Paths to Persistence - ERIC · Paths to Persistence: An Analysis of Research on Program Effectiveness at Community Colleges NEW AGENDA SERIES™ Thomas R. Bailey and Mariana Alfonso

Paths toPersistence:

An Analysis ofResearch on

ProgramEffectiveness at

CommunityColleges

NEW AGENDA SERIES™

Thomas R. Bailey and Mariana Alfonso

Community College Research Center, Teachers College, Columbia UniversityV o l u m e 6 • N u m b e r 1 • J a n u a r y 2 0 0 5

AcknowledgmentsThis report was funded by Lumina Foundation for Education as part of a national partnership initiativecalled Achieving the Dream: Community Colleges Count. The Community College Research Center wasfounded as a result of a grant from the Alfred P. Sloan Foundation, which continues to support its work.The authors wish to thank Thomas Brock, Linda Hagedorn, Katherine Hughes, Mort Inger, Melinda Karp,Richard Kazis, Lauren Koch, John Lee, Carol Lincoln, Vanessa Morest, Lisa Rothman, Sarah Rubin, WendySchwartz, Kate Shaw, Karen Stout, Nikki Thompson, Richard Voorhees and Pat Windham for their helpand advice.

Executive summary1

Introduction5

The state of research on persistence and completion7

Conceptual perspectives on retention and completion11

Empirical studies of the effect of institutional management and practices 15

Conclusions and implications for practice 23

References31

Endnotes33

About the authors35

Table of contents○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Executive summary○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

PATHS TO PERSISTENCE 1

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Research about and at

community colleges must

play a central role in

any strategy to increase

student success.

C ommunity colleges are designed tobe open-door institutions, and theyenroll a much wider variety ofstudents than baccalaureate-granting colleges. Community

colleges have always played a crucial role inproviding access to college. During the lastdecade, however, educators and policy-makershave shifted their attentionalso to the success ofstudents once they entercommunity college. As aresult, accreditationagencies and state regulatorsare increasingly scrutinizingmeasures of studentoutcomes such as persis-tence and completion rates.

At the same time,national initiatives byfoundations and the U.S. Department of Educationare focused on developing policy and institutionalpractices that will improve success rates forcommunity college students. This report has beenwritten as part of one of those initiatives. In 2003,Lumina Foundation for Education joined eight

other organizations to launch Achieving theDream: Community Colleges Count. Achievingthe Dream is based on the premise that researchabout and at community colleges must play acentral role in any strategy to increase studentsuccess.

This report presents a critical analysis of thestate of the research on the effectiveness of four

types of practices inincreasing persistence andcompletion at communitycolleges: 1) advising,counseling, mentoring andorientation programs;2) learning communities;3) developmental educa-tion and other services foracademically under-prepared students; and4) college-wide reform. We

use this analysis to draw substantive lessons abouteffective institutional practices, to identifypromising areas for future research, to evaluate thestate of program-effectiveness research atcommunity colleges, and to make recommenda-tions for improving related research.

2 PATHS TO PERSISTENCE

Substantive lessonsAmong the practices and strategies that we

examined, learning communities appear to havethe most support grounded in research. In alearning community, students go through theprogram as a cohort, and their instruction istypically organized around themes. The learningcommunity model’s positive effects on persistenceand graduation are consistent with the mostinfluential theoreticalperspectives used to studyretention. Empiricalresearch also suggestspositive effects. Researchconclusions point out thatcounseling, advising anddevelopmental educationare all crucial for community college students, butresearch has been less helpful in identifying themost effective design and organization for theseservices. Major national research projects in theseareas are now a decade old, and the policy anddemographic environments have changedsignificantly in the intervening years. Studies ofbroad, college-wide change and the institutional-ization of pilot programs are particularly wanting.We lack good conceptualizations as well asempirical measurement of these types of strategies.

Improving researchResearch on program effectiveness at commu-

nity colleges can be improved by addressingproblems in four areas. First, the large majority ofthe research on program effectiveness in highereducation is limited to studies of four-yearcolleges. Insights obtained from this research donot necessarily translate to effective practices forthe part-time, working and adult population thatcharacterizes community colleges. Second, thenational data sets that allow comprehensiveanalysis of the experience of postsecondarystudents do not include data on the types ofspecific institutional practices and policies thatcolleges use to increase student success. Third,

methodological problems thwart definitiveconclusions about the effectiveness of communitycollege policies and practices. Fourth, thedissemination and discussion of research oncommunity colleges are inadequate. Reports aredifficult to obtain and usually include too littleinformation to allow a judgment about the validityof the conclusions. Faculty and even researchers atcommunity colleges rarely have the opportunity to

discuss research on programeffectiveness in a way thatallows them to understandthe strengths, weaknessesand lessons of the studies.

To strengthen ourability to choose and assesspolicies and practices that

will help increase the success of communitycollege students, we suggest the following changesin community college research:

Theoretical issues: The dominant theoreticalperspective on retention and completion, thestudent integration or engagement model, wasdeveloped based primarily on four-year collegemodels with particular emphasis on full-time,traditional-aged, residential students. Empiricaltests of these models have not yielded strongsupport for their application to communitycolleges. Researchers have begun to take intoaccount commuter students, but the particularcharacteristics of community colleges and theirstudents are still neglected.

Data availability: Because insufficient nationaldata exist on institutional practices, most program-effectiveness research is based on samples fromsingle institutions. While these can be useful, theirconclusions are difficult to generalize becauseeffects may be based on particular features of thecollege being studied. As much as possible,national-level databases, such as those created bythe National Center for Education Statistics,should include programmatic detail to allowresearch on the effectiveness of common practices

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Most program-effectiveness

research is based on samples

from single institutions.

used to improve student success. Collaborativeprojects within community college districts andstates, supplemented with increasingly availablestate- or district-level student record data, alsowould be helpful in understanding persistence andstudent success and in overcoming some of thelimitations of single-campus studies. Even withsingle-institution studies, research that makes useof transcript-based longitudinal data can providemany important insights.

Empirical issues: There are some excellentstudies on program effectiveness at communitycolleges; in general, however, empirical researchneeds to do a better job of exploiting availabledata and of employing quantitative and qualitativemethodologies for effectiveness research. Too fewstudies that measure arelationship betweencommunity collegeprograms and studentoutcomes are designed inways that support conclu-sions about whether thatrelationship is a causal one.The following suggestionscan strengthen the reliabil-ity and validity of researchon these issues.

■ Empirical research must, as much aspossible, control for measured studentbackground characteristics. Scores on entryassessment tests or information on the highschool academic record are particularlyimportant.

■ Under some circumstances, statisticaltechniques can account for unmeasuredcharacteristics, such as motivation, thatmight influence student outcomes. Thesetechniques are not now used in communitycollege research.

■ Random-assignment designs address manyof the most difficult methodologicalproblems; thus, their conclusions areparticularly useful and influential. However,

because such studies are costly anddifficult to administer, they are infrequent.

■ Finally, every study of a program mustinclude a detailed description of thecharacteristics of the program and of theprocess through which students enter thatprogram. This information gives readersessential background that allows them tointerpret the research results and judge thevalidity of the conclusions.

Research at the community college: Althoughuniversity researchers must pay more attention tocommunity colleges, research will have a funda-mental influence on the colleges only when itplays a more prominent role on the campusesthemselves. Reformers refer to this as developing a

“culture of evidence” inwhich institutional researchfunctions play a moreprominent role and facultyand administrators aremore fully engaged withdata and research about thesuccess of their students,using those data to makedecisions. We present sixsuggestions for developingthat culture:

1. Colleges must assess the resources andskills needed for effective institutionalresearch, recognizing that research is aninvestment. As with even the mostrewarding investment, its payoff emergesonly over time.

2. Colleges must recognize that assessmentsof program effectiveness are difficult andinvolve a continuum of activities andanalyses.

3. Projects should combine quantitativeresearch on student outcomes withqualitative research to elicit insights fromstudents about those outcomes.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Research will have a

fundamental influence on

the colleges only when it

plays a more prominent role

on the campuses.


4. Colleges, states and college associationsmust provide more opportunities for facultyand administrators to engage in theresearch process and to discuss evidenceabout student outcomes.

5. Colleges and states must develop moresystematic methods to publicize anddisseminate research findings.

6. Collaboration among institutionalresearchers at different colleges andbetween college-level and state-levelresearchers should be promoted.

This report argues that much needs to be doneto strengthen research on community college

program effectiveness. Too often, researchprovides ambiguous or weak guidance concerningmany policies and programs designed to improvestudent retention and success. Of course, despitethese uncertainties, colleges must continue tomove forward and act on the best availableknowledge — even if that knowledge is limited oropen to alternative interpretations. Our overallrecommendation is that colleges search for thebest information they can find and monitorprogress as thoroughly and rigorously as possible.The interaction between research and practiceinvolves a continuous conversation within andamong the colleges, and with outside researchersand policy-makers, as practitioners try to improvetheir performance in a constantly changingenvironment.



Introduction○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Community college students

have low persistence and

completion rates.

