Student,Neighborhood,and School Factors and Their Association With College Readiness: Exploring the Implementation of a Race- and Socioeconomic-Based Student Assignment Plan

Abstract

In response to the Meredith v. Jefferson County Board of Education Supreme Court decision, Jefferson County Public Schools (JCPS) reconfigured the district’s race-based student assignment and implemented a race- and socioeconomic-based student assignment plan. Using hierarchical linear multiple regression, this study examined students’ backgrounds and school composition factors within a race- and socioeconomic-based assignment plan to determine their relationship with college and career readiness as measured by the ACT. We found that student race, participation in the National School Lunch Program (NSLP), ACT PLAN performance, school composition, and neighborhood category were the largest and most consistent factors for predicting college and career readiness. African American students were at a disadvantage in each model compared with White students. Participation in NSLP was negatively associated with student performance compared with non-NSLP students, and PLAN scores positively predicted ACT performance. School composition was found negatively associated with ACT performance (Title I compared with non–Title I schools), and Category 1 and Category 2 neighborhoods were negatively associated with ACT scores compared with Category 3 neighborhoods. We conclude by discussing implications for policy, practice, and future research.

Keywords

desegregation student assignment diversity urban schools

School districts use student assignment to direct students to their appropriate grade-level schools that serve the area in which the students’ parents or guardians reside (Frankenberg, 2013; McDermott, Frankenberg, & Diem, 2014; Tefera, Frankenberg, Siegel-Hawley, & Chirichigno, 2011). Students’ backgrounds and school composition are key factors in student assignment plans utilized to develop schools that are racially and socioeconomically diverse as well as promote student achievement (Borman et al., 2004; Diem & Frankenberg, 2013; Frankenberg, 2013; Frankenberg, Lee, & Orfield, 2003; Kahlenberg, 2006, 2012; Orfield & Frankenberg, 2011; Orfield, Frankenberg, & Garces, 2008; Potter, Quick, & Davies, 2016; Reardon, Yun, & Kurlaender, 2006; Tefera et al., 2011). Research suggests that school districts seeking integrated student assignment while also promoting student achievement should account for students’ backgrounds and school composition (Coleman et al., 1966; Kahlenberg, 2006, 2012; Konstantopoulos & Borman, 2011; Orfield & Frankenberg, 2011; Orfield et al., 2008; Potter et al., 2016; Reardon et al., 2006).

Although released from its desegregation court order, Jefferson County Public Schools (JCPS) in Louisville, Kentucky, found itself a defendant in the U.S. Supreme Court case, Meredith v. Jefferson County Board of Education (2006) case. The plaintiff contended that the JCPS student assignment plan created a proportion system for maintaining racial diversity that violated the students’ right to equal protection as guaranteed by the 14th Amendment.

Like all school districts, JCPS is tasked with preparing students to adapt to a technologically progressing and academically sophisticated society beyond their secondary education—to be college or career ready upon completion of the 12th grade (Camara, 2013; Conley, 2007). Measuring college-readiness aids in determining whether students have a high probability of academic success in entry-level, credit-bearing courses in postsecondary education (Camara, 2013; Conley, 2007). Complicating matters, parents and students in JCPS can seek educational opportunities related to increased college readiness using intradistrict transfer procedures. Petitioning parents and students seek admission to other schools, typically applying to affluent, non–Title I schools that have historically produced higher results in mathematics and reading compared with their lower performing district counterparts (Lauen, 2007; Phillips, Hausman, & Larsen, 2012; Tefera et al., 2011). Drawing upon JCPS district data, this study sought to answer the following research question: To what extent do student, neighborhood, and school factors predict college readiness in English language arts and mathematics?

Background of the Problem

The development of student assignment plans and the composition of schools are the result of political movements, federal legislation, local legislation, and court opinions (Borman et al., 2004). In 1868, the U.S. Congress ratified the 14th Amendment, guaranteeing equal protection—or a guarantee of common privileges, rights, and protection—of law for all citizens, including recently freed slaves. Despite the fact that the 14th Amendment expanded the protection of civil rights for American citizens, several state governments enacted legislation to segregate White and African American citizens. The legal foundation for such segregation was the landmark Plessy v. Ferguson (1896) case, which granted legal segregation of African American citizens from White citizens so long as facilities and resources were equal to those accessible by White citizens. Even though the Plessy decision focused on segregation in railcars, public school districts used the legal precedent in developing student assignment plans that resulted in heavily segregated schools. In 1954, Brown v. Board of Education re-examined the segregation of children in public schools in relation to equal protection under the 14th Amendment. The 9-to-0 Supreme Court decision overturned Plessy v. Ferguson and determined that segregation created an environment of inferiority and unequal protection in accordance with the 14th Amendment.

After the Brown v. Board of Education (1954) decision, courts mandated that school districts in the United States implement student assignment plans concentrated on the desegregation of student populations. This court order was often satisfied using race-based assignment plans establishing thresholds for the percentage of White and ethnic-minority students for schools within a district. JCPS was among the school districts ordered to implement a desegregation student assignment plan in 1975. After 25 years of desegregation efforts, JCPS was released from the court order per the ruling of Hampton v. Jefferson County Board of Education (2000).

Parents v. Seattle School District and Meredith v. Jefferson County

After release from the court order for meeting the desegregation requirements, JCPS enacted an assignment plan allowing students to apply for a choice of schools. This choice included a filter to ensure schools remained integrated and diverse. Student residence, school enrollment capacity, and race served as factors for determining student choice enrollments. The use of race as a standard for enrollment resulted in a legal challenge in Meredith v. Jefferson County Board of Education (2006). The petitioner argued that the JCPS student assignment plan created a proportion system for maintaining racial diversity that violated the students’ right to equal protection as guaranteed by the 14th Amendment. The U.S. Supreme Court heard the Meredith case in conjunction with Parents Involved in Community Schools v. Seattle School District Number 1 (2007). As in the Meredith case, the Parents Involved in Community Schools case was rooted in a provision of Seattle’s student assignment plan, wherein race served as a tiebreaker for determining assignment.

Students who applied for open enrollment in the Seattle School District were subjected to enrollment criteria based on the amount of applications and student race. Schools were not to exceed the racial thresholds matching the district’s demographics of 60% ethnic minority and 40% White. A suit alleging violation of the Equal Protection Clause of the 14th Amendment was filed after previous courts ruled that the district had a compelling interest in upholding racial diversity. In a 5-to-4 decision, the U.S. Supreme Court ruled the racial tiebreaker to be unconstitutional, as it violated the Equal Protection Clause in the 14th Amendment. The use of White and ethnic-minority designations for race did not meet the “narrow tailoring” standard in creating racial diversity as mandated in Grutter v. Bollinger et al. (2003). The holdings of the Meredith and Parents Involved in Community Schools cases resulted in a new student assignment plan for JCPS. In an effort to retain desegregated schools, JCPS implemented a student assignment plan that accounted for race and socioeconomic status of students.

