Regional College-Going
Patterns of Philadelphia Public High School Graduates: The Role of High
School Curriculum
Introduction
The Philadelphia
Context
Obtaining a postsecondary education is increasingly critical for young
people. With an increasing number of jobs requiring, at minimum, some
postsecondary education, the economic and social disadvantages associated
with lack of postsecondary education intensifying, and the number of students
aspiring to postsecondary education growing (Callan, 2001; Education Trust,
2001; Heller, 2001; Labaree, 1997), getting students enrolled in college
has become an increasingly important goal of educational systems.
As a result, college-going rates and patterns have become increasingly important indicators for use when considering the effectiveness of educational systems. In recognition of this, during the Spring of 2002, the School District of Philadelphia (SDP) and Temple University's Office of School and Community Partnerships partnered through the federal GEAR-UP program to develop a K-16 data-sharing consortium in the Philadelphia region. This partnership was initiated in order to examine the regional college-going patterns of Philadelphia public high school graduates and to provide a K-16 context for informing SDP's current efforts to improve secondary education.1
Increasing access to and participation in postsecondary education for the students that SDP serves (primarily minority and/or low-income students who have been historically underrepresented in the postsecondary arena) is an important goal of the District. This was the goal during the previous eras of reform (including the reform eras occurring during the years 1995-2001, the time period to which the data used in this study pertains), and has remained the case throughout the current reform efforts. In the last two years, these reform efforts have included: a reorganization of SDP leadership, achieved through a managed state takeover of the District; the hiring of private, non-profit, and university partners as educational management organizations; changes in the management and/or organizational structures of approximately 75 schools; the development of a new District-wide standardized curriculum for grades K-9; the implementation of a new District-wide standardized assessment system; and numerous other changes. More significantly for the analysis of college-going rates and patterns, these reform efforts have also included a renewed focus on reforming both the organization and the curriculum of SDP high schools. Significant capital commitments for the building of new high schools and improvements to existing high schools; plans to reorganize several existing high schools; the initiation of efforts to develop standardized secondary course offerings across high schools; new mandates to increase the number of Advanced Placement and International Baccalaureate courses across high schools; and the hiring of new counselors across high schools are all reforms that are relevant to a discussion of the college-going patterns of Philadelphia public high school graduates.
To inform these efforts to reform secondary education in Philadelphia and to understand their likely impact on regional college-going rates and patterns, it is helpful to first examine national college-going rates and the national demographic trends that impact them. It is also helpful to situate the discussion within the context of the ongoing national debate regarding the relative impact various factors have on college-going rates. These include factors that are internal to K-12 systems, such as academic preparation in secondary schools, as well as factors that are external to K-12 systems, including the socioeconomic pressures impacting students and families and the financial aid policies that alternately relieve and intensify these pressures. With this foundation, an analysis of the Philadelphia Regional K-16 Data-Sharing Consortium can more effectively answer the questions that this paper ultimately addresses: What are the relationships among various high school-level factors that influence college-going rates in Philadelphia? And, which of these high school-level factors most strongly influence regional college-going rates in Philadelphia?
College-Going
Rates
College-going rates are an important measure for use by educational institutions
and researchers. The California Postsecondary Education Commission, for
example, states that college-going rates can play a critical role in evaluating
educational systems, refining educational policy, and planning for future
growth. They report,
In addition to projecting enrollment demand, examining college-going rates can facilitate a better understanding of a wide variety of educational matters when considered in conjunction with other indicators. For instance, college-going rates, when examined along with other information, provide a basis by which education analysts, policy makers, and the public can make assessments about the institutional efficiency of our public schools, the effectiveness of student academic development programs, the significance of college awareness programs, the efficacy of admissions policies, and the accessibility of financial aid programs (CPEC, 1999, p. 6).
Increasingly, college-going rates are also recognized as one important indicator of the effectiveness of educational systems, both at the K-12 and the postsecondary level.
The National Center for Education Statistics reports that nationally 63.3% of 2000 high school completers enrolled in college the October after completing high school (NCES, 2002a). Some analysts, however, stress that college-going rates should be determined by considering the number of high school graduates who enroll in a postsecondary institution within a given number of years of high school graduation (often two) rather than those who enroll directly from high school. Nationally, it is estimated that 75% of high school graduates currently meet this criteria (Choy, Horn, Nunez, & Chen, 2000; Education Trust, 1999; NCES, 1998), with that number expected to rise to 80% by 2006 (Education Trust, 1999).
Demographic
Trends
Demographic trends provide an important frame for understanding college-going
rates. One such trend that bears great relevance is the changing number
of high school graduates over time. Heller (2001) describes this trend
nationally: "In 1971, the United States had 2.9 million high school
graduates; the number peaked at 3.2 million in 1977 with the tail end
of the baby boom, then declined to a low of 2.5 million in 1994"
(p. 12). Holsworth (2000) describes a similar trend for Pennsylvania,
in which the number of high school graduates declined 31.6% between 1976
and 1999. On the other hand, the School District of Philadelphia experienced
a slight increase in its number of graduates between 1990 and 2001 (School
District of Philadelphia, 2003a).
Despite a shrinking national pool of college-eligible persons, the number of students enrolling in postsecondary education in the United States has increased significantly. Heller (2001) reports that the number of undergraduates in the United States grew steadily during the 1970's and 1980's and reached a peak of 12.4 million in 1991 before leveling off. He attributes these enrollment increases to a tremendous (156 percent) increase in the number of 'nontraditional' college attendees (students outside the cohort of 18- to 24-year olds) and to a less-dramatic (32 percent) increase in the enrollment of 18- to 24-year old students. In addition, college-bound students nationally are applying to more colleges than ever before (Astin, Oseguera, Sax, & Korn, 2002), thus increasing their options for postsecondary attendance. It seems that these trends are present in Pennsylvania, as well. While state-wide actual college enrollment numbers for Pennsylvania are not available, Holsworth (2000) reports that the percentage of Pennsylvania high school graduates who self-reported that they would be attending an IHE as a degree-seeking student increased steadily from 41.8% in 1976 to 71.8% in 1999. Prior to the existence of the Philadelphia Regional K-16 Data-Sharing Consortium, there existed no comprehensive data on the college-going rates of SDP graduates.