C ommunity colleges are designed tobe open-door institutions. Theyenroll a much wider variety ofstudents than baccalaureate-granting colleges: Minority

students, first-generation college students, studentswith lower levels of academic achievement in highschool, and students from low-income families areall significantly overrepresented in communitycolleges when compared with their enrollment inbaccalaureate-granting institutions. During thedecades of rapid expansionof community colleges,these considerations ofaccess were paramount, butmore recently educators,policy-makers, researchersand foundations have allincreasingly turned theirattention to the actualexperience of students enrolled in these institu-tions. This focus has revealed that communitycollege students have low persistence andcompletion rates. Of all first-time college studentswho entered a community college in 1995, only 36percent earned a certificate, associate’s or

bachelor’s degree within six years. Although manystudents who did not complete degrees may havemet other personal goals, policy-makers andeducators judge these rates to be too low (Bailey &Leinbach, 2005). Moreover, completion rates forAfrican-American, Hispanic, Native American andlow-income students are lower than the overallnumbers, indicating inequitable racial and incomegaps.1

National initiatives by foundations and theU.S. Department of Education are focused on

developing policy andinstitutional practices thatwill improve retention,completion and othermeasures of success forcommunity collegestudents. These initiativesalso aim to reduce theachievement gaps between

students from different racial/ethnic and incomegroups. This report has been written as part of oneof those initiatives. In 2003, Lumina Foundationfor Education joined with eight other organiza-tions to launch Achieving the Dream: CommunityColleges Count.2 In early 2005, Achieving the


Dream is working with 27 community colleges infive states to help them increase retention,completion and success for community collegestudents, particularly those in groups that havebeen underserved in higher education, such as low-income students, students of color and first-generation college students. Research and data areat the heart of all aspects of Achieving the Dream.First, colleges are expected to use institutionallyrelevant data and research to measure the currentlevels of students’ success. Data are to be disaggre-gated by race, ethnicity and income (whenpossible) and to be used to diagnose the causes ofthe problems they identify. Moreover, participatingcolleges are expected to select strategies andinterventions for which there is empirical evidenceof effectiveness. Finally, the initiative is designed tohelp the colleges strengthen their own capacity toconduct, and particularly to use, institutionalresearch on an ongoing basis to identify problemsand choose and assess alternative solutions. Thisbroader objective is articulated in the admonitionto develop a “culture of evidence,” rather thanrelying on a “culture of anecdote” in whichcommunity colleges justify themselves simply bytelling encouraging stories about individuals whoovercome daunting barriers to succeed at commu-nity colleges.

The goal of this report is to provide back-ground information and analysis to support abroad effort to strengthen the use of data andresearch in improving student outcomes atcommunity colleges. It starts by presenting anoverview of the state of research on the effective-ness of institutional programs and policiesdesigned to improve community college studentoutcomes, particularly persistence and completion.The next section discusses the most commontheoretical perspectives used in the study ofretention and completion in higher education andquestions the appropriateness of applying thoseperspectives to community college students. Wethen review research on effectiveness of commu-nity college programs in four broad areas: studentservices and advising, developmental education,learning communities and institution-wide reform.In addition to drawing programmatic lessons fromthis literature, we use these analyses to supportgeneralizations about the state of program-effectiveness research on community colleges. Weend with conclusions and recommendations forresearch about community colleges and for reformsin research done on community college campuses— reforms that can help develop a “culture ofevidence” at the colleges.


The state of research onpersistence and completion

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

A wealth of research exists onpersistence and completion inhigher education. In 1991, ErnestPascarella and Patrick Terenzinipublished an 800-page volume

reviewing almost 3,000 studies on How CollegeAffects Students, including many studies of retentionand completion. Since then, research on the topichas continued in journals and in unpublishedreports. But this vastlandscape of papers andreports yields relatively fewconcrete insights about ourspecific topic: the effects ofinstitutional policies oncommunity college retention and completion.

This dearth of insight results largely from fourproblems. First, most of this research is about four-year colleges. Second, available national (or evenmulti-college) data do not have good measures ofinstitutional practices designed to promoteretention and completion. Third, flawed method-ology often thwarts efforts to properly assess

institutional practices. Fourth, the disseminationand discussion of research reports on communitycolleges are inadequate. We shall examine each ofthe four problems in turn.

1. Overemphasis on the four-year college:The vast majority of the research on studentretention and completion is concerned with four-year colleges. In concluding their definitive review,

Pascarella and Terenzini(1991) state that their workwas “based almost exclu-sively on samples oftraditional college studentswho are age 18 to 22, who

attend four-year institutions full-time, and who liveon campus” (p. 632). This review is now almost 15years old. There is no question that communitycolleges have attracted more attention during thelast decade, but this new interest is not reflected inpublished research. A review of articles publishedin five mainstream higher education journals3

between 1990 and 2003 by Townsend, Donaldson

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Most of this research is

about four-year colleges.


Community college students

tend to be older, are more

likely to be working, and

are more likely to interrupt

their enrollments.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

and Wilson (2004) found that only 8 percent ofthe 2,321 articles even mentioned communitycolleges.

The lack of research on community colleges isa particularly serious problem when it comes to thestudy of retention. Much of the research andthinking on retention has been based on theconcepts of student engagement and integrationwith the college. These concepts are likely to bemost powerful for residential students, whorepresent a small minority of the student popula-tion at community colleges, which are primarilycommuter schools. But what differentiates thecommunity college student body is the predomi-nance of part-time students: Only 36 percent ofcommunity college studentsattend full-time, while 71percent of four-year collegestudents are enrolled on afull-time basis.4 In addition,community college studentstend to be older, are morelikely to be working, and aremore likely to interrupt theirenrollments. Policiesdesigned to retain 18-year-old students living in dormsare not likely to be as effective for part-time,working students and especially for adults withfamilies and full-time jobs.

2. Lack of data on institutional policies: Theprimary source for national data on institutionalcharacteristics is the Integrated PostsecondaryEducation Data System (IPEDS) collected by theNational Center for Education Statistics (NCES).IPEDS includes data on more than 1,000 commu-nity colleges, but it has little detail on the types ofinstitutional practices colleges use to improveretention — student services, pedagogic strategies,organizational innovations, etc. Therefore, moststudies of the effects of institutional practices arebased on student surveys and often involve only asingle institution. They are of mixed quality and,in any case, produce results that are difficult to

generalize. To the extent that results depend onanswers provided by students, the studies also failto measure institutional practices directly. Incontrast, there are much better national dataavailable to study the relationship betweenindividual characteristics and retention and othereducational outcomes. This is interesting research,but it does not answer our question — how doinstitutional factors and policies influenceretention and completion?

3. Methodological problems: Evaluations ofinstitutional practices are notoriously difficultbecause of problems with the attribution ofcausality. Most practices that are studied involve

discrete programs; somestudents are in thoseprograms, and others arenot. Studies of theeffectiveness of theprograms generally consistof a comparison betweenthose two groups ofstudents, but these types ofcomparisons often do notprovide enough informa-tion to make a valid

judgment. As long as there is some non-randomprocess by which students enroll or are chosen forsuch a program, it may be that any differencesbetween participants and non-participants resultfrom the selection process, not from the programitself. Thus, even though a program shows positiveresults, those results may not hold for other groupsof students. There is no question that causalityproblems can be difficult, but there are techniquesthat can be used to address them. As this reportwill show, much of the research on communitycolleges fails to make use of these techniques.

4. Inadequate dissemination and discussion:Methodological problems are compounded by themethods generally used in community colleges todisseminate and discuss institutional policy andinitiatives. Results are often posted on Web sites,


frequently in the form of PowerPoint presenta-tions, or are presented at conferences andconventions, such as those held by the AmericanAssociation of Community Colleges or the Leaguefor Innovation in the Community College. Mostof these studies, however, are unpublished andtherefore may not undergo a rigorous vetting andquality-control process. Full reports often are

difficult to track down and rarely provide enoughmethodological detail for the reader to make ajudgment about the reliability of the results. Ingeneral, community college institutional research-ers and practitioners get little chance to discussresearch findings in a way that might allow a morecomprehensive understanding of the results andimplications of existing research.



Conceptual perspectiveson retention and completion

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Research on student

integration has a profound

causality problem.

I n this section, we discuss in more detail thetheoretical approaches to studying studentretention and how they are used for theanalysis of community colleges. VincentTinto’s (1975, 1993) student integration model

forms the conceptual basis of much of the researchon persistence and gradua-tion. It also has someattractive implications forinstitutional policy. Tinto’smodel is designed to helpcolleges understand whystudents leave, so theinstitutions can designactivities to better serve students’ needs andthereby increase retention and graduation rates.Tinto states that students’ departures from aninstitution “reflect the character of the individual’ssocial and intellectual experiences within theinstitution. Specifically, they mirror the degree towhich those experiences serve to integrateindividuals into the social and intellectual life ofthe institution” (1993, p. 51). This model differen-

tiates between social integration, which is measuredby such factors as interaction with faculty andparticipation in extracurricular activities, andacademic integration, which is usually measured bygrades or other indications of academic achieve-ment. This perspective implies that institutions

should develop processesand activities that fosterboth types of integrationamong college students.

Although this modelresonates with educators andportrays an attractivecollege environment,

empirical analysis has been difficult. Research onstudent integration has a profound causalityproblem. Studies of four-year colleges show thatstudents who participate in student organizationsor interact with faculty persist and graduate athigher rates. Still, it does not follow that gradua-tion rates will increase if every student joins astudent organization or interacts with faculty.Students may interact with faculty because they and


Academic advising should

be designed to increase

goal commitment.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

the faculty share values and an orientation towardacademic activity. For example, Pascarella andTerenzini (1991) reported the results of a studythat concluded: “educational aspirations are morelikely to influence contact with faculty thancontact with faculty is to influence educationalaspirations” (p. 395). This problem is compoundedwhen studies are done on single institutions. Inthese studies, students often are asked to fill outquestionnaires, and their levels of social oracademic integration are scored based on theiranswers (for example, see Napoli & Wortman,1998). But if the students are all at the sameinstitution, then presumably they all have access tothe same services or activities that might promoteintegration. If the students share the sameenvironment, then measures of integration aremeasures of individual characteristics, notinstitutional characteristics.Thus, the types of studiesthat show a relationshipbetween social integrationand retention do not implythat introducing policies topromote integration willnecessarily increaseretention.