Race- Versus Socioeconomic-Based Student Assignment

School districts seeking student assignment that promotes diversity have the option of race- or socioeconomic-based student assignment (Diem, 2012; Frankenberg, 2013; Frankenberg et al., 2003; Kahlenberg, 2006, 2012; Orfield, 2001; Orfield & Frankenberg, 2011; Orfield et al., 2008; Potter et al., 2016; Reardon et al., 2006; Tefera et al., 2011). Proponents of race-based assignment plans consider integrated schools beneficial to students and communities by promoting racial understanding, reducing prejudice, and properly preparing students for a diverse workforce (Orfield, 2001; Orfield et al., 2008; Orfield, Kucsera, & Siegel-Hawley, 2012; Orfield, Siegel-Hawley, & Kucsera, 2014; Tefera et al., 2011). Those who support race-based plans seek an education that reaches beyond curriculum by creating a setting wherein students learn about cultural values along with curriculum (Diem, 2012; Diem & Frankenberg, 2013; Frankenberg, 2013; Frankenberg et al., 2003; Orfield, 2001; Orfield & Frankenberg, 2011; Orfield et al., 2008; Orfield et al., 2014; Reardon et al., 2006; Tefera et al., 2011). Specifically, race-based plans establish thresholds for school populations to create diverse student bodies and prevent schools from becoming segregated (Diem, 2012; Diem & Frankenberg, 2013; Frankenberg, 2013; Frankenberg et al., 2003; McDermott et al., 2014; Orfield, 2001; Orfield & Frankenberg, 2011; Orfield et al., 2008; Orfield et al., 2012; Orfield et al., 2014; Reardon et al., 2006; Tefera et al., 2011).

School districts in Boston, Massachusetts; Raleigh, North Carolina; and Louisville, Kentucky, implemented race-based plans to prevent the establishment of ethnic-minority or all-White schools resulting from segregated neighborhoods feeding into neighborhood schools (McDermott et al., 2014). Race-based student assignment plans are necessary in establishing and maintaining integrated schools, and plans developed without the consideration of race isolate students and have negative educational and societal implications (Diem, 2012; Diem & Frankenberg, 2013; Frankenberg, 2013; Frankenberg et al., 2003; McDermott et al., 2014; Orfield, 2001; Orfield & Frankenberg, 2011; Orfield et al., 2008; Orfield et al., 2012; Orfield et al., 2014; Reardon et al., 2006; Tefera et al., 2011). Opponents of using a race-only filter contend that student poverty is a more accurate predictor of student achievement (Kahlenberg, 2006, 2012; Potter et al., 2016; Reardon et al., 2006; Rumberger & Palardy, 2005).

Similar to race-based plans, socioeconomic-based student assignment establishes diverse school populations based on student socioeconomic status instead of race (Diem, 2012; Kahlenberg, 2006, 2012; Potter et al., 2016; Reardon et al., 2006). Previous and current socioeconomic-based student assignment plans employed filters to identify low-income students such as parent income, family qualification for government food stamp programs, and student participation in the National School Lunch Program (NSLP; Diem, 2012; Kahlenberg, 2006, 2012; Potter et al., 2016; Reardon et al., 2006). Socioeconomic-based plans are constructed on the ideology that higher economic students excel academically, are college-minded, are supported by parents who are actively involved in academics and schools, and have access to more highly qualified teachers (Kahlenberg, 2006, 2012; Potter et al., 2016; Reardon et al., 2006; Rumberger & Palardy, 2005). Access to higher performing peers helps establish an academic-oriented environment, promotes academic growth among peers, and reduces distractions in classrooms often associated with low-socioeconomic students (Kahlenberg, 2006, 2012; Potter et al., 2016; Reardon et al., 2006; Rumberger & Palardy, 2005). Proponents of socioeconomic-based student assignment plans consider socioeconomic integration an effective tool for establishing racial integration, while also promoting student achievement by placing high-poverty students in low-poverty schools where peers and parents promote academic success (Kahlenberg, 2006, 2012; Potter et al., 2016; Reardon et al., 2006; Rumberger & Palardy, 2005). Thus, socioeconomic-based plans promote racial diversity, but focus more on academic achievement associated with higher economics than racial integration (Kahlenberg, 2006, 2012; Potter et al., 2016).

According to Kahlenberg (2006), about 40 school districts use or have used socioeconomic student assignment plans including La Crosse, Wisconsin; Raleigh, North Carolina; and San Francisco, California. Beginning in the 1970s, La Crosse implemented one of the earliest forms of socioeconomic-based student assignment plans, mandating school populations to stay between 15% and 45% NSLP eligible (Kahlenberg, 2006). The basis for assigning students by socioeconomic status was to integrate the district’s two high schools, previously segregated into affluent and less advantaged populations (Kahlenberg, 2006). The school board overturned the plan 20 years later (Kahlenberg, 2006).

Just after La Crosse transitioned to race-based student assignment plan in the 1990s, Raleigh transitioned to a socioeconomic-based assignment plan in 2000 (Kahlenberg, 2006). A socioeconomic plan that directed populations to comprise no more than 40% of students eligible for NSLP and no more than 25% below grade level replaced the previous race-based plan requiring schools to contain 15% to 40% ethnic-minority populations (Kahlenberg, 2006). The changes to the student assignment in Raleigh also occurred after the election of new board members (McDermott et al., 2014). In 2001, San Francisco dropped a race-based assignment plan and implemented a socioeconomic plan that accounted for nonracial factors such as NSLP eligibility, participation in public housing, and mother’s education (Kahlenberg, 2006). Opponents of socioeconomic-based plans contend that the filters used to measure poverty are not necessarily accurate of student poverty and fail to align consistently with student race (Diem, 2012; Diem & Frankenberg, 2013; Frankenberg, 2013; Frankenberg et al., 2003; McDermott et al., 2014; Orfield, 2001; Orfield & Frankenberg, 2011; Orfield et al., 2008; Orfield et al., 2014; Reardon et al., 2006). Although a debate exists surrounding the effectiveness of race-based versus socioeconomic-based student assignment plans, this study focused on the relationships between students’ backgrounds and school composition in a district implementing a student assignment plan constructed on student race and socioeconomic status.

JCPS Student Assignment in the Wake of the Meredith Decision

In compliance with the Meredith decision, JCPS designed and implemented a student assignment plan that took into account the diversity of students and the 540 county neighborhoods. The plan, however, was race-based and failed to comply with the Court order of using factors beyond race such as student socioeconomic status. In cooperation with JCPS, Orfield and Frankenberg (2011) recommended a student assignment plan with a neighborhood diversity index, designed to achieve diversity by reaching beyond race by including the neighborhood socioeconomic element.

The diversity index—comprising parent income, parent education, and the percentage of White students for each neighborhood—acts as a filter by assigning one of three possible socioeconomic diversity designations to each neighborhood in Jefferson County (Orfield & Frankenberg, 2011). Table 1 displays the categories of the diversity index and the factors used to calculate and assign the diversity indicator for each neighborhood in the JCPS student assignment plan.

Table 1.

Jefferson County Public Schools Student Assignment Diversity Categories.

Category	Factor
Category	Annual income	Parent education	Race (% White)
1	< $42,000	< 3.5	< 73
2	$42,000-62,000	3.5-3.7	73-88
3	> $62,000	> 3.7	> 88

Note. Parent education based on a weighted score of 3.5 = some college or associate’s degree and 4 = bachelor’s degree.

Table 2 displays the educational weights assigned to the parent education factor of the diversity index (Orfield & Frankenberg, 2011). Weighted scores were assigned to each education attainment level, placing increased values on higher levels of education (Orfield & Frankenberg, 2011).

Table 2.

Parent Education Attainment Weight.

Weight	Education
1.0	Finished Grade 8 or less
2.0	Did not finish high school
3.0	Finished high school
3.5	Some college or associate’s degree
4.0	Bachelor’s degree
5.0	Master’s or professional degree
6.0	Doctorate

Source. Orfield and Frankenberg (2011).

Orfield and Frankenberg (2011) recommended that JCPS use the following formula to calculate the diversity index for each neighborhood:

\begin{matrix} Neighborhood Socioeconomic Designation = 1 + . 23 (Income Integer) \\ + . 33 (Parent Education Integer) \\ + . 33 (Race Integer) . \end{matrix}

When the neighborhood socioeconomic designation formula is utilized, students from a neighborhood with an income less than $42,000, an education weight less than 3.5, and a racial composition of 73% to 88% White would receive a neighborhood designation of 2.22, or Neighborhood Socioeconomic Designation = 1 + .23(1) + .33(1) + .33(2) = 2.22. The weighted averages displayed in Table 3 were calculated for each of Jefferson County’s 540 neighborhoods, and an overall category label was assigned to differentiate each neighborhood (Orfield & Frankenberg, 2011). A weighted average of 2.22 falls within Category 2 in the JCPS student assignment plan.