Furthermore, the number and proportion of non-White students in the United States enrolling in college has increased steadily. Evangelauf (1992), Heller (2001), and Perna (2001) all report a recent disproportionate increase in minority postsecondary enrollment. The enrollment growth of Hispanics, Blacks, Asian and Pacific Islanders, American Indians, and Alaskan Natives have all outpaced the enrollment growth of White students. Carnevale and Fry (2002) predict that this trend will continue, with the vast majority (approximately 80%) of growth in the college-eligible population between 2000 and 2015 expected to come from members of minority groups. Although White students still represent 71.4% of those enrolled in postsecondary institutions (NCES, 2002b), their share of the total postsecondary population is diminishing. While the number of minority students who are enrolling in postsecondary institutions and their proportional share of all postsecondary enrollment are growing, Evangelauf (1992) warns that these increases are primarily attributable to the overall growth of the United States' minority population; the college-going rates of Black and Hispanic students remained relatively stable during the 1980's. Differences in college-going trends based on income levels also exist. College-going rates of students from all income groups increased between 1970 and 1997. The gap between these rates for students from the lowest and highest income quartiles remained flat, however, at 32%, with 85 % of high school graduates from families earning more than $75,000 going to college, and only 53 % of graduates from families earning less than $25,000 doing so (Fitzgerald & Delaney, 2002). Students from low-income families are more likely to enroll in college now than they were 30 years ago, but they remain proportionally less likely to do so than their higher-income peers; this gap has not changed significantly.
Despite the declines in the number of high school graduates, the proportion of graduates enrolling in postsecondary education has increased dramatically over the past 30 years. Along with that rise, the racial composition of the United States' postsecondary student population has become increasingly diverse. Students from low-income groups are more likely than they were 30 years ago to enroll in college, but are still significantly less likely to do so than their higher-income peers. These national trends provide a backdrop for understanding the college-going patterns of Philadelphia public high school graduates. They also frame the way that researchers have investigated the causes and effects of changing college-going rates and patterns.
Factors
Influencing College-Going Rates
Burd (2002) reports in the January 25, 2002 edition of the Chronicle of
Higher Education that "The question of which is the primary obstacle
[to college participation]-money or preparation-has created a deep rift
among researchers who have been studying federal higher-education policy
for decades"(p. A18). This divide significantly marks the existing
literature that explores the factors influencing college-going rates and
patterns.
Financial aid policies. Many researchers focus their attention on the role that federal, state, and institutional financial aid policies play in shaping college-going rates and patterns. Through their analyses of these economic and policy trends, these researchers examine the impact of socioeconomic factors such as race, gender, social class, and family income on college-going rates and patterns. They posit that structural forces play a crucial role in determining low-SES students' access to college (or lack thereof.) Unfortunately, financial aid information for graduates of Philadelphia public high schools was unavailable for this study. Nevertheless, the existing research dealing with federal, state, and institutional financial aid policies provides an adequate, if imperfect, frame for contextualizing the socioeconomic variables included in this study - high schools' minority enrollment and percentages of students eligible for a free or reduced lunch. Callan (2001), Gladieux (2002), and Lee (2002) outline two dominant trends in federal policy over the last two decades. First, the primary form of federal aid made available to college-eligible students has shifted from need-based grants to loans. In addition, the real value of federal Pell Grants, long the bulwark of financial assistance for low-income students, has steadily decreased (Heller, 2001; Gladieux, 2002). Second, there has been a federal shift in emphasis from directly appropriating funds to help needy low-income students attend college towards providing tax credits and tax incentives to provide relief that helps 'academically deserving' (and predominantly middle- and upper-class) students more easily afford college. The 1997 enactment of the Hope Tax Credit, providing relief for the cost of the first two years of college tuition, and the Lifelong Learning Tax Credit, for tuition expense after the first two years, are examples of this shift (Gladieux, 2002).
In recent years, state aid to students has undergone similar shifts with similar results. State aid programs have moved both their focus and their dollars away from the need- and non-need-based grants towards awards based on merit (Heller, 2002, p. 65). This has impacted college-going patterns by rewarding and supporting those (usually middle- and upper-income) students already most likely to attend college to the disadvantage of students who are economically needy and/or from racial/ethnic groups that are historically underrepresented in the postsecondary arena.
Furthermore, McPherson and Shapiro (2002) argue that "colleges and universities have also shifted their practices in providing student aid-both need-based and merit-based-to target more benefits to middle- and upper-income students"(p.73). As Gose (1998) reports, even urban institutions that have historically served the largely low-income, working class, minority populations in which they are embedded have altered their recruitment and admissions patterns. Institutions such as Temple University (the primary four-year destination for School District of Philadelphia students, according to the Philadelphia Regional K-16 Data-Sharing Consortium) are intensifying efforts to "woo" middle-class, suburban students who "are better prepared academically and come from wealthier families than city students" (Gose, 1998, p. A61). These trends in institutional, state, and federal aid policy have occurred in combination with a changing national economy and rising postsecondary tuition rates. The result has been seen in the costs of both tuition and college attendance as percentages of total family income rising markedly over the last three decades, with the greatest burden being felt by low-income students (Heller, 2001). The impact of these policy trends has been to significantly shift the objective of federal student aid from increasing access to postsecondary education for low-income students to increasing the affordability of postsecondary education for middle class students (Lee, 2002). Researchers focusing on the roles that federal, state, and institutional financial aid policies have on college-going rates and patterns argue that socioeconomics, particularly family income level, play a critical role in determining access to postsecondary education.
