Informing PInforming PInforming Public Public Public ...

Integrating Higher EducationFinancial Aid and Financing Policies

Supported by a grant from Lumina Foundation for Education

I n f o r m i n g PI n f o r m i n g PI n f o r m i n g PI n f o r m i n g PI n f o r m i n g Pu b l i c Pu b l i c Pu b l i c Pu b l i c Pu b l i c Po l i c y :o l i c y :o l i c y :o l i c y :o l i c y :F i n a n c i a l A i d a n dF i n a n c i a l A i d a n dF i n a n c i a l A i d a n dF i n a n c i a l A i d a n dF i n a n c i a l A i d a n d

S t u d e n t PS t u d e n t PS t u d e n t PS t u d e n t PS t u d e n t Pe r s i s t e n c ee r s i s t e n c ee r s i s t e n c ee r s i s t e n c ee r s i s t e n c e

A u g u s t 2 0 0 3

Informing Public Policy:Financial Aid and

Student Persistence

August 2003

Donald E. Heller

Supported by a grant from Lumina Foundation for Education

Western Interstate Commission for Higher Education

The Western Interstate Commission for Higher Education (WICHE) is a public, interstate agencyestablished to promote and facilitate resource sharing, collaboration, and cooperative planningamong the Western states and their colleges and universities. Member states are:

Alaska Idaho OregonArizona Montana South DakotaCalifornia Nevada UtahColorado New Mexico WashingtonHawaii North Dakota Wyoming

WICHE’s broad objectives are to:

• Strengthen educational opportunities for students through expanded access to programs,• Assist policymakers in dealing with higher education and human resource issues through

research and analysis, and• Foster cooperative planning, especially that which targets the sharing of resources.

This publication was prepared by the Policy Analysis and Research Unit, which is involved in theresearch, analysis, and reporting of information on public policy issues of concern in the WICHEstates.

This report is available free of charge online at http://wiche.edu/Policy/Changing_Direction/Pubs.htm.

For additional inquiries, please contact Caroline Hilk at (303) 541-0224 or [email protected].

Copyright © August 2003 by theWestern Interstate Commission for Higher Education

P.O. Box 9752Boulder, Colorado 80301-9752

Telephone: (303) 541-0200An Affirmative Action/Equal Opportunity Employer

Printed in the United States of AmericaPublication Number 8A21

Contents

Acknowledgements ............................................................................................................................... vForeword ............................................................................................................................................. viiExecutive Summary ..............................................................................................................................ixIntroduction ..........................................................................................................................................1

About the Changing Direction Project ...........................................................................................1A Brief History of State and Institutional Financial Aid ..................................................................1Objectives of this Study ..................................................................................................................3

Brief Review of the Research on Financial Aid and Persistence ...........................................................5Analysis of Institutional Grant Awards in the 1999-2000 Academic Year ..........................................9

Introduction ....................................................................................................................................9National Data ..................................................................................................................................9

The Relationship between Institutional and State Grants, Student Persistence, and Degree Attainment .................................................................................................................. 18

Description of the Models ........................................................................................................... 18Models of Persistence and Degree Attainment ........................................................................... 20Summary of Changes in the Awarding of Institutional and State Grants, 1995-1996 to 1999-2000 ....................................................................................................... 27

Conclusion ......................................................................................................................................... 28Summary of Findings ................................................................................................................... 28Policy Implications for States and Public Higher Education Institutions .................................... 29Questions and Discussion Items for State and Institutional Leaders ......................................... 30

Biography ........................................................................................................................................... 33Appendix A - NPSAS Sample Definition ............................................................................................ 35Endnotes ............................................................................................................................................ 37

TablesTable 3.1 Institutional Grant Awards by Sector, 1999-2000 ........................................................... 10Table 3.2 Institutional Grant Awards by Attendance Status, 1999-2000 ........................................ 11Table 3.3 Institutional Grant Awards by Income Quartile, 1999-2000 ............................................ 12Table 3.4 Institutional Grant Awards by Age, 1999-2000 ............................................................... 13Table 3.5 Institutional Grant Awards by Dependency Status, 1999-2000 ....................................... 14Table 3.6 Average Student Cost of Attendance in High Aid and Low Aid States, 1999-2000 ........ 16Table 3.7 Institutional Grant Awards by Sector in High Aid and Low Aid States, 1999-2000 ........ 17Table 4.1 Variables used in Multivariate Models .............................................................................. 19Table 4.2 Models of Persistence/Attainment in 1996-1997 ............................................................ 21Table 4.3 Models of Persistence/Attainment in 2000-2001 ............................................................ 23Table 4.4 Models of Bachelor’s Degree Attainment through 2000-2001 ........................................ 24Table 4.5 Institutional and State Grant Awards by Sector, 1995-1996 to 1999-2000 ................... 26Table A.1 NPSAS:2000 Institutional Sample Strata ........................................................................... 36

iv

x Acknowledgements

This project could not have been undertakenwithout the generous support of the WesternInterstate Commission for Higher Education. For50 years, this organization has had a mission ofpromoting public policies that assure accessand excellence in higher education. Inparticular, I would like to thank Cheryl Blanco,senior program director, and DemaréeMichelau, project coordinator, for theirassistance and guidance in the formulation andwriting of the report. They were an invaluableresource in helping complete this project in avery short period of time.

In addition, I would like to thank the membersof the review panel who read and commentedon the first draft of this report. Theseindividuals gave up part of their valuable timein the summer to attend a meeting to discussthe report, and they too provided extremelyhelpful suggestions and feedback. The reviewpanel members included:

Brenda Albright, Consultant

Cheryl Blanco, Director, Policy Analysis and Research,Western Interstate Commission for Higher Education

Debra Cheshier, Director of Educational Policy,Planning and Improvement Center, MissouriDepartment of Higher Education

Patrick Dallet, Deputy Director, Council for EducationPolicy Research and Improvement

Julie Davis Bell, Education Program Director, NationalConference of State Legislatures

Nancy Goldschmidt, Associate Vice Chancellor ofPlanning and Accountability, Oregon University System

v

Dennis Jones, President, National Center for HigherEducation Management Systems

Jacqueline King, Director, Center for Policy Analysis,American Council on Education

Keith King, Majority Leader, Colorado House ofRepresentatives

Charles Lenth, Senior Associate, State Higher EducationExecutive Offices

David Longanecker, Executive Director, WesternInterstate Commission for Higher Education

Larry Matejka, Executive Director, Illinois StudentAssistance Commission

Demi Michelau, Project Coordinator, Western InterstateCommission for Higher Education

Shirley Ort, Associate Provost and Director, Universityof North Carolina

Christine Walton, Education Policy Associate, NationalConference of State Legislatures

Thomas Wickenden, Associate Executive Director forAcademic and Student Affairs, Arizona Board ofRegents

John Wittstruck, Senior Research Associate, MissouriDepartment of Higher Education

David Wright, Senior Research Analyst, State HigherEducation Executive Officers

Donald E. HellerUniversity Park, PennsylvaniaAugust, 2003

vi

x Foreword

Financial aid issues have emerged in publicpolicy in recent years as states and their highereducation institutions struggle to determinehow to best allocate limited resources across awide range of financial aid programs. Not onlymust states and institutions consider the needsof their citizens and the state, they must alsolook at the distribution of funds amongprograms that serve needy students and thosethat serve meritorious students or those withspecial skills.

However a state or institution decides to awardits financial aid, there is the hope that bygranting aid the student will persist in collegeand graduate. The actual granting of money isdone with the expectation that the additionalmoney will either make it possible for thestudent to stay enrolled until graduation or itwill entice the student to stay at that particularinstitution or in that state. The major objectivefrom the state and institutional perspective is toretain the student by providing financialassistance.

The relationship between financial aid andpersistence is a tenuous one. And measuring orstudying that relationship is made even morefragile by the push and tug of a myriad of otherfactors in a student’s life. In addition to lookingat financial aid in terms of influencing studentbehavior, states must make choices throughpublic policy about the most effectivemechanism for delivering the aid—throughstate-based programs or through institution-based programs.

This study by Dr. Donald Heller, PennsylvaniaState University, was commissioned throughChanging Direction: Integrating HigherEducation Financial Aid and Financing Policies,a national initiative designed to help states andkey constituents examine how to structurefinancial aid and financing policies andpractices to maximize participation, access, andsuccess for all students and to promote moreinformed decision making on issuessurrounding financial aid and financing inhigher education. Changing Direction servespolicymakers in the legislative and executivebranches of state government and their staffs,higher education researchers, state executiveagencies, governing and coordinating boards,educators, college and university leaders, andbusiness and corporate leaders.

This report is one of a series of documentsproduced to foster greater understanding of keyissues related to establishing strongeralignment of financial aid and financingpolicies. Four complementary reports are:

x Policies in Sync: Appropriations, Tuition,and Financial Aid for Higher Education —A set of four commissioned papers that lookinto a system comprised of integratedfinancial aid and financing policies.

x Linking Tuition and Financial Aid Policy:The State Legislative Perspective —Summary of survey responses fromlegislative leaders in the US on the degree ofalignment between tuition and financial aidpolicymaking, their role in the policymaking

vii

process, and their degree of satisfactionwith the process.

x Tuition and Fees Policies in the Nation’sPublic Community Colleges – An analysis oftuition and fees policies among publiccommunity colleges in the U.S. withimplications for public policy.

x Integrating Financial Aid and FinancingPolicies: Case Studies from Five States - Acollection of case study reports fromArizona, Connecticut, Florida, Missouri, andOregon as they have tried to align statehigher education policies related to financialaid and financing

The Changing Direction project has beensuccessful in large part because of WICHE’scollaboration with the American Council onEducation (ACE) and the State Higher EducationExecutive Officers (SHEEO). ACE’s Center forPolicy Analysis and SHEEO have long-standingreputations for high-quality work on a widerange of issues, with a history of specializing infinancial aid and financing issues. WICHE andits partners also collaborate closely with thenational Conference of State Legislatures

viii

(NCSL), a national, bipartisan organization thatbrings even more visibility to the project andprovides additional expertise concerning thestate legislative role in creating integratedhigher education policy. The cooperationamong the organizations has been especiallyvaluable to this project.

WICHE is most grateful to Lumina Foundationfor Education, a private, independentfoundation that strives to help people reachtheir potential by expanding access and successin education beyond high school, for itsgenerous support of this project. Without theFoundation’s assistance and encouragement,this project would not be possible.

David LonganeckerExecutive DirectorWestern Interstate Commission for HigherEducation

x Executive Summary

Changing Direction: Integrating HigherEducation Financial Aid and Financing Policies,a project of the Western Interstate Commissionfor Higher Education (WICHE), examines how tostructure financial aid and financing policiesand practices to maximize college participation,access, and success for all students. This study,Informing Public Policy: Financial Aid andStudent Persistence, was commissioned byWICHE to help support one of the ChangingDirection project goals: that of helping toinform decision-making on issues related tofinancial aid and student success.

The primary objective of this report is to shedlight on the topic of institutionally and state-funded grants to students attending publichigher education institutions. This study usesdata from a nationally representative survey ofthe U.S. Department of Education to analyze thecharacteristics of students receiving thesegrants and whether the use of the grants haschanged in recent years. It also examineswhether the institutional and state grant awardsare related to student persistence and degreeattainment. Bivariate and multivariate statisticalmethods are used to perform these analyses.

Data presented in this report indicate that whilethe number of institutional grant awards andaverage award size increased between the1995-1996 and 1999-2000 academic years,the size of the awards did not keep pace withthe increase in tuition costs at four-yearinstitutions, though it did at communitycolleges. In addition, institutional aid in the

latter year was much more likely to be awardedwithout consideration of financial need: while in1995-1996, 42 percent of institutional grantdollars provided to undergraduates in publicinstitutions was awarded without means testing,four years later almost two-thirds of all dollarswere awarded in this fashion. This trend echoesthat of the growth of state-sponsored meritscholarship programs. Combining state andinstitutional awards, 28 percent of the totaldollars were awarded without consideration offinancial need in 1995-1996, a proportion thatgrew to 46 percent in 1999-2000.

The analyses of the distribution of institutionalgrant awards by student characteristics indicatesome distinct patterns. Full-time, dependent,and traditional college-age students (under 24)were more likely to receive awards, and theawards received were larger than those to part-time, independent, and older students. Whilelower-income students were more likely toreceive an institutional grant, the overall size ofthe average grant received was larger formiddle- and higher-income students.

In examining institutional grant awards instates with relatively greater spending on state-funded aid, compared to their lower-spendingpeers, different patterns were found. Studentsin the low-aid states were more likely to receivean institutional grant, and institutions in thesestates made much greater use of non-needgrants than did those in the high-aid states.The average non-need grant tended to belarger than need-based awards.

ix

The multivariate analysis examines how avariety of variables were related to thepersistence and degree attainment of studentswho began their college careers in publicinstitutions in the 1995-1996 academic year(many of those variables were also found inearlier research to be related to persistence).The analyses confirmed the prior research:academic factors are the strongest predictors ofwhether a student successfully navigates his orher way through college to the ultimateattainment of a degree or certificate.