Table 3.

JCPS Neighborhood Socioeconomic Designation.

Category	Weight range
Category	Minimum	Maximum
1	1.00	1.99
2	2.00	2.99
3	3.00	> 3.00

Note. JCPS = Jefferson County Public Schools.

Orfield and Frankenberg (2011) reported that 30% of the neighborhoods were rated as Category 1, 46% Category 2, and 24% Category 3. Orfield and Frankenberg also recommended that JCPS divide the county into 13 clusters to account for school diversity and bus travel time. Each cluster comprised neighborhoods that would balance school diversity. In 2012, JCPS adopted an altered version of the plan presented by Orfield and Frankenberg by implementing a race- and socioeconomic-based assignment plan, while continuing the use of magnet programs, traditional schools, and school choice. Student assignment in JCPS is now based on the residential diversity index, resulting in a diversity weight of 1.4 to 2.5 for its 16 comprehensive high schools (JCPS, 2012).

However, the composition of schools is influenced beyond the student assignment plan, as JCPS has retained school choice options as well as magnet and traditional programs. Students attending JCPS secondary schools have the opportunity to use a school choice procedure to access a variety a schools and programs. The 16 comprehensive secondary schools are separated into three networks, with five schools of study divided among them. Students have the option of enrolling in one of the schools of study listed below or they may attend their default home school, which is already assigned a school of study:

Business and Finance; Information Technology;

Communications, Electronic and Print Media, Visual and Performing Arts;

Engineering, Architecture, Construction;

Human Services (Law/Government Service, Fire, Police, EMS), Education, International Studies, Heavy Equipment Science;

Medical Arts and Science, Allied Health, Environmental Science.

Also available to students are district-wide magnet high schools, traditional structure programs, and district-wide magnet programs, but enrollment is only granted by submitting all of the required application materials for review. Students may attend a school beyond their network or the district-wide programs by applying to any of the 16 comprehensive schools through open enrollment. Students who apply through open enrollment are subjected to meeting the application standards established by the individual comprehensive schools, and the different schools or their principals determine admissions decisions. Requisite application materials vary, but generally require the submission of an application, recommendation letters or forms, and academic records (transcripts, attendance records, and standardized test scores). Schools may also require an essay, a writing sample, or auditions. For example, Central High School, a magnet high school with programs in technology, law, nursing, pre-medicine, veterinary science, dentistry, and Montessori, states that its application criteria require a minimum grade point average (2.5 or better), a minimum attendance rate of 95%, a review of standardized test scores from Grade 6 to Grade 8, behavior, teacher recommendations, and a writing sample. With regard to the latter, students are required to respond to one of the prompts on the online form (Central High School, n.d.). Manual High School, a magnet high school consisting of programs in performing arts, visual arts, journalism, communication, math, science, technology, and the high school university program, requires that applicants submit to a competitive process. Admissions are based on a review of student test scores, extracurricular involvement, academic achievement, personal essays, attendance, and teacher recommendations. For visual arts applicants and journalism and communication applicants, an onsite writing or drawing exercise is also required. Youth Performing Arts School (YPAS) applicants must submit to an audition (duPont Manual High School, n.d.).

Students enrolling in magnet or traditional programs decrease the number of students assigned to the 16 comprehensive high schools included in this study. For example, a school assigned 100 students from Category 3 neighborhoods may only receive 75 of the students if 25 enroll in magnet or traditional programs. A change in student population could result in a school having a larger or smaller population of NSLP-eligible students than those assigned by JCPS. The blend of school choice, along with magnet and traditional programs, resulted in the de facto segregation of schools (see Table 4), disadvantaging poorer and ethnic-minority students (Diem, 2012; Diem & Frankenberg, 2013; Frankenberg, 2013; Frankenberg et al., 2003; Lauen, 2007; Orfield, 2001; Orfield et al., 2012; Orfield et al., 2014; Phillips et al., 2012; Tefera et al., 2011).

Table 4.

2015 Jefferson County Public Schools Comprehensive School Demographics and ACT Performance.

School	% of students		2015 ACT composite M score	% students meeting benchmarks
School	White	National School Lunch Program eligible	2015 ACT composite M score	English	Math	Reading
Atherton	66.9	40.7	21.6	76.6	40.6	55.7
Ballard	59.2	35.2	21.9	70.4	52.6	52.0
Doss	38.2	72.5	16.1	23.7	7.6	13.1
Eastern	65.6	32.2	21.4	72.9	38.1	51.9
Fairdale	60.4	68.9	17.3	36.7	14.8	20.5
Fern Creek	46.4	64.7	17.7	45.6	13.7	21.1
Iroquois	28.9	81.9	15.2	20.3	5.5	7.8
Jeffersontown	48.9	59.7	17.9	43.0	18.4	23.3
Moore	46.5	61.4	17.5	34.8	16.0	19.8
PRP	64.6	63.0	17.8	40.9	21.6	24.9
Seneca	39.4	69.9	16.7	35.5	8.3	15.7
Shawnee	46.4	77.8	16.0	24.4	5.9	7.6
Southern	51.0	67.6	16.6	35.6	10.0	17.1
Valley	60.1	69.2	16.1	20.1	3.6	13.7
Waggener	40.1	68.7	17.0	31.0	12.6	18.4
Western	24.8	78.5	15.7	22.7	7.8	9.7

This study is significant, as the findings may inform policy in JCPS’s policies on student assignment and school choice as they relate to district college-readiness rates. Specifically, an understanding of student background and school factors contained within the JCPS student assignment plan may guide student assignment that continues to establish desegregated schools while also promoting growth in student college readiness. The effects of student and school factors may also apply to individual schools to clarify local college-readiness achievement rates. Finally, the findings may guide school professional development to prepare teachers adequately for the noncognitive factors present in students and school populations.

Data Sources and Methods

The context and source of data in this study is JCPS, an urban school district centered in Louisville, Kentucky. JCPS is the largest school district in the Commonwealth of Kentucky and one of the largest urban school districts in the United States. We drew upon data from the 2014 to 2015 school year. The instrument chosen to measure college readiness is the ACT college-readiness assessment, administered annually to all 11th-grade students in JCPS. The ACT exam is the last of a three-test series, preceded by the ACT Explore and ACT PLAN, which measure student trajectories for college readiness (ACT, 2013a, 2013b, 2014). Developed for 11th- and 12th-grade students, the ACT serves as a measure of college readiness because it measures knowledge of curriculum through the final years of a secondary education (ACT, 2014). It includes four multiple-choice exams concentrated on English, Mathematics, Reading, and Science curricula Postsecondary institutions used the results in admissions decisions (ACT, 2014). The composite score produced by the ACT exam ranges from 1 to 36. Scored independently, the English, Mathematics, Reading, and Science sections produce a raw score that is adapted to the scale score of 1 to 36 (ACT, 2014). The scale scores are then averaged to produce the total exam composite score (ACT, 2014).

