial composition on mathematics outcomes. The meta-analysis included 25 empirical studies with 98 regression coefficients and led to a conclusion that school racial isolation has a small but substantively meaningful negative effect on school-level mathematics achievement. These researchers also noted that the effects of racial segregation at school level were stronger in secondary schools in comparison to elementary schools and the achievement gaps widened as students moved up to higher grade levels.

Texas schools experienced the dual pressures of court-ordered desegregation decrees and dramatic demographic shifts resulting from suburbanization, immigration, and rapid population growth. The combination of these factors makes it possible that African American students in Texas nowadays enroll in schools more diverse in ethnicity. Therefore, Hanushek, Kain, and Rivkin (2009) used data on fourth through seventh graders in Texas public schools and found that the percentage of African American students at the school level had a negative effect on mathematics achievement growth for African Americans students in the upper half of the ability distribution. The racial composition effects for high ability African American students appear to be much stronger than those for Caucasian, Hispanic, and low ability African American students. Nonetheless, Hanushek et al. (2009) cautioned that the policy implications of their findings are unclear because school quality was not measured in the study and

because of both the imbalance in the distribution of students across jurisdictions and the possibility that expanded exposure to nonblacks following additional desegregation activity could have a much different effect on achievement than that estimated from the current distribution of students among schools. (p. 29)

Achievement Gap, Ethnicity, Socioeconomic Status, and School Characteristics

Ethnicity

The academic achievement gap between minority students (specifically African American and Hispanic students) and European American students has been researched extensively ever since the Coleman Report in 1966 but not yet reached consensus. Sufficient studies have pointed out the fact that African American students and Hispanic American students had significantly lower academic achievement in comparison to European American and Asian American students from elementary school (Fryer & Levitt, 2004, 2006) to middle school (Clotfelter et al., 2009) and high school (Rampey et al., 2008). The known gap in achievement still exists (Barton & Coley, 2009).

Quite a few theories exist to explain the experiences of African American students in relation to the well-known achievement gap (Ford et al., 2008). The Attitude-Achievement Paradox, first suggested by Mickelson (1990), gives structure to the internal conflicts that African Americans face concerning education (Ford et al., 2008). It suggests that African American students will verbalize that education is important, when in truth they have little faith in the ability of hard work to significantly change their lives (Ford et al., 2008). In effect, they view the opportunity structure as not rewarding them for diligent work in educational pursuits based on their life experiences (Mickelson, 1990).

Ogbu’s (1987) theory of Peer Pressures and Steele’s (1997, 1999) theory of Stereotype Threat are also pivotal theories within the body of African American achievement gap research. These theories suggest that African American students do not want to “act White” and would not perform to their greatest ability out of the fear of the negative stereotype regarding academic ability (Ford et al., 2008). To illustrate this point, the Barton and Coley (2009) shows that gaps still persist in participation in the Advanced Placement (AP) program between European American and African American students.

The Secondary Resistance theory (Ogbu, 1987) attempted to explain the low academic achievement of African American students by these students’ resistance to values and beliefs of the majority, who were the perceived oppressor of their culture in historical context. African American students are referred to as an involuntary minority. The ability of minority students to show achievement or not is then based on the history of that minority group, their subordination or exploitation by the majority group, the individual’s ability to express themselves, and their ability to overcome these historical conflicts. The school, the society, and the community are all noted as factors in a minority student’s ability to achieve academically. Sometimes students are able to overcome these factors to achieve.

Some theories applicable to African American students are not relevant for Hispanic students because of their status as voluntary minorities within American culture. One of the greatest challenges for Hispanic students is the point at which they enter the American public school. Immigrant students may enter American schools with a different knowledge base and different prior experiences than American students. Studies have shown that when Hispanic students enter kindergarten, their skills are well below those of non-Hispanic White students in reading and math (Reardon & Galindo, 2009). Of entering Hispanic students in Washington State, only 47% have reached grade levels in reading and 27% in math (Valadez, 2008). There is also considerable variance in reading and math achievement among Hispanic subgroups. These differences persist through fifth grade, but become less pronounced (Reardon & Galindo, 2009).

Surprisingly, Hispanic subgroups with lower levels of math and reading skills when they enter kindergarten show the most narrowing of the gap in achievement over time. Reardon and Galindo (2009) suggest that part of the gains shown by these students in achievement is concerned with English acquisition over time and focused instruction in certain programs that are effective with English language learners. With comparison to the Black–White achievement gap that widens over time, the Hispanic–White achievement gap narrows. Keeping Hispanic students in school past the dropout age is also a challenge. In the United States, Latinos have the highest dropout rate than any other major ethnic group at 54% (Green & Winters, 2002).