In contrast to the theoretical emphasis onsocial integration, Bean and Metzner (1985)explicitly developed a model of attrition fornontraditional students, both at four- and two-yearinstitutions. Their contention was that socialintegration would play a much smaller role amongthese students and that outside “environmental”variables would be more important. Thesevariables included finances, hours of employment,outside encouragement and family responsibilities.They also suggested that “goal commitment” and“intent to leave” were important for nontraditionalstudents and that these students are more focusedon the economic benefits of their education.

Since Bean and Metzner (1985) place a greatdeal of importance on environmental factorsoutside the college’s control, their approach wouldappear to leave less potential for an institutional

response. In their model, the two variables that areunder the control of the colleges are academicadvising and course availability. Presumably,academic advising should be designed to increasegoal commitment and influence the student’s intentto leave. Course availability is certainly a logicaldeterminant of attrition, especially for nontradi-tional students, who generally have a moreinstrumental view of their college education.

More recently, Braxton, Hirschy andMcClendon (2004) have attempted to extendTinto’s model to “commuter universities andcolleges.” Unlike residential colleges, the authorssay, “commuter colleges and universities lack well-defined and -structured communities for studentsto establish membership” (p.35). The authors’recommendations focus first on building studentinvolvement in the classroom through learning

communities. This is alogical emphasis since theclassroom is the placewhere commuter studentshave the most contact withthe college. Braxton andhis co-authors also suggestthat colleges need to try to

connect with parents and spouses, since “signifi-cant others” have more day-to-day influence oncommuter students than they do on residentialstudents. Braxton and his co-authors also empha-size practical considerations such as providingcourses at convenient times and locations,developing jobs on campus (to ease the college/work conflict), and providing day care.

Their work is an important advance in that itexplicitly takes account of the special needs ofnon-residential students. Nevertheless, the analysisdoes not differentiate between two- and four-yearcolleges, and the authors suggest that this issomething that remains to be done. Overall, thebook leans more toward a four-year collegeperspective, partly because it relies on a body ofresearch that is dominated by four-year collegestudies. For example, the book contains nodiscussion of issues related to developmental


Many studies suggest

that academic and social

integration have positive

effects on persistence.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

education or to students with explicit occupationalgoals. Community college faculty and administra-tors would generally agree that the problems ofacademically underprepared students and thedevelopmental programs designed to help thosestudents need to be central components of anyframework for understand-ing and improving commu-nity college retention.

What does empiricalresearch suggest about thesemodels? Although manymethodological problemspersist, many studies dosuggest that academic andsocial integration havepositive effects on the persistence of four-yearcollege students, especially residential students(Braxton, Hirschy & McClendon 2004; Cabrera,Nora & Castañeda, 1993). In an extensive reviewof research on college student outcomes, Pascarellaand Terenzini (1991) found that both thefrequency and quality of interactions with peersand faculty and the participation in extracurricularactivities — measures of social and academicintegration into the college life — contributepositively to students’ persistence at baccalaureate-granting institutions. However, the effect is weakerwhen factors such as student characteristics, pre-college experiences and other college experiencesare taken into account.

Interestingly, Tinto’s model (1997, 1993) hashad more influence on community college researchthan Bean’s and Metzner’s framework (1985),despite the latter’s attempts to address theproblems of nontraditional students. Many studiesof community college attrition are explicitly set upas tests of the student integration model, butresearch on social engagement by communitycollege students is at best mixed. Pascarella, Smartand Ethington (1986) found that academic andsocial integration did have positive effects onpersistence, but they used a specialized sample,which included first-time college students enrolledin community colleges in 1971 and who aspired to

a bachelor’s degree.5 In their review of integrationresearch on community college, Bean and Metzner(1985) concluded that “social integration is rarely amajor factor in attrition decisions” at commuterinstitutions (p. 520). In their 1991 review ofempirical research on undergraduate student

attainment, Pascarella andTerenzini contended thatthese models do not workas well for commutercolleges as for residentialcolleges, stating that “witha few exceptions, theweight of evidence is clearthat various measures ofsocial integration (includ-

ing interaction with faculty, interaction with peersand extracurricular involvement) show little if anypositive relationship with persistence at commuterinstitutions. This lack of a positive relationshipholds regardless of the specific measure of socialintegration used and irrespective of whether or notstudent background characteristics were taken intoaccount in the study design” (p. 402).

Subsequent research, not reviewed byPascarella and Terenzini, continued to find mixedeffects. Bers and Smith (1991) and Napoli andWortman (1998) found small positive effects,Borglum and Kubala (2000) found no effect, andNora, Attinasi and Matonak (1990) found anegative effect of social integration on persistence.But all of these results came from single-institutionstudies that determined levels of integration basedon answers to surveys administered to the students;these studies do not measure the influence of policiesto promote social integration on persistence.

In a recent review of empirical tests of Tinto’sintegration model, Braxton, Hirschy andMcClendon (2004) found strong empirical supportfor the application of the model to residentialcolleges and universities. For commuter universi-ties, they found “modest” support for the role ofsocial and academic integration in promoting“commitment to the institution” if not persistenceitself (pp. 16-17). With respect to the model’s


usefulness for community colleges, these authorsstated: “Given this configuration of support [thepattern of empirical support that they found intheir review], the explanatory power of Tinto’stheory to account for student departure in two-year colleges remains undetermined and open toempirical treatment” (pp. 17-18).

What are the implications of these theoreticalperspectives for institutional practice? In particular,the concept of social engagement is probably mostrelevant for a college experience typified by theresidential liberal arts college with multifacetedinteractions inside and outside the classroomamong students and between students andprofessors. Given the nature of their students andthe large number of part-time faculty, trying toreproduce the liberal arts/residential ideal may notbe the best strategy for community colleges ortheir students. The empirical record is certainlyconsistent with this conclusion.

With respect to institutional practice, Bean andMetzner (1985) emphasize the importance ofacademic advising, presumably to influence students’goal commitment and “intent to leave.” We reviewthe research on advising later in this report. Theavailability of courses, another variable in the Beanand Metzner model, has not played an importantrole in subsequent empirical research — perhaps

because most studies are about a single institution— yet this ought to be a promising direction forresearch. Braxton, Hirschy and McClendon (2004)also emphasize course availability, and they addday care. Other factors that might be particularlyimportant for community college students includeconvenient transportation, high-quality onlineeducation, applied pedagogies and well-designedinternships. These features may be more importantto community college students than the nature oftheir relationships with professors or participationin student organizations.6 Thus, researchers shouldcertainly be looking for other institutionalvariables that do a better job of matching theneeds of the typical part-time, working communitycollege student — needs that are as much practicalas they are psychological.

The one place where the engagement modelmay be most relevant at the community college isin the classroom. This, after all, is where evencommuter students interact with faculty andpotentially with other students. Designing theclassroom experience to promote more meaningfulinteraction among students and teachers is onepromising strategy for community colleges; we willexamine efforts to do this later, during ourdiscussion of learning communities.


Empirical studies of the effect ofinstitutional management and practices

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

M uch research in the academicliterature has been concernedwith testing various theories ofstudent persistence andcompletion. As we have

emphasized, however, little of this work directlytests the effects of institutional practice on studentoutcomes. Another genre of research asks a moredirect set of questions to determine whetherstudents who participate in a particular type ofprogram or activity persist or graduate at higherrates. In this section we turn to studies that takethis more straightforward approach.

Although there are many studies on a long listof possible program activities, we focus here onfour large categories: student services (such asadvising, counseling, mentoring and orientationprograms), learning communities, developmentaleducation and college-wide reform. Our purpose isto draw substantive conclusions and to illustratethe status of research on the effectiveness ofpractices at community colleges.

Advising, counseling, mentoring andorientation programs

Colleges have been experimenting with varioustypes of advising and student services for decades.According to Pascarella’s and Terenzini’s summaryon the effect of such programs, “the most consis-tently effective program format appears to be afirst-semester freshman seminar that meets as aregular class with an assigned instructor. Thepurpose of this type of seminar is to orient thestudent to the institution and its programs and toteach important academic survival skills”(Pascarella and Terenzini, 1991, p. 403). Muraskinand Wilner (2004), in their review of institutionalpractices, also concluded that freshman-yearprograms were effective. Nevertheless, both sets ofauthors acknowledged that participation in theprograms was voluntary. Therefore, the positiveassociation might be influenced by initial studentcharacteristics and not the service itself. Still, theconsistency of the findings gives more weight tothe positive conclusions.


Studies on counseling and

advising primarily concern

four-year colleges.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Evidence on the effectiveness of other types ofadvising and counseling is more mixed. Pascarellaand Terenzini (1991) concluded that whenstudents’ pre-college characteristics were takeninto account, relatively short-term orientationprograms had a trivial and statistically insignificantdirect effect on persistence. They also found mixedresults from research on the effects on persistenceof the amount and quality of academic advising.Muraskin and Wilner (2004) reported thatpersisting students tended to express highersatisfaction with counseling services than did thosewho left, although research on the impact of suchservices was also mixed. The authors also sug-gested, though, thatmeasurement of counselingeffectiveness may bedistorted because studentswho use counseling mayhave come to the collegewith more problems thanthose who do not use it.This is a situation in which failing to account fordifferences in the characteristics of programparticipants and non-participants may understatethe effectiveness of a program.

Early intervention for academically weakcommunity college students, through counselingor other student support services, is thought toimprove their persistence and academic perfor-mance (Grubb, 2003; Summers, 2003). Summers(2003), in a review of the literature on the impactof counseling on attrition, indicated that studieshave found that counseling increases the retentionof students who are identified as highly likely todrop out.