Academic preparation. The persistence of low-income students' comparatively low college-going rates is one of the most troubling-and most researched-facets of higher education (Burd, 2002). While some researchers focus on socioeconomic factors and structural forces when explaining this phenomenon, others focus on factors that are internal to K-12 educational systems. Of particular interest to many such researchers is students' academic preparation while in high school.
The National
Center for Educational Statistics is a primary proponent of this alternative
explanation for trends in national college-going rates and patterns. NCES
(1998a) analyzed data from the National Education Longitudinal Study of
1988 (NELS:88) to examine access to postsecondary education of 1992 high
school graduates by 1994, two years after high school graduation. They
reported,
High school graduates whose parents have low levels of income and education are able to attend four-year colleges at the same rates as students from middle-income families, if they do what four-year colleges expect them to do. That is, if low-income students have an academic record and aptitude test scores which demonstrate even the minimal qualifications for admission to a four-year institution, if they take a college entrance examination, and if they submit an application for admission, the majority of low-income students enroll in postsecondary education, and over 83 percent attend a four-year college or university (NCES, 1998a, p.1)
NCES (1998b)
gauged academic preparation as an index of criteria that includes high
school grade point average, class rank, standardized test scores, academic
coursework, and completion of steps necessary for admission to a four-year
institution (taking a college entrance examination and submitting an application.)
Students were classified on a range from 'very highly qualified' to 'marginally
or not qualified.' They reported the following relationship between this
index and college-going rates:
Among those seniors classified as marginally or not qualified for regular
four-year college admission, half entered postsecondary education, but
only 15 percent enrolled in a four-year college or university. Among those
seniors who were minimally qualified, three-fourths enrolled in some postsecondary
education and 35 percent attended a four-year institution. With each higher
level of academic qualification, the total proportion of students enrolled
in postsecondary education increases…56 percent of the somewhat qualified,
73 percent of the highly qualified, and 87 percent of the very highly
qualified high school graduates enrolled in four-year institutions. (NCES,
1998b, p. 3)
In their analysis, academic preparation, as gauged by the index of criteria described above, impacted college-going (and four-year college-going) rates above and beyond other factors, including income levels. The clear suggestion is that factors related to academic preparation and achievement are more significant determinants of college-going than the structural factors described above. Indeed, NCES (1998c) reported "nearly two-thirds of the low-income 1992 high school graduates were enrolled in postsecondary education within two years of high school graduation despite the financial burden" (p.2).
In an analysis
of earlier NELS data (for the graduating high school class of 1982), the
College Board focused specifically on the impact that students' high school
coursework had on their likelihood of attending college; this study found
similar results. "When students are divided into groups according
to their degree of participation in [high school mathematics, laboratory
sciences, and foreign languages coursework], students who have taken more
courses are more likely to go on to college than those students who have
taken fewer courses" (College Board, 1990, p. 33). Their findings
with regard to specific coursework are worth noting:
Students who took one year or more of algebra were two to three-and-one-half times as likely to attend college as students who did not take algebra.
Almost 82 percent of the students who took at least one year of geometry attended some college within four years of graduation, whereas only 41 percent of students without geometry went on to college.
Students taking at least one year of laboratory science courses attended any college directly following high school at a rate of 65 percent, compared to a rate of 31 percent for students with less than one year of lab science.
Students with two years or more of a foreign language were far more likely to…attend some college within four years of graduation [86 percent, compared to 44 percent of students with fewer foreign language courses.] (College Board, 1990, pp. 35-37)
Academic preparation significantly impacted college-going rates and patterns. Better-prepared, higher-achieving high school graduates were more likely than their worse-prepared, lower-achieving peers to participate in postsecondary education. Other researchers, including Ozden (1996), St. John (1991), and Adelman (1999) offer evidence that supports the conclusion that academic preparation and achievement have the most significant impact on students' college-going rates (and, some argue more importantly, their degree-completion rates).
In addition to academic preparation, researchers have also examined the impact of other factors that are more directly dependent on the internal workings of K-12 educational systems. The College Board (1990) found that students' postsecondary aspirations were closely related to their actual postsecondary paths. In the Board's analysis, more than 85 percent of students expecting to obtain a bachelor's degree attended college within four years of high school graduation, while only 40 percent of those who did not think they would obtain a bachelor's degree did so. Sophomores who intended to complete college were also five times as likely to proceed directly from high school to a four-year institution as their counterparts.
This paper's analysis wades into the literature's divide regarding the factors that most influence college-going rates and patterns. I employ a model for predicting college-going rates using high school-level predictor variables that gauge both socioeconomic factors and factors related to academic preparation. This model provides initial insight into the relative impact that these factors have on college-going rates. The results of this analysis suggest that academic preparation, constructed in this study as a high school's ability to successfully guide students through, at the least, a minimally rigorous high school strongly influences the likelihood that the high school will send its graduates on to regional postsecondary institutions. This is a significant finding given the scope of the ongoing debate in the literature regarding the factors that most influence college-going rates and patterns.
This paper also contributes to the literature by focusing on a unique local environment. I provide a broad portrait of college-going patterns in a regional K-16 education system that includes a wide variety of postsecondary institutions and a large, urban school district that serves a high-minority, high-poverty, student body. The conclusions and policy implications that emerge should thus be of interest to educators and policymakers who have stakes in similar districts across the nation. These conclusions and policy implications should also be of great interest to those interested in the School District of Philadelphia. A thorough examination of college-going rates and patterns is important when attempting to frame, understand, and inform the current reform efforts underway in the School District of Philadelphia.