We drew upon data from students enrolled in the 16 comprehensive high schools in JCPS in 2015. During the 2014 to 2015 school year, JCPS enrolled approximately 97,000 students district wide, with 20,450 of the students being educated in the 16 comprehensive high schools used within this study (JCPS, 2014a). The 16 schools comprised 42% African American students, 46.6% White students, and 11.4% classified as other ethnicities (JCPS, 2014a). The average enrollment for the 16 schools was approximately 1,100 students, but ranged from 504 to 2,066 students (JCPS, 2014a). Approximately 4,500 Grade-11 students who attended the 16 comprehensive schools completed the ACT exam in March 2015, which produced an English, Mathematics, and Reading score for each student. Data initially consisted of 4,494 Grade-11 students obtained with permission from JCPS. However, the final sample comprised 3,818 students who met all of the study’s criteria, having removed 676 students missing data (e.g., ACT PLAN scores, residential information). We analyzed student backgrounds, neighborhood socioeconomic data, and school poverty data in relation to ACT results produced by 11th-grade students.

In terms of analytical strategies, we utilized simple descriptive analysis and hierarchical linear multiple regression (HLMR). We used HLMR as the analytical approach because it can determine the contribution of student, neighborhood, and school factors to predict college readiness (Ho, 2014; Osborne, 2016; Petrocelli, 2003; Stevens, 2007, 2009). Specifically, we used three variable blocks to examine college readiness. Block 1 comprised student-level factors, Block 2 comprised school factors, and Block 3 contained neighborhood socioeconomic factors. Proper entry of the variable blocks into the HLMR model is directed by logical or theoretic foundations of the literature (Ho, 2014; Osborne, 2016; Petrocelli, 2003; Stevens, 2007, 2009); therefore, the analysis was conducted after consideration of two different orders of entry. Table 5 displays the variables of the study, their level of measurement, and the measurement score for each variable.

Table 5.

Independent and Dependent Variables.

Variable type	Variable	Level of measurement	Measurement indicator
Independent
Block 1: Student background	Gender	Nominal	0 = Female 1 = Male
	Race	Nominal	• African American• Latinx• White
	Special education status	Nominal	0 = No, 1 = Yes
	National School Lunch Program eligibility	Nominal	0 = No, 1 = Yes
	ACT PLAN Mathematics	Interval	Total score = 1-32
	ACT PLAN English	Interval	Total score = 1-32
	ACT PLAN Reading	Interval	Total score = 1-32
Block 2: School	School composition	Nominal	0 = Non–Title I,1 = Title I
Block 3: Neighborhood	Neighborhood socioeconomic designation	Nominal	• Category 1• Category 2• Category 3
Dependent
Mathematics	ACT Mathematics	Interval	Total score = 1-36
English language arts	ACT English	Interval	Total score = 1-36
	ACT Reading	Interval	Total score = 1-36

We used HLMR to address the research question on the degree to which student, school, and neighborhood factors predict college readiness. Student variables were gender, race (African American, Latinx, and White), special education participation, NSLP status, and the ACT PLAN score from the separate PLAN test section that matched the ACT dependent variable (e.g., ACT PLAN Reading was used when examining ACT Reading). The school composition factor included whether a school was designated Title I; schools were required to have a population greater than 40% poverty to be designated as Title I (JCPS, n.d.). Neighborhoods were represented by Category 1, 2, or 3 and comprised parent education, family income, and neighborhood ethnic-minority rate (Orfield & Frankenberg, 2011). The three dependent variables were scores on ACT English, Mathematics, and Reading exams.

We conducted the HLMR analysis by sequentially entering variable blocks into the model to measure their relationships with the dependent variables. Specifically, Block 1 included the aforementioned student variables (gender, race, special education participation, NSLP status, and PLAN scores). Block 2 included the school variable of Title I designation. Finally, Block 3 included neighborhood category (Category 1, Category 2, and Category 3).

Prior to conducting the regression analysis, we examined the specific model assumptions. These included multicollinearity, linearity, homoscedasticity, outliers, and independence of error terms. According to Stevens (2007, 2009), moderate-to-high correlations between the model predictors may create problems within the regression model. Specifically, the intercorrelation among model predictors may result in a reduction in the predictive utility of independent variables on the dependent variable, which results in a reduction of the R², which serves as a measure of model quality—the proportion of Y variability explained by the model (Ho, 2014; Osborne, 2016; Petrocelli, 2003; Shavelson, 1996; Stevens, 2007, 2009). A variance inflation factor of less than 10 is desired for each factor (Ho, 2014; Osborne, 2016; Petrocelli, 2003; Shavelson, 1996; Stevens, 2007, 2009). In this study, variance inflation factor values ranged from 1.025 to 2.549, suggesting that the data and analysis were robust to violations of multicollinearity.

The results of the HLMR analysis are based on the linear relationship of predictors to a dependent variable (Ho, 2014; Osborne, 2016; Petrocelli, 2003; Shavelson, 1996; Stevens, 2007, 2009). To satisfy the assumption, the residuals were entered into a normal probability plot and checked for linearity to the fitted line. The diagonal residual plot was aligned with the fitted line, which satisfied the assumption of linearity.

Homogeneity of variance, or equal variance, between the variables is observable on the residuals plot. Specifically, the scores of the population should appear normally distributed at the 0 line of the residuals plot to avoid homoscedasticity. The residuals appear normally distributed and clustered tightly in the analyses, with a minor curve to the residuals in the mathematics scatterplot. Thus, it can be stated that assumption of homoscedasticity was met.

Outlier data for the predicting factors and dependent variables may overly influence the R² (Stevens, 2007, 2009). Cook’s distance is used to measure the influence of outliers on the R², and although Stevens (2007, 2009) suggested a maximum of 2 standard deviations for Cook’s distance, the value for each residual should fall below 1. Upon examination of the Cook’s distance results, the maximum values were 0.007 (SD = 0.000) for ACT English, 0.014 (SD = 0.001) for ACT Mathematics, and 0.009 (SD = 0.000) for the ACT Reading analyses. All cases remained in the study since the values were below 1.

The predicted values produced by the model are required to be independent to avoid an error caused by autocorrelation (Ho, 2014; Stevens, 2007, 2009). A Durbin–Watson score between 1.5 and 2.5 is required to satisfy the independence of error assumption, but a score closer to 2.0 is desired (Ho, 2014; Stevens, 2007, 2009). The ACT English analysis (1.977), the ACT Mathematics analysis (2.068), and the ACT Reading analysis (1.991) produced Durbin–Watson scores between the required 1.5 and 2.5 limits. The Durbin–Watson test confirmed each analysis avoided autocorrelation.

Descriptive Analysis

Table 6 reports the percentage of student subgroups across neighborhoods represented in this study. Gender groups were split evenly, with 50.2% of the students reported as female (n = 1,915) and 49.8% reported as male (n = 1,903). This varied only slightly when separating students by neighborhood diversity. In particular, there were three categories of neighborhoods in the JCPS student assignment plan designed by Orfield and Frankenberg (2011); neighborhoods denoted as Category 1 were 51.4% male (n = 508), Category 2 neighborhoods were 50.4% female (n = 1,026), and neighborhoods denoted as Category 3 were 51.5% female (n = 408). Student race for the total sample was primarily White at 57.9% (n = 2,210), with 34.7% reported as African American (n = 1,325), and 7.4% reported as Latinx (n = 283).

Table 6.

Percentage of Student Demographics Represented by Neighborhood Diversity Index.

Student group	Total sample	Category 1neighborhood	Category 2neighborhood	Category 3neighborhood
Gender
Female	50.2	48.6	50.4	51.5
Male	49.8	51.4	49.6	48.5
Race
African American	34.7	72.4	25.0	12.5
Latinx	7.4	6.5	8.9	4.8
White	57.9	21.1	66.1	82.7
Special education	8.7	12.9	8.2	4.7
National School Lunch Program rate	58.4	82.9	59.3	25.5
Enrollment rate at Title I school	48.3	58.9	53.0	10.6

Table 7 reports ACT PLAN scores obtained by the students during the 2013 to 2014 school year. The results of the total sample showed an average score of 15.3 for PLAN English, 16.0 for PLAN Mathematics, and 15.8 for PLAN Reading sections. However, a difference in the results was revealed when the scores were separated by neighborhood diversity index. Category 1 neighborhoods scored below the sample average by 1.6 to 2 average points for each PLAN test section, and Category 3 neighborhoods scored above the sample average by 2.2 to 2.5 average points. The reported means revealed an achievement gap between neighborhoods of almost 4 average points. PLAN scores from students in Category 2 neighborhoods also were found to be below the average Category 3 neighborhood scores, but the Category 2 scores aligned with the total sample average for each PLAN test.