Institutional grants, however, were still found tobe related to persistence and attainment evenafter controlling for the demographic,academic, institutional, and college cost factorsincluded in the models. Students who receivedan institutional need-based grant of $1,200 intheir first year of college (the average grantaward) were 6 percentage points more likely topersist into their second year than werestudents who did not receive an institutionalneed-based award.

Awarding of institutional aid early in a student’scareer was an important predictor of much laterpersistence or attainment of a credential(certificate, associate’s degree, or bachelor’sdegree). Students who received a $1,200 need-based grant in their first year in college were 6percentage points more likely to persist thannonrecipients; those who received the averagenon-need award ($2,000) were also 6percentage points more likely to persist until orattain some type of credential by the 2000-2001 year. Institutional non-need grants hadthe same relationship with attainment of abachelor’s degree by that year. The finding thatwork-study awards were also related topersistence and attainment by the sixth year

x

also echoes that of other researchers. State aidwas also found to be related to persistence intothe second year (non-need grants) andbachelor’s degree attainment (need-basedgrants).

The analyses in this report demonstrate theimportant role that both institutional and stateaid can play in promoting persistence anddegree attainment. Even controlling for otherfactors influential on these outcomes, grantsfrom institutions and the state – aid awards thatare under the direct control of state and/orinstitutional policymakers – are predictors ofpostsecondary success.

Institutional grants have become an importantpart of the fabric of a financial aid system inthis country that has been described by onerecent observer as “a hodgepodge of programsinvolving a number of participants with diverseinterests.”1 In the 1999-2000 year, publichigher education institutions in this countryawarded $2.5 billion in institutional grants toundergraduate students, while the states,through their grant programs, awarded another$2.1 billion to undergraduates in publicinstitutions. These two sources combinedprovided more than the $4.5 billion awarded inPell Grants, the primary federal financial aidprogram.

The states and their higher educationinstitutions have a large amount of resourcesavailable to help offset the costs of attendingcollege for their students (and supplement theassistance available from the federalgovernment and private sources). How thesetwo sources of aid are coordinated – or moreappropriately, whether they are coordinated – isgoing to vary from state to state, depending

largely upon the higher education governancestructure in each. States with more centralizedcontrol over public higher education institutionsor systems have more opportunities to ensurethat state and institutional financial aidprograms work in tandem to accomplish thestate’s goals regarding higher education access,persistence, and degree attainment.

Whether their state has a strong, centralizedhigher education governance structure or amore decentralized configuration, there are anumber of steps decision makers should gothrough to determine how best to use thelimited resources available to promote the

xi

persistence and degree attainment of publiccollege and university students. Thesediscussions should be engaged in by a broadarray of constituents who have responsibility forestablishing the goals of public highereducation in the state, as well as for carryingout the implementation of those goals.Legislators, executive branch educationadvisors, higher education governing orcoordinating boards, system heads, campusleaders, leaders of the business sector, andcommunity organizations – all can play animportant role in helping establish objectivesand devising programs and strategies foraccomplishing them.

xii

x Introduction

About the Changing Direction Project

Changing Direction: Integrating HigherEducation Financial Aid and Financing Policiesexamines how to structure financial aid andfinancing policies and practices to maximizeparticipation, access, and success for allstudents. With funding support from LuminaFoundation for Education, the WesternInterstate Commission for Higher Education(WICHE) embarked on a multiyear project withthe goal of better, more informed decision-making on issues surrounding financial aid andfinancing in higher education.

Changing Direction provides a venue forpolicymakers and educators from all regions ofthe country to critically examine strengths andweaknesses of public policies and develop newmodels by looking at emerging trends, theirpotential impact on higher education, and thepolicy implications related to issues of financialaid, finance, cost of education, and access.While this necessarily involves all sources ofassistance and financing – federal, state, local,and institutional – the project focuses on statepolicies and practices. It addresses currentpractices and policies, with emphasis onexploring innovative, creative, perhaps untestedapproaches to national- and state-levelchallenges. Changing Direction servespolicymakers in the legislative and executivebranches of state government and their staffs,higher education researchers, state executiveagencies, governing and coordinating boards,

educators, college and university leaders, andbusiness and corporate leaders.

WICHE’s primary partners in this initiative arethe State Higher Education Executive Officersand the American Council on Education. WICHEis also collaborating closely on the project withthe National Conference of State Legislatures. 2

A Brief History of State and InstitutionalFinancial Aid

State support of higher education in the UnitedStates began with public allocations to private,largely church-chartered institutions.3 Thissupport was often in the form of the granting ofpublic lands and authorization for the runningof lotteries to benefit the institution. Many stategovernments in the late 18th and early 19thcenturies began to provide direct financialassistance from general tax revenues to supporta number of private colleges and universities.Spurred in part by the Morrill Act of 1862, statesupport for higher education expanded throughthe 19th and 20th centuries, with most statesfocusing their direct appropriations primarily onpublic institutions.

As early as the first half of the 20th century,some states began to develop state-sponsoredfinancial aid programs. The TrumanCommission on Higher Education, in its reviewof the financing of higher education, singledout the New York state scholarship program(the nation’s first large-scale state program),

1

which awarded grants based on performance inthe regents examinations.4 While theCommission did encourage states to developtheir own scholarship programs to help meetthe goal of equal opportunity in highereducation, its primary recommendation was thecreation of a federally funded program becauseof its belief that the states would not or couldnot step up to their obligation:

Irrespective of, and in addition to whateverprogram of grants-in-aid the FederalGovernment may decide to adopt, thisCommission urges generous extension ofState scholarship provisions. Nevertheless,it is realistic to concede that in theimmediate future many States will not feelthat they can afford to embark upon sucha program. . . . In other words, howeverintrinsically desirable it is to extend sucha program within the States, this Comm-ission believes that such scholarshipswould not represent a sufficientlycomprehensive or adequate attack uponthe problem; and especially would this betrue in the less prosperous states.5

Notwithstanding the pessimism of the TrumanCommission, a number of states did join NewYork in developing their own scholarshipprograms. By the end of the 1960s, theCarnegie Commission on Higher Educationreported that there were 19 state-runscholarship programs, and in the 1969-1970academic year they awarded a total of almost$200 million in grants to 488,000 students.6

The programs ranged in size from Maine’s,which appropriated $61,000 and served150 students, to the oldest program in NewYork, where $59 million was divided among263,000 students.

It took the federal government almost twodecades to respond with legislation to helpaccomplish the goal of equality in educationalaccess established by the Truman Commission.The Higher Education Act of 1965 establishedthe federal government’s first broad-basedstudent assistance programs. An importantfeature of the first reauthorization of the HigherEducation Act in 1972 was the creation of theState Student Incentive Grant (SSIG) program,which provided federal matching funds forstate-run, need-based grant programs. Thisproved to be a critical catalyst to thedevelopment and expansion of the stateprograms. While in 1969, 19 statesappropriated just under $200 million for theseprograms, by 1974, this had expanded to36 states and $423 million.7 By 1979, everystate (and the District of Columbia) reported atleast one grant program, and the totalappropriated funds had increased to over$800 million.8 A 1975 survey conducted by theNational Association of State ScholarshipPrograms commented that “growth representedin ’74-75 and ’75-76 in the historical summarytable above, to a large degree, is a response tothe new SSIG Program which permits up to a$1,500 annual student award (equal shares of$750 Federal/State) in this new form of State/Federal partnership.”9 State grant programscontinued to expand in the 1980s and 1990s.As of the 2001-2002 fiscal year, 48 states (allbut Alaska and South Dakota) had programsawarding a total of over $5 billion in grants toundergraduate students.10

The advent of institutional grants in publiccolleges and universities is a relatively newphenomenon. There has been little researchconducted on the topic (more research existson the use of institutional aid by private

2

colleges and universities), but there are somedata available from the National Center forEducation Statistics (NCES). Through itsIntegrated Postsecondary Education DataSystem (IPEDS) surveys, which by law are to becompleted by every Title IV eligible institution inthe country annually, NCES has trackedinstitutional spending on financial aid since1987.11 In FY 1987, public institutions awarded$486 million in institutional grants andscholarships. By 1996, this total increased294 percent, to over $1.9 billion.12 In contrast,institutional spending in private institutionsincreased 227 percent during this same period,and the Consumer Price Index increased39 percent.

State and institutional grants to undergraduatesin public institutions in the 1999-2000academic year totaled $4.7 billion, or over41 percent of the total grants received bystudents in these colleges and universities. Incontrast, federal Pell Grants represented40 percent of the total grants (the remaining19 percent were from private sources). Stateand institutional grants have become animportant part of the complex mechanism thenation uses for funding higher education.Understanding how these sources of aid areused – and in particular, how they are used topromote the persistence and degree attainmentof students – is an important policy issue forstates and higher education institutions alike.

Objectives of this Study

This study was commissioned by WICHE inorder to help support one of the ChangingDirection project goals of “more informeddecision-making on issues surrounding

financial aid and financing in highereducation.”13 As the public as well aspolicymakers at both the federal and statelevels scrutinize higher education to ensure itsaccountability for use of public resources,higher education and state leaders needinformation and analyses to ensure that thoseresources are being used effectively andefficiently. Concerns have been raised that whilethe country has made great progress inimproving overall access to higher education,we have been less successful in ensuring thatstudents are successful once they are in college.And “success” has generally been defined asattaining some form of degree or credentialbefore leaving postsecondary education.

The primary objective of this report is to shedlight on the topic of institutionally and state-funded grants to students attending publichigher education institutions. While there hasbeen a fair amount of research conducted onstate financial aid, much less is known abouthow institutional grants are used in publiccolleges and universities. This study uses datafrom a nationally representative survey of theU.S. Department of Education to analyze thecharacteristics of students receiving thesegrants and whether the awarding of thesegrants is related to persistence and degreeattainment. A secondary objective is tofamiliarize state and institutional policymakerswith the type of data available from the federalgovernment that could be used for conductingtheir own analyses of student persistence.

Following this introduction, the report providesa brief summary of the research on collegeretention and degree attainment, with a focuson the role of financial aid. The subsequentsection summarizes the use of institutionally

3

funded grants by public colleges anduniversities in 1999-2000, the most recent yearfor which data are available. Section 4 thenprovides a multivariate analysis of therelationship between a number of factors –including characteristics of the studentsthemselves and of the institutions they attend –to examine how financial aid is related topersistence in the context of otherdeterminants. The final section of the reportsummarizes the key findings of the study andpresents some questions for policymakers atthe state and institutional level. Thesequestions are designed to help stimulatediscussion of the relationships between stateand institutional financial aid policy by theparticipants in the Changing Direction project.

4

x Brief Review of theResearch on FinancialAid and Persistence

The last three decades have seen a largeamount of research on the factors that affectpersistence in college. These studies utilize avariety of methods and techniques; qualitative,quantitative, case studies, descriptive,multivariate. Entire books have been written onthe topic.14 Scholarly journal articles andmonographs, too numerous to list, have alsoaddressed the topic. These studies have found anumber of factors that help determine whethera student persists through a college program.Socioeconomic factors, academic aptitude andpreparation (both in high school and college),and institutional factors have all been found tohelp promote or hinder retention.15

In recent years, the relationship betweenfinancial aid and persistence has received moreattention. As the price of college has increased(as measured against the ability of students andtheir parents to pay), the role of financial aidboth in promoting access to college and inhelping to ensure students stay enrolled oncethere has received much more attention. Thediversification of the student body to includegroups who historically had rarely participatedin higher education – including the poor,minority and handicapped students, andstudents older than the traditional college-going ages of 18 to 24 – has also helped toraise the visibility of the importance of financialaid in promoting student success.

There are a number of difficulties in conductingresearch on this relationship. First, the term“financial aid” encompasses a wide variety of

programs from a number of different sources.Financial aid generally is provided in one ofthree forms, each of which can have a numberof varieties: grants, scholarships, or tuitionwaivers, which are subsidies toward tuition thatare considered gifts and do not have to berepaid by the recipient; loans, whose repaymentmay or may not be subsidized by the federal ora state government; and work-study assistance,which requires that the student perform someform of work, usually on campus, for whichthey are paid a wage that is largely subsidizedby the federal government (and in some cases,by state governments).

A second difficulty is the longitudinal nature ofthe act of persisting in college and attaining adegree. Unlike cross-sectional studies, whichgenerally only require an interaction betweenresearcher and subjects at a single point intime, research on persistence requires thatstudents be followed for some period of time.Longitudinal research tends to be expensive toconduct because of the logistical difficulties ofstaying in contact with a group of subjects andrevisiting them on a periodic basis to obtainmore information.