Student Support Services (SSS), funded underthe federal TRIO programs, is perhaps the mostwidespread student services initiative. SSS wasevaluated in the mid-1990s, and the study foundthat “freshman-year SSS participants increasedtheir grade point averages by 0.15 in the first yearand 0.11 in the second year of college. SSSparticipation also increased retention to the secondyear of college at the same institution by 7 percent

and retention to the third year in any institution by3 percent” (Muraskin, 1997, p. 1). Interestingly, thestudy also found that effects increased withincreased exposure to SSS activities. The authoridentified peer tutoring, workshops and culturalevents as effective components, with peer tutoringshown as particularly effective. This was acomprehensive study, and the author was careful tocompare the characteristics of participants to non-participants; nevertheless, he cautioned thatunmeasured motivation still might have influencedboth enrollment and program effects. Finally, theevaluation did not focus particularly on communitycolleges. Indeed, only one of the five colleges that

were used in a follow-upbenchmarking study was acommunity college(Muraskin, 1997).

Thus, the studies oncounseling and advisingprimarily concern four-yearcolleges. Research on these

services, and on institutional practices in general, ismuch scarcer for community colleges. For example,a recent review of attrition research at communitycolleges by Summers (2003) cited only twounpublished single-institution studies on studentsupport services, although it reported positiveeffects of “matriculation” services such as assess-ment, orientation and counseling.

The experience of the Community College ofDenver (CCD) has been particularly influential inthe discussion of policy and practice in communitycolleges. A book by Roueche, Ely and Roueche(2001) provides a description of many of thepractices and strategies that the college used.CCD’s counseling and academic support servicesare organized in a comprehensive unit called theAcademic Support Center (ASC). The collegereports that the class withdrawal rate was 7.8percent for students receiving ASC support, whilethe overall campus rate was 12.4 percent. Althoughthis finding is encouraging, the reports do notinclude information that would allow a judgmentabout how students got access to these services


○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Learning communities

typically organize instruction

around themes.

and the comparability of the ASC students withother students at the college.

Research on counseling and student servicessuggests that such programs can play an importantrole in retention and graduation. At the same time,this body of research supports two generalizationsabout research on community college practices.First, although formal counseling would seem to beparticularly important for community collegestudents (who are likely to have fewer informalsources of information and guidance than studentsat four-year institutions), the large majority ofresearch on any particularhigher education practice islimited to four-yearcolleges. Second, whilethere are some encouragingresults from the smalleramount of communitycollege research, thematerial published from that research does notpermit a judgment about the validity of itsconclusions.

Learning communitiesIn the last 15 years, educators in both two- and

four-year institutions have experimented withlearning communities as a means of engaging andmotivating students. Learning communitiestypically organize instruction around themes, andstudents go through such programs as cohorts.Learning communities are designed to providemore coherent and engaging experiences thantraditional courses, and to give students andfaculty more opportunities for increased intellec-tual interaction and shared inquiry (Knight, 2002;Smith et al., 2004; Tinto, 1997; Tinto & Love,1995). The learning community model is particu-larly interesting for community colleges because itis one way that these commuter institutions canengage with their students in a more intensive waythan normally occurs in the classroom (Braxton,Hirschy & McClendon, 2004; Fogarty & Dunlap,2003; Grubb & Associates, 1999). If the student

integration model does apply to communitycolleges, it would probably be implemented withclassroom-oriented approaches such as learningcommunities. In fact, many community collegeshave adopted various forms of learning communi-ties, specifically as a strategy to develop a morecoherent intellectual environment and forgestronger links with the diverse and fragmentedcommunity college student body (Fogarty &Dunlap, 2003; Smith et al., 2004).

On the other hand, the design of learningcommunities may discourage nontraditional

students from participating.Because a learningcommunity often requires acohort of students toattend several classes as agroup, it may be difficultfor working students orstudents attending part

time to participate. Because learning communitiesrequire close coordination among professors, theyappear to be most effective when regular, full-timefaculty members are used rather than part-timeadjuncts. Yet, full-time faculty are more likely toteach during the day, rather than at night whenmany nontraditional students attend classes. Thesefactors suggest that learning communities mayattract more middle-class, traditional-age studentsamong those enrolled in community colleges.7

In 2003, the National Learning CommunitiesProject at The Evergreen State College inOlympia, Wash., published an extensive review ofmore than 100 studies of the effectiveness oflearning communities (Taylor et al., 2003). What isremarkable about this review is the length towhich the authors went to find unpublished as wellas published research. As a result, this publicationgives an unusually comprehensive picture of thestate of research on this topic, including, indeedpredominantly, studies conducted by institutionalresearchers at colleges. The authors concludedthat “a preponderance of studies indicate thatlearning communities strengthen student retentionand academic achievement” (p. iii).


Only 32 of the 119 studies, or

about one-quarter, covered

community colleges.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Thus, research on the effects of learningcommunities is encouraging, yet the comprehen-sive review of effectiveness studies also illustratesmany of the challenges to community collegeresearch (Taylor et al., 2003). For example, thelarge majority of such studies were unpublished,single-institution assessments; only one-seventhwere published in journals or books. Therefore,while there is a tremendous amount of material onlearning communities, much of it is difficult toobtain. Some of the reported studies based theirassessments on comparisons between learningcommunity and non-learning communitystudents, controlling forentering academic charac-teristics. But withoutknowing more about theprocess used to recruit andenroll students into learningcommunities, it is difficult to judge whether thesecontrols were adequate to account for initialdifferences between learning community andcomparison students.

Not surprisingly, learning communities in four-year colleges have received much more attentionthan those in community colleges; only 32 of the119 studies, or about one-quarter, coveredcommunity colleges. Taylor et al. (2003) chose 17studies that were “deemed notable for the qualityof the assessment study and the manner in which itwas reported” (p. 4). Of these, only one was abouta community college program. The review doesnot discuss issues associated with the design andeffectiveness of learning communities at commu-nity colleges — in particular, how effective theinnovation is for the more typical part-timecommunity college student.

The best-known evaluation of communitycollege programs is the 1997 article by Tintopublished in the Journal of Higher Education,“Classrooms as Communities: Exploring theEducational Character of Student Persistence,”which discussed learning communities at SeattleCentral Community College. Tinto found that

participation in a learning community did increasethe probability of quarter-to-quarter persistence.Based on a qualitative analysis, he argued thatlearning communities promote persistence byfacilitating the creation of supportive peer groupsamong students, encouraging shared learning, andgiving students the opportunity to activelyparticipate in knowledge creation. This study useda multivariate methodology that controlled forpossibly confounding characteristics. Tinto alsorecognized the potential distortion in the modelcaused by student self-selection, and he presented

an argument for explainingwhy it was not a problem.8

In contrast to the resultsfrom the Seattle study, anearlier study of learningcommunities at LaGuardiaCommunity College byTinto and Love (1995)

found that participation in learning communitiesdid not significantly increase the probability ofpersistence.

In 2003, the research organization MDRCstarted an evaluation of learning communities atKingsborough Community College. The studyinvolves full-time freshmen between ages 17 and34, most of whom had applied directly toKingsborough after missing the applicationdeadline for the City University of New York(CUNY) system.9 Other students who had lowplacement test scores were also invited toparticipate. Students were assigned, through arandom selection process, to learning communitiesor to a control group that received servicesgenerally available to students at the college. Thecourse-taking patterns of both groups were trackedthrough the second semester. Preliminary andunpublished results suggest that the learningcommunity students had passed more courses(including basic English classes) and had highergrade point averages than the comparison group,with differences that were statistically significant.There were other encouraging, although notstatistically significant, differences.10 In summary,


Almost one-fifth of

traditional-aged community

college students never

complete 10 credits.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

learning communities are an attractive strategywith encouraging empirical support. The learningcommunity approach also is one strategy that iscommon in community colleges and that isconsistent with the dominant theoretical perspec-tive on student retention. That is, learningcommunities offer the potential for more in-classengagement with commuterstudents, who may not havea chance to participate insocial and other extracur-ricular activities at thecollege. The use of learningcommunities should beexpanded while outcomescontinue to be tracked. Inthe future, researchers needto better account for the student recruitment andenrollment process in their analyses; also, theyshould examine the particular problems associatedwith scheduling and formatting learning communi-ties so they can be most effective and convenientfor nontraditional community college students.

Developmental education and services foracademically underprepared students

Improving the effectiveness of developmental(remedial) education is perhaps the most importantissue confronting community colleges today. Mostentering community college students arrive withacademic skills that do not allow them to partici-pate effectively in at least some college-levelcourses (Perin, forthcoming). Almost one-fifth oftraditional-aged community college students nevercomplete 10 credits,11 and that number is probablyhigher for older students. Many of these studentsleave because of academic problems and, indeed,many students never successfully complete alldevelopmental courses that are deemed necessary.Thus, as Summers (2003) pointed out: “[M]anyinstitutions’ primary strategy for reducing attritionis the early identification of students likely to dropout and the development and implementation ofintervention services for those students” (p. 64).

Therefore, colleges offer a variety of services forstudents with weak academic skills.

Based on their review of the literature onacademic achievement, Pascarella and Terenzini(1991) suggested that institutions can aid theacademic adjustment of poorly prepared studentsby providing extensive instruction in academic

skills, advising, counselingand comprehensive supportservices. Although theyreported findings primarilyfrom studies of four-yearcolleges, Pascarella andTerenzini contended thattheir findings have beenreplicated in severalnational studies and that

the results hold even after controlling for impor-tant student and institutional characteristics.