Table 7.

Descriptive Statistics of ACT PLAN Scores Across Neighborhoods, 2013-2014.

Test and neighborhood	M	SD	Minimum	Maximum
ACT PLAN English
Total sample	15.3	4.3	3	32
Neighborhood 1	13.5	3.5	3	30
Neighborhood 2	15.3	4.1	6	32
Neighborhood 3	17.8	4.4	6	30
ACT PLAN Mathematics
Total sample	16.0	4.1	2	32
Neighborhood 1	14.4	3.2	2	30
Neighborhood 2	16.0	3.9	3	32
Neighborhood 3	18.2	4.4	3	32
ACT PLAN Reading
Total sample	15.8	4.2	4	30
Neighborhood 1	14.0	3.5	4	27
Neighborhood 2	15.7	4.1	4	30
Neighborhood 3	18.1	4.3	6	30

Note. Total sample N = 3,818; Neighborhood 1 (n = 989); Neighborhood 2 (n = 2,036); Neighborhood 3 (n = 793).

Table 8 reports the descriptive statistics for the ACT scores produced during the 2014 to 2015 school year. An examination of ACT performance for the total sample revealed that students averaged 17.3 for the ACT English section, 17.8 for ACT Mathematics, and 18.4 for the ACT Reading section.

Table 8.

Descriptive Statistics of ACT Scores Across Neighborhoods, 2014-2015.

Test and neighborhood	M	SD	Minimum	Maximum.
ACT English
Total sample	17.3	6.1	0	36
Neighborhood 1	14.2	4.8	0	34
Neighborhood 2	17.1	5.8	3	35
Neighborhood 3	21.5	6.0	5	36
ACT Mathematics
Total sample	17.8	4.3	0	35
Neighborhood 1	16.1	3.1	0	31
Neighborhood 2	17.6	4.0	0	35
Neighborhood 3	20.6	5.0	11	35
ACT Reading
Total sample	18.4	5.8	0	36
Neighborhood 1	15.9	4.6	0	35
Neighborhood 2	18.3	5.6	4	36
Neighborhood 3	22.0	5.8	4	36

Note. Total sample N = 3,818; Neighborhood 1 (n = 989); Neighborhood 2 (n = 2,036); Neighborhood 3 (n = 793).

The results of the ACT exam were similar to ACT PLAN performance when disaggregating the data by neighborhood category. Category 1 neighborhoods scored 1.7 to 3.5 average points below the sample average on the ACT exams, and Category 3 neighborhoods scored 2.8 to 4.2 average points above the sample average. The reported means for ACT scores revealed an achievement gap similar to PLAN scores between Category 1 and Category 3 neighborhoods. Similar to the ACT PLAN, Category 2 neighborhoods were found to produce average ACT scores below Category 3 neighborhoods, but the scores for students in Category 2 neighborhoods aligned with the total sample average. Finally, the minimum scores for the ACT test sections were notably different across neighborhood categories. Category 1 neighborhoods had the lowest minimum scores, ranging from 0 to 3 for each ACT exam section, whereas Category 3 neighborhoods had comparably higher minimum ACT exam section scores, ranging from 4 to 11.

Correlational analyses were conducted to measure the relationship between the ACT PLAN and the dependent variables of ACT English, ACT Mathematics, and ACT Reading. Correlations between .60 and .79 are considered strong, and values 0.80 to 1.00 are very strong (Stevens, 2007, 2009). Overall, the correlations among the scores were high, with coefficients exceeding .73. Specifically, a high correlation was found between ACT PLAN English and ACT English (r = .83) and between ACT PLAN Mathematics and ACT Mathematics scores (r = .79). A slightly lower correlation coefficient was found between ACT PLAN Reading and ACT Reading with a high correlation of .73. The strong correlational associations of .73 and above indicated a positive, linear association between the PLAN and the ACT scores. Simply, an increase in ACT PLAN scores is likely to result in a similar increase in ACT scores. The measures of central tendency revealed an achievement gap on ACT performance between neighborhoods. We now move on to inferential analysis.

HLMR Analysis

We utilized HLMR to address the research question on the degree to which student, school, and neighborhood factors predict college readiness. First, we report the HLMR analysis predicting ACT English scores. We then report the HLMR analysis for ACT Mathematics scores. Finally, HLMR analysis for ACT Reading scores.

ACT English Results

Table 9 reports the results of the HLMR analysis of student, school, and neighborhood factors on ACT English scores. As shown, variable Block 1 comprising student variables was found to be statistically significant F(6, 3,811) = 1,563.307, p < .001, and accounted for 71.1% of the variance in ACT English scores. Subsequently, the addition of variable Block 2 explained an additional 0.9% (cumulative 72%) of the variance of ACT English scores and was statically significant, F(7, 3,810) = 1,397.855, p < .001. Finally, the entry of Block 3 was found to be statistically significant, F(9, 3,808) = 1,108.067, p < .001, and explained an additional 0.4% (cumulative 72.4%) of the variance in ACT English scores.

Table 9.

Hierarchical Linear Multiple Regression Predicting ACT English Scores.

Variable	R ²	Adj. R²	ΔR²	ΔF	Estimates
Variable	R ²	Adj. R²	ΔR²	ΔF	B	SE	β
Model 1	.711*	.711*
(Constant)					1.414*	.272
Gender					−0.401*	.108
Race: African American					−0.993*	.127
Race: Latinx					−0.111	.214
Special education					−1.374*	.196
NSLP					−0.734*	.121
PLAN English					1.105*	.014	.772*
Model 2	.720*	.719*	.009*	117.763*
(Constant)					2.412*	.283
Gender					−0.374*	.106
Race: African American					−0.984*	.125
Race: Latinx					−0.024	.211
Special education					−1.406*	.193
NSLP					−0.505*	.121
PLAN English					1.069*	.014	.746*
School composition					−1.227*	.113
Model 3	.724*	.723*	.004*	27.010*
(Constant)					3.133*	.299
Gender					−0.365*	.106
Race: African American					−0.799*	.135
Race: Latinx					0.025	.209
Special education					−1.388*	.191
NSLP					−0.324*	.122
PLAN English					1.058*	.014	.739*
School composition					−0.964*	.118
Neighborhood 1					−1.252*	.190
Neighborhood 2					−1.016*	.148

Note. NSLP = National School Lunch Program.

p < .05.

As shown in Table 9, there were several significant model predictors after the inclusion of the final variable block. Specifically, in terms of demographics, the coefficients for student gender, African Americans, special education enrollment, NSLP status, and ACT PLAN English scores contributed significantly to the regression model. The coefficients revealed a negative association for males (−0.365) compared with females, African American students (−0.799) compared with White students, special education students (−1.388) compared with regular education students, and for students participating in the NSLP (−0.324) compared with students not eligible for the NSLP. The negative coefficients indicated that female, White, regular education, and non-NSLP students scored higher than their male, White, regular education, and NSLP participant student counterparts.

The PLAN English scores, however, were found to have a positive relationship with ACT English scores (0.739). For every standard deviation increase in PLAN English scores (SD = 4.3), the ACT English scores increased by 4.51 points. The coefficients for school composition and neighborhood were also reported as significant to the ACT English model. The Title I school coefficient was found to have a negative relationship with ACT English scores (−0.964) compared with students who attended non–Title I schools, which indicated higher scores for students in non–Title I schools.