Research Methods
Data Collection
During the spring of 2002, the School District of Philadelphia (SDP) formally
requested that approximately 60 area institutions of higher education
(IHE's) report the numbers of SDP public high school graduates who applied
to, were accepted by, and enrolled at their respective institutions as
first-time, non-transfer freshmen in each year from 1995 through 2001.
Each IHE was asked to provide this information for all of the 43 public
high schools in SDP. All postsecondary application, acceptance, and enrollment
data used in this study derives from this request. IHE's were selected
for inclusion in the initial request based upon the following criteria:
1) informal conversations with SDP administrators and high school counselors
indicated that SDP graduates applied to the IHE in significant numbers;
2) the IHE is in close geographical proximity to the City of Philadelphia;
and/or 3) inclusion of the IHE helps to provide a representative sample
of types of postsecondary institutions. Forty-six of the 60 IHE's from
whom this request was made provided the requested information2.
Of the 46, 21 provided the requested information for all seven specified
years; 21 institutions provided the requested information for six of the
seven specified years; two institutions provided the requested information
for four of the seven specified years; one institution provided the requested
information for three of the seven specified years; and one institution
provided the requested information for one of the seven specified years.
These forty-six institutions do not represent the entire spectrum of postsecondary destinations for SDP graduates. The seven-year data set resulting from this request is accordingly not complete. The 46 participating regional colleges and universities do, however, represent a broad variety of institutional types, including public, private, state, state-related, and historically Black institutions, as well as two- and four-year institutions. Therefore, they provide a strong foundation from which policymakers and researchers can begin to understand the regional college-going patterns of Philadelphia public high school graduates and the contours of the Philadelphia-area K-16 educational system. As a more comprehensive listing of the actual institutions attended by SDP graduates becomes available, analysis of regional college-going rates and patterns can be enhanced by their inclusion, and the conclusions that are able to be drawn will have greater resonance.
Secondary data on Philadelphia high schools was collected from two sources. High school enrollment numbers, numbers of graduates by high school, and percentage of graduates from each high school self-reporting plans to enroll in a postsecondary education were all obtained from the Pennsylvania Department of Education (PDE) website for K-12 statistics (Pennsylvania Department of Education, 2003). This secondary data on high schools is available for 37 of the 43 high schools included in SDP's request to area IHE's. High schools' classification, racial composition, percentage of students eligible for free/reduced lunch, and percentage of June graduates enrolled in a college-preparatory curriculum are all made available to the public and to educational researchers by the School District of Philadelphia (2003b). This secondary data on high schools is available for 34 of the 43 high schools included in SDP's request to area IHE's. It is important to note that no individual student-level data has been collected from any source for this study. The unit of analysis for all data is the individual high school, and each variable is intended to measure school-level characteristics.
Independent
Variables
The secondary data on Philadelphia high schools described above was used
to construct six independent variables. They are:
1. High School Classification: Each high school was classified as either "Comprehensive," "Area Vocational & Technical," or "Special Admit." These three classifications are organized hierarchically, with Comprehensive high schools as a category being the least selective and Special Admit high schools as a category being the most selective. Comprehensive high schools are those that provide students from their immediate geographical vicinity with four years of academic and humanities course offerings. Area Vocational & Technical high schools are those that provide full-time students with four years of both academic and vocational/technical education and selectively admit students based in part on academic criteria. Special Admit high schools are those that have specialized programs (e.g., arts, international affairs, etc.) and have established admissions requirements, based at least in part on academic criteria, for accepting students (School District of Philadelphia, 2003c).
2. High School Size: This continuous independent variable was constructed as the average size of each individual high school's total enrollments in grades 9-12 over the years 1999-2001. High school enrollment numbers, taken each year as of October 1st, are reported by SDP to PDE on the Public School Enrollment Survey (PDE-4035).
3. High School Minority Enrollment: This continuous independent variable was constructed as each individual high school's percentage of non-White students enrolled in grades 9-12 at the end of the 2002 school year. "Minority," in this instance, is used to describe African-American, Black, Latino, Hispanic, and Asian students. 2002 minority enrollment information has been used as a proxy for "minority enrollment" over the years 1999-2001 because this information was not available for each year during this time span and because anecdotal evidence strongly indicates that there has been little change in most high schools' racial composition over this time period.
4. High School Poverty: This continuous independent variable was constructed as each individual high school's percentage of students enrolled in grades 9-12 at the end of the 2002 school year who were eligible for free or reduced lunch. In keeping with accepted educational research practice, "eligible for free/reduced lunch" is used as a proxy for poverty. 2002 high school poverty information has been used as a proxy for "high school poverty" over the years 1999-2001 because this information was not available for each year during this time span and because anecdotal evidence strongly indicates that there has been little change in the aggregate income levels of the most high schools' student bodies over this time period.
5. High School Curriculum: This continuous independent variable was constructed as each individual high school's percentage of June 1999 graduates who 'minimally' completed at least two years of college-preparatory math, science, and the same foreign language. This definition of a 'college-preparatory curriculum' was chosen for three reasons: 1) it reflects the minimum level of curricular rigor with which a high school student can maintain a reasonable expectation of postsecondary success; 2) it reflects a progression through a sequence of increasingly more challenging coursework in specific subjects; and 3) these courses reflect the current research consensus regarding 'gatekeeper' courses critical for postsecondary success (College Board, 1990; Adelman, 1999). While the data available for construction of this 'High School Curriculum' variable does not allow for a comprehensive measure of a school's curricular rigor, it does effectively gauge a high school's capacity for successfully guiding students through a curriculum that is reasonably likely to serve as a solid foundation for postsecondary success. 1999 high school curriculum information has been used as a proxy for high school curriculum information over the years 1999-2001 because information was not available for each year during this time span. Anecdotal evidence strongly indicates that there was little change in high schools' curriculum during this three-year period.