Some more recent results are mixed. Many findthat students who enroll initially in developmentalcourses graduate at lower rates than do studentswho start in regular credit courses (Muraskin &Wilner, 2004), although, once again, many ofthese studies analyzed students at four-yearcolleges. In a study of college transcripts, Adelman(1998) found that the more remedial coursesstudents were required to take, the less likely theywere to earn a degree. Among students whoattended two-year and/or four-year institutions andearned more than 10 credits, 45 percent of thosewho took two remedial courses earned either anassociate’s or bachelor’s degree by the time theywere 30 years old, compared with 60 percent ofthose who took no remedial courses. Students whoare judged to have low reading skills in particularare more likely to need extensive remediation andless likely to earn a degree. Another study foundthat students at the Community College of Denverwho started college in developmental classes werefound to graduate at the same rate as students whostarted in regular courses (Roueche et al., 2001).

Researchers studying the effectiveness ofdevelopmental education face particularly seriousmethodological challenges. On average, students


It is hard to identify a causal

relationship between remedial

education and subsequent

educational attainment.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

who attend developmental education classes startout with weaker academic skills. As a result, it ishard to identify a causal relationship betweenremedial education and subsequent educationalattainment. Even if students who start in develop-mental classes appear to do more poorly thanother students, it is still possible that theremediation was effective; without it, the studentsmight have done even worse. Comparing similarstudents who do and do not participate indevelopmental education is difficult because manystates and individual institutions mandateremediation for studentswith low assessment scores.

A study by Bettinger andLong (2004) is of interestbecause it uses a statisticaltechnique designed toidentify the causal relation-ship in these types of cases,although, once again, it is astudy restricted to four-yearinstitutions. The Bettinger-Long study focused on a large sample of first-time,full-time students of traditional age enrolled inOhio’s non-selective public four-year colleges inthe fall of 1998. The results suggest that studentsplaced in remediation are more likely to withdrawfrom college, but they also indicate that participa-tion in remedial courses does not seem to decreasethe likelihood of transferring to more selectiveinstitutions or attaining a bachelor’s degree.

Some studies compare outcomes for differenttypes of developmental programs. The best knownof these types of studies is the National Study ofDevelopmental Education, conducted by theNational Center for Developmental Education(Boylan, Bliss & Bonham, 1997). The programmaticimplications of this and other studies of develop-mental education were subsequently published inWhat Works: Research-Based Best Practices inDevelopmental Education (Boylan, 2002).12 Thisstudy tested the relative effectiveness of central-ized programs (compared with decentralizedprograms), programs with tutorial services with

trained tutors, programs with advising andcounseling, and those programs that includedevaluation on several outcome measures: first-termGPA, cumulative GPA, retention in developmentcourses and success (earning a D or better) in mathand English developmental courses. In general, thisstudy found more positive results for four-yearcolleges than for two-year colleges, perhapsbecause four-year students may have had strongerinitial skills.13 The study did find some positive andstatistically significant effects for communitycolleges on some of the outcomes listed above.

The National Study(Boylan et al., 1997)yielded useful informationabout developmentaleducation, and itsrecommendations are inaccord with the experienceof many practitioners.However, the publishedreports exclude informa-tion that would be helpful

for evaluating the effectiveness of remediation. Forexample, it would be interesting to know themagnitude of the effects. Also, because eachprogram characteristic is analyzed separately, thereis no way to determine if a particular programcomponent is effective or whether it is effectiveonly in combination with other components.Finally, it would also be important to knowwhether there are any other institutional featuressuch as size, college organization, financialcondition or typical student characteristics thatmight influence student outcomes and also berelated to a particular program feature.

Boylan (2002) lists 33 features that he suggestsare “best practices.” The book references fiveinstitutions judged to have successful developmen-tal programs for benchmarking purposes, but theevidence supporting the 33 practices comesprimarily from other studies and reports, not fromthe five chosen sites.14 Much of the research in thisstudy on developmental/remedial education iseither based on unpublished documents or material


○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

No program, however

well designed, can

work in isolation.

in the National Center for DevelopmentalEducation’s own publications. Thus, while many ofthese practices are widely used and consideredeffective by practitioners, it is difficult to judgewhether the evidence supporting these practices isdefinitive.

Given the pervasiveness of developmentaleducation and the controversy surrounding itspractice, it is surprising that there is still so muchuncertainty about the most effective approaches toworking with students with weak academic skills.Grubb (2001), in his review of research ondevelopmental education, argued that learningcommunity formats for developmental educationdid appear to have positive benefits, but he foundless evidence on the effectiveness of otherapproaches to remedialeducation. To be sure, theresearch is complicated bydiverse state regulations andby local variation in thecriteria used to place studentsin developmental programs,including differing “cutoff”scores on assessment tests.Further, many colleges use informal processes toalter these regulations on a case-by-case basis(Perin, forthcoming). But this overall complexityalso offers researchers opportunities to makecomparisons among similar students who do anddo not enroll in formal remediation or who enrollin different types of developmental courses. Asmore detailed transcript-based data becomeavailable, considerable progress can be made onunderstanding the characteristics and effects ofthese strategies.

College-wide reformSo far we have reported on research about

individual programs. Many practitioners believetheir programs are successful for the students whotake part in them. However, although there arecertainly positive indications for all of thesepractices, we have suggested that an objective look

at the empirical evidence and the methodologiesused to test the effectiveness of these programspresents a more ambiguous picture. And, thoughmany people believe in the effectiveness ofindividual programs, there is much more skepti-cism about whether they can be “taken to scale,”that is, applied as institutional reforms. Comple-tion rates at community colleges are low, andimproving them significantly will probably requirethe successful expansion of pilot programs and thestrengthening of related programs and services. Noprogram, however well designed, can work inisolation. An excellent developmental or counsel-ing program in a college with generally ineffectiveteaching may ultimately have no effect on studentcompletion rates. We have found virtually no

research that attempts todefine and assess programinstitutionalization orbroader college-widereforms.

There have beeninitiatives designedespecially to bring aboutreform throughout an

institution. The best-known initiative for commu-nity colleges is the Learning College movement.Published work on this model (O’Bannion, 1997)presents useful accounts of the processes throughwhich colleges have brought about importantchanges, but so far no rigorous assessment of thisstrategy has been published.

While community college practitioners believethat few colleges have been able to bring pilotprograms to scale successfully, many are convincedthat the Community College of Denver (CCD)has, over the past 20 years, succeeded in bringingabout fundamental reforms in the basic ways thatthe college operates (Roueche et al., 2001). Thesereforms followed a systematic planning andbenchmarking process and have included majorchanges in organization, teaching methods,counseling and student services, relationships tothe community, and organizational philosophy.The study of CCD (Roueche et al., 2001) reported


increases in the number of graduates and in thenumber of graduates of color and concluded that“cohort tracking indicated no significant differencein student success on the basis of race, ethnicity,age, or gender” (p. 23). This study also reportedthat many CCD students of color transferredsuccessfully, that there were high rates of studentsatisfaction with programs and instructors, andthat surveyed employers demonstrated unanimoussatisfaction with the skills of CCD graduates.These are encouraging results, although, as is truefor much of the research we have described in this

analysis, the information reported in the study isnot adequate to judge whether the measuredoutcomes were caused by the described programchanges or by other factors.15 Given the impor-tance of the CCD case, further investigation iswarranted. Careful investigations of otherinstitutional change efforts also are important.Nevertheless, the CCD experience suggests thatinstitution-wide reform can have an effect, andthat perhaps a focus on individual programs maybe less effective.


Conclusions and implications for practice○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Existing research provides the

most support for learning

communities.

T his report has examined the researchliterature on effective institutionalpractices for improving retentionand completion at communitycolleges in order to draw substantive

lessons about those practices and to assess theoverall state of research inthe area. We will first reviewthe substantive conclusionsbased on our review of theempirical research, includingsuggestions for additionalresearch. We then presentconclusions and suggestionsabout the research itself, discussing theoretical,data and methodological issues and outliningsuggestions for strengthening the research thattakes place at community colleges.

Substantive questionsWe reviewed the research on the effectiveness

of four types of practices in increasing persistenceand completion at community colleges: 1) ad-vising, counseling, mentoring and orientationprograms; 2) learning communities; 3) develop-

mental education and other services for academi-cally underprepared students; and 4) broader,institution-wide reform.

Among the practices and strategies that weexamined, existing research provides the mostsupport for learning communities. The positive

effects of learningcommunities on persis-tence and graduation areconsistent with the mostinfluential theoreticalperspectives used to studyretention, and empiricalresearch also suggests

positive effects. Thus learning communities offeran approach to connecting more intensively withcommunity college students, who often spendlittle time on campus outside of classes. Oneimportant area for future research involvesinvestigation of learning communities for part-time, nontraditional students.

Researchers and practitioners agree thatcounseling and advising are crucial for communitycollege students. The national evaluation of thefederally funded Student Support Services programhas many useful conclusions, but that study is now

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○


Many issues concerning the

organization and pedagogy

of developmental education

warrant further investigation.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

a decade old, and the policy and demographicenvironments have changed significantly in theintervening years. Many important researchquestions remain to be answered concerning thebest design of student services, including therelationship between counseling and advising andthe extent to which student services should becentralized or distributed among wider groups offaculty and staff.

Developmental education also has receivedconsiderable attention, and there is widespreadagreement that it is one of the most important andchallenging issues facing community colleges.Although researchers face difficult methodologicalproblems, available data and current empiricaltechniques allow for significant improvement inthe reliability of conclusions. Moreover, the majornational study on develop-mental education is now 10years old. Many issuesconcerning the organizationand pedagogy of develop-mental education warrantfurther investigation. Forexample: Should theseservices be centralized ordecentralized? Candevelopmental instruction be integrated intoregular classes? Under what circumstances shoulddevelopmental studies be considered prerequisitesto other classes? How effective are alternatives tospecialized developmental classes such as tutoringcenters?