A third complexity with conducting research onfinancial aid and student persistence is thechanging nature of financial aid programs. Mostaid programs, whether operated by the federalor state governments, private groups, or highereducation institutions, often change theireligibility rules and award levels on an annualbasis. This makes it difficult to assess the

5

impact of a particular form of aid because itbecomes a moving target; just as studentbehavior (that is, whether they persist ordropout) changes, so are the sources and typesof financial aid.

Even given these problems, however, a goodbody of research has developed to help usbetter understand this relationship. Quantitativeresearch conducted by such scholars as EdwardSt. John, Michael Paulsen, and Larry Leslie andPaul Brinkman have used multiple datasets andstatistical techniques to try to tease out theeffects of financial aid, separate from those ofthe other factors, on persistence and degreeattainment.16 A number of government agenciesand nongovernmental policy organizations havealso sponsored research in this area.17

In general, the research on the relationshipbetween financial aid and college persistencehas found positive effects: that is, the receipt ofa financial aid award is positively related tohigher rates of persistence. These findings holdtrue even in those multivariate analyses thatcontrol for many of the other factors related topersistence and degree attainment. In a reviewof much of this literature, Pascarella andTerenzini concluded that “while the findingswere somewhat mixed, students who receivefinancial aid are as likely to persist in college asthose who do not, even when academic abilitiesare taken into account.”18 They ascribe thiseffect to aid’s ability to “level the financialplaying field” for students from differentincome backgrounds. The implication is thatwithout the financial aid, these students wouldbe less likely to persist than their peers who didnot receive aid.

These positive effects of financial aid onpersistence have been reported for all threeforms of aid: grants, loans, and work-study. Insome studies the effects are more pronouncedon within-year persistence (staying enrolledfrom the first semester to the second) than onyear-to-year persistence. As with reviews of theresearch on the relationship between financialaid and initial college entry, grants have oftenbeen found to be the best type of aid forpromoting persistence.19 While work-study hasalso been found to be positively related topersistence, there is still some question whetherthe effect is primarily because of the financialassistance work-study provides or because ofits value in helping students become moreintegrated into their college campuses.

There are a number of areas of need in the bodyof research on the relationship betweenfinancial aid and persistence. Most of the workon student loans was conducted using cohortsof students that attended college before thelarge expansion in loan limits (andconsequential increase in student indebtedness)that occurred in the 1992 reauthorization of theHigher Education Act. In the nine years sincethe increase in loan limits in thatreauthorization (the most recent for which dataare available), borrowing in the federal studentloan programs increased 168 percent. In theprior nine years leading up to the 1992reauthorization, student indebtednessincreased only 91 percent.20 Research usingmore recent cohorts of students is needed toassess whether student loans are still positivelyrelated to persistence, given these increaseddebt burden levels. Congress is set to examinewhether loan limits need to be raised again inthe current reauthorization, yet some

6

organizations have questioned whether studentloan burdens are already too high.21

A second area is an examination of the effectson persistence of different forms of grants.Historically, the great majority of grantsprovided by states and public institutions wereawarded using financial need as the primarycriterion. Yet in both state and institutionalgrant programs, the recent trend has been inthe use of merit aid, or aid awarded withoutconsideration of financial need (see Section 4 ofthis report for more on this issue). Few studieshave examined directly the differing effects ofneed-based versus merit grants on persistence,though some recent research on related topicsmay help shed light on this question.22 Thisreport does begin to address this issue,however.

7

8

x Analysis ofInstitutional GrantAwards in the 1999-2000 Academic YearIntroduction

Since 1986, the National Center for EducationStatistics (NCES) of the U.S. Department ofEducation has conducted a series of studiesexamining the use of financial aid in Americancolleges and universities. The NationalPostsecondary Student Aid Studies (NPSAS),conducted every three years through 1995, andthen every four years, is a cross-sectional,nationally representative survey of over 50,000college students across the country. Thesurveys include information about the students’socioeconomic backgrounds, academicperformance in high school and college,financing of college, and reasons for attendingcollege, as well as information about thecolleges the students attend.23

In addition to the cross-sectional data, NPSASincludes in alternating surveys longitudinal dataabout students who were beginning theirpostsecondary careers or completing abaccalaureate degree. These longitudinalsurveys, which include varying numbers offollow-up questionnaires, can be used toanswer questions about the persistence ofstudents in college, as well as about the post-baccalaureate experiences of graduatingstudents.

Data from the NPSAS:2000 survey, the mostrecent available, were used to analyze the useof institutional grant awards to undergraduatesby public institutions in the 1999-2000 year.24

In that year, these institutions enrolled over 12million undergraduate students in academic orvocational programs. Analyses for allinstitutions nationally are first presented,followed by an analysis of states that have largestate-run grant programs compared to stateswith much smaller programs.

National DataAid awards by institutional sector

Table 3.1 presents data on institutional grantawards for all public institutions and by sector.As described earlier, over 12 millionundergraduates were enrolled in publicinstitutions in 1999-2000. Approximately 12percent of all students received some form ofinstitutional grants, with 7 percent receiving aneed-based award and 6 percent receiving anon-need award.25 The average grant (forstudents who received a grant) totaledapproximately $1,800, with the need awardsaveraging $1,200 and the non-need awardsaveraging $2,300 each.26

9

Not surprisingly, doctoral-granting institutionswere most likely to give institutional grants tostudents and gave the largest grants, onaverage. These institutions generally charge thehighest tuition in the public sector, and in thosestates where the institutions have the discretionto utilize tuition revenue as they see fit (oftenwithin some broad constraints imposed by stateregulations), these revenues can be used tofund institutional aid programs through “tuitionrecycling.” These doctoral institutions are alsothe most likely to be engaged in fund raisingand development programs which can be usedto create endowed or annually fundedscholarship programs.27

While the average grant at community collegestotaled approximately $750, the mean grant atdoctoral-granting institutions was almost fourtimes as large. Only 8 percent of communitycollege students received institutional grants,

while almost one in five students attendingdoctoral institutions received assistance. In allthree sectors, non-need grants were larger thanneed-based grants.

Aid awards by student characteristics

Four characteristics of the students were usedto analyze differences in aid awards amongthese groups: attendance status, income,dependency status, and age. Thesecharacteristics have been found in the priorresearch to be influential in whether studentsreceive financial aid. For example, students whoattend college less than half time are noteligible for federal Pell Grants and for many ofthe state-funded financial aid programs. Manypublicly funded aid programs, including PellGrants and approximately 75 percent of stategrant awards, impose financial needs-testing

Table 3.1 Institutional Grant Awards by Sector, 1999-2000

% of students Mean grantCategory receiving grant amount

All publicAll grants 11.7% $1,791Need grants 6.5 1,213Non-need grants 5.7 2,296

Community collegesAll grants 8.0 757Need grants 5.3 422Non-need grants 2.3 879

4-year non-doctoral grantingAll grants 13.9 1,783Need grants 5.8 1,064Non-need grants 8.4 2,137

4-year doctoral grantingAll grants 19.2 2,820Need grants 9.4 2,185Non-need grants 10.6 3,023

10

on recipients. And many aid programs restricttheir awards to financially dependent studentsand/or students in traditional age groups.

The NPSAS surveys record three characteristicsof student attendance status: whether theyattended college for the entire academic year;whether they attended full time or less than fulltime; and whether they attended one institutionor more than one institution during the year.They also record each student’s attendancestatus separately in the fall semester. I havedivided students into three categories: thosewho attended a single institution full time forthe entire academic year; those who begancollege in the fall semester and attended asingle institution during the year, but less thanfull time for the full year; and all other students,who are excluded from the analysis.

The rationale for the first group is that full-timestudents attending college for the entire yearare most likely to be recipients of financial aid.

The rationale for excluding the third group isthat the NPSAS data are less reliable forstudents who either attended more than oneinstitution during the year or were not enrolledin college in the fall semester. This third groupincludes approximately 20 percent of the 12million undergraduates enrolled in publicinstitutions in 1999-2000.

Table 3.2 presents grant data for theapproximately 80 percent of all students inpublic institutions who fell into the first twocategories described above. Students whoattended college full time for the entireacademic year were more likely to receive bothneed-based and non-need grants than werestudents who attended less than full time, fullyear. Grants for the full-time students were alsolarger on average, which is what one wouldexpect given the enrollment intensity of andcosts faced by these students.

Table 3.2 Institutional Grant Awards by Attendance Status, 1999-2000

Category # of grants % of students Mean grant

All students (9,631,379)All grants 1,296,148 13 % $1,889Need grants 720,059 7 1,283Non-need grants 632,572 7 2,409

Full-time, full-year (4,153,750)All grants 867,735 21 2,298Need grants 440,376 11 1,650Non-need grants 470,408 11 2,694

Less than Full-time/full-year (5,477,629)All grants 428,413 8 1,060Need grants 279,683 5 704Non-need grants 162,164 3 1,583

11

To examine the pattern of grant awards byincome, I divided all students (public andprivate institutions combined) in the NPSASsample into income quartiles, based on theirparents’ income (in 1998) for dependentstudents, and their own income (and that oftheir spouse, if married) for independentstudents. The analyses include students in thelowest quartile, the two middle quartiles

combined, and the highest quartile, based onthese income levels.28

Approximately 21 percent of students in publicinstitutions were in the lowest income quartile,52 percent were in the middle two quartiles,and 27 percent were from the highest incomegroup.

Dependent Students Independent StudentsLowest quartile Below $31,646 Below $9,000Middle two quartiles $31,646 to $85,583 $9,000 to $41,999Highest quartile Above $85,583 Above $41,999

Table 3.3 presents the number of grants andaverage grant award for students in eachincome group. Students in the lowest quartilewere approximately two and three times morelikely to receive any type of institutional grant

than were students in the middle and upperincome groups, respectively. The average sizeof the total grant received increased, however,as you go up the income ladder.

Table 3.3 Institutional Grant Awards by Income Quartile, 1999-2000

Category # of grants % of students Mean grant

All students (12,144,763)All grants 1,414,630 12% $1,791Need grants 789,500 7 1,213Non-need grants 686,238 6 2,296

Lowest quartile (2,516,995)All grants 497,614 20 1,526Need grants 353,286 14 1,156Non-need grants 164,168 7 2,139

Middle two quartiles (6,304,261)All grants 700,895 11 1,828Need grants 378,318 6 1,156Non-need grants 356,302 6 2,367

Highest quartile (3,323,507)All grants 216,121 7 2,282Need grants 57,896 2 1,936Non-need grants 165,768 5 2,300

12

For need-based grants, students in the lowestincome group were more than twice as likely asthose in the middle groups, and seven timesmore likely than upper quartile students, toreceive an award. However, while students inthe first two groups received average awards ofthe same size, higher income students receivedawards approximately 67 percent larger. Thisresult is likely driven by the fact that higherincome students generally attend moreexpensive schools, and thus their financial needis increased by a higher cost of attendance.

Non-need grants were awarded to students ineach of the three groups at approximately equalrates – from 5 to 7 percent of each group

received a non-need award. The average size ofthe awards was also approximately the same foreach group.

For an analysis of aid awards by age, a cut pointof age 24 (as of December 31, 1995) was used.This is the age below which students aregenerally considered dependent by federalfinancial aid rules.29 Table 3.4 reflects the factthat younger students, who are most likelydependent students, are more likely to receiveinstitutional grants and receive larger grants onaverage. While those under age 24 represented55 percent of all undergraduates in publicinstitutions, they received 72 percent of allgrants shown in Table 3.4.

Table 3.4 Institutional Grant Awards by Age, 1999-2000

Category # of grants % of students Mean grant

All public (12,144,763)All grants 1,414,630 12 % $1,791Need grants 789,500 7 1,213Non-need grants 686,238 6 2,296

Under 24 (6,738,850)All grants 1,016,081 15 2,107Need grants 519,472 8 1,419Non-need grants 544,277 8 2,578

24 and above (5,405,913)All grants 398,549 7 987Need grants 270,028 5 817Non-need grants 141,961 3 1,217

13

The pattern for dependency status is similar(Table 3.5). As noted above, students below theage of 24 can be considered independent. Thus,when one includes those students under theage of 24 who are independent with those 24and above (who must, by federal rules, be

independent), the balance shifts to 53 percentof students falling into the independentcategory. These students received only 37percent of the institutional grant awards,however, and the awards they received weresmaller, on average, than those of theirdependent peers.