6. College Aspirations: This continuous independent variable was constructed as the percentage of each individual high school's June graduates in the years 1999-2001 who, on the yearly High School & Beyond Survey, self reported plans to enroll in a two- or four-year degree-granting college or university, or a specialized associate degree-granting institution. In keeping with accepted educational research practice, the results of the High School & Beyond Survey were chosen as the best indicator of graduates' postsecondary plans and expectations for themselves.
Dependent
Variable
The secondary data on Philadelphia high schools was also used to construct
the dependent variable used in this study, the preliminary three-year
regional college-going rate. The preliminary three-year regional college-going
rate of 36 high schools was determined by contrasting each high school's
average yearly number of June graduates over the years 1999-2001 with
each high school's average yearly number of graduates who enrolled in
the Fall semester at any of the 46 consortium IHE's over the years 1999-2001.
Limitations
of the Data
The dependent variable is referred to as the preliminary three-year
regional college-going rate because there are some limitations associated
with its calculation. It should be considered as a broad indicator, not
as an exact rate. First, there is some lack of consistency in the enrollment
numbers reported by Consortium IHE's3.
Second, the use of high school-level, and not individual student-level,
data is a defining characteristic of this study. As a result of this reliance
on high-school level data, this study involves a relatively small number
of cases: 43. This decision also makes calculating an exact college-going
rate impossible, and impacts the conclusions that can be drawn from this
data. For example, it is not possible to use this study to draw conclusions
about how students' race/ethnicity, income level, or high school
coursework impacts their likelihood of enrolling in an area postsecondary
institution. It is, however, both possible and appropriate to use this
study to draw conclusions about the influence that a high school's
minority enrollment, poverty level, and course-taking patterns have on
the college going rate of that school's graduates. Although the majority
of the extant research on this matter deals with factors that impact individual
students' enrollment rates, the high school-level data in this study should
be of great interest and value to policymakers, educators, counselors,
parents, and students alike.
In addition, the manner in which the independent variable High School Curriculum is constructed in this study should be carefully considered when drawing conclusions based on this study's analyses. Due to the lack of student-level data employed in this study and the author's limited access to SDP records, High School Curriculum cannot reasonably be considered as an encompassing measure of curricular rigor. Instead, it should be considered as a measure of high schools' ability to successfully guide students through, at the least, a minimally rigorous college preparatory curriculum, and conclusions should be drawn accordingly.
Analyses
My first stage of analysis was determining the ranges, means, standard
deviations, and/or frequencies for each of the variables included in this
study. Next, I determined the Pearson correlation coefficients between
each independent variable and the dependent variable, and among the independent
variables. This stage of analysis indicated statistically significant
relationships between the dependent variable and all but one of the independent
variables (high school size). It also indicated, however, a high degree
of multicolinearity among the independent variables involved.
In order to determine the independent effects that each independent variable has on the preliminary three-year regional college-going rates, I then conducted an ordinary regression analysis. This stage of analysis provided insight into the explanatory power of the model I constructed. It also indicated significant predictive power for one independent variable (High School Curriculum), controlling for all other independent variables.
Due to the evident multicolinearity problem, I also sought to determine if the model would prove more efficient if I were to eliminate some independent variables based on their unique contributions to the model. I achieved this by conducting a backwards stepwise regression and assessing whether there was a significant change in the R- Squared values of the ever-more parsimonious models. By limiting the number of predictor variables in this stage of my analysis, I was able to account for somewhat less of the variance among high schools' preliminary three-year regional college-going rates, but I was able to lessen the multicolinearity problem described above, thus gaining additional insight into the relative predictive power of each of my independent variables.
Results
Table 1 displays descriptive statistics for the five continuous independent variables and one continuous dependent variable as well as a frequency count for the one ordinal independent variable used in this study. High schools range in size from 254 to 3346 students enrolled. High schools' percentage minority enrollment ranges from 34% to 100% and their percentages of students who are eligible for free or reduced lunch ranges from 39% to 86%. High schools' percentages of graduates completing a college-preparatory curriculum ranges from 0% to 100%, while their percentages of graduates who self-report plans to enroll in postsecondary education ranges from 20% to 96%. Clearly, there is a tremendous amount of variance among Philadelphia public high schools along each of these indicators. There is also noticeable variance among Philadelphia public high schools' preliminary three-year regional college-going rates. While the minimum case on this variable, 3.7%, is a low outlier and is missing other relevant information pertaining to the independent variables, there remains wide variance between the next lowest case, 17.1%, and the maximum case, 48.8%.
Table
1 Descriptive Statistics
(Click on the above link to display the table in a new
window)
Additionally, many of these indicators have a strong impact on the dependent variable and/or are tightly intertwined with each other. Table 2 provides the answer to the first research question addressed by this paper, What are the relationships among various high school-level factors that influence college-going rates in Philadelphia? Table 2 shows the Pearson correlation coefficients among all six independent variables and between each independent variable and high schools' preliminary three-year regional college-going rate. Of the six independent variables employed in this study, all but High School Size are highly correlated with the dependent variable. As expected, High School Classification (.487), High School Curriculum (.750), and College Aspirations (.674) are all significantly positively correlated with high schools' preliminary three-year regional college-going rates, while High School Minority Enrollment (-.478) and High School Poverty Level (-.690) are both significantly negatively correlated with the dependent variable.
The multicolinearity present among the independent variables is also apparent at this stage of analysis. As expected, there are strong, statistically significant, positive correlations between High School Poverty Level and High School Minority Enrollment (.699) and between High School Curriculum and College Aspirations (.707). High School Classification is strongly positively correlated with both High School Curriculum (.501) and College Aspirations (.540); Special Admit and other selective high schools are more likely to have high percentages of their graduates at the least minimally academically prepared for and personally motivated for postsecondary education.