Other questions that need attention include:How can programs best be combined? Whatfactors promote the institutionalization ofsuccessful pilot programs? Combining differentreforms into comprehensive college-wideinitiatives is an attractive prospect, but we are farfrom any concrete understanding about how suchprograms should be combined. Further, moving asuccessful pilot program to college-wide scale maybe more difficult than implementing the pilotalone. Researchers can make important theoreticaland empirical contributions in this area by

developing models for organizational change atcommunity colleges, by working out methods tomeasure that change, and by assessing theeffectiveness of those strategies.

All of the substantive areas we analyzedillustrated weaknesses in community collegeresearch on program effectiveness. These weak-nesses include an overemphasis on research onfour-year colleges; lack of appropriate data oninstitutional practices; methodological problems,especially having to do with identifying causalrelationships; and inadequate reporting, dissemina-tion and discussion of research. We presentsuggestions in each of these areas below:

Theoretical issues: We have emphasized inthis report that the dominant theoretical perspec-

tive on retention, thestudent integration model,is most appropriate fortraditional four-yearstudents, particularly thoseliving on campus.Empirical assessments ofthe model for communitycolleges have beeninconclusive at best.

Researchers have begun to focus on applying theseconcepts to “commuter” universities and colleges,but more work still needs to be done to takeaccount of the particular characteristics ofcommunity colleges and their students. Futurework should pay more attention to the needs ofpart-time, working students; to convenience andaccessibility; to the particular problems of studentswith significant academic deficiencies; and to thefocused occupational goals of many communitycollege students.

Data availability: Because insufficient dataexist on institutional practices, most program-effectiveness research is based on single-institutionsamples. While these can be useful, their conclu-sions are difficult to generalize because effects maybe based on particular features of the college being


○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

The most difficult empirical

problem in the assessment of

retention practices is the

attribution of causality.

studied. As much as possible, national-leveldatabases, such as those created by the NationalCenter for Education Statistics, should includeprogrammatic detail to allow research on theeffectiveness of common practices used to improvestudent success. Collabora-tive projects within commu-nity college districts andstates, supplemented withincreasingly available state-or district-level studentrecord data, also will behelpful in understandingstudent persistence and inovercoming the limitations ofsingle-campus studies. Even with single-institutionstudies, research that makes use of transcript-basedlongitudinal data can provide many importantinsights. Current proposals to collect unit recorddata at the federal level may provide data thatwould be extremely useful in studying studentsuccess at community colleges. Concerns aboutstudent privacy may prevent the collection ofthese data, however.

Empirical issues: The most difficult empiricalproblem in the assessment of retention practices isthe attribution of causality. Most of the programsthat we have examined are voluntary, so compari-sons of outcomes for participants and non-participants need to be interpreted carefully. Aredifferences in outcomes the result of programeffects or of initial differences between the twogroups?

There are several steps that researchers cantake to purge the analysis of this distortion.Multivariate analysis can control for manycharacteristics. Perhaps the most important is somemeasure of pre-program academic ability: If aprogram enrolls more successful students, then itmay not be surprising that program participantsare more successful than comparison students.Pascarella and Terenzini (1991) generally discussedthe strength of such controls in the studies thatthey reviewed, but many studies — especially

unpublished, single-institution studies — fail totake this step, or at least fail to report it. However,even if available variables are included in ananalysis, there may be unmeasured differencesbetween the two groups. Students who sign up for

a program may be moremotivated than those whodo not, even if theirmeasured characteristics aresimilar. Indeed, thisunobserved motivation maybe the reason they soughtassistance in the first place.Alternatively, counselorsmay encourage enrollment

for students whom they judge might “benefit” froma program.

What can researchers do to address thisproblem of comparison group differences?Fundamentally, the solutions require an analysis ormanipulation of the process through whichstudents enter a program. One approach is forprogram organizers to enroll a group chosen atrandom from a pool of applicants, therebyeliminating any systematic differences between thetwo groups of applicants.16 Unfortunately, thisstrategy is expensive. For practical reasons, suchstudies cannot usually form the basis of anexpanded and invigorated research function onmost campuses.

A second approach involves conducting astatistical analysis of the selection process andusing those results to adjust the measurement ofthe program effect (see, for example, Bettinger &Long, 2004). This is an increasingly popularapproach in program evaluation, but for the mostpart, it has simply not been used in the communitycollege literature. The disadvantage is that thisapproach requires particular types of data that areoften unavailable and a fairly high level ofstatistical skill. In addition, the results aresometimes difficult to interpret.

When these approaches are not possible, inaddition to controlling for relevant measurablecharacteristics such as academic achievement,


The nation’s community

college leaders have come to

agree that colleges must be

more “data-driven.”

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

researchers must include a detailed description ofthe process through which students entered aprogram so that the reader can make a judgmentabout the potential effect that the recruitment andenrollment process might have had on the analysis.If the selection process tends to enroll moremotivated students, then the measured programeffect may be exaggerated. Conversely, theenrollment process might result in an understate-ment of the program effect. For example, ifcounselors consciously send students with themost serious problems to a program, and ifstudents in that program do as well as otherstudents with similarmeasured characteristics,then it may be reasonable toconclude that the programwas effective. Since many ofthe studies of programsinvolve one or a very smallnumber of institutions, thereis no reason why researcherscannot provide this type ofinformation, yet few of the studies reviewed heredid so.

Research at the community college: Themajority of retention research published inmainstream education journals is written byuniversity-based academics, and these researchershave so far largely overlooked the communitycollege sector. Though beyond the scope of thisreport, it is an interesting point that researchers —who are usually preoccupied by the experience ofminority, low-income and other underservedstudents — continue to neglect a sector of highereducation that enrolls nearly half of all under-graduates and even larger percentages ofunderserved students.

Nevertheless, while it is important thatuniversity-based researchers pay more attention tocommunity colleges, most research on effectivepractices at the colleges will likely be carried out atthe colleges themselves. This is evident from theextensive review published by the National

Learning Communities Project discussed earlier.Most of those studies were single-institutionprojects published by the individual institutionsthemselves. Moreover, during the last severalyears, the nation’s community college leaders havecome to agree that colleges must be more “data-driven” and shift from a “culture of anecdote” to a“culture of evidence.” The research on which thisreport is based supports a major initiative begun in2003 called Achieving the Dream: CommunityColleges Count. That initiative is rooted in theprinciple that colleges should base their practiceon systematic information and research. This

implies a rise in theprominence of institutionalresearch and an increase inthe extent to which facultyand administrators use dataand research to enhancethe success of theirstudents. The growingemphasis on analysis ofstudent outcomes at both

the state and the local levels is one encouragingdevelopment. Statewide data systems and moresophisticated cohort tracking on the campusesoffer many promising opportunities for usefulanalysis.

Although, in principle, a shift toward “data-driven decision making” is not controversial, thereis no consensus about exactly what such a shiftentails or what constitutes “evidence.” As we haveseen, there is already some research on the issuesof interest here, but in many cases, even in well-known and influential reports, methodologies arepreliminary or are inadequately described. Forexample, samples and variable definitions are notwell defined. Statements are made about whatsuccessful colleges do without defining how thoseexemplars were chosen. Reports provide too littledetail about the programs and, especially, abouthow student participants are recruited andselected. Many studies look at several componentsof an overall program — for example, counselingand mandatory placement in remediation — one at


○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Administrators have few

opportunities to develop a

deep understanding of

outcome data.

a time and then report the results for each one. Butit may be possible that those components tend togo together, so it is not clear whether one or bothelements have the effect, if there is an effect. Ofcourse, it is possible that many of the details of themethodology are simply omitted from writtenreports in an attempt to make them more readable,but the reader has no way of knowing that.

Another barrier to data-driven decision makingis simply the lack of time or opportunity to make itwork. Because workloads are typically heavy atcommunity colleges, and because the resourcesdevoted to institutional research are often scant,faculty and administrators have few opportunitiesto develop a deep understanding of outcome dataand forge a common plan to improve outcomes.Indeed, for many community college facultymembers and administrators, the best opportuni-ties to address these issues occur in brief paneldiscussions at conferences — such as those held bythe American Association of Community Collegesor the League for Innovation in the CommunityCollege — or at one-time faculty workshops.Unfortunately, these eventstypically feature researchersasserting conclusions, not athorough discussion ofmethods. Definitive anddubious results all arepresented in a similar way, soparticipants can come awaywith a distorted view of thereliability of researchfindings.

Moreover, there is no widespread agreementabout what constitutes useful analysis. In givingcolleges advice about how to conduct evaluationson developmental education, one influential expertstated: “Evaluation of developmental educationdoes not require the use of complicated statistics.Program outcomes can be accurately describedusing nothing more than percentages, bar graphs,and pie charts” (Boylan, 2002, p. 42). These typesof descriptive statistics are useful to begindiscussion and investigation, and they are

important to provide a basis for wide engagementof faculty and administrators in discussions ofpolicies and practices. Presenting research findingsin a way that is understandable to a broad audienceis key to their effectiveness. Nevertheless,evaluating the effectiveness of educationalprograms is difficult, and simple descriptiveinformation can’t paint a full picture. Such datashould be seen as only a first step in understandingprogram effects.

In summary, developing a “culture of evidence”in community colleges involves a commitment tocarry out thoughtful research — which often mustbe complex — and an ability to engage faculty,administrators and even students in meaningfuldiscussions about the implications of that research.We present six suggestions for developing thatculture:

1. Colleges must devote more resources andskills to research. Some communitycolleges have sophisticated researchdepartments with well-trained personnel,

but many institutionalresearch depart- mentsare staffed by part-timeresearchers who haveother primary responsibili-ties. A college’s emphasison institutional researchdemonstrates the depth ofits leaders’ commitment tomanagement based oninformation and research.

While colleges certainly need researcherswith quantitative skills, qualitative researchbased on interviews, focus groups andobservations also are important tounderstand and interpret quantitativefindings.