The analysis in this section has shown thatthere are differences in the proportion ofstudents receiving an institutional grant, andthe average size of those grants, dependingupon the student’s characteristics. However, itis important to note that a number of factorsinfluence the awarding of institutional grants.Besides those examined here – which can workin tandem or in conflict in influencing grantawards – other factors include the cost of theinstitution attended and the availability andawarding of other financial aid. As noted above,higher-income students tend to attend moreexpensive institutions; thus, all other thingsbeing equal, one would expect this factor to

Table 3.5 Institutional Grant Awards by Dependency Status, 1999-2000

Category # of grants % of students Mean grant

All public (12,144,763)All grants 1,414,630 12% $1,791Need grants 789,500 7 1,213Non-need grants 686,238 6 2,296

Dependent (5,761,830)All grants 896,150 16 2,189Need grants 447,834 8 1,515Non-need grants 490,086 9 2,618

Independent (6,382,933)All grants 518,480 8 1,104Need grants 341,666 5 818Non-need grants 196,152 3 1,492

lead them to qualify for more need-based aidthan if the cost of attendance was more closelyequivalent for students from all income groups.Similarly, full-time students will generallyqualify for more aid than part-time students.

High Aid Versus Low Aid States

In order to examine whether there weredifferences in the use of institutional grants instates that had large state aid programs, ascompared to states with smaller state-fundedprograms, 10 states were selected for eachcategory. The selection was made based on a

14

ranking of the states on how much they spenton need-based grants to undergraduates forevery 18- to 24-year-old resident in the statein the 1999-2000 academic year.30 The 10highest-spending and 10 lowest-spendingstates are included in the two categories. Thepurpose of this analysis is to help answer the

question of whether institutions in states thatdid not have large state-funded aid programssubstitute their own grants to make up for theabsence of state support, in comparison tostates with larger state grant programs.

Each category contains the following ten states:

Need-based Aid All AidSpending per National Spending per National18-24 pop. ranking 18-24 pop. ranking

High aid statesNew York $375 1 $386 1Illinois 301 2 321 2Pennsylvania 274 3 274 5Vermont 264 4 269 6Minnesota 259 5 259 7New Jersey 253 6 282 4Massachusetts 202 7 205 9Iowa 185 8 187 13Indiana 183 9 186 14Connecticut 143 10 145 19

Low aid statesSouth Dakota $0 51* 0 51Georgia 0 50 318 3Alaska 0 49 22 42Louisiana 3 48 144 20Wyoming 3 47 3 50Hawaii 4 46 4 49Alabama 4 45 17 44Mississippi 5 44 67 32Idaho 6 43 8 47Arizona 6 42 6 48

* The NASSGAP ranking includes the District of Columbia along with the 50 states.

15

Together, these 20 states represented 37percent of all undergraduates enrolled in publicinstitutions in the 1999-2000 academic year:the high-aid states enrolled 26 percent of allstudents nationally, and the low-aid statesenrolled 11 percent.

When comparing financial aid awarded by statesand institutions, it is helpful to have anunderstanding of the relative prices of highereducation institutions in the states. The NPSASdata include information about each student’scost of attendance, adjusted for theirattendance intensity. Table 3.6 presents thisinformation for each sector in the two groups ofstates. While the average price of community

Table 3.6 Average Student Cost of Attendance in High Aid and Low Aid States,1999-2000

Sector High aid states Low aid states

All public $7,716 $7,190Community colleges 5,331 5,0354-year non-doctoral granting* 9,377 8,1604-year doctoral granting* 11,929 10,226

* Estimates between the two groups of states are statistically different.

colleges in each group were statistically nodifferent from one another, the high-aid stateshad higher prices in the two four-year sectors.31

.Table 3.7 shows the comparison of institutionalgrant awards between the two groups of states.When examining the public sector as a whole,institutions in the low-aid states were morelikely to award institutional grants to studentsthan were colleges and universities in the high-aid states. The difference was driven by muchlarger proportions of non-need grant awards inthe low-aid states. Institutions in the high-aidstates, however, awarded larger grants onaverage than institutions in the low-aid states.

16

In community colleges, students in the low-aidstates were more likely to receive institutionalgrants, while in both four-year sectors, theproportion of students receiving grants in thetwo groups of states were approximately equal.In all three sectors, students in the low-aidstates were more likely to receive non-needgrants than their counterparts in the high-aidstates. In the four-year doctoral-grantingsector, a higher proportion of students receivedneed-based grants in the high-aid states thanin the low-aid states.

Table 3.7 also compares the mean need-basedand non-need grant in each sector for eachgroup of states to see if the two types of grantsare statistically different from one another. Inboth groups of states across all three sectors,the two types of grants were different from oneanother, as they were in four-year nondoctoralinstitutions in the high-aid states and four-yeardoctoral institutions in the low-aid states (theaverage non-need awards are larger in eachcomparison).

Table 3.7 Institutional Grant Awards by Sector in High Aid and Low Aid States,1999-2000

Top 10 high aid states Bottom 10 low aid states

% receiving % receivingCategory grants Mean grant grants Mean grant

All publicAll grants* 7% $2,443 11% $1,884Need grants* 3 1,847 † 3 1,071†

Non-need grants* 4 2,719 † 9 2,070†

Community collegesAll grants 3 773 8 1,008Need grants 1 463 3 621†

Non-need grants 2 915 6 1,118†

4-year non-doctoral grantingAll grants 9 1,983 10 1,877Need grants 4 964† NA NANon-need grants* 5 2,718 † 9 2,031

4-year doctoral grantingAll grants 17 3,376 16 2,625Need grants* 9 2,633 4 1,639†

Non-need grants 9 3,600 12 2,827†

* Estimates of the two groups of states for this category of grant in this sector are statistically different.† Estimates of the need and non-need grants for this group of states in this sector are statisticallydifferent.NA Too few observations to produce a reliable estimate.

17

x The Relationshipbetween Institutionaland State Grants,Student Persistence,and Degree Attainment

Description of the Models

In order to untangle the effects of the manyvariables that affect student persistence anddegree attainment (described in Section 2), thissection of the report performs a multivariateanalysis of this relationship with a special focuson the role of institutional and state grants. Thedata used for this analysis are from theBeginning Postsecondary Students LongitudinalStudy (BPS), a survey conducted by the NationalCenter for Education Statistics (2003).32

Students attending public two-year and four-year institutions were included in the models.

The BPS survey is based on the NationalPostsecondary Student Aid Study of 1995-1996and extracted approximately 12,000 studentswho were beginning their postsecondarycareers. The students were surveyed in 1998and again in 2001 regarding their continuingeducational progress and other factors. Whilethe BPS surveys provide a wealth of financial aidinformation about students in their freshmanyear of college, only limited information abouttheir receipt of financial aid (and college costs)is available for subsequent years. Nevertheless,the survey is a valuable tool for examining therelationship between financial aid and studentpersistence.

The models used in this analysis group thevariables used as predictors into categories,including: student demographic factors; studentacademics; institutional sector; tuition cost; andstudent aid. Table 4.1 provides the variablesused in the models. For each outcome, thevariables are added in blocks by category, inorder to show the sequential effects of eachgroup of predictors in each model.

While the BPS survey is rich in information, ithas a number of limitations. Because data werecollected both from students as well as theirinstitutions attended in the 1995-1996academic year, there is a great deal ofinformation available, including about financialaid awards, for that year. In the two follow-upsurveys, however, data were collected only fromstudent interviews as well as from some federaldatabases. Thus, there is only limitedinformation available about receipt of financialaid and college costs in students’ subsequentyears in college. However, because students’experiences in their first postsecondary yearsare so critical to whether they ultimately attain adegree or not (most students who drop out doso in there first year or between the first orsecond year), the BPS survey is a useful resourcefor measuring the effects of a number ofvariables.

18

The BPS surveys utilized a complex, multistagesampling design, with the sample stratified byinstitutional sector, and students sampledwithin institutions. Appropriate statisticalmethods have been utilized to account for thiscomplex design, including the use of properweights and Huber/White standard errors togenerate estimates that are nationally-representative.34

Three different outcomes are modeled: whetherthe student persisted from the firstpostsecondary year (1995-1996) to the second(1996-1997); persisted into the sixth year(2000-2001) or ever attained a certificate ordegree (associate’s or bachelor’s); or attained abachelor’s degree by 2001. For the first twooutcomes, attainment of a degree or certificateis counted as a positive outcome. The models inthis section use logistic regression, an

Table 4.1 Variables used in Multivariate Models

Category VariablesOutcome Attainment or persistence through 96-97

Attainment or persistence through 00-01Attainment of a bachelor’s degree by 00-01

Demographic Age as of 12/31/95Parental education level of highest parent (less than HS, HS grad, college or beyond)Income quartile (1994)Race (white, black, Hispanic, Asian American/Pacific Islander, Native American)

Academic Attendance intensity (full-time, part-time, mixed full-time/ part-time)High school merit index (ACT/SATscore, curricular rigor, GPA)Expected highest degree (95-96)College GPANumber of stopouts

Institutional sector Institution as beginning student (community college or 4-year)

Tuition cost9 Net cost of attendance after aid 95-96Financial aid Years of work-study received

Years receiving any aidYears receiving Pell GrantInst. need-based grant amount (95-96)Inst. non-need grant amount (95-96)State need-based grant amount (95-96)State non-need grant amount (95-96)

Note: Varying measures (depending upon what variables are available in the BPS dataset) areused for different outcomes.

19

appropriate multivariate technique when theoutcome is dichotomous in nature.

The effect of each predictor on persistence isexpressed as a delta-p statistic, recommendedby Cabrera and Petersen as a method forexpressing the relationship between a unitchange in a predictor and the estimatedpercentage point change in the outcome.35 Forexample, a delta-p value of 0.025 indicates thata one unit change in the predictor is related to a2.5 percentage point increase in the likelihoodthat a student would persist (or attain acredential). For categorical variables the effectof the predictor is estimated as compared abaseline, or referent, group within the category.In measuring the relationship between race andpersistence, for example, the referent group iswhite students.

Models of Persistence and DegreeAttainmentAttainment or Persistence into Second Year(1996-1997)

Table 4.2 presents the results of the models ofpersistence into the 1996-1997 academic year.Students were counted as a positive outcome ifthey enrolled at any time during this year, orattained some type of credential by the end ofthis year. Model 1 includes only the student’sdemographic variables; models 2 through 5 addacademic variables, the institutional sector, netcost of attendance, and the financial aidinformation, respectively.36 Included at thebottom of the table for each model are: thepercentage of students in the model with apositive outcome; the number of observations;the number of observations weighted up tonational levels; a pseudo-R2 statistic, which is a

measure of the proportion of total variance inthe outcome explained by the model; and thepercentage of cases that the model properlyclassifies as a positive or negative outcome (i.e.,properly predicts a student as a persister ornonpersister using a cutoff of 0.5).

For example, Model 1 indicates that for everyyear older a student was, his or her probabilityof being enrolled in 1996-1997 declined by 1.2percentage points, controlling for the otherdemographic variables in the model.37 Studentsin the highest income quartile had a probabilityof persisting that was 8.1 percentage pointsgreater than students in the lowest quartile,again, controlling for the other demographicfactors. Eighty-seven percent of the 3,234students (in public institutions) in the surveypersisted into their second year, representing845,802 students nationally. This first modelexplains only 4.3 percent of the variance in theoutcome (pseudo-R2 statistic), but properlyclassifies over 90 percent of the cases38

Model 2 adds students’ academic characteristicsto the demographic variables included in Model1. With the addition of the academic measures,age and income remain as significant predictorsof persistence among the demographicvariables. Students who when interviewedduring their first year in college reported thatthey expected to ultimately attain at least abachelor’s degree were 7.6 percentage pointsmore likely to persist than students who did notexpect to attain that level of education. Higherability students were also more likely to persist;every one point of GPA was related to a 6.0percentage point increase in the probability ofpersistence, and students with higher meritmeasures in high school also were more likelyto persist. Stopping out of college, however,

20

Table 4.2 Models of Persistence/Attainment in 1996-1997

Variable Model 1 Model 2 Model 3 Model 4 Model 5Demographic

Age as of 12/31/95 -0.012 -0.011 -0.012 -0.011 -0.010Income – highest quartile(compared to lowest quartile) 0.081 0.072 0.073 0.069 0.087Race – Asian/Pacific Islander(compared to white) 0.099

AcademicExpected degree – BA orgreater (compared to lessthan a BA) 0.076 0.076 0.076 0.076GPA in 1995-96 (0 - 4 scale) 0.060 0.059 0.059 0.049HS merit index – middle tohigh merit (compared tolow merit) 0.084 0.080 0.081Number of stopouts through96-97 – 1 or more (comparedto none) -0.123 -0.118 -0.116 -0.099

Tuition costNet cost of attendance 95-96after aid ($00s) 0.001

Financial aidNumber of years receivingany aid – 1 or more(compared to none) 0.089Institutional need-basedgrants in 95-96 ($00s) 0.005State non-need grants in95-96 ($00s) 0.010

Percentage persisting orattaining 87.0% 87.0% 87.0% 87.0% 87.0%Number of observations inmodel 3,234 3,234 3,234 3,234 3,234Weighted observations(nationally) 845,802 845,802 845,802 845,802 845,802Pseudo-R2 0.043 0.194 0.196 0.196 0.239Percentage of cases properlyclassified 91.0% 90.7% 90.8% 90.7% 91.4%

21

was strongly related to a failure to make it intothe second year. Students who stopped out atleast once were 12.2 percentage points lesslikely to persist into 1996-1997 than werestudents who had no stopout spells.39 Theaddition of the academic factors greatlyimproved the ability of the model to predictpersistence; the demographic and academicfactors jointly explained 19 percent of thevariance in students’ persistence.40

The addition of neither the institutional sectornor net cost of attendance were significantpredictors in Models 3 and 4. Model 5, whichadded information about financial aid, didpresent some new predictors of persistence andimproved the overall fit of the model slightly. Ifa student received any type of aid in either orboth of her first two years of college, she was8.9 percentage points more likely to persist(compared to students receiving no aid in bothyears). Students receiving institutional need-based grants were one-half percentage pointmore likely to persist for every $100 in grantaid (that is, a student receiving a $1,000 grantwould be 5 percentage points more likely topersist than one receiving no institutional needgrant). Students receiving state-sponsorednon-need grants were 1 percentage point morelikely to persist (for every $100 in grant aid).