Perhaps most
striking, though, are the extremely strong negative correlations between
High School Poverty and High School Curriculum (-.734) and High School
Poverty and College Aspirations (-.704). Those high schools with high
percentages of their student bodies eligible for free or reduced lunch
are very unlikely to have high percentages of their graduates at the least
minimally academically prepared for and personally motivated for postsecondary
education.
Table
2 Correlations
(Click on the above link to display the table in a new
window)
*** Correlation is significant at the .001 level (2-tailed)
** Correlation is significant at the .01 level (2-tailed)
* Correlation is significant at the .05 level (2-tailed)
Taken together as an predictive model, the six independent variables employed in this study explain a great deal of the variance among high schools' preliminary three-year regional college-going rates. The R Squared for this six-variable model is .635, as Table 3 demonstrates. F-testing of this model indicates that it is a significant predictor with regard to the dependent variable when p < .001.
In order to determine the independent effects each predictor variable has on the dependent variable, I conducted an ordinary regression analysis, the results of which are reported in Table 4. This analysis establishes that of the six predictor variables employed in this model, the strongest-and only statistically significant-independent predictor of high schools' preliminary three-year regional college-going rates is High School Curriculum.
The Beta coefficient for this variable is .440, which is significant at p < .05. None of the other independent variables approached significance in this regression analysis.
Given the multicolinearity problems described above, I decided to also conduct a backwards stepwise regression, sequentially removing independent variables from this initial six-variable model based on their individual contributions. By creating ever-more parsimonious models and assessing the changes in the R-Squared of each successive model, I was able to simultaneously develop a more efficient predictive model given the limitations of the data. I was also able to gain further insight into the level of variance in the dependent variable for which High School Curriculum independently accounts. As Table 5 demonstrates, High School Curriculum does indeed seem to account for a significant proportion of this variance above and beyond the other independent variables.
As the original six-variable model is made increasingly sparse through successive removal of the least influential variables, the resulting models still explain the vast majority of the variance present in the dependent variable. By the time Model 4 is created, three independent variables remain: High School Classification, High School Poverty Level, and High School Curriculum. The R-Squared for this model is .624. The least influential of these three variables, High School Classification, is then removed to create Model 5. Model 5, which includes only High School Poverty Level and High School Curriculum, has an R-Squared of .605. The R-Squared change from Model 4 to Model 5 is not statistically significant. A more efficient model has been created, given what is known about the strong correlations between High School Classification and both High School Poverty Level and High School Curriculum.
Table
5. Stepwise Regression
(Click
on the above link to display the table in a new window)
The next step in this regression is the removal of the less influential of the two remaining independent variables, High School Poverty Level. The resulting single-variable model, Model 6, has an R-Squared of .563. The R-Squared change from Model 5 to Model 6, while the most dramatic of any of the regression steps taken, is not statistically significant. Disentangling High School Poverty Level from High School Curriculum does not significantly impact the predictive power of the resulting single-variable model, providing further evidence that High School Curriculum does indeed account for the variance in high schools' preliminary three-year regional college-going rates above and beyond the other independent variables included in this study.
Discussion & Implications
Discussion
Of great interest to both local and national researchers and policymakers
is the second research question addressed by this paper: which high school-level
factors most strongly influence college-going rates in Philadelphia? The
results of this paper suggest this response: High School Curriculum plays
a critical role in determining the college-going rates of Philadelphia
public high schools. Those high schools that are able to successfully
guide a high percentage of their graduates through, at the least, a minimally
rigorous college preparatory sequence, including two years of college-preparatory
math, science, and the same foreign language, are likely to send a high
percentage of their graduates to Philadelphia's regional institutions
of higher education. The results of this study suggest that the impact
of this High School Curriculum variable on high schools' preliminary three-year
regional college-going rates is felt above and beyond the impact of other
factors, including high school classification, high school size, high
school minority enrollment, high school poverty level, and the college
aspirations of high school graduates. Increasing students' access to and
successful completion of a minimally rigorous curriculum at all high schools-including
neighborhood comprehensive high schools that have high minority enrollments
and high poverty levels-is likely to increase these students' access to
and participation in regional postsecondary institutions as well.
This study's
conclusions directly support the findings of other researchers. Choy et
al. (2000), Adelman (1999), NCES(1998a, 1998b, 1998c), Adelman (1998),
Ozden (1996), St. John (1991), and the College Board (1990) all make related
arguments: measures of high school graduates' academic preparation and
achievement, including high school curriculum completed, strongly influence
college-going rates. The results of this paper should not, however, be
taken as a refutation of the argument made by Heller (2001, 2002), Lee
(2002), Callan (2001) and others that structural factors-such as the changing
nature of federal, state, and institutional student financial aid-impact
college access and college-going rates. Conceptualizing academic preparation
and access to financial aid as mutually exclusive creates a false dichotomy
that has dangerous policy implications. It would be counterintuitive to
better prepare all students for college while removing the financial aid
infrastructure that so often makes college attendance possible. Instead,
efforts to maintain and build upon the United States' historical successes
in making postsecondary education widely financially accessible should
serve as a platform for efforts to make postsecondary education more widely
academically accessible. Indeed, the best justification for a strong student
financial aid infrastructure that supports all students' postsecondary
attendance may well be a rising tide of students who are academically
prepared to make the most of it.
Given the strengths of the data upon which these analyses are based and the various statistical angles from which they are evident, the conclusions drawn above are done so reasonably. Given the limitations inherent in the data and the fact that the independent variable High School Curriculum is not able to comprehensively operationalize and measure curricular rigor, however, alternate interpretations should be considered. For example, this study proposes a uni-directional relationship between a high school's ability to successfully guide students through, at the least, a minimally rigorous college preparatory curriculum and that school's likelihood of sending its graduates to regional colleges and universities. This interpretation should be tested in future studies.