2. Colleges must recognize that assessing theeffectiveness of practices is difficult andinvolves a continuum of activities andanalyses that range from simple descriptive


comparisons to more time-consuming andexpensive controlled analyses andexperiments. Randomized experiments areconsidered the “gold standard,” but theyare costly and difficult to implement. Shortof the most ambitious methodologies,colleges can make agreat deal ofprogress byemploying thought-fully designedapproaches that usethe increasinglyavailable data at thecampus and statelevels.

3. Projects should combine quantitativeresearch on student outcomes withqualitative research to elicit insights fromstudents about those outcomes. Studentperspectives are crucial for interpretingquantitative findings. Analysis of studentlongitudinal data may show that manystudents never complete developmentalmath classes, for example, but focus groupsand interviews with those students canprovide insights into why that happens andwhat steps might be successful in improv-ing the class completion rates.

4. Colleges, states and college associationsmust provide more opportunities for facultyand administrators to discuss evidenceabout student outcomes. The typical shortpresentations with a few minutes of “Q&A”give a distorted view of research results.Moreover, thorough college-wide reformbased on evidence requires broad-basedparticipation. Research cannot be theprovince of a few specialists who reportonly to the top administration and board.

5. Colleges and states must develop moresystematic methods to publicize and

disseminate useful research findings fromstate and institutional research offices.State- and college-level researchersfrequently analyze and assess policies andpractices, but the results of this work areseldom published in ways that are useful to

wider audiences. Manyreports of “best practices”fail to provide enoughbackup information anddata for readers to make aninformed judgment aboutthe practices’ relativeeffectiveness. Assessmentresults are often onlyavailable in PowerPoint

slides, institutional reports, various types oftestimony, or in the files of institutionalresearchers.

6. Collaboration among academic, institu-tional and state-level researchers should bepromoted. Researchers working to improvethe performance of community collegesface formidable problems. They will havemore chance of success if they use a varietyof methodologies and if they combineresearch based on national, state and localdata sets, as well as specific institutionaland state-level knowledge.

We have criticized the quantity and quality ofresearch on the effects of institutional practices oncommunity college retention and completion rates,but we also are pragmatists. We recognize thatcommunity colleges are ongoing operations andthat program improvement cannot await definitiveresearch results. Educators, policy-makers andstudents must move forward based on the bestinformation available, even if that information issubject to alternative interpretations. Still, aspolicy-makers, private funders and the public haveincreasingly turned their attention to communitycolleges, we have the opportunity to strengthenthe available research and provide more useful

Educators, policy-makers and

students must move forward

based on the best

information available.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○


information as administrators and faculty memberstry to improve students’ educational outcomes.

We suggest that, in planning activities, collegessearch for the best information they can find. Andthey should search critically, recognizing that allresearch is not the same and that even the mostdefinitive studies, such as those using random-assignment methodologies, have limitations. At thesame time, they should do what they can tomonitor progress and do so as thoroughly and

rigorously as possible. The interaction betweenresearch and practice should not be seen as asearch by experts for the final and definitiveanswer to the question “What works?” Rather, it isa constant and continuous process — a conversa-tion within and among the colleges and with out-side researchers and policy-makers, using the bestpossible data and the most appropriate methodolo-gies, as practitioners try to improve their practicein a constantly changing environment.



○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

References

Adelman, C. (1998). The kiss of death? Analternative view of college remediation. NationalCrosstalk, 6(3).

Bailey, T.R., & Leinbach, D.T. (2005). Is studentsuccess considered institutional failure? Theaccountability debate at community colleges. NewYork, NY: Community College Research Center,Teachers College, Columbia University.

Bean, J.P., & Metzner, B.S. (1985). A conceptualmodel of nontraditional undergraduate studentattrition. Review of Educational Research, 55(4),485-540.

Bers, T.H., & Smith, K.E. (1991). Persistence ofcommunity college students: The influence ofstudent intent and academic and social integration.Research in Higher Education, 32(5), 539-558.

Bettinger, E., & Long, B.T. (2004). Shape up or shipout: The effects of remediation on students at four-year colleges (No. w 10369). Cambridge, MA:National Bureau of Economic Research.

Borglum, K., & Kubala, T. (2000). Academic andsocial integration of community collegestudents: A case study. Community CollegeJournal of Research and Practice, 24(7), 567-576.

Boylan, H.R. (2002). What works: A guide toresearch-based best practices in developmentaleducation. Boone, NC: Appalachian StateUniversity, Continuous Quality Improvement

Network with the National Center for Develop-mental Education.

Boylan, H.R., Bliss, L.B., & Bonham, B.S. (1997).Program components and their relationship tostudent performance. Journal of DevelopmentalEducation, 20(3), 2-8.

Braxton, J.M., Hirschy, A.S., & McClendon, S.A.(2004). Understanding and Reducing CollegeStudent Departure. ASHE-ERIC Higher Educa-tion Report, 30(3).

Cabrera, A.F., Nora, A., & Castañeda, M.B. (1993).College persistence: Structural equationsmodeling test of an integrated model of studentretention. Journal of Higher Education, 64(2),123-139.

Fogarty, J., & Dunlap, L., with Dolan, E., Hesse,M., Mason, M., & Mott, J. (2003). Learningcommunities in community colleges. NationalLearning Communities Project MonographSeries. Olympia, WA: The Evergreen StateCollege, Washington Center for Improving theQuality of Undergraduate Education, incooperation with the American Association ofHigher Education and the American Associationof Community Colleges.

Grubb, W.N., & Associates (1999). Honored butinvisible: An inside look at teaching in communitycolleges. New York, NY: Routledge.


Grubb, W.N. (2001, February). From black box toPandora’s box: Evaluating remedial/developmentaleducation. New York, NY: Columbia University,Teachers College, Community College ResearchCenter.

Grubb, W.N. (2003). Using community colleges to re-connect disconnected youth. Menlo Park, CA: TheWilliam and Flora Hewlett Foundation.

Knight, W.E. (2002, June). Learning communities,first year programs and their effectiveness: The role ofthe IR office. Paper presented at the 42nd AnnualForum of the Association for InstitutionalResearch, Toronto, Canada.

Muraskin, L. (1997). “Best practices” in studentsupport services: A study of five exemplary sites.Washington, DC: U.S. Department of Educa-tion, Planning and Evaluation Service.

Muraskin, L., & Wilner, A. (2004). What we knowabout institutional influences on retention. Washing-ton, DC: JBL Associates.

Napoli, A.R., & Wortman, P.M. (1998). Psycho-social factors related to retention and earlydeparture of two-year community collegestudents. Research in Higher Education, 39(4),419-455.

Nora, A., Attinasi, L.C., & Matonak, A. (1990).Testing qualitative indicators of precollegefactors in Tinto’s attrition model: A communitycollege student population. Review of HigherEducation, 13(3), 337-356.

O’Bannion, T. (1997). A learning college movement forthe 21st century. The American CommunityCollege Association and the American Councilon Education Series on Higher Education.Washington, DC: Oryx Press.

Pascarella, E., & Terenzini, P. (1991). How collegeaffects students: Findings and insights from twentyyears of research. San Francisco, CA: Jossey-Bass.

Pascarella, E., Smart, J., & Ethington, C. (1986)Long-term persistence of two-year collegestudents, Research in Higher Education, 24(1),47-71.

Perin, D. (forthcoming). Can community collegesprotect both access and standards? The problemof remediation. Teachers College Record.

Roueche, J.E., Ely, E.E., & Roueche, S. D. (2001).In pursuit of excellence: The Community College of

Denver. Washington, DC: Community CollegePress.

Smith, B.L., MacGregor, J., Mathews, R.S., &Gabelnick, F. (2004). Learning communities:Reforming undergraduate education. San Francisco,CA: Jossey-Bass.

Summers, M.D. (2003). ERIC Review: Attritionresearch at community colleges. CommunityCollege Review, 30(4), 64-84.

Taylor, K., with Moore, W.S., MacGregor, J., &Limblad, J. (2003). Learning community researchand assessment: What we know now. NationalLearning Communities Project MonographSeries. Olympia, WA: The Evergreen StateCollege, Washington Center for Improving theQuality of Undergraduate Education, incooperation with the American Association forHigher Education.

Tinto, V. (1976). Dropout from higher education:A theoretical synthesis of recent research.Review of Educational Research, 45, 89-125.

Tinto, V. (1993). Leaving college: Rethinking thecauses and cures of student attrition. Chicago, IL:University of Chicago Press.

Tinto, V. (1997). Classrooms and communities:Exploring the educational character of studentpersistence. Journal of Higher Education, 68(6),599-623.

Tinto, V., & Love, A.G. (1995). A longitudinal studyof learning communities at LaGuardia CommunityCollege. University Park, PA: National Center onPostsecondary Teaching, Learning, andAssessment.

Townsend, B.K., Donaldson, J., & Wilson, T.(2004). Marginal or monumental? Visibility ofcommunity colleges in selective higher educationjournals. Paper presented at the Conference ofthe Council for the Study of CommunityColleges, Minneapolis, MN.

U.S. Department of Education (2003). 1995-96Beginning Postsecondary Students LongitudinalStudy, second follow-up (BPS: 96/01). NCES2003-160. Washington, DC: National Centerfor Education Statistics.

U.S. Department of Education (n.d.). IntegratedPostsecondary Education Data System (IPEDS).Washington, DC: National Center for EducationStatistics.


○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Endnotes

1. Completion rates in this paragraph are based onthe authors’ calculations from the BeginningPostsecondary Student Longitudinal Study of1996-2001 (BPS96).