While the addition of the financial aidinformation helped improve the models, someof the academic and demographic factors stillremained as important predictors of persistenceinto the second year. Degree expectations andacademic performance in college still remainedstrong predictors of persistence, even aftercontrolling for financial aid awards. However,when financial aid was controlled for, gradepoint average in the first year was not quite asstrong a predictor of persistence, and stopping

out did not have as deleterious an effect onpersistence.

Attainment or Persistence into Sixth Year (2000-2001)

The next set of models (Table 4.3) predictpersistence into the 2000-2001 academic year, orattainment of a credential (certificate, associate’sdegree, or bachelor’s degree) by the spring of2001. The models used to predict this outcomeinclude some measures of students’ academic andother experiences from the base year (1995-1996)through 2000-2001. This additional informationadds to the explanatory power of the models; thefully-specified model (number 5) explains 37percent of the variation in persistence patterns,compared to only 24 percent of the variation in themodels of persistence into the second year.

As with the models of persistence into the secondyear, academic factors jointly considered were themost important predictors of persistence into2000-2001, even academic factors earlier in astudent’s college career. In the fully-specifiedmodel, students who attended college exclusivelypart-time through 1998 were 19.6 percentagepoints less likely to persist into 2000-2001 thanwere students who were full-time through 1998.Similarly, students whose attendance intensity wasa mix of both part-time and full-time were 10.2percentage points less likely to persist. Gradeswere also important predictors of persistence;students who achieved grades higher than C’s(average through 2000-2001) were at least 20percentage points more likely to persist. The onlydemographic factor that was found to be related topersistence through the fully-specified model wasrace; black students were 26 percentage pointsless likely to persist into 2000-2001 than whites,even controlling for the other factors in the model.

22

23

Table 4.3 Models of Persistence/Attainment in 2000-2001

Variable Model 1 Model 2 Model 3 Model 4 Model 5Demographic

Race – black (comparedto white) -0.169 -0.190 -0.201 -0.192 -0.260Race – Asian/Pacific Islander(compared to white) 0.151 0.131

AcademicAttendance intensity through97-98 – always part-time(compared to always full-time) -0.354 -0.327 -0.309 -0.196Attendance intensity through97-98 – mixed PT/FT (comparedto always full-time) -0.132 -0.113 -0.114 -0.103Cum. GPA through 00-01 –mostly B’s or B’s and C’s(compared to mostly C’s and below) 0.215 0.215 0.215 0.206Cum. GPA through 00-01 – mostlyA’s or A’s and B’s (compared tomostly C’s and below) 0.229 0.229 0.229 0.225

Tuition costNet cost of attendance 95-96 afteraid ($00s) 0.002

Financial aidNumber of years receivingwork-study – 1 or 2(compared to none) 0.125Number of years receivingwork-study – 3 or more(compared to none) 0.160Number of years receivingany aid – 3 or more(compared to none) 0.179Institutional need-basedgrants in 95-96 ($00s) 0.005Institutional non-need grantsin 95-96 ($00s) 0.003

Percentage persisting or attaining 76.3% 76.3% 76.3% 76.3% 76.3%Number of observations inmodel 3,382 3,382 3,382 3,382 3,382Weighted observations(nationally) 1,129,929 1,129,929 1,129,929 1,129,929 1,129,929Pseudo-R2 0.026 0.317 0.319 0.319 0.370Percentage of cases properlyclassified 80.8% 87.6% 87.5% 87.5% 87.9%

points less likely to receive a degree. Theacademic factors are very strongly related todegree attainment. Students who attendedcollege their first three years other than full-time were 16.7 percentage points less likely toattain a degree within six years. Stopouts werealso an important predictor of whether astudent received a bachelor’s degree; studentswho had at least one stopout were 36.6percentage points less likely to attain a degreethan students who did not stopout.Grades clearly play a strong role in predictingattainment, with students earning grades aboveC (again, average through 2000-2001) beingover 50 percentage points more likely topersist.16 Another interesting finding is that thestudent’s high school merit index (constructedfor the BPS survey) – which was not a predictorof persistence into the second or sixth years –was a predictor of ultimate degree attainment.The higher was this index (which is in a rangeof 1 to 4), the greater was the probability thestudent received a bachelor’s degree. Inaddition, students who began their collegecareers in a 4-year institution were 25percentage points more likely to ultimatelyattain a bachelor’s degree than were studentswho began in a community college.17

Table 4.4 Models of bachelor’s degreeattainment through 2000-2001Variable Model 1 Model 2Model 3 Model 4 Model 5DemographicAge as of 12/31/95 “0.033

Income – highest quartile (compared to lowestquartile) 0.160Financial aid did play a role in predictingstudents’ bachelor’s degree attainment.Students who received one or two years ofwork-study were 13.9 percentage points more

likely to attain a degree than students whoreceived no work-study, while students withthree or more years of work-study were 22.4percentage points more likely to receive abachelor’s degree. Students who received anytype of aid from any source for three or moreyears were 20.4 percentage points more likelyto attain a degree.18 Students who received oneor two years of Pell Grants were 10.4percentage points less likely to receive a degreethan students who received no Pell Grants. Thisresult, which at first glance appearscontradictory to much of the research reviewedin section 2, may be due to the fact that manystudents who received a Pell Grant for only oneor two years may have been enrolled in sub-baccalaureate programs.Students receiving institutional non-needgrants in their first year were also more likely toattain a degree, with an increase in theirpredicted rate of 0.3 percentage points forevery $100 in grants. Students receiving need-based state aid saw a similar boost in theirpredicted degree attainment rate, with anincreased probability of 0.6 percentage pointsfor every $100 in state grants.19

Summary of Changes in the Awarding ofInstitutional and State Grants, 1995-1996 to1999-2000One limitation of the multivariate analyses inthis section is that they use data from a cohortof students who began college almost eightyears ago.20 Much has changed in the financingof higher education in the ensuing period. Forexample, tuition prices at public institutionsrose substantially from 1995-1996 to 2002-2003; the average cost of attendance at a public4-year increased 39 percent and at communitycolleges it increased 37 percent (College Board,2002a). In addition, the provision of state grantaid to undergraduates has increased at even a

Table 4.4 Models of Bachelor’s Degree Attainment through 2000-2001

Variable Model 1 Model 2 Model 3 Model 4 Model 5Demographic

Age as of 12/31/95 -.033Income – highest quartile(compared to lowest quartile) 0.160Race – black (compared to white) -0.129 -0.109 -0.108 -0.135Race – Hispanic (compared towhite) -0.115Race – Asian/Pacific Islander(compared to white) 0.209 0.192

AcademicAttendance intensity through97-98 – other than full-time(compared to always full-time) -0.238 -0.179 -0.179 -0.167Number of stopouts through 00-01– 1 or more (compared to none) -0.370 -0.370 -0.369 -0.365Cum. GPA through 00-01 –mostly B’s or B’s and C’s(compared to mostly C’s and below) 0.548 0.544 0.544 0.529Cum. GPA through 00-01 –mostly A’s or A’s and B’s(compared to mostly C’s and below) 0.561 0.562 0.561 0.551HS merit index – middle merit(compared to low merit) 0.184 0.141 0.140 0.121HS merit index – middle to highmerit (compared to low merit) 0.283 0.209 0.209 0.168HS merit index – high merit(compared to low merit) 0.297 0.215 0.215 0.189

Institutional sectorStarted in public 4-year(compared to community college) 0.297 0.295 0.251

Tuition costNet cost of attendance 95-96after aid ($00s) 0.001

Financial aidNumber of years receivingwork-study – 1 or 2 (comparedto none) 0.139Number of years receivingwork-study – 3 or more(compared to none) 0.224Number of years receivingany aid – 3 or more(compared to none) 0.204Number of years receivingPell Grants – 1 or 2(compared to none) “0.104Institutional non-need grantsin 95-96 ($00s) 0.003State need-based grants in95-96 ($00s) 0.006

Percentage attaining bachelor’s degree 41.5% 41.5% 41.5% 41.5% 41.5%Number of observations in model 3,382 3,382 3,382 3,382 3,382Weighted observations (nationally) 1,129,929 1,129,929 1,129,929 1,129,929 1,129,929Pseudo-R2 0.049 0.416 0.444 0.444 0.464Percentage of cases properly classified 54.3% 81.3% 81.9% 82.0% 82.1%

24

Financial aid plays an important role inpredicting persistence into later years, also.Students who received one or two years ofcollege work-study support were 12.5percentage points more likely to persist(controlling for all other factors in the model)than students who received no work-study.Similarly, students who received even morework-study were 16 percentage points morelikely to persist than non-work-studyrecipients.41 Students who received any form ofaid (from any source) in three or more yearswere 17.9 percentage points more likely topersist than students receiving no aid in theircollege careers. Students receiving institutionalgrants in their first year in college were morelikely to persist through to 2000-2001.Recipients saw a 0.5 percentage point increasein their likelihood of persisting for every $100in need-based grants and a 0.3 percentagepoint increase for every $100 in non-needgrants.

Attainment of a Bachelor’s Degree by 2000-2001

The third outcome examined is the attainmentof a bachelor’s degree by June 2001 (Table 4.4).Consistent with the other outcomes modeled,academic factors as a group are the mostimportant determinants of whether a studentwho started his postsecondary career in apublic institution ultimately attains a bachelor’sdegree. The fully specified model, containing allfive categories of predictors, explains 46percent of the variation in degree attainment.

Overall, almost 42 percent of all beginningstudents in 1995-1996 ultimately attained abachelor’s degree by 2001. In the fully-

specified model (Model 5), the onlydemographic factor related to degreeattainment was race; black students, controllingfor all other factors, were 13.5 percentagepoints less likely to receive a degree. Theacademic factors are very strongly related todegree attainment. Students who attendedcollege their first three years other than full-time were 16.7 percentage points less likely toattain a degree within six years. Stopouts werealso an important predictor of whether astudent received a bachelor’s degree; studentswho had at least one stopout were 36.6percentage points less likely to attain a degreethan students who did not stopout.