Category 1 (−1.252) and Category 2 neighborhoods (−1.016) also had a negative association compared with Category 3 neighborhoods. Thus, students in Category 2 neighborhoods scored higher than those in Category 1 neighborhoods, but students in Category 3 neighborhoods outperformed students in Category 1 and 2 neighborhoods. The overall model accounted for 72.4% of the variance of ACT English scores.

ACT Mathematics Results

Table 10 reports the results of the HLMR analysis of student, school, and neighborhood factors on ACT Mathematics scores. As reported in Table 10, variable Block 1 comprising student variables was found to be statistically significant, F(6, 3,811) = 1,084.222, p < .001, and accounted for 63.1% of the variance in ACT Mathematics scores. In succession, the addition of variable Block 2 explained an additional 0.9% (cumulative 64%) of the variance of ACT Mathematics scores and was statically significant, F(7, 3,810) = 969.066, p < .001. Finally, the entry of Block 3 was found to be statistically significant, F(9, 3,808) = 763.775, p < .001, and explained an additional 0.4% (cumulative 64.4%) of the variance in ACT Mathematics scores.

Table 10.

Hierarchical Linear Multiple Regression Predicting ACT Mathematics Scores.

Variable	R ²	Adj. R²	ΔR²	ΔF	Estimates
Variable	R ²	Adj. R²	ΔR²	ΔF	B	SE	β
Model 1	.631*	.630*
(Constant)					5.956*	.220
Gender					−0.119	.086
Race: African American					−0.599*	.100
Race: Latinx					−0.365*	.169
Special education					−0.088	.156
NSLP					−0.602*	.095
PLAN Mathematics					0.782*	.012	.743*
Model 2	.640*	.640*	.010*	103.375*
(Constant)					6.669*	.228
Gender					−0.073	.085
Race: African American					−0.578*	.098
Race: Latinx					−0.285	.167
Special education					−0.073	.085
NSLP					−0.424*	.095
PLAN Mathematics					0.757*	.012	.719*
School composition					−0.908*	.089
Model 3	.644*	.643*	.003*	16.917*
(Constant)					7.145*	.242
Gender					−0.065	.084
Race: African American					−0.541*	.107
Race: Latinx					−0.264	.166
Special education					−0.114	.153
NSLP					−0.324*	.097
PLAN Mathematics					0.750*	.012	.713*
School composition					−0.759*	.094
Neighborhood 1					−0.606*	.151
Neighborhood 2					−0.685*	.118

Note. NSLP = National School Lunch Program.

p < .05.

Table 10 also reports the significant model predictors after the inclusion of the final variable block. Regarding demographics, the coefficients for African American referent race subgroup, NSLP status, and ACT PLAN Mathematics scores contributed significantly to the regression model. The coefficients revealed a negative association for African American students (−0.541) compared with White students and for students participating in the NSLP (−0.324) compared with non-NSLP participants. The negative coefficients indicated that White and non-NSLP students scored higher than African American and NSLP-participant students. The ACT PLAN Mathematics scores were found to have a positive relationship with ACT Mathematics scores (.713). For every standard deviation increase in ACT PLAN Mathematics scores (SD = 4.1), the ACT Mathematics scores increased by 3.07 points.

The coefficients for school composition and neighborhood were also reported as significant to the ACT Mathematics model. The Title I school coefficient was found to have a negative influence on ACT Mathematics scores (−0.759) compared with non–Title I schools, as did residing in Category 1 (−0.606) or Category 2 neighborhoods (−0.685) when compared with Category 3 neighborhoods. The negative coefficients indicated that students in non–Title I schools scored higher than students in Title I schools. In addition, Category 2 neighborhoods scored higher than Category 1 neighborhoods, but Category 3 neighborhoods scored higher than Category 1 and 2 neighborhoods. The overall model accounted for 64.4% of the variance of ACT Mathematics scores.

ACT Reading Results

Table 11 reports the results of the HLMR analysis of student, school, and neighborhood factors on ACT Reading scores. As shown, variable Block 1 comprising student variables was found to be statistically significant, F(6, 3,811) = 816.087, p < .001, and accounted for 56.2% of the variance in ACT Reading scores. Subsequently, the addition of variable Block 2 explained an additional 1% (cumulative 57.2%) of the variance of ACT Reading scores and was statistically significant, F(7, 3,810) = 728.469, p < .001. Finally, the entry of Block 3 was found to be statistically significant, F(9, 3,808) = 571.253, p < .001, and explained an additional 0.2% (cumulative 57.4%) of the variance in ACT Reading scores.

Table 11.

Hierarchical Linear Multiple Regression Predicting ACT Reading Scores.

Variable	R ²	Adj. R²	ΔR²	ΔF	Estimates
Variable	R ²	Adj. R²	ΔR²	ΔF	B	SE	β
Model 1	.562*	.562*
(Constant)					5.405*	.313
Gender					−0.314*	.125
Race: African American					−1.181*	.146
Race: Latinx					−0.458	.247
Special education					−1.241*	.226
NSLP					−0.969*	.138
PLAN Reading					0.908*	.016	.664*
Model 2	.572*	.572*	.010*	89.304*
(Constant)					6.339*	.325
Gender					−0.279*	.124
Race: African American					−1.157*	.145
Race: Latinx					−0.355	.245
Special education					−1.258*	.224
NSLP					−0.720*	.139
PLAN Reading					0.875*	.016	.640*
School composition					−1.236*	.131
Model 3	.574*	.573*	.002*	9.551*
(Constant)					6.865*	.347
Gender					−0.273*	.123
Race: African American					−1.059*	.157
Race: Latinx					−0.324	.244
Special education					−1.248*	.223
NSLP					−0.596*	.142
PLAN Reading					0.868*	.016	.635*
School composition					−1.056*	.137
Neighborhood 1					−0.813*	.222
Neighborhood 2					−0.734*	.173

Note. NSLP = National School Lunch Program.

p < .05.

As shown in Table 11, there were several significant model predictors after the inclusion of the third variable block. Specifically, the demographic coefficients for student gender, African Americans, special education, NSLP status, and ACT PLAN Reading scores contributed significantly to the regression model. The coefficients revealed a negative association for males (−0.273) compared with female students, African American students (−1.059) compared with White students, special education students (−1.248) compared with regular education students, and for students participating in the NSLP (−0.596) compared with non-NSLP students. The negative coefficients indicated that female, White, regular education, and non-NSLP students scored higher than their male, White, regular education, and NSLP participant student counterparts. The PLAN Reading scores, however, were found to have a positive relationship with ACT Reading scores (0.635). For every standard deviation increase in PLAN reading scores (SD = 4.2), the ACT Reading scores increased by 3.68 points.

The coefficients for school composition and neighborhood were also reported as significant to the ACT Reading model. The Title I school coefficient was found to have a negative relationship with ACT Reading scores (−1.056) compared with non–Title I schools, as did residing in Category 1 (−0.813) or Category 2 neighborhoods (−0.734) compared with Category 3 neighborhoods. The negative coefficients indicated that students in non–Title I schools scored higher than students in Title I schools. Different from the previous models, students in Category 1 neighborhoods scored higher than those in Category 2 neighborhoods, but students in Category 3 neighborhoods continued to score higher than students in Category 1 and 2 neighborhoods. The overall model accounted for 57.4% of the variance of ACT Reading scores.