Implications
This paper's conclusions also lend initial support to the School District
of Philadelphia's current high school reform policies, which include a
focus on making a more rigorous curriculum available to all high school
students. The results of this study suggest that effective strategies
for improving high schools' ability to successfully guide students through,
at the least, a minimally rigorous college preparatory curriculum will
result in regional postsecondary educational options becoming more widely
accessible to all students. SDP's current efforts to improve curricular
offerings, to ensure that all students are aware of and are placed into
these rigorous courses, and to staff these courses with teachers who can
help all students towards successful completion appear to qualify as such
strategies. Future research and evaluation should test this. The outcomes
of Philadelphia's efforts to improve access to both a college preparatory
high school curriculum and postsecondary education will no doubt be of
great interest to school districts across the country, particularly those
that feed large urban K-16 systems.
These conclusions also have policy implications for the higher education side of the K-16 equation, both locally and nationally. These conclusions suggest that colleges and universities in the Philadelphia region can increase students' access to their institutions by working more closely with SDP to improve schools' ability to successfully guide students through a minimally rigorous college-preparatory curriculum. K-16 philosophy includes numerous strategies for postsecondary involvement in this effort, including: alignment of curriculum and assessment; improvement of teacher preparation efforts, so that more SDP teachers are prepared to teach rigorous curriculum to SDP students; and further integration into a regional K-16 system that facilitates a renewed focus on college-going for all students. When implemented thoroughly and carefully, these strategies have pointed towards increased college-going among all students in urban areas including New York City and El Paso, Texas. The responsibility for improving the college-going rates of Philadelphia public school graduates should by no means be considered the sole responsibility of the School District of Philadelphia.
Finally,
this paper points to important directions for future research into college-going
rates and patterns, both locally and nationally. In both instances, increased
efforts should be made to more closely examine the links between students'
secondary and postsecondary performance along a range of indicators. While
this study points to a link between high schools' ability to successfully
guide students through, at the least, a minimally rigorous college-preparatory
curriculum and that school's likelihood of sending its graduates to regional
colleges and universities, the High School Curriculum variable used in
this study is imperfect. Efforts to more accurately operationalize and
measure curricular rigor will be welcome additions that can test and provide
substance to the findings of this study. For example, an analysis of the
relationship between students' high school coursework and their performance
on college placement exams will allow researchers and policymakers to
more precisely examine, understand, and evaluate the effectiveness of
educational systems across the K-16 spectrum. Additionally, a mixed-method
study able to more accurately analyze the actual content and instruction
in high schools and classrooms and their impact on college-going would
be quite valuable. Throughout the future research, every effort should
be made to consider the large group of students who are excluded from
most studies involving college-going rates (including this one): high
school non-completers. These students need to be included in our analyses
of the factors influencing college-going rates and patterns if we are
to truly use these measures to expand access to and participation in postsecondary
education to all students. Ultimately, this should be the primary goal
of both local and national researchers and policymakers who are analyzing
college-going rates and patterns.
References
Adelman, C. (1999). Answers in the tool box: Academic intensity, attendance patterns, and bachelor's degree attainment. Washington, D.C.: United States Department of Education.
Astin, A. W., Oseguera, L., Sax, L. J., & Korn, W. S. (2002). The American freshman: Thirty -five year trends. University of California, Los Angeles: Higher Education Research Institute, Graduate School of Education & Information Studies.
Burd, S. (2002, January 25). Rift grows over what keeps low-income students out of college. The Chronicle of Higher Education, 48, pp. A18-A19.
California Postsecondary Education Commission. (1999). College-going rates: What are they and what do they tell us? [Online update]. Retrieved January 19, 2003 from the World Wide Web: http://www.cpec.ca.gov/HigherEdUpdates/ Update1999/ UP99-1.ASP
Callan, P. M. (2001). Reframing access and opportunity: Problematic state and federal higher education policy in the 1990s. In D. Heller (Ed.). The states and public higher education philosophy (pp. 83-99). Baltimore, MD: The Johns Hopkins University Press.
Carnevale, A. P. & Fry, R A. (2002). The demographic window of opportunity: College access and diversity in the new century. In D. Heller (Ed.) Condition of access: Higher education for lower income students (pp. 137-152). Westport, CT: American Council on Education and Praeger Publishers.
Choy, S. P., Horn, L. J., Nunez, A., & Chen, X. (2000). Transition to college: What helps at-risk students and students whose parents did not attend college. New Directions for Institutional Research, 107, pp. 4-63.
The College Board. (1990). Changing the odds: Factors increasing access to college. New York City: The College Entrance Examination Board.
The Education
Trust (1999). Ticket to nowhere: The gap between leaving high
school and entering college and high-performance jobs. Washington,
D.C.: The Education Trust.
The Education
Trust (2001). Youth at the crossroads: Facing high school and
beyond. Prepared for the National Commission on the High School Senior
Year by the Education Trust, Inc.
Evangelauf, J. (1992, January 22). Minority-group enrollment at colleges rose 10% from 1988 -1990, reaching record levels. The Chronicle of Higher Education, 39, A33, A37.
Fitzgerald, B. K. & Delaney, J. A. (2002). Educational opportunity in America. In D. E. Heller (Ed.) Condition of Access: Higher Education for Lower Income Students (pp. 4-24). Westport, CT: American Council on Education and Praeger Publishers.
Gladieux, L. E. (2002). Federal student aid in historical perspective. In Donald E. Heller (Ed.) Condition of Access: Higher Education for Lower Income Students (pp. 45-57). Westport, CT: American Council on Education and Praeger Publishers.
Gose, B. (1998, December 11). Temple U. raises standards to woo suburban students. The Chronicle of Higher Education, 44, pp. A61-A63.