In a study of undocumented immigrant Latino students, those who had higher levels of personal and environmental protective factors (e.g., supportive parents, friends, and participation in school activities) reported higher levels of academic success than those lacking the same protective factors. This rise in academic achievement seems to occur in the presence of protective factors despite specific risk factors (e.g., elevated feelings of societal rejection, low parental education, and high employment hours during school; Perez et al., 2009).

Parent involvement in schools can also be seen through a lens of ethnicity. Parental involvement is considered problematic to teachers when parents of minority children do not participate in their children’s schooling (Barton & Coley, 2009). Parents of African American or Hispanic students are much less likely to attend a school event or act as a volunteer. Another research finding with ethnicity implications is Reed’s (2009) conclusion that educational federalism creates an expectation gap between ethnic groups when proficiency cut scores for standardized tests are compared. In fourth grade, the median African American student faces a lower proficiency cut score in math and reading than the median European American student. This translates into lower expectations for African American fourth graders. The demographic differences do seem to disappear when eighth grade, the only other grade level compared, proficiency cut scores are examined (Reed, 2009). Although evidence of an expectations gap is found in this study, the author makes it clear that this gap in expectations is not monolithic (Reed, 2009).

Socioeconomic status

Socioeconomic status (SES) can mean several things when it comes to public schools. SES of a student’s family, their community, the community where the school is located and the SES of the school itself are all important factors in determining the success of the child. Coleman et al. (1966) found that a child’s family background and a school’s economic makeup are determining factors in the achievement gap between ethnicities. SES is often linked to ethnicity, as Fram et al. (2007) discovered when assessing schools in the United States. High-ethnic minority schools were found to be in high-poverty areas and low-ethnic minority schools in low-poverty areas. Children with single parents disproportionately attended high-ethnic minority schools as did children whose mothers became pregnant as a teenager. Also, children in high-ethnic minority schools disproportionately had mothers with lower levels of education and lower SES (Fram et al., 2007). Parents with lower household incomes are also less likely to attend school meetings or volunteer at their child’s school (Barton & Coley, 2009). Teachers are also much more likely to report that lack of parental involvement is a moderate or serious problem in high-poverty schools (Barton & Coley, 2009).

In communities where there is higher social capital, there are higher levels of academic achievement in students (Woolley et al., 2008). The relationship between social capital and academic achievement increases in influence as children get older. This is because older students often have more contact with community members. Inversely, poor neighborhood conditions are associated with decreases in academic achievement (Woolley et al., 2008). Regardless of the school SES, community SES, race or ethnicity, children from similar SES situations perform similarly. If students begin school with an achievement gap in a specific subject area, that gap will persist throughout their school career if not remedied (Foster & Miller, 2007). Sirin (2005) found evidence to the contrary, citing that SES was a stronger predictor of academic achievement for White students than minority students. The relationship between family SES and academic achievement was also weakest for urban schools, which are thought to be higher in minority student enrollment (Sirin, 2005). Reardon and Galindo’s (2009) meta-analysis noted an interaction between ethnicity and SES; by 5th grade some studies show that SES is not a factor for African American students’ low achievement, while it is still a major factor for Hispanics. These findings call for the need to examine the interaction factors of ethnicity and SES both at the student and school levels.

School characteristics

Perhaps least emphasized in studies concerning achievement gaps are school-level factors. The school environment is conceivably the most influential factor on a student’s achievement and the arena where all other individual student characteristics interact. A classroom, whether at the elementary or high school levels, comprised basic fundamental elements. Teachers, students, knowledge capital, and relationships react to hopefully make the classroom a place of learning, growth, and achievement.

Studies have shown that students in high-ethnic minority schools had lower levels of academic achievement than students in low-ethnic minority schools (Fram et al., 2007; Sass et al., 2010). High-ethnic minority schools tend to be schools with low social economic status. In a study of longitudinal data of Grades 3–8 students in North Carolina, Clotfelter et al. (2009) found that the Black–White achievement gap was smaller at the low end of the achievement distribution but much larger at the upper ends and explained that this phenomenon could be due to accountability programs such as NCLB. Low-performing schools could relocate limited resources to students who were failing to raise the achievement at the bottom of the distribution. Clotfelter and his colleagues found that socioeconomic factors could explain a sizable portion (30%) of Black–White test score gaps.

As the findings from previous results are not consistent, we approach the same topic with a large urban school district in the Southeastern United States. The research questions that guided this study were as follows:

How do school-level factors impact the student academic achievement gap?