Summary of Findings

The results of the analyses conducted in this study revealed that student, school, and neighborhood factors were associated with student performance on the ACT English, Mathematics, and Reading exams. Student race, participation in NSLP, ACT PLAN performance, school composition, and neighborhood category were found to be the largest and most consistent factors for predicting college readiness. In addition, student gender and special education factors were significant, but only for the ACT English and Reading analyses. The analyses of ACT scores identified common significant coefficients for each of the models. Specifically, African American students were found to have a disadvantage in each model compared with White students, participation in NSLP was negatively associated with student performance compared with non-NSLP students, and PLAN scores positively predicted ACT performance. School composition was found negatively associated with ACT performance (Title I compared with non–Title I schools), and Category 1 and Category 2 neighborhoods were negatively associated with ACT scores compared with Category 3 neighborhoods. Gender and participation in special education were also negatively associated with ACT for males and special education students, but those coefficients were only significant to ACT English and Reading scores.

The model summary for each analysis determined that student backgrounds, neighborhood diversity, and school composition have a significant association with ACT performance and explained high levels of variance in ACT English scores (R² = .724), ACT Mathematics scores (R² = .644), and ACT Reading scores (R² = .574). The addition of the school and neighborhood factors to each model explained only a small amount of variance; however, the factors accounted for approximately 1.5 to 2 points in ACT performance.

Discussion

By implementing a race- and socioeconomic-based student assignment plan, JCPS has made progressive strides toward integrating school populations that move beyond race by considering student, family, and neighborhood factors. However, school choice practices have shaped the racial and socioeconomic composition of student bodies and resulted in schools with vastly different levels of ethnic-minority and poverty populations. JCPS (n.d.) reported that school populations comprising above 40% low-income students were designated as Title I, and 10 of the 16 comprehensive high schools in this study acquired the Title I designation. These Title I schools enrolled students from low-income and low-parental education neighborhoods at rate almost 50% higher than high-income and high-parental education neighborhoods. Moreover, students who attended JCPS Title I schools were shown to have a negative association with ACT scores.

According to Schwartz (2010), school poverty levels between 35% and 85% similarly influenced student performance in mathematics and reading, and school poverty levels below 20% were the most beneficial to high-poverty students. Poverty rates below 35% reduced learning barriers such as misbehavior that were associated with higher poverty levels (Schwartz, 2010). Thus, when implementing a race- and socioeconomic-based student assignment plan that promotes diversity and achievement, JCPS may benefit from establishing a policy requiring schools not to exceed a 35% poverty threshold. The use of a poverty threshold may require limitations to school choice practices currently employed by JCPS. Furthermore, an examination of the screening and enrollment policies for the non–Title I magnet and traditional school programs may be necessary in implementing a poverty threshold in the comprehensive high schools. Although JCPS cannot control the student background factors brought to school by students from the 540 district neighborhoods, a change in school choice practices and the effective use of their race- and socioeconomic-based student assignment plan may work to mitigate the negative relationship of school poverty with college and career readiness as measured by ACT scores.

Beyond examining student assignment and school choice practices, JCPS may benefit from exploring and evaluating student support efforts in schools with higher poverty populations. Although high-socioeconomic schools have been found to attract and retain more effective teachers and have more effective practices and processes for improving instruction (Rumberger & Palardy, 2005), impoverished or ethnic-minority students assigned to low-poverty and low-ethnic-minority schools may experience anxiety and feelings of environmental rejection (Angrist & Lang, 2004; Crosnoe, 2009). Thus, academic advantages may only be a temporary benefit as the students struggle to meet the academic demands in unfamiliar environments (Angrist & Lang, 2004; Crosnoe, 2009). In addition, the examination of teacher assignment and professional development practices may be required since high-poverty schools tend to have less experienced teachers, less stable teaching staffs, and lower teacher qualifications (Borman et al., 2004; Clotfelter, Ladd, Vigdor, & Wheeler, 2007; Orfield & Lee, 2005; Sass, Hannaway, Xu, Figlio, & Feng, 2012). The teacher assignment and transfer provisions in the collective bargaining agreement¹ between the Jefferson County Board of Education (JCBE) and the Jefferson County Teachers Association (JCTA) grants transfer priority to more experienced and tenured teachers (JCBE, 2013). Changes to teacher transfer policies and procedures may be required to prevent experienced teachers from leaving high-poverty schools. In addition, the contract also requires teachers to earn a requisite number of professional development credits per school year (JCBE, 2013). An examination of the professional development practices may be necessary to ensure that teachers are receiving relevant training on providing effective support to students whose background factors negatively influence their college readiness performance.

Finally, JCPS should continue the practice of assessing the trajectory of student college readiness prior to administering the ACT to 11th-grade students. ACT (2013b, 2014) previously reported the ACT PLAN results as a significant predictor of ACT performance; hence, the inclusion of PLAN results in this study. Sharing the results of this study on the relationship of student background and school factors coupled with individual student college readiness scores may aid in assessing college-readiness concerns for students. Although JCPS has traditionally provided ACT EXPLORE and PLAN results to high schools, an understanding of the influence of student background and school composition factors may be beneficial in developing individualized student support plans focused on improving college readiness (Angrist & Lang, 2004; Crosnoe, 2009). While Coleman et al. (1966) concluded that student backgrounds destabilize the effects of schools, the more recent research of Rumberger and Palardy (2005) and Konstantopoulos and Borman (2011) concluded that schools might counter the influence of backgrounds by using instructional resources, personnel support, and higher expectations for student achievement. Therefore, the early identification of student performance deficits may allow teachers to develop a plan to reinforce English, Mathematics, or Reading comprehension strategies, thus counterbalancing the influence of background factors.

Several limitations exist within our study that limit the generalizability of the results. First, this study focused on a single urban school district and data from a single academic year. Relating the results to other school districts with different student, neighborhood, and school demographics may produce varying results on their influence on college readiness. Furthermore, other factors such as the effectiveness of ACT preparation programs used by districts and schools were not included. A future examination of the effectiveness of college preparatory resources may further our understanding of school factors and their relationship to student outcomes.

Next, this study included data only from 11th-grade students enrolled in the 16 comprehensive JCPS high schools. We excluded students enrolled in magnet or traditional programs. The student selection processes used by magnet and traditional programs results in student populations that are not reflective of the 16 comprehensive schools, as not all students qualify for enrollment. Students who do not meet the standards for enrollment are not considered. Instead, this study focused on the comprehensive schools for which performance in middle school is not required for enrollment. A future study on the influence of magnet and traditional programs on college readiness may provide additional insights regarding the effectiveness of those programs on college readiness.

Finally, student factors beyond those included in this study may provide further insight and clarification into the influences of achieving college readiness. Although we used neighborhood categories, specific neighborhoods within the same category may have varying levels of influence on college readiness between neighborhoods. Moreover, even though school composition was utilized, Title I schools may influence college-readiness rates differently. Teacher qualifications, years of experience, and effectiveness between schools may also provide clarification into variance of college-readiness rates. The inclusion of more specific student- and school-level data and the use of hierarchical linear modeling may provide a deeper insight into specific influences on student college readiness.

Footnotes

Authors’ Note

An early version of this paper was presented at the 2017 Annual Meeting of the University Council for Educational Administration in Denver, Colorado.

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) received no financial support for the research, authorship, and/or publication of this article.

Notes

ORCID iD

W. Kyle Ingle

Author Biographies

Gregory Herberger is an assistant principal at Eastern High School in the Jefferson County Public School (Louisville, Kentucky), where he has worked as a teacher or administrator since 2006. He earned his PhD in Educational Leadership and Organizational Development at the University of Louisville in 2016.

Jason Immekus is an associate professor in the Department of Educational Leadership, Evaluation, and Organizational Development at the University of Louisville. His research interests include psychometrics, quantitative methods, and program evaluation. His specific research focuses on the investigation of the psychometrics properties of various instruments (e.g., cognitive, non-cognitive).

W. Kyle Ingle is an associate professor and assistant chair in the Department of Educational Leadership, Evaluation, and Organizational Development at the University of Louisville. His research interests include human resource functions in education, the politics of education, and economic evaluations of education programs.