Grades clearly play a strong role in predictingattainment, with students earning grades aboveC (again, average through 2000-2001) beingover 50 percentage points more likely topersist.42 Another interesting finding is that thestudent’s high school merit index (constructedfor the BPS survey) – which was not a predictorof persistence into the second or sixth years –was a predictor of ultimate degree attainment.The higher was this index (which is in a rangeof 1 to 4), the greater was the probability thestudent received a bachelor’s degree. Inaddition, students who began their collegecareers in a four-year institution were 25percentage points more likely to ultimatelyattain a bachelor’s degree than were studentswho began in a community college.43

Financial aid did play a role in predictingstudents’ bachelor’s degree attainment.Students who received one or two years ofwork-study were 13.9 percentage points morelikely to attain a degree than students whoreceived no work-study, while students withthree or more years of work-study were 22.4

25

percentage points more likely to receive abachelor’s degree. Students who received anytype of aid from any source for three or moreyears were 20.4 percentage points more likelyto attain a degree.44 Students who received oneor two years of Pell Grants were 10.4

percentage points less likely to receive a degreethan students who received no Pell Grants. Thisresult, which at first glance appearscontradictory to much of the research reviewedin Section 2, may be due to the fact that manystudents who received a Pell Grant for only one

Table 4.5 Institutional and State Grant Awards by Sector, 1995-1996 to1999-2000

1995-1996 1999-2000 % Change% receiving Mean % receiving Mean % receiving Mean

All publicAll inst. grants 9.8% $1,496 11.7% $1,791 19.2% 19.7%Inst. need grants 7.0 1,201 6.5 1,213 (7.7) 1.0Inst. non-need grants 3.1 2,020 5.7 2,296 85.2 13.7All state grants 8.8 1,310 12.6 1,390 43.6 6.1State need grants 8.2 1,262 9.5 1,346 15.9 6.7State non-need grants 0.8 1,571 3.4 1,413 346.7 (10.1)

Community collegesAll inst. grants 7.2 581 8.0 757 11.2 30.3Inst. need grants 5.6 526 5.3 422 (4.7) (19.8)Inst. non-need grants 1.6 777 2.3 879 48.4 13.1All state grants 5.5 779 9.7 958 75.6 23.0State need grants 5.4 762 6.8 931 25.4 22.2State non-need grants * * 3.1 970 “ “

4-year non-doctoral grantingAll inst. grants 11.3 1,608 13.9 1,783 23.3 10.9Inst. need grants 7.4 1,259 5.8 1,064 (22.2) (15.5)Inst. non-need grants 4.3 2,039 8.4 2,137 95.6 4.8All state grants 14.7 1,384 17.7 1,538 20.4 11.1State need grants 14.0 1,375 14.8 1,472 5.8 7.1State non-need grants 0.9 1,250 3.2 1,716 250.5 37.3

4-year doctoral grantingAll inst. grants 15.1 2,481 19.2 2,820 27.3 13.7Inst. need grants 10.3 2,054 9.4 2,185 (8.2) 6.4Inst. non-need grants 5.8 2,823 10.6 3,023 84.0 7.1All state grants 12.4 1,819 15.4 1,936 23.6 6.4State need grants 10.8 1,759 11.0 1,785 1.5 1.5State non-need grants 2.0 1,807 4.8 2,093 145.9 15.8

* Too few observations to estimate.

26

or two years may have been enrolled in sub-baccalaureate programs.

Students receiving institutional non-needgrants in their first year were also more likely toattain a degree, with an increase in theirpredicted rate of 0.3 percentage points forevery $100 in grants. Students receiving need-based state aid saw a similar boost in theirpredicted degree attainment rate, with anincreased probability of 0.6 percentage pointsfor every $100 in state grants.45

Summary of Changes in theAwarding of Institutional and StateGrants, 1995-1996 to 1999-2000

One limitation of the multivariate analyses inthis section is that they use data from a cohortof students who began college almost eightyears ago.46 Much has changed in the financingof higher education in the ensuing period. Forexample, tuition prices at public institutionsrose substantially from 1995-1996 to 2002-2003; the average cost of attendance at a publicfour-year increased 39 percent and atcommunity colleges it increased 37 percent.47

In addition, the provision of state grant aid toundergraduates has increased at even a fasterrate, 76 percent from 1995-1996 to2001-2002.48

While the most recent NPSAS data are stillapproximately four years old, they are still morerecent than the 1995-1996 data, and thus, canhelp shed light on how the use of financial aidhas changed since the BPS cohort examined inthis section. Table 4.5 shows the percentage ofstudents receiving both institutional and stategrants, and the average grant amount (for

students receiving grants) from the NPSAS:1996and NPSAS:2000 surveys.49 Also shown is thechange over the two periods in the proportionof students receiving grants and the averagegrant amount. Among the notable changes wasthe increase in the proportion of studentsreceiving both institutional and state non-needgrants. Across all public institutions, and withineach sector, these categories of grants saw thefastest growth in their use.

Overall, institutional grant spending forundergraduate students in the public sectorincreased 41 percent, from almost $1.8 billionin 1995-1996 to $2.5 billion in 1999-2000.50

State grant spending in public institutionsincreased 51 percent to $2.1 billion during thissame period.51 In the earlier year, 42 percent ofinstitutional grant dollars and 10 percent ofstate grant dollars were awarded withoutconsideration of financial need. By 1999-2000,these proportions increased to 62 percent and27 percent, respectively.

During this period, tuition prices increased 20percent at four-year public institutions and 15percent in community colleges.52 The increasein both institutional and state average grantamount (need and non-need combined) did notkeep pace with the tuition increase in four-yearinstitutions, but did in community colleges.Tuition prices have again begun to rise morerapidly under the current recession. In thethree years from 1999-2000 to 2002-2003,tuition prices in both public college sectorsincreased another 19 percent.

27

x Conclusion

Summary of Findings

Institutional grant aid in the public sector usedto be a little-known phenomenon, restrictedlargely to use in funding graduate stipends andfellowships. In recent years, however publiccolleges and universities have begun to makemore use of institutional grant awards toundergraduate students.

Data presented in Sections 3 and 4 of thisreport indicate that while the number ofinstitutional grant awards and average awardsize increased between the 1995-1996 and1999-2000 academic years, the size of theawards did not keep pace with the increase intuition costs at four-year institutions but did atcommunity colleges. In addition, institutionalaid in the latter year was much more likely to beawarded without consideration of financialneed; while in 1995-1996 42 percent ofinstitutional grant dollars provided toundergraduates in public institutions wasawarded without means-testing, four years lateralmost two-thirds of all dollars were awarded inthis fashion. This trend echoes that of thegrowth of state-sponsored merit scholarshipprograms. Combining state and institutionalawards, 28 percent of the total dollars wereawarded without consideration of financial needin 1995-1996, a proportion growing to 46percent in 1999-2000.

The analyses of the distribution of institutionalgrant awards in Section 3 indicate some distinctpatterns. Full-time, dependent, and traditional

college-age students (under 24) were morelikely to receive awards, and the awardsreceived were larger than part-time,independent, and older students. While lower-income students were more likely to receive aninstitutional grant, the overall size of the grantreceived was larger for middle- and higher-income students.

In examining institutional grant awards instates with relatively larger spending on state-funded aid compared to their lower-spendingpeers, different patterns were found. Studentsin the low-aid states were more likely to receivean institutional grant, and institutions in thesestates made much greater use of non-needgrants than did those in the high-aid states.The average non-need grant tended to belarger than need-based awards.

The multivariate models in Section 4 utilizedthe Beginning Postsecondary Students Survey toexamine how a variety of variables – includingmany of those found in earlier research to berelated to persistence – were related to thepersistence and degree attainment of studentswho began their college careers in publicinstitutions in the 1995-1996 academic year.The analyses confirmed the prior research thatacademic factors are the strongest predictors ofwhether a student successfully navigates herway through college to the ultimate attainmentof a degree or certificate.

Institutional grants, however, were still found tobe related to persistence and attainment even

28

after controlling for the demographic,academic, institutional, and college cost factorsincluded in the models. Students who receivedan institutional need-based grant of $1,200 intheir first year of college (the average grantawarded, as indicated in Table 4.5) were 6percentage points more likely to persist intotheir second year than were students who didnot receive an institutional need-based award.

Awarding of institutional aid early in a student’scareer was an important predictor of much laterpersistence or attainment of a credential(certificate, associate’s degree, or bachelor’sdegree). Students who received a $1,200 need-based grant in their first year in college were 6percentage points more likely to persist thannonrecipients; those who received the averagenon-need award ($2,000) were also 6percentage points more likely to persist until orattain some type of credential by the 2000-2001 year.53 Institutional non-need grants hadthe same relationship with attainment of abachelor’s degree by that year. The finding thatwork-study awards were also related topersistence into and attainment by the sixthyear also echoes that of other researchers. Stateaid was also found to be related to persistenceinto the second year (non-need grants) andbachelor’s degree attainment (need-basedgrants).

The findings of this study provide valuableevidence to inform the policy debate regardingwhether financial issues or academics are whatlimits college access and persistence,particularly for certain groups of students.54

This study demonstrates that both are relatedto persistence.

Policy Implications for States andPublic Higher Education Institutions

The analyses in this report demonstrate theimportant role that both institutional and stateaid can play in promoting persistence anddegree attainment. Even controlling for otherfactors influential on these outcomes, grantsfrom institutions and the state – aid awards thatare under the direct control of state and/orinstitutional policymakers – are predictors ofpostsecondary success.

There are also important limitations to the useof the BPS survey for modeling studentpersistence. For example, there is very littleinformation in BPS to gauge the degree to whichstudents are engaged in their campus andcurricular studies, a characteristic that has beenfound to be important in promoting studentretention.55 Similarly, there is no information inBPS regarding students’ participation infederally funded pre-college preparationprograms, such as Talent Search and UpwardBound, or in-college assistance programs, suchas Student Support Services or McNair Scholars.Also, as mentioned earlier, detailed financial aidinformation is available only for students’ firstyear in college.

While the evidence here indicates thatinstitutional and state grants in general help topromote persistence, it is difficult to ascribeprecise meaning to the differing effects ofneed-based versus non-need grants. Thesemodels are relatively parsimonious, and furtherinvestigation would have to be conducted todetermine if in fact either type of grant is truly a“better” vehicle for helping students pay forcollege, or whether the findings in this studyare at least in part due to these grants acting as

29

proxies for some unmeasured characteristics ofstudents or their institutions that are alsorelated to persistence. It is important forpolicymakers to note that the research on theimpact of financial aid on college participation(summarized in section 2) has found that grantsare most effective when targeted at studentswho need the financial assistance in order toattend and persist in college. Further studiescould utilize both national data, as from the BPSsurvey, as well as data obtained from individualstate or institutional databases. Models ofpersistence and attainment are best used whenthey contain detailed financial aid informationfor students in every year in college.56

As described earlier, institutional grants havebecome an important part of the fabric of afinancial aid system in this country that hasbeen described by one recent observer as “ahodgepodge of programs involving a number ofparticipants with diverse interests.”57 In the1999-2000 year, public higher educationinstitutions in this country awarded $2.5 billionin institutional grants to undergraduatestudents. The states, through their grantprograms, awarded another $2.1 billion toundergraduates in public institutions. Thesetwo sources combined provided more than the$4.5 billion awarded in Pell Grants, the primaryfederal financial aid program.

The states and their higher educationinstitutions have a large amount of resourcesavailable to help offset the costs of attendingcollege for their students (and supplement theassistance available from the federalgovernment and private sources). How thesetwo sources of aid are coordinated – or moreappropriately, whether they are coordinated – isgoing to vary from state to state, depending

largely upon the higher education governancestructure in each. States with more centralizedcontrol over public higher education institutionsor systems have more opportunities to ensurethat state and institutional financial aidprograms work in tandem to accomplish thestate’s goals regarding higher education access,persistence, and degree attainment.

Based on the research conducted on therelationship between college costs and financialaid, and college access and persistence(summarized in Section 2), the analysis of thechanges in institutional grant awards from1995-1996 to 1999-2000 and the patterns ofawards in the latter year may pose a troublingtrend for disadvantaged students. Thesestudents are more likely to be enrolled parttime, attend community colleges, and havehigher financial need.58 They are also less likelyto qualify for merit grants than their higher-income peers.59

Questions and Discussion Items forState and Institutional Leaders

Whether in a state with a strong, centralizedhigher education governance structure or astate with a more decentralized configuration,there are a number of steps states should gothrough to determine how best to use thelimited resources available that can be focusedon promoting the persistence and degreeattainment of public college and universitystudents. These discussions should be engagedin by a broad array of constituents who haveresponsibility for establishing the goals ofpublic higher education in the state, as well asfor carrying out the implementation of thosegoals. Legislators, executive branch education

30

advisors, higher education governing orcoordinating boards, system heads, campusleaders, leaders of the business sector, andcommunity organizations – all can play animportant role in helping establish objectivesand devising programs and strategies foraccomplishing them.

The following is a list of questions and issues tohelp state and institutional policymakers begina dialogue on how state resources can best beused to promote the persistence and degreeattainment of postsecondary students.

x What are the state’s overall educationalattainment goals? Does the state need morepeople with shorter-duration credentials,such as certificates or vocational training?Or does the state need more people withbachelor’s degrees? In what fields are theseskills needed?

The answers to these questions may helppolicymakers decide whether to emphasizeaccess and persistence in communitycolleges or four-year institutions. Differentfinancial aid and other retention strategiesmay be best used to promote one goal overthe other. For example, the evidence hereindicates that students who begin theirpostsecondary careers in a four-yearinstitution (and stay there) are more likely toearn a bachelor’s degree.

x How will the state ensure degree andcertificate holders will ultimately stay in thestate and contribute to the economy?

The issue of state “brain drain” of the mostacademically talented students has risen in

visibility on the economic andpostsecondary agendas in many states inrecent years.60 Many states have respondedto this concern in their financial aid policiesby implementing merit scholarships,designed to encourage the most ablestudents to attend college in the state. Yetthere is scant evidence that these studentsstay in the state and contribute to theeconomy after college graduation. Statepolicymakers should examine whether otherstate policies – such as loan forgivenessprograms or state income tax credits – maybe more effective at encouraging holders ofvalued job skills to stay in the state.

x To what degree does the state have ahistory of providing a significant level ofcentralized (state-run) grants toundergraduate students? If there is little orno history, is there political will to fund anew program or expand existing ones?