Is there a statistically significant difference between African American students and Caucasian students in their achievement in reading and mathematics test scores as well as the annual growth rate from elementary to high schools?

How does the change in school environment from isolated schools to diverse schools impact African American students’ achievement in reading and mathematics test scores and the annual growth rate from elementary to high schools?

Method

Sample Selection

All students who have valid data through Grades 3–10 on reading and mathematics End-of-Grade (EOG) scores and high school Algebra I, Geometry, Algebra II, and English End-of-Course (EOC) scores selected for this study. The data include all students (n = 3,182) who started the third grade in the 1998–1999 academic year and graduated from high schools (12th Grade) in the 2007–2008 academic year within the school district. Among them are 1,764 (55%) females and 1,418 (45%) males. The ethnicity backgrounds of the students are diverse: 1,288 (41%) African American, 1,648 (52%) Caucasian, 142 (5%) Asian, 72 (2%) Hispanic, 22 (0.7%) Multiracial, and 10 (0.3%) Native American. These students are from 19 highs schools, 30 middle schools, and 85 elementary schools.

Data Analytical Procedure

Mickelson (2001) defined a racially isolated African American school as one in which the enrollment of African American students exceeded by more than 15% the school district population percentage of African American students. According to Mickelson (2001), this procedure is identical to the practice of the federal district court in 1999, and it was also consistent with the school district board policy. In the current study, the percentage of African American students in the school district ranged from 42.10 to 43.40 from 1999 to 2008. Following the same procedure as Mickelson (2001), 58.5% was used as a cut-off percentage of African American students to determine whether a particular school is racially isolated (at least 58.5%) or racially diverse (less than 58.5%). There are eight possible combinations of the school change from elementary to high schools: (1) diverse schools in elementary, middle, and high (n = 1,895); (2) isolated schools in elementary, middle, and high (n = 200); (3) diverse schools in elementary and middle but isolated in high (n = 284); (4) diverse schools in elementary, isolated schools in middle, and diverse schools in high (n = 209); (5) diverse schools in elementary, but isolated schools in middle and high (n = 348); (6) isolated schools in elementary but diverse schools in middle and high (n = 99); (7) isolated schools in elementary, diverse schools in middle, and isolated schools in high (n = 89); and (8) isolated schools in elementary and middle but diverse schools in high (n = 58).

To determine which track of school change makes a difference for the students, analysis of variance (ANOVA) was run for elementary school reading EOGs (sum of Grades 3–5), elementary school mathematics EOGs (sum of Grades 3–5), middle school reading EOGs (sum of Grades 6–8), middle school mathematics EOGs (sum of Grades 6–8), high school mathematics EOCs (sum of Algebra I, Algebra II, and Geometry), and high school English EOC. Results from the ANOVAs were used to determine which tracks of students were used for individual growth curve analysis with hierarchical linear models (HLM). Simple models were run to examine each predictor variable at a time, and then, all statistically significant predictors in the simple models were included in the complete model.

Results

Table 1 shows the statistics for reading and mathematics standardized test scores by each school change track.

Table 1.

Descriptive Statistics for African American Students on School Change Tracks.

Tracks	Elementary		Middle		High
Tracks	Read	Math	Read	Math	English	Math
1 (n = 1,895)	469.07 (21.51)	574.90 (26.37)	700.93 (19.93)	831.07 (28.03)	63.29 (5.81)	213.46 (37.63)
2 (n = 200)	447.31 (21.10)	550.69 (24.82)	678.06 (22.19)	801.23 (23.98)	58.08 (8.31)	192.08 (49.03)
3 (n = 284)	452.49 (23.24)	553.11 (26.56)	685.58 (21.89)	808.39 (27.72)	59.24 (5.81)	205.99 (49.22)
4 (n = 209)	450.79 (20.87)	555.26 (26.36)	684.22 (20.75)	808.93 (26.73)	59.29 (5.79)	189.30 (53.27)
5 (n = 348)	444.32 (24.28)	546.46 (26.93)	678.07 (22.29)	800.80 (25.31)	57.50 (6.37)	196.46 (53.09)
6 (n = 99)	458.67 (25.91)	562.72 (29.54)	689.81 (23.40)	815.29 (30.07)	60.71 (6.82)	208.39 (48.67)
7 (n = 89)	452.33 (21.84)	556.21 (25.26)	685.28 (22.04)	808.08 (26.85)	58.90 (5.77)	201.89 (53.51)
8 (n = 58)	451.43 (20.31)	555.67 (25.17)	680.98 (20.95)	807.38 (27.19)	59.16 (5.79)	187.17 (49.93)
Total (n = 3182)	461.20 (24.16)	565.78 (28.75)	693.37 (22.82)	820.84 (30.20)	61.43 (6.51)	207.04 (44.48)