Heller, D. E. (2001). Trends in the affordability of public colleges and universities: The contradiction of increasing prices and increasing enrollment. In D. E. Heller (Ed.), The States and Public Higher Education Philosophy (pp. 11-38). Baltimore, MD: The Johns Hopkins University Press.
Heller, D.E.
(2002). State aid and student access: The changing picture. In D. E. Heller
(Ed.) Condition of Access: Higher Education for Lower Income Students
(pp. 59-72). Westport, CT: American Council on Education and Praeger Publishers.
Holworth, K. B. (2000). Actual and projected high school graduates
by county for Pennsylvania, 1999-2011. Prepared for the Pennsylvania
State University Budget Office, University Park, Pennsylvania.
Labaree, D. (1997). How to succeed in school without really learning: The credentials race in American education. New Haven: Yale University Press.
Lee, J. B. (2002). An issue of equity. In D.E. Heller (Ed.) Condition of Access: Higher Education for Lower Income Students (pp. 25-41). Westport, CT: American Council on Education and Praeger Publishers.
McPherson, M. S. & Schapiro, M.O. (2002). Changing patterns of institutional aid: impact on access and education policy. In D. E. Heller (Ed.) Condition of Access: Higher Education for Lower Income Students (pp. 73-94). Westport, CT: American Council on Education and Praeger Publishers.
National Center for Education Statistics. (1998a). Access to postsecondary education for the 1992 high school graduates: highlights [Online report]. Retrieved March 7, 2003 from the World Wide Web: http://nces.ed.gov/pubs98/access/98105-1.html.
National Center for Education Statistics. (1998b). Access to postsecondary education for the 1992 high school graduates: qualification for admission to four-year colleges [Online report]. Retrieved March 26, 2003 from the World Wide Web: http://nces.ed.gov/pubs98/access/98105-10.html
National Center for Education Statistics. (1998c). Access to postsecondary education for the 1992 high school graduates: financial aid and college costs [Online report]. Retrieved March 7, 2003 from the World Wide Web: http://nces.ed.gov/pubs98/access/981058.html
National Center for Education Statistics. (2002a). Percentage of high school completers who were enrolled in college the October after completing high school, by family income and race/ethnicity: October 1972-2000 [Online table]. Retrieved March 7, 2003 from the World Wide Web: http://nces.ed.gov/programs/coe/2002/section3/tables/t20_1.asp.
National Center for Education Statistics. (2002b). Number and percent of students enrolled in postsecondary institutions, by disability status and selected student characteristics: 1995 -96. [Online table]. Retrieved March 6, 2003 from the World Wide Web: http://nces.ed.gov/pubs2002/digest2001/tables/dt212.asp
Oklahoma State Regents for Higher Education. (2002). Oklahoma educational indicators program, high school to college-going rates: for Oklahoma high school graduates to Oklahoma colleges. Oklahoma City, OK: Oklahoma State Regents for Higher Education.
Ozden, Y. (1996). Have efforts to improve higher education opportunities for low-income youth succeeded? Journal of Student Financial Aid, 26(3), pp. 19-39.
Pennsylvania Department of Education. (2003). Pennsylvania K-12 statistics [Online tables]. Retrieved March 1, 2003 from the World Wide Web:http://www.pde. state.pa.us/k12statistics/lib/k12statistics
Perna, L. W. (2000). Racial and ethnic group differences in college enrollment decisions. New Directions for Institutional Research, 107, pp. 65-81.
School District of Philadelphia. (2003a). [School District of Philadelphia High School Graduates, 1990-2001]. Unpublished raw data.
School District of Philadelphia. (2003b). [School District of Philadelphia High School Demographics]. Unpublished raw data.
School District of Philadelphia. (2003c). High School Resources Book. Philadelphia, PA: School District of Philadelphia.
St. John, E. P. (1991). What really influences minority attendance? Sequential analyses of the high school and beyond sophomore cohort. Research in Higher Education, 32(2), pp. 141-158.
APPENDIX I: PARTICIPATING IHE'S
Arcadia University
Bloomsburg University of Pennsylvania
California University of Pennsylvania
Cheyney University of Pennsylvania
Clarion University of Pennsylvania
Community College of Philadelphia
Drexel University
East Stroudsburg University of Pennsylvania
Edinboro University of Pennsylvania
Haverford College
Howard University
Indiana University of Pennsylvania
Kutztown University of Pennsylvania
La Salle University
Lincoln University
Lock Haven University of Pennsylvania
Mansfield University of Pennsylvania
Millersville University of Pennsylvania
Morgan State University
Neumann College
New York University
Penn State University (19 campuses)
Shippensburg University of Pennsylvania
Slippery Rock University of Pennsylvania
Temple University
University of the Arts
Villanova University
West Chester University of Pennsylvania
Notes
1 The author oversaw development of this partnership. back
2
See Appendix 1 for a complete listing of participating IHE's back
3
In reporting the numbers of 'first-time, non-transfer freshmen,' consortium
IHE's did not employ a single time frame for inclusion of students. That
is, IHE's did one of two things: either reported on 'traditional college-aged
students,' usually defined as students directly removed from high school
or removed by two to five years from high school; or reported on students
of all ages, indicating that the overwhelming majority (usually 85% or
more) of their freshmen classes were of traditional college age. As a
result, it is not entirely clear how total is the overlap between those
enrolling in consortium IHE's in the Fall semester of 1999-2001 and those
who graduated from high school in the Spring of years prior to 1999-2001.
The totality of this overlap is also affected by the construction of the
formula used to calculate college-going rates; 1999 graduates have three
years in which to enroll, while 2001 graduates have only one. Also, it
should be noted that the enrollment data reported for the only two-year
institution in the consortium includes both full- and part-time enrollees,
while the enrollment data reported for all of the four-year institutions
includes only full-time enrollments. back
View
the PDF version
Note: You will need Acrobat
Reader to view or print these notes in Portable Document Format (PDF).