State-funded grants are distinct frominstitutional grants in that they are generallyawarded in the form of a voucher; that is,they are portable and students can use themat any institution in the state.61 Institutionalgrants, on the other hand, are not portable;they can generally be used only at theinstitution making the award, or within thesystem offering the grant if it is a multi-campus system.

x To what degree are campus leaders willingto use institutionally funded grants (eitherfrom restricted funds or from recycledtuition revenue) to help accomplish broaderstate goals? Or are institutional grants usedexclusively for more narrow, enrollment-management objectives?

31

Many public institutions, particularly in thefour-year sector, operate in a competitiveadmissions environment. Thus, withoutintervention from outside sources, they arelikely to use their own aid to promoteinstitutional objectives. This is an entirelyrational approach for campus leaders,particularly in those states where campusheads report to their own board of trusteesor regents, rather than to a broader,statewide governing or coordinating board.In the public four-year sector (combiningdoctoral and nondoctoral grantinginstitutions), overall spending on need-based grants increased 2 percent from1995-1996 to 1999-2000, while spendingon non-need grants increased 121 percentin the same period (calculated from NPSASdata used to construct Table 4.5). This risein the use of institutional merit aid, relativeto need-based grants, is likely an indicatorof the increased competitiveness in whichmost of these institutions find themselves.Research by others has shown thatinstitutional merit aid is increasingly beingawarded to higher-income students. 62

x How should resources available to helppromote persistence and attainment bedistributed among the already-provenstrategies?

As described earlier in this section, statesand institutions combined awarded $4.7billion in grants to undergraduates in the1999-2000 academic year, an amountexceeding that available in the federal PellGrant program. Countless other amountsare spent on other strategies to promotepersistence and attainment. State andinstitutional leaders should examine all of

these policies as a group in decided how toallocate funding among them. Somestrategies may be best used for certaintypes of students or in certain institutions;others will work better for different groups.Too often funding decisions for groups ofprograms with similar purposes are made inisolation, leading to difficulty in maximizingthe ability of these programs to work inconsort with one another.

32

Donald E. Heller came to Penn Statein 2002 from the University of Michigan, wherehe earned a national reputation for his studiesof higher education finance, tuition pricing,financial aid, and student access. Dr. Hellerearned an Ed.D. in Higher Education from theHarvard Graduate School of Education, andholds an Ed.M. in Administration, Planning, andSocial Policy from Harvard and a B.A. inEconomics and Political Science from TuftsUniversity. Before his academic career, he spenta decade as an information technology managerat the Massachusetts Institute of Technology.Dr. Heller teaches and conducts research onissues relating to higher education economics,public policy, and finance, as well as academicand administrative uses of technology in highereducation. The primary focus of his work is onissues of access and choice in postsecondaryeducation, examining the factors and policiesthat help to determine whether or notindividuals attend college, and what type ofinstitution they attend. He has consulted onhigher education policy issues with universitysystems and policymaking organizations inCalifornia, Colorado, Massachusetts, Michigan,New Hampshire, and Tennessee.

Dr. Heller’s research has been published inscholarly journals including The Journal ofHigher Education, The Review of HigherEducation, Change, and The Journal of StudentFinancial Aid, and his work has been reportedon by national media including The New YorkTimes, The Wall Street Journal, The Los AngelesTimes, The Washington Post, Newsweek, U.S.News & World Report, Business Week, TheChronicle of Higher Education, CNN HeadlineNews, and Marketplace Radio. He is the editorof the books The States and Public HigherEducation Policy: Affordability, Access, andAccountability (Johns Hopkins University Press,2001), and Condition of Access: HigherEducation for Lower Income Students (ACE/Praeger, 2002).

Dr. Heller received the 2002 Promising Scholar/Early Career Achievement Award from theAssociation for the Study of Higher Education, ascholarly society with 1,400 members dedicatedto higher education as a field of study. He wasalso the recipient in 2001 of the Robert P. HuffGolden Quill Award from the NationalAssociation of Student Financial AidAdministrators, for his contributions to theliterature on student financial aid.

33

34

35

Appendix A

x NPSAS SampleDefinition

In defining institutional eligibility for inclusionin the NPSAS:2000 sample, NCES establishedthe following criteria:

To be eligible for NPSAS:2000, aninstitution was required, during the1999–2000 academic year, to:x Offer an educational program

designed for persons who hadcompleted secondary education.

x Offer more than just correspondencecourses.

x Offer at least one academic,occupational, or vocational programof study lasting at least threemonths or 300 clock hours.

x Offer courses that were open tomore than the employees ormembers of the company or group(for example, a union) thatadministered the institution.

x Be located in the 50 states, theDistrict of Columbia, or Puerto Rico.

x Be other than a U.S. ServiceAcademy.

x Have a signed Title IV participationagreement with the U.S. Departmentof Education.63

These criteria made 6,422 postsecondaryinstitutions eligible for inclusion in the survey.From this universe of institutions, 1,082, or16.8 percent of the institutions were selectedfor participation in NPSAS:2000. Usable datawere obtained from 999 institutions.

The institutional sample was stratified into 22groups, based on institutional control, highestdegree level, and the proportion of studentsearning bachelor’s degrees in education (thetop 20 percent of institutions based on thisranking were designated “high education”institutions, and the remaining were designated“low education”). Table A.1 presents the 22institutional strata.

Student eligibility for inclusion in theNPSAS:2000 sample was defined as:

Those who attended a NPSAS-eligibleinstitution during the 1999–2000academic year and who were:x Enrolled in either: an academic

program; at least one course forcredit that could be applied towardfulfilling the requirements for anacademic degree; or an occupationalor vocational program that requiredat least three months or 300 clockhours of instruction to receive adegree, certificate, or other formalaward.

x Not concurrently enrolled in highschool.

x Not enrolled solely in a GED or otherhigh school completion program.64

Table A.1 NPSAS:2000 Institutional Sample Strata

Stratum # Control Level (highest degree) Education1 Public Less than 2-year2 Public 2-year3 Public Baccalaureate High4 Public Baccalaureate Low5 Public Master’s High6 Public Master’s Low7 Public Doctorate High8 Public Doctorate Low9 Public First professional High10 Public First professional Low11 Private, non-profit Less than 2-year12 Private, non-profit 2-year13 Private, non-profit Baccalaureate High14 Private, non-profit Baccalaureate Low15 Private, non-profit Master’s High16 Private, non-profit Master’s Low17 Private, non-profit Doctorate High18 Private, non-profit Doctorate Low19 Private, non-profit First professional High20 Private, non-profit First professional Low21 Private, for-profit Less than 2-year22 Private, for-profit 2-year or more22 Private, for-profit 2-year or more

36

Within institutions, students were grouped intoseven strata based on their degree level,whether they were completing a baccalaureatedegree, and major: undergraduate businessmajors receiving a bachelor’s degree in 1999-2000; other majors receiving a bachelor’sdegree; other (nongraduating) undergraduates;

master’s; doctorate; other graduate degree; andfirst professional.65

Approximately 63,000 students from the 999institutions were included in the NPSAS:2000sample. Of this total, 49,930 wereundergraduate students.

37

x Endnotes1. R. B. Archibald, “Redesigning the Financial AidSystem: Why Colleges and Universities Should SwitchRole with the Federal Government” (Baltimore, MD: TheJohns Hopkins University Press, 2002), 1.2. This section has been copied from the WesternInterstate Commission for Higher Education (2003).3. Parts of this section have been adapted from D. E.Heller, “The Policy Shift in State Financial Aid Programs,”in J. C. Smart (ed.), Higher Education: Handbook ofTheory and Research (New York: Agathon Press, 2002-a), vol. 17, 221-261. Readers interested in more of thehistory of state support for higher education can referto this article.4 . President’s Commission on Higher Education, HigherEducation for American Democracy (New York: Harperand Brothers: 1947).5. President’s Commission, vol. II, 47.6. Carnegie Commission on Higher Education, TheCapitol and the Campus: State Responsibility forPostsecondary Education – A Report andRecommendations (New York: McGraw-Hill, 1971).7. R. H. Fenske and J. D. Boyd, “State Need-BasedCollege Scholarship and Grant Programs: A Study oftheir Development, 1969-1980” (New York: CollegeEntrance Examination Board, 1981).8. National Association of State Scholarship and GrantPrograms (various years), NASSGP/NASSGAP AnnualSurvey Report (Deerfield, IL; Harrisburg, PA; and Albany,NY: Illinois State Scholarship Commission; PennsylvaniaHigher Education Assistance Agency; and New YorkState Higher Education Services Corp.).9. J. D. Boyd, “State/Territory Funded Scholarship/GrantPrograms to Undergraduate Students with FinancialNeed to Attend Public or Private Post-secondaryEducational Institutions: Seventh Annual Survey, 1975-76 Academic Year” (Deerfield: Illinois State ScholarshipCommission, 1975) 2.10. National Association of State Student Grant and AidPrograms, NASSGAP 33rd Annual Survey Report onState-sponsored Student Financial Aid, 2001-2002Academic Year (Albany: New York State HigherEducation Services Corp., 2003).11. Unfortunately, NCES has been slow in releasing thedata, so the most recent time-series data available endin 1996.12. Quantum Research Corp., CASPAR database system,online data file (Bethesda, MD: Author, 2003). Note: theIPEDS data covers both undergraduate and graduatestudents. Data from the National Postsecondary StudentAid Study for that same academic year (1995-1996,described in more detail in Section 4) show a total of$1.8 billion awarded by public institutions to

undergraduates, so the IPEDS data are likely a fairlyaccurate measure of institutional spending.13. Western Interstate Commission for HigherEducation, “What is Changing Direction? A ProjectOverview,” Western Policy Exchanges (May 2003), 1.14. See, for example:l J. M. Braxton, Reworking the Student Departure

Puzzle (Nashville, TN: Vanderbilt University Press,2000)..

l E. T. Pascarella and P. T. Terenzini, How CollegeAffects Students: Findings and Insights fromTwenty Years of Research (San Francisco, CA:Jossey-Bass Publishers, 1991).

l V. Tinto, Leaving College: Rethinking the Causesand Cures of Student Attrition, 2nd ed. (Chicago,IL: University of Chicago Press, 1993).

15. A review of the research on the relationship amongthese other factors and persistence is beyond the scopeof this study.16. See, for example:l E. P. St. John, “Price Response in Persistence

Decisions: An Analysis of the High School andBeyond Sophomore Cohort,” Research in HigherEducation 31, no. 4 (1990). 387-403.

l E. P. St. John, S. Hu, M. Clements, and E. H. Asker,“Keeping Indiana’s Public Colleges Affordable:The Effects of State Grants on Persistence by Full-time Students,” paper presented at the NASSGAP/NCHELP Research Network Conference, Savannah,GA (May 1999).

l E. P. St. John, M. B. Paulsen, and J. B. Starkey, “TheNexus between College Choice and Persistence,”Research in Higher Education 37, no. 2 (1996),175-220.

l E. P. St. John and J. B. Starkey, “An Alternative toNet Price,” Journal of Higher Education 66, no. 2(1995), 156-186.

l E. P. St. John and J. B. Starkey, The Impact ofStudent Aid on Persistence in Washington HigherEducation: Logistic Analyses of the 1991, 1992,and 1993 Cohorts (Olympia, WA: State ofWashington Higher Education CoordinatingBoard, 1996).

l M. B. Paulsen, “Recent Research on the Economicsof Attending College: Returns on Investment andResponsiveness to Price,” Research in HigherEducation 39, no. 4 (1998), 471-489.

l M. B. Paulsen and E. P. St. John, “Social Class andCollege Costs: Examining the Financial Nexusbetween College Choice and Persistence,” Journalof Higher Education 73 no. 2 (2002), 189-236.

l L. L. Leslie and P. T. Brinkman, “Student PriceResponse in Higher Education,” Journal of HigherEducation 58, no. 2 (1987), 181-204.

38

l L. L. Leslie and P. T. Brinkman, The EconomicValue of Higher Education (New York: AmericanCouncil on Education/Macmillan Publishing,1988).