Note. 1 = diverse schools in elementary, middle, and high; 2 = isolated schools in elementary, middle, and high; 3 = diverse schools in elementary and middle but isolated in high; 4 = diverse schools in elementary, isolated schools in middle, and diverse schools in high; 5 = diverse schools in elementary, but isolated schools in middle and high; 6 = isolated schools in elementary but diverse schools in middle and high; 7 = isolated schools in elementary, diverse schools in middle, and isolated schools in high; 8 = isolated schools in elementary and middle but diverse schools in high; numbers in parentheses are standard deviations.

ANOVA showed statistically significant differences between the eight tracks with respect to reading in elementary schools, F(7, 3,174) = 91.75, p < .001, partial η² = .17; mathematics in elementary schools, F(7, 3,174) = 85.39, p < .001, partial η² = .16; reading in middle schools, F(7, 3,174) = 95.14, p < .001, partial η² = .17; mathematics in middle schools, F(7, 3,174) = 98.23, p < .001, partial η² = .18; English in high schools, F(7, 3,170) = 67.17, p < .001, partial η² = .13; and mathematics in high schools, F(7, 3,174) = 19.01, p < .001, partial η² = .04. Post hoc multiple comparisons using Scheffe’s method to control Type I error rate suggested that Track 1 students (students who stayed in diverse schools from elementary to high schools) had significantly higher reading and mathematics scores than students of any other tracks in elementary schools as well as in middle schools. In high schools, students in Track 1 had significantly higher English scores than students of any other tracks and significantly higher mathematics scores than Track 2 (student who stayed in isolated schools in elementary, middle, and high schools), Track 4 (students who stayed in diverse schools in elementary schools, isolated schools in middle schools, and diverse schools in high schools), Track 5 (students who stayed in diverse schools in elementary schools, but isolated schools in middle and high schools), and Track 8 (students who stayed in isolated schools in elementary and middle schools but diverse schools in high schools). The common feature of Tracks 2, 4, 5, and 8 is that the students stayed in isolated middle schools. This result suggests that middle school might be a critical period of time for academic achievement gap. Furthermore, the consistent significant difference between Track 1 and Track 5 students suggests that staying in isolated middle and high schools might be enlarging the achievement gap. Table 1 also shows that Track 2 and Track 5 students had the lowest mean scores on all measures.

As Tracks 3, 4, 6, and 7 were never shown to be statistically significantly different from each other on any of the measures, these tracks were merged into one group and treated as the comparison group (coded as “0”). Students in Tracks 1, 2, 5, and 8 were coded as “1,” respectively, for each variable, to represent each of these tracks so that the coefficient represents the gap between each of Tracks 1, 2, 5, and 8 and the comparison group (students in Tracks 3, 4, 6, and 7). What is in common between Tracks 3, 4, 6, and 7 is that students stayed in isolated schools only once (either in elementary schools, or in middle schools, or in high schools). What is in common between Tracks 2, 5, and 8 is that students stayed in isolated schools at least twice (both elementary and middle schools or both middle and high schools). As is known, Track 1 students are those who stayed in diverse schools in elementary, middle, and high schools. As the sample sizes for students of other ethnicities are small and the focus of this study is to investigate the achievement gap between African American and Caucasian students, students of other ethnic groups were not included in the following analysis. This reduced the sample size to 2,936, of which 1,288 (44%) were African American and 1,648 (56%) were Caucasian.

North Carolina introduced new scales (second version) for mathematics in 2000–2001 and for reading in 2002–2003, so the published conversion table between the old and new scales was used to put all scores into the second version scales so that the change in scores can be compared on the same scale. High-school Algebra I, Geometry, and Algebra II EOC scores were combined and converted to a developmental scale to follow the linear trend of EOG mathematics tests. Similarly, high-school English EOC scores were converted to a developmental scale to follow the linear trend of EOG reading tests. Table 2 shows the population means and standard deviations scores (Grades 3–8) published by North Carolina Department of Public Instruction (“The North Carolina Mathematics Tests Technical Report,” 2006; “North Carolina Reading Comprehension Tests Technical Report,” 2009) and rescaled high-school mathematics and English developmental scores.

Table 2.

North Carolina Student Population Means and Standard Deviations (Grades 3–8) in Reading and Mathematics Tests (Second Version).