References

ACT. (2013a). EXPLORE technical manual. Iowa City, IA: Author. Available from http://www.act.org

ACT. (2013b). PLAN technical manual. Iowa City, IA: Author. Available from http://www.act.org

ACT. (2014). ACT technical manual. Iowa City, IA: Author. Available from http://www.act.org

Angrist

J. D.

Lang

(2004). Does school integration generate peer effects? Evidence from Boston’s METCO program. American Economic Review, 94, 1613-1634.

Borman

K. M.

Eitle

T. M.

Michael

Eitle

D. J.

Lee

Johnson

Shircliffe

(2004). Accountability in a postdesegregation era: The continuing significance of racial segregation in Florida’s schools. American Educational Research Journal, 41, 605-631.

Brown v. Board of Education of Topeka, 347 U.S. 483, 74 S. Ct. 686, 98 L. Ed. 873 (1954).

Camara

(2013). Defining and measuring college and career readiness: A validation framework. Educational Measurement: Issues and Practice, 32(4), 16-27.

Central High School. (n.d.). Apply to central high school. Retrieved from https://schools.jefferson.kyschools.us/High/Central/page12.html

Clotfelter

Ladd

H. F.

Vigdor

Wheeler

(2007). High-poverty schools and the distribution of teachers and principals. North Carolina Law Review, 85, 1345-1379.

10.

Coleman

J. S.

Campbell

Hobson

McPartland

Mood

Weinfield

York

(1966). Equality of educational opportunity. Washington, DC: National Center for Education Statistics.

11.

Conley

D. T.

(2007). Redefining college readiness. Available from https://www.epiconline.org/

12.

Crosnoe

(2009). Low-income students and the socioeconomic composition of public high schools. American Sociological Review, 74, 709-730.

13.

Diem

(2012). The relationship between policy design, context, and implementation in integration plans. Education Policy Analysis Archives, 20, 1-36.

14.

Diem

Frankenberg

(2013). The politics of diversity: Integration in an era of political and legal uncertainty. Teachers College Record, 115(11), 1-30.

15.

duPont Manual High School. (n.d.). Manual admissions information. Retrieved from http://dupontmanual.com/magnetapp.html

16.

Frankenberg

(2013). The role of residential segregation in contemporary school segregation. Education and Urban Society, 45, 548-570.

17.

Frankenberg

Lee

Orfield

(2003). A multicultural society with segregated schools: Are we losing the dream? Available from http://civilrightsproject.ucla.edu/

18.

Grutter v. Bollinger, 539 U.S. 306, 123 S. Ct. 2325, 156 L. Ed. 2d 304 (2003).

19.

Hampton v. Jefferson County Board of Education, 102 F. Supp. 2d 358 (W.D. Ky. 2000).

20.

(2014). Handbook of univariate and multivariate data analysis with IBM SPSS (2nd ed.). Boca Raton, FL: Taylor & Francis.

21.

Jefferson County Board of Education. (2013). JCBE-JCTA agreement 2013-2018. Retrieved from https://www.jefferson.kyschools.us/sites/default/files/forms/jcta_0.pdf

22.

Jefferson County Public Schools. (2012). Student assignment. Available from http://www.jcpsky.net

23.

Jefferson County Public Schools. (2014a). Data books: Current enrollment by race and gender 2014-2015. Available from http://www.jcpsky.net

24.

Jefferson County Public Schools. (n.d.). Title I, II, IV, & Programmatic Support. Retrieved from https://www.jefferson.kyschools.us/department/title-i

25.

Kahlenberg

R. D.

(2006). A new way on school integration. Available from http://www.tcf.org

26.

Kahlenberg

R. D.

(2012). The future of school integration: Socioeconomic diversity as an education reform strategy. New York, NY: The Century Foundation.

27.

Konstantopoulos

Borman

G. D.

(2011). Family background and school effects on student achievement: A multilevel analysis of the Coleman data. Teachers College Record, 113, 97-132.

28.

Lauen

L. L.

(2007). Contextual explanations of school choice. Sociology of Education, 80, 179-209.

29.

McDermott

K. A.

Frankenberg

Diem

(2014). The “post-racial” politics of race: Changing student assignment policy in three school districts. Educational Policy, 29, 504-554. doi:10.1177/0895904813510775

30.

Meredith v. Jefferson County Board of Education, 548 U.S. 938 (2006).

31.

Orfield

(2001). Schools more separate: Consequences of a decade of resegregation. Available from http://civilrightsproject.ucla.edu/

32.

Orfield

Frankenberg

(2011). Diversity and educational gains: A plan for a changing county and its schools. Available from http://civilrightsproject.ucla.edu/

33.

Orfield

Frankenberg

Garces

L. M.

(2008). Statement of American social scientists of research on school desegregation to the U.S. Supreme Court in Parents v. Seattle School District and Meredith v. Jefferson County. The Urban Review, 40, 96-136.

34.

Orfield

Kucsera

Siegel-Hawley

(2012). E pluribus . . . separation: Deepening double segregation for more students. Available from http://civilrightsproject.ucla.edu/

35.

Orfield

Lee

(2005). Why segregation matters: Poverty and educational inequality. Available from http://civilrightsproject.ucla.edu/

36.

Orfield

Siegel-Hawley

Kucsera

(2014). Sorting out deepening confusion on segregation trends. Available from http://civilrightsproject.ucla.edu/

37.

Osborne

J. W.

(2016). Regression and linear modeling: Best practices in modern methods. Thousand Oaks, CA: SAGE.

38.

Parents Involved in Community Schools v. Seattle School District No. 1, 551 U.S., 127 S. Ct. 2738, 168 L. Ed. 2d 508 (2007).

39.

Petrocelli

J. V.

(2003). Hierarchical multiple regression in counseling research: Common problems and possible remedies. Measurement and Evaluation in Counseling and Development, 36(1), 9-22.

40.

Phillips

K. J.

Hausman

Larsen

E. S.

(2012). Students who choose and the schools they leave: Examining participation in intradistrict transfers. The Sociological Quarterly, 53, 264-294.

41.

Plessy v. Ferguson, 163 U.S. 537, 16 S. Ct. 1138, 41 L. Ed. 256 (1896).

42.

Potter

Quick

Davies

(2016). A new wave of school integration: Districts and charters pursuing socioeconomic diversity. Available from http://www.tcf.org

43.

Reardon

S. F.

Yun

J. T.

Kurlaender

(2006). Implications of income-based school assignment policies for racial school segregation. Educational Evaluation and Policy Analysis, 28, 49-75.

44.

Rumberger

Palardy

(2005). Does segregation still matter? The impact of student composition on academic achievement in high school. Teachers College Record, 107, 1999-2045.

45.

Sass

T. R.

Hannaway

Figlio

D. N.

Feng

(2012). Value added of teachers in high-poverty schools and lower poverty schools. Journal of Urban Economics, 72, 104-122.

46.

Schwartz

(2010). Housing policy is school policy: Economically integrative housing promotes academic success in Montgomery County, Maryland. Available from http://www.tcf.org

47.

Shavelson

(1996). Statistical reasoning for the behavioral sciences (3rd ed.). Needham Heights, MA: Allyn & Bacon.

48.

Stevens

J. P.

(2007). Intermediate statistics: A modern approach (3rd ed.). New York, NY: Taylor & Francis.

49.

Stevens

J. P.

(2009). Applied multivariate statistics for the social sciences (5th ed.). New York, NY: Taylor & Francis.

50.

Tefera

Frankenberg

Siegel-Hawley

Chirichigno

(2011). Integrating suburban schools: How to benefit from growing diversity and avoid segregation. Available from http://civilrightsproject.ucla.edu/