17. See Lee and St. John; and the following:l L. K. Berkner, S. Cuccaro-Alamin, and A. C.

McCormick, Descriptive Summary of 1989-90Beginning Postsecondary Students: 5 Years Laterwith an Essay on Postsecondary Persistence andAttainment (Washington, DC: U.S. Department ofEducation, National Center for EducationStatistics, 1996).

l L. K. Berkner, S. He, and E. F. Cataldi, DescriptiveSummary of 1995-96 Beginning PostsecondaryStudents: Six Years Later, NCES 2003151(Washington, DC: U.S. Department of Education,National Center for Education Statistics, 2002).

l S. P. Choy, Students Whose Parents Did Not Go toCollege: Postsecondary Access, Persistence, andAttainment, NCES 2001-126 (Washington, DC:U.S. Department of Education, National Center forEducation Statistics, 2001).

l S. P. Choy, Access and Persistence: Findings from10 years of Longitudinal Research on Students(Washington, DC: American Council on Education,Center for Policy Analysis, 2002).

l Institute for Higher Education Policy and TheEducation Resources Institute, The Next Step:Student Aid for Student Success (Washington, DC:Authors, 1995).

l J. B. Lee and E. St. John, Student Financial Aid andthe Persistence of Recipients at WashingtonColleges and Universities: Analysis of Results(Olympia, WA: State of Washington HigherEducation Coordinating Board, 1996).

l T. G. Mortenson, The Impact of Financial Aid onCollege Retention for Minority Students(Washington, DC: Council for Opportunity inEducation, 1999).

l United States General Accounting Office,Restructuring Student Aid Could Reduce Low-income Student Dropout Rate, GAO/HEHS-95-48(Washington, DC: Author, 1995).

l C. C. Wei, and L. Horn, Persistence andAttainment of Beginning Students with PellGrants, NCES 2002-169 (Washington, DC: U.S.Department of Education, National Center forEducation Statistics, 2002).

18. Pascarella and Terenzini, 420.19. See Leslie and Brinkman, 1987; also see, forexample:l D. E. Heller, “Student Price Response in Higher

Education: An Update to Leslie and Brinkman,”Journal of Higher Education 68, no. 6 (1997),624-659.

l G. A. Jackson and G. B. Weathersby, “IndividualDemand for Higher Education, Journal of HigherEducation 46, no. 6 (1975), 623-652.

20. College Board, Trends in Student Aid, 2002(Washington, DC: Author, 2002-b).21. S. Burd, “Congress Should Freeze Federal LoanLimits, Report Says,” The Chronicle of Higher Education(2 May 2003), A30.22. See for example the studies of state merit programsin D. E. Heller and P. Marin (eds.), Who Should We Help?The Negative Social Consequences of Merit Scholarships(Cambridge, MA: Harvard Civil Rights Project, 2002).23. J. A. Riccobono, M. B. Cominole, P. H. Siegel, T. J.Gabel, M. W. Link, and L. K. Berkner, 1999-2000National Postsecondary Student Aid Study (NPSAS:2000)Methodology Report, NCES 2002-152 (Washington, DC:U.S. Department of Education, National Center forEducation Statistics, 2002).24. Institutions included in the analyses here includecommunity colleges and four-year institutions, andexcludes less than two-year public institutions (whichenrolled fewer than 1 percent of undergraduatesenrolled in public institutions that year). See AppendixA for more information about institutions and studentseligible to be included in the NPSAS surveys.25. Some students receive both types of awards.“Grants” includes grants, scholarships, fellowships, andtuition waivers. In the NPSAS survey, “need” awards areprovided using solely financial need criteria or criteriathat utilize a combination of financial need and othermeasures. “Non-need” awards include aid that wasprovided without consideration of financial need. SeeNational Center for Education Statistics. NationalPostsecondary Student Aid Study of 1999-2000Electronic Codebook for Windows , computer data file(Washington, DC: U.S. Department of Education, 2000).26. It is important to note that the data presented hereincludes all students enrolled in these institutions,including both full-time and part-time enrollees. Thus,the average awards and percentage of studentsreceiving awards will be less than in a recent reportreleased by NCES. See L. Horn, and K. Peter, WhatColleges Contribute: Institutional Aid to Full-TimeUndergraduates Attending 4-year Colleges andUniversities, NCES 2003-157 (Washington, DC: U.S.Department of Education, National Center for EducationStatistics, 2003). That report analyzed data on full-timestudents at four-year institutions only and becausethese students are more likely to receive institutionalaid, the figures in that report are higher.27. The NPSAS data do not distinguish betweenunrestricted institutional grants (those funded fromtuition revenues) versus restricted grants (those fundedfrom endowed scholarships or designated gifts).28. Dependent college students in public institutionsare, on average, from higher-income families than all

39

Americans. While the median family income of thesestudents in the NPSAS sample was $53,276, the medianincome of all families in the U.S. was $46,737 in 1998.See United States Bureau of the Census, “Regions—Families (All Races) by Median and Mean Income: 1953to 2001,” online data file, (http://www.census.gov/hhes/income/histinc/f06.html) (Washington, DC:Author, 2003).29. In general the only cases in which those under age24 can be considered independent are: orphans orwards of the court; military veteran; married; or had atleast one legal dependent other than a spouse. Allstudents over the age of 24 are considered independentstudents.30. National Association of State Student Grant and AidPrograms, NASSGAP 31st Annual Survey Report 1999-2000 Academic Year (Albany: New York State HigherEducation Services Corp., 2001). The ranking ofspending on need-based aid (as compared to all aidspending for undergraduates) was used because themajority of state aid is awarded based on need.31. The reason why the high-aid states have an overallcost of attendance so close to that of the low-aid states– even though the four-year institutions are so muchmore expensive in the former group – is that there areproportionally more community college students in thehigh-aid states. Since their cost of attendance is lowerthan students attending four-year institutions, thishelps drive down the average cost. Tests comparing themean amounts in Tables 3.6 and 3.7 were conductedusing a Wald test and probability cutoff of pd” .05. Formore information about prices paid by students indifferent college sectors, see College Board, Trends inCollege Pricing, 2002 (Washington, DC: Author, 2002-a); and L. Horn, C. C. Wei, and A. Berker, What StudentsPay for College: Changes in Net Price of CollegeAttendance between 1992-93 and 1999-2000, NCES2002-174 (Washington, DC: U.S.Department of Education, National Center for EducationStatistics, 2002).32. National Center for Education Statistics, “BeginningPostsecondary Students Longitudinal Study RestrictedUse Files, computer data file (Washington, DC: U.S.Department of Education, 2003).33. The BPS dataset contains a number of variablesrelated to cost in the base year (1995-1996), includingthe total cost of attendance, net price (after financialaid), and level of unmet need (cost of attendance lessfamily contribution and all aid). The literature on therelationship of college costs and persistence points tothe importance of looking not just at the sticker price ofcollege, but at the net price and/or unmet need. Thelatter two measures are strongly correlated, and thus,cannot be included in the models due to problems withcollinearity. Because more complete information wasavailable on net cost, this variable was used in themodels to represent costs faced by students. See

Paulsen and St. John; St. John; and St. John, Paulsen, andStarkey.34. Stata Corp., Stata Survey Data Reference Manual,Release 8 (College Station, TX: Stata Press 35, 2003).35. A. F. Cabrera, “Logistic Regression Analysis inHigher Education: An Applied Perspective,” in J. C. Smart(ed.), Higher Education: Handbook of Theory andResearch, vol. X (New York: Agathon Press, 1994), 225-256). Also see T. Petersen, “A comment on presentingresults from logit and probit models,” AmericanSociological Review 50, no. 1 (1985), 130-131.36. The delta-p statistics only for those variables thatwere statistically significant at a level of p < .05 areshown in the tables (that is, the coefficient wassignificantly different from zero).37. For clarity, the outcome discussed in this sectionrefers to “persistence,” but this includes attainment of acredential (degree or certificate through the yearindicated).38. When the mean of a dichotomous outcome is highlyskewed toward either the successful or unsuccessfuloutcome, it is common for logistic regression models tocorrectly classify a high percentage of cases.39. The BPS survey defines a stopout as “a break inenrollment of five or more consecutive months.” SeeNational Center for Education Statistics, 2003.40. The order in which the blocks of variables are addedinto the models does not matter; the large gain inpredictive ability added by the academic factors wouldoccur no matter to which model they were added.41. This finding should be interpreted with somecaution, however. In order to receive work-study (or anyaid) for three or more years, a student has to beenrolled for at least three years. A student enrolled forat least three years has, by definition, cleared theimportant hurdle of persisting into the second year. Astudent who received no work-study (or no aid) mayhave dropped out after one year, and thus, could notreceive a bachelor’s degree.42. It is interesting to note that the biggest boostcomes not from making stellar grades, but from making“good” grades. Compared to students with mostly C’sand below, those with mostly B’s or B’s and C’s have apredicted persistence rate 52.9 percentage pointshigher. Students with mostly A’s or A’s and B’s haveonly a slightly higher increase in persistence.43. Note that the models included the students’ degreeexpectations when they first enrolled in college. So thisfinding is important as it indicates that even controllingfor degree expectations (i.e., for all students whoexpected to ultimately attain a bachelor’s degree) andother factors, students who began in communitycolleges were at a disadvantage.44. See note 41 for a caution in interpreting thisfinding.

40

45. In order to examine whether there were any majorchanges in the results if the sample was restricted onlyto those students who said they aspired to a bachelor’sdegree or greater in their first year in college, Model 5in Tables 4.3 and 4.4 was tested with this restrictedsample. The restricted sample produced resultssubstantively the same as those shown in Tables 4.3and 4.4.46. The next wave of the NPSAS survey in the 2003-2004 academic year will include a new BPS cohort.47. College Board, 2002-a.48. National Association of State Scholarship and GrantPrograms.49. The institutional grant figures for 1999-2000 arethe same as those found in Table 3.1.50. These totals are not shown in the table, but werecalculated by multiplying the average grant award bythe number of students nationally who received awards.51. Overall, state grant awards to undergraduatesincreased 40 percent, indicating that students in publiccolleges received a larger share of state grants(compared to their peers in private institutions) in1999-2000 than four years earlier (see NationalAssociation of State Scholarship and Grant Programs).As a point of reference, the Consumer Price Indexincreased 9.6 percent over these four years. SeeNational Center for Education Statistics, Digest ofEducation Statistics, 2001 (Washington, DC: U.S.Department of Education, 2002).52. College Board, 2002-a.53. While the BPS dataset does not have data oninstitutional aid in students’ second and subsequentyears in college, it is reasonable to assume that astudent receiving either a need-based award or a non-need grant would have a high probability of receivingsuch an award in later years. Thus, the effect describedhere is likely not just that of the award made in the firstyear of college, but is also likely a proxy, or marker, oflater receipt of institutional aid.54. S. Burd, “Rift Grows over What Keeps Low-incomeStudents out of College,” The Chronicle of HigherEducation (25 January 2002), A18.55. Pascarella and Terenzini.56. Ideally, a database used to model persistence anddegree attainment should contain information aboutthe financial aid offers received and college costs facedby those students who do not persist. This informationwould provide even better measures of theeffectiveness of financial aid in promoting persistence.57. Archibald, 1.58. See the following:l Advisory Committee on Student Financial

Assistance, Access Denied: Restoring the Nation’sCommitment to Equal Educational Opportunity(Washington, DC: U.S. Department of Education.

Advisory Committee on Student FinancialAssistance, 2001).

l Advisory Committee on Student FinancialAssistance, Empty Promises: The Myth of CollegeAccess in America (Washington, DC: U.S.Department of Education, 2002).

l P. T. Terenzini, A. F. Cabrera, and E. M. Bernal,Swimming against the Tide: The Poor in AmericanHigher Education (New York: College Board,2001).

59. Heller and Marin.60. See the following:l J. D. Creech, State-Funded Merit-based

Scholarship Programs (http://www.sreb.org/Main/LatestReports/accountbench/scholarship/scholarship.html) (Atlanta, GA: Southern RegionalEducation Board, 1999).

l D. E. Heller, “The Policy Shift in State Financial AidPrograms,” in J. C. Smart (ed.), Higher Education:Handbook of Theory and Research, vol. 17 (NewYork: Agathon Press, 2002-a), 221-261.

l D. E. Heller, “State Merit Scholarship Programs:An Introduction,” in D. E. Heller and P. Marin(eds.), Who Should We Help? The Negative SocialConsequences of Merit Scholarships (Cambridge,MA: Harvard Civil Rights Project, 2002-b), 15-23).

l L. G. Tornatzky, D. O. Gray, S. A. Tarant, and C.Zimmer, Who Will Stay and Who Will Leave?(Research Triangle Park, NC: SouthernTechnology Council, 2001).

61. Many state programs provide some restrictions onhow students can use the grants. Some require thestudents to use them only at in-state institutions;others distinguish between public and privateinstitutions within the state (or offer differing awardlevels for attending each sector). But in all, there issome degree of portability to the grants.62. M. S. McPherson and M. O. Schapiro, The StudentAid Game: Meeting Need and Rewarding Talent inAmerican Higher Education (Princeton, NJ: PrincetonUniversity Press, 1997).63. Riccobono et al., 12.64. Riccobono et al., 13.65. Not all student strata existed within eachinstitutional stratum; for example, the public two-yearsector would contain individuals only in studentstratum 3.

PO Box 9752Boulder, CO 80301-9752303.541.0200 www.wiche.edu