	Grade Level
Subject Area	3	4	5	6	7	8	High
Reading/English	245.21 (10.15)	250.00 (10.01)	253.92 (9.61)	255.57 (10.41)	256.74 (10.96)	259.35 (11.13)	262.35 (11.13)
Mathematics	248.30 (9.90)	252.90 (10.70)	256.00 (12.80)	260.00 (11.80)	263.40 (12.50)	267.10 (12.80)	270.10 (12.80)

Note. Numbers in parentheses are standard deviations.

Results from HLM are presented in Tables 3 and 4. Interpretations for each model are presented separately in the following paragraph. Models in reading/English tests are followed by models in mathematics.

Table 3.

Estimated Parameters of Growth Curve HLM Models of Reading Achievement Tests.

Parameters	Simple models				Complete model
Parameters	Coefficient	SE	t	p	Coefficient	SE	t	p
Intercept					249.75	0.48	522.27	<.001
Male	0.21	0.35	0.60	.55
African	−9.64	0.31	−31.18	<.001	−8.09	0.44	−18.58	<.001
Track 1	7.75	0.33	23.29	<.001	1.60	0.47	3.38	.001
Track 2	−6.12	0.66	−9.27	<.001	−0.86	0.71	−1.21	.226
Track 5	−6.98	0.57	−12.20	<.001	−1.77	0.61	−2.90	.004
Track 8	−4.39	1.11	−3.95	<.001	0.45	1.16	0.39	.696
Slope					2.71	0.07	41.26	<.001
Male	−0.13	0.04	−3.06	.003	−0.16	0.04	−3.80	<.001
African	−0.13	0.04	−3.01	.003	−0.10	0.06	−1.67	.094
Track 1	0.11	0.05	2.55	.011	0.01	0.06	0.12	.902
Track 2	−0.27	0.12	−2.22	.026	−0.21	0.12	−1.73	.083
Track 5	−0.07	0.07	−0.87	.383
Track 8	−0.34	0.14	−2.42	.016	−0.29	0.14	−2.09	.036

Table 4.

Estimated Parameters of Growth Curve HLM Models of Mathematics Achievement Tests.

Parameters	Simple models				Complete model
Parameters	Coefficient	SE	t	p	Coefficient	SE	t	p
Intercept					252.15	0.51	490.03	<.001
Male	1.89	0.44	4.33	<.001	1.04	0.31	3.33	.001
African	−12.23	0.38	−32.22	<.001	−10.72	0.49	−21.92	<.001
Track 1	9.05	0.41	21.88	<.001	1.57	0.47	3.34	.001
Track 2	−6.34	0.78	−8.14	<.001	−1.02	0.66	−1.55	.121
Track 5	−8.35	0.69	−12.15	<.001	−1.55	0.73	−2.13	.033
Track 8	−3.91	1.33	−2.94	.004	2.24	1.34	1.67	.095
Slope					3.43	0.04	95.27	<.001
Male	−0.02	0.06	−0.30	.761
African	0.40	0.07	6.06	<.001	0.42	0.07	5.95	<.001
Track 1	−0.12	0.07	−1.81	.070
Track 2	−0.23	0.14	−1.65	.098
Track 5	0.26	0.12	2.23	.026	0.02	0.13	0.17	.863
Track 8	−0.63	0.21	−2.92	.004	−0.85	0.22	−3.83	<.001

Reading Achievement

Growth curve modeling with HLM did not find any significant differences between male and female students with respect to their third-grade reading EOG scores, t(2,930) = 0.60, p = .55. However, male students had significantly lower annual growth rates than female students, t(20,536) = −3.80, p < .001. Female students had an annual gain of 2.71, whereas male students had an annual gain of 2.55. African American students had significantly lower performance than their Caucasian peers at Grade 3, t(2,930) = −18.58, p < .001. On average, Caucasian students’ third-grade EOG reading test score is 249.75, whereas African American students’ EOG reading test score is 241.66. The annual growth rate was not statistically significantly different from each other between African American students and Caucasian students from elementary to high schools, t(20,536) = −1.67, p = .09 when gender and track were controlled although the simple model suggested that African American students experienced a significantly lower annual growth rate in comparison to Caucasian students when ethnicity only is considered.

When students in each track were compared to each other, students in Track 1 had significantly higher third-grade EOG scores than their counterparts in the comparison group, while students in Tracks 2, 5, and 8 had significantly lower third-grade EOG scores than their counterparts in the comparison group. Nevertheless, when ethnicity was controlled, students in Track 2 and Track 8 were no longer significantly different from the comparison group in third-grade EOG reading scores, but students in Track 1 still had significantly higher third-grade EOG scores, and students in Track 5 still had significantly lower third-grade EOG scores in comparison to the comparison group. When ethnicity and gender were both controlled, only students in Track 8 had significantly lower annual growth rate than the comparison group, t(20,536) = −2.09, p = .04. This is to say that within Caucasian and African American students and within male and female students, the gap between students who stayed in isolated elementary and middle schools and those who stayed in either isolated elementary or isolated middle schools was growing from elementary to high schools. Put it in another way, staying in both isolated elementary and middles schools had a significant negative impact on the students’ reading achievement for both Caucasian and African American students and for both male and female students.

Mathematics Achievement

Male students had significantly higher third-grade mathematics EOG scores than female students, t(2,929) = 3.33, p = .001. This difference was kept the same from elementary to high schools as the annual growth rates for male and female students were not statistically significantly different from each other, t(20,541) = −0.30, p = .76. African American students had significantly lower performance than their Caucasian peers at Grade 3, t(2,929) = −21.92, p < .001. On average, Caucasian students’ third-grade EOG mathematics test score is 252.15, whereas African American students’ EOG mathematics test score is 241.43. However, African American students experienced significantly a higher annual growth rate than Caucasian students from elementary to high schools, t(20,541) = 5.95, p < .00,1 in Tracks 5 and 8.

Students in Track 1 had significantly higher third-grade mathematics EOG scores than the comparison group whether or not gender and ethnicity were controlled. This means that Track 1 students had higher start levels than the comparison group within or across ethnicity and gender. This difference was kept the same from elementary to high schools as the growth rate for Track 1 students was not statistically significantly different from that of the comparison group, t(20,541) = −1.81 p = .070. Students in Track 2 had significantly lower third-grade EOG scores than the comparison group when gender and ethnicity were not controlled but the difference was no longer significantly different from zero when gender and ethnicity were controlled. This is to say that within male and female students and within African American and Caucasian students, Track 2 students were not significantly different from the comparison group in their third-grade EOG mathematics scores. The difference was also kept the same from elementary to high schools as the growth rate for Track 2 students was not statistically significantly different from that of the comparison group, t(20,541) = −1.65 p = .098.

Students in Track 5 had significantly lower third-grade mathematics EOG scores than the comparison group when gender and ethnicity were not controlled as well as when gender and ethnicity were controlled. This means that Track 5 students had lower start levels than the comparison group within or across ethnicity and gender. However, this difference became smaller from elementary to high schools as the growth rate for Track 5 students was statistically significantly higher than that of the comparison group, t(20,541) = 2.23, p = .026. When ethnicity was not controlled; however, the difference in annual growth rate between Track 5 students and the comparison group was no longer significantly different from zero, t(20,541) = 0.17 p = .863. This means that African American students in Track 5 had lower start levels than African American students in the comparison group and this gap was kept the same.

Students in Track 8 had significantly lower third-grade mathematics EOG scores than the comparison group when gender and ethnicity were not controlled, but this difference was no longer significantly different from zero when gender and ethnicity were controlled. This means that within African American and Caucasian students and within male and female students, Track 8 students were not significantly different from the comparison at third grade in mathematics achievement. However, the growth rate for Track 8 students was significantly lower than the comparison group when ethnicity was controlled as well as when ethnicity was not controlled. This means that students in Track 8 experienced a significantly lower annual growth rate regardless of their ethnicity.

Isolated Middle School Effect

Results from HLM indicated that experience of isolated middle school is detrimental to student’s learning growth from Grade 3 to Grade 9. Next, the fixed effects of middle school experience were examined.

A principal component analysis was run to generate a factor score for reading and mathematics achievement from Grade 3 to Grade 9. A single construct explained 73.05% of the total variance of all reading and mathematics EOG scores from Grade 3 to Grade 8 and EOC scores at Grade 9. The construct was named achievement score. As this is a factor score, the mean of the score is 0 with a standard deviation of 1. The minimum value was −2.89 and the maximum value was 2.50. Students in Tracks 2, 4, 5, and 8 were put into a single group because all these tracks share a common feature: isolated experience in the middle school. A dummy variable was created to put the students into two groups: those who experienced isolated middle schools and those who did not.

A two-way ANOVA was run to examine the interaction effect of ethnicity and isolated middle school experience as well as the main effects of ethnicity and isolated middle school experience. Table 5 shows the descriptive statistics for each group.

Table 5.

Descriptive Statistics of Achievement Score for Students Classified by Ethnicity and Middle School Experience.

	Isolated Middle School (n = 670)		Diverse Middle School (n = 2,262)		Total
	M	SD	M	SD	M	SD
Caucasian (n = 1,647)	0.32	0.97	0.50	0.82	0.49	0.83
African American (n = 1,285)	−0.70	0.86	−0.57	0.81	−0.63	(0.84)
Total (n = 2,932)	−0.52	0.96	0.15	0.96	0.00	1.00

The interaction effect on the achievement score between ethnicity and middle school experience was not statistically significant, F(1, 2,928) = 0.42, p = .52, partial η² < .001. The main effect of ethnicity, F(1, 2,928) = 516.72, p < .001, partial η² = .15, and isolated middle school experience, F(1, 2,928) = 11.80, p = .001, partial η² = .01, were both found to be statistically significant. That is to say, African American students had significantly lower achievement scores than Caucasian students and students who had isolated middle school experience had significantly lower achievement scores than students who had diverse middle school experience.

Discussion

It is not surprising that African American students in this study were found to have significantly lower performance in third-grade reading and mathematics EOG scores than Caucasian students because this is consistent with the literature (Barton & Coley, 2009; Fram et al., 2007). What this study contributes to the literature is that African American students’ annual growth rate from elementary to high schools was compared to that of Caucasian students by considering gender and their experiences during elementary, middle, and high schools with respect to whether or not they stayed in diverse or isolated schools. Students in Track 8, who stayed in isolated schools in both elementary and middle schools, were not significantly different from the comparison group at the third grade in both reading and mathematics achievement but experienced significantly lower annual growth rate than the comparison group in both reading and mathematics achievement. The comparison group is consisted of students who stayed in diverse schools in elementary or middle schools or students who stayed in isolated schools at either elementary or middle school levels. Thus, we may conclude that isolated school experience in both elementary and middle schools has a negative effect on the student’s learning in both reading and mathematics.

Another interesting result is the different trend of growth in reading and mathematics for African American students. African American students had a significantly lower annual growth rate in reading but a significantly higher annual growth rate in mathematics in comparison to Caucasian students. This suggests that African American students are catching up with Caucasian students in mathematics but following further behind Caucasian students in reading as they move from elementary to high schools.

The same result from two different statistical approaches (ANOVA and HLM) that Track 1 students were consistently higher achieving in reading and mathematics from elementary to high schools than students of other tracks, especially those in Tracks 2, 4, 5, and 8, suggests that diverse school environment is beneficial to student learning in reading and mathematics and experience in middle school is critical to the development of academic achievement.

Based upon findings from this study, school policy makers should consider creating a diverse school environment for all students regardless of race. All students would benefit in their academic learning in reading/English and mathematics from elementary to high schools when the students are not segregated in schools.

Footnotes

Author’s Note

Chuang Wang and David K. Pugalee are also affiliated to University of Macau, China. And Xitao Fan is now affiliated to The Chinese University of Hong Kong, Shenzhen, China.

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was partially funded by University of North Carolina Faculty Research Grant.

Author Biographies

Chuang Wang is distinguished professor of Quantitative Research Methods and dean of the Faculty of Education at the University of Macau in China. His expertise is educational statistics and program evaluation.

Xitao Fan is presidential chair professor of Educational Research and dean of the School of Humanities and Social Science in the Chinese University of Hong Kong (Shenzhen) in China. His expertise is statistical modeling and educational measurement.

David K. Pugalee is professor and director of the STEM Center at the University of North Carolina at Charlotte in the United States. His expertise is mathematics education.

References

Barton

P. E.

Coley

R. J.

(2009). Parsing the achievement gap II: Policy information report. Educational Testing Service.

Barton

P. E.

Coley

R. J.

(2010). The Black-White achievement gap: When progress stopped. www.ets.org/research/pic

Billings

S. B.

Deming

D. J.

Rockoff

J. E.

(2012). School segregation, educational attainment and crime: Evidence from the end of busing in Charlotte-Mecklenburg (NBER working paper no. 18487). www.nber.org/papers/w18487

Braun

Chapman

Vezzu

(2010). The black-white achievement gap revisited. Educational Policy Analysis Archives, 18(2), 1–96.

Brown v. Board of Education (1954). Retrieved from https://supreme.justia.com/cases/federal/us/347/483/

Clotfelter

C. T.

Ladd

H. F.

Vigdor

J. L.

(2009). The academic achievement gap in Grades 3 to 8. The Review of Economics and Statistics, 91, 398–419.

Coleman

J. S.

Campbell

E. O.

Hobson

C. J.

McPartland

Mood

A. M.

Weinfield

F. D.

York

R. L.

(1966). Equality of educational opportunity. U.S. Office of Education, National Center for Educational Statistics.

Ford

D. Y.

Grantham

T. C.

Whiting

G. W.

(2008). Another look at the achievement gap: Learning from the experiences of gifted black students. Urban Education, 43(2), 216–239.

Foster

Miller

(2007). Development of the literacy achievement gap: A longitudinal study of kindergarten through third grade. Language, Speech, & Hearing Services in Schools, 38(3), 173–181.

10.

Fram

Miller-Cribbs

Van Horn

(2007). Poverty, race, and the contexts of achievement: Examining educational experiences of children in the U.S. South. Social Work, 52(4), 309–319.

11.

Fryer

R. G.

Levitt

S. D.

(2004). Understanding the Black-White test score gap in the first two years of school. The Review of Economics and Statistics, 86, 447–464.

12.

Fryer

R. G.

Levitt

S. D.

(2006). The Black-White test score gap through third grade. American Law and Economics Review, 8, 249–281.

13.

Green

Winters

(2002). Public school graduation rates in the United States. Manhattan Institute for Public Policy.

14.

Hanushek

E. A.

Kain

J. F.

Rivkin

S. G.

(2009). New evidence about Brown v. Board of Education: The complex effects of school racial composition on achievement. Journal of Labor Economics, 27, 349–383.

15.

Hoxby

C. M.

Weingarth

(2005). Taking race out of the equation: School reassignment and the structure of peer effects. Department of Economics, Harvard University.

16.

Mickelson

R. A.

(1990). The attitude-achievement paradox among black adolescents. Sociology of Education, 63(1), 44–61.

17.

Mickelson

R. A.

(2001). Subverting Swann: First- and second-generation segregation in the Charlotte-Mecklenburg Schools. American Educational Research Journal, 38(2), 215–252.

18.

Mickelson

R. A.

Bottia

M. C.

Lambert

(2013). Effects of school racial composition on K-12 mathematics outcomes: A metaregression analysis. Review of Educational Research, 83, 121–158.

19.

National Governors Association Center for Best Practices. (2009). Closing the achievement gap.

20.

The North Carolina mathematics tests technical report. (2006). North Carolina Department of Public Instruction.

21.

North Carolina reading comprehension tests technical report. (2009). North Carolina Department of Public Instruction.

22.

Ogbu

J. U.

(1987). Variability in minority school performance: A problem in search of an explanation. Anthropology & Education Quarterly, 18(4), 312–334.

23.

Perez

Espinoza

Ramos

Coronado

Cortes

(2009). Academic resilience among undocumented Latino students. Hispanic Journal of Behavioral Sciences, 31(2), 149–181.

24.

Rampey

B. D.

Dion

G. S.

Donahue

P. L.

(2008). NAEP 2008 trends in academic progress (NCES 2009-479). National Center for Education Statistics, Institute of Educational Sciences, U.S. Department of Education.

25.

Reardon

S. F.

Galindo

(2009). The Hispanic-white achievement gap in math and reading in the elementary grades. American Educational Research Journal, 46(3), 853–891.

26.

Reed

D. S.

(2009). Is there an expectations gap? Educational federalism and the demographic distribution of proficiency cut scores. American Educational Research Journal, 46(3), 718–742.

27.

Sass

T. R.

Hannaway

Figlio

D. N.

Feng

(2010). Value added of teachers in high-poverty and lower-poverty schools (Working paper No. 52). National Center for Analysis of Longitudinal Data in Education Research

28.

Sirin

S. R.

(2005). Socioeconomic status and academic achievement: A meta-analytic review of research. Review of Educational Research, 75(3), 417–453.

29.

Steele

C. M.

(1997). A threat in the air: How stereotypes shape the intellectual identities and performance of women and African Americans. American Psychologist, 52, 613–629.

30.

Steele

C. M.

(1999). Thin ice: “Stereotype threat” and Black college students. Atlantic Monthly, 284(2), 50–54.

31.

Valadez

J. R.

(2008). Shaping the educational decisions of Mexican immigrant high school students. American Educational Research Journal, 45(4), 834–860.

32.

Woolley

M. E.

Grogan-Kaylor

Gilster

M. E.

Karb

R. A.

Gant

L. M.

Reischl

T. M.

Alaimo

(2008). Neighborhood social capital, poor physical conditions, and school achievement. Children & Schools, 30(3), 133–145.