2. For more information on Achieving the Dream,see www.achievingthedream.org. The nine partnerorganizations are: American Association ofCommunity Colleges; Community CollegeLeadership Program, University of Texas-Austin;Community College Research Center, TeachersCollege, Columbia University; The FuturesProject, Brown University; Jobs for the Future;Lumina Foundation for Education; MDC Inc.;MDRC; Public Agenda.

3. The mainstream journals reviewed were TheJournal of Higher Education, Research in HigherEducation, The Review of Higher Education, Journal ofCollege Student Development and the NASPAJournal. Of these journals, the Journal of HigherEducation had the highest community college sharewith 13 percent.

4. Authors’ calculations from BPS96.

5. Fifty-three percent completed a bachelor’sdegree within nine years. This baccalaureate

completion rate is much higher than typicalcontemporary cohorts.

6. Some of the for-profit institutions emphasizeconvenience over engagement. For example, theUniversity of Phoenix has been known for usinglocations and schedules to facilitate the participa-tion of working adults and has chosen not topromote student involvement by developingextracurricular campus activities.

7. We thank Kate Shaw for this point.

8. Although Tinto addressed the selection issue, hedid not provide enough information to make ajudgment about the extent to which it is aproblem.

9. Thomas Brock, the principal investigator,reported that Kingsborough institutional researchindicated that these types of students are often theleast prepared and have low retention rates.

10. Personal communication with Thomas Brock,the principal investigator of the MDRC study. Thisstudy is scheduled to be published in early 2005after results from a second cohort of Kingsboroughstudents become available.


11. Authors’ calculation from the NationalEducation Longitudinal Study (NELS).

12. Since this study only examined students placedin developmental programs or courses, it could notmeasure the effect of developmental programscompared with placement into mainstreamcourses.

13. This type of information is not provided bythis study.

14. The study itself presents no evidence of theeffectiveness of the developmental programs at thefive chosen sites. Indeed, the sites were chosen noton the basis of their outcomes (which are notreported), but on the basis of the programs thatthey used.

15. For example, the levels of satisfaction arereported for one point in time, but the reader doesnot know what those rates were before theintroduction of the college-wide reforms.Enrollments and graduation rates improved, but wealso know that recruitment policies at the collegechanged, resulting in a larger number of tradi-tional-aged students; thus, the changing character-istics of the student body may have affected someof the results. Additionally, the study did notindicate how “student success” was defined andwhat control variables were used in the cohortanalysis.

16. The MDRC study discussed earlier takes thisapproach.

About the authors○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○


Thomas R. Bailey is the George and Abby O’Neill Professor of Economics and Education in the Depart-ment of International and Transcultural Studies at Teachers College, Columbia University. He holds adoctorate in labor economics from the Massachusetts Institute of Technology, and is an expert on theeconomics of education, educational policy, community colleges and the educational and trainingimplications of changes in the workplace. In 1996, with support from the Alfred P. Sloan Foundation,Bailey established the Community College Research Center (CCRC) at Teachers College, which is nowthe leading independent research center devoted to the study of community colleges. It conducts a largeportfolio of qualitative and quantitative research based on fieldwork at community colleges and analysis ofnational- and state-level data sets. Since 1992, Bailey also has been the director of the Institute onEducation and the Economy at Teachers College, an interdisciplinary policy research center that focuses itsattention on the interaction between education and the economy. His articles have appeared in a widevariety of policy-oriented and academic journals, and he has authored or co-authored several books on theemployment and training of immigrants and the extent and effects of on-the-job training.

Mariana Alfonso is a postdoctoral research associate in public policy at Brown University’s TaubmanCenter for Public Policy and American Institutions. She holds a doctorate in economics and educationfrom Teachers College, Columbia University, and a bachelor’s degree in economics from UniversidadNacional de Córdoba, in Argentina. Her research focuses on issues of access and attainment at communitycolleges. Before coming to Brown, Alfonso worked as a senior research assistant at CCRC, where sheparticipated in projects that analyzed the impact of occupational postsecondary education on degreeattainment, and issues of higher education access and attainment by low-income and minority students.

This monograph is based on research done to support a partnership effort called Achieving theDream: Community Colleges Count. Achieving the Dream is a national initiative to increase thesuccess of community college students, particularly those in groups that have been underserved in

higher education.

It is a large-scale, multi-year initiative whose creation and implementation are being shared by severalorganizations committed to improving education and social policy. Achieving the Dream is funded byLumina Foundation, managed by MDC Inc., and includes the following partner organizations:

■ American Association of Community Colleges■ Community College Leadership Program, University of Texas-Austin■ Community College Research Center, Teachers College, Columbia University■ The Futures Project, Brown University■ Jobs for the Future■ MDRC■ Public Agenda

In the coming years, the initiative expects to attract other organizations, including additional funders.

In its initial phase, 27 community colleges in five states are participating in Achieving the Dream. Thesecolleges all have made a commitment to improve the success of underserved students on their campuses, touse data to guide their decisions, and to broadly share the lessons they learn so that students in otherinstitutions and states can benefit.

For more information on the initiative, visit www.achievingthedream.org.

○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○


Also available from Lumina Foundation for Education○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Powerful Partnerships: Independent Colleges

Share High-impact Strategies for

Low-income Students’ Success

Richard Ekman, Russell Garth and John F. Noonan, Editors

October 2004

Fifty Years of College Choice: Social, Political and

Institutional Influences on the Decision-making Process

Jillian Kinzie, Megan Palmer, John Hayek, Don Hossler,

Stacy A. Jacob and Heather Cummings

September 2004

When Saving Means Losing: Weighing the Benefits of

College-savings Plans

Roberto M. Ifill and Michael S. McPherson

July 2004

Expanding College Access:

The Impact of State Finance Strategies

Edward P. St. John, Choong-Geun Chung, Glenda D. Musoba,

Ada B. Simmons, Ontario S. Wooden and Jesse P. Mendez

February 2004

Unintended Consequences of Tuition Discounting

Jerry Sheehan Davis

May 2003

Following the Mobile Student; Can We Develop the

Capacity for a Comprehensive Database to Assess

Student Progression?

Peter T. Ewell, Paula R. Schild and Karen Paulson

April 2003

Meeting the Access Challenge: Indiana’s

Twenty-first Century Scholars Program

Edward P. St. John, Glenda D. Musoba,

Ada B. Simmons and Choong-Geun Chung

August 2002

Unequal Opportunity: Disparities in College

Access Among the 50 States

Samuel M. Kipp III, Derek V. Price and Jill K. Wohlford

January 2002

Hope Works: Student Use of

Education Tax Credits

Barbara A. Hoblitzell and Tiffany L. Smith

November 2001


Learning in the Fast Lane: Adult Learners’ Persistence

and Success in Accelerated College Programs

Raymond J. Wlodkowski, Jennifer E. Mauldin

and Sandra W. Gahn

August 2001

Debts and Decisions: Student Loans and Their

Relationship to Graduate School and Career Choice

Donald E. Heller

June 2001

Funding the “Infostructure:” A Guide to Financing

Technology Infrastructure in Higher Education

Jane V. Wellman and Ronald A. Phipps

April 2001

Discounting Toward Disaster: Tuition

Discounting, College Finances, and Enrollments

of Low-Income Undergraduates

Kenneth E. Redd

December 2000

College Affordability: Overlooked Long-Term

Trends and Recent 50-State Patterns

Jerry Sheehan Davis

November 2000

HBCU Graduates:

Employment, Earnings and Success After College

Kenneth E. Redd

August 2000

Student Debt Levels Continue to Rise

Stafford Indebtedness: 1999 Update

Patricia M. Scherschel

June 2000

Presidential Essays: Success Stories —

Strategies that Make a Difference at

Thirteen Independent Colleges and Universities

Allen P. Splete, Editor

March 2000

Are College Students Satisfied?

A National Analysis of Changing Expectations

Lana Low

February 2000

Fifty Years of Innovations in Undergraduate Education:

Change and Stasis in the Pursuit of Quality

Gary H. Quehl, William H. Bergquist and Joseph L. Subbiondo

October 1999

Cost, Price and Public Policy:

Peering into the Higher Education Black Box

William L. Stringer, Alisa F. Cunningham, with

Jamie P. Merisotis, Jane V. Wellman and Colleen T. O’Brien

August 1999

Student Indebtedness:

Are Borrowers Pushing the Limits?

Patricia M. Scherschel

November 1998

It’s All Relative: The Role of Parents in

College Financing and Enrollment

William L. Stringer, Alisa F. Cunningham,

Colleen T. O’Brien and Jamie P. Merisotis

October 1998


Lumina Foundation for Education, a private,independent foundation, strives to help peopleachieve their potential by expanding access and

success in education beyond high school. Throughresearch, grants for innovative programs andcommunication initiatives, Lumina Foundationaddresses issues surrounding financial access andeducational retention and degree or certificateattainment — particularly among underserved studentgroups, including adult learners. Focusing on these areas, the Foundation framesissues and explores new solutions through fact-basedresearch. Because we strive to be a credible andobjective source of information on issues affectinghigher education, Lumina Foundation encouragesoriginal sponsored research. The results of that research,and therefore the content of this and other LuminaFoundation® publications, do not necessarily representthe views of the Foundation or its employees. Believing that published research may have thelongest-term impact on higher education, theFoundation publishes and disseminates articles, researchreports and books. We prefer topics and approachesthat are more practical than theoretical, and whichemphasize pragmatic tools that will assist institutionsand public policy-makers.

Additional information about the Foundation’sresearch and grant-making programs may be obtainedfrom:

Leah Meyer AustinSenior Vice President for Research and [email protected]

Additional information about Lumina Foundation®

programs and communications may be obtained from:Susan O. ConnerExecutive Vice President for Impact [email protected]

Lumina Foundation for EducationP. O. Box 1806Indianapolis, IN 46206-1806800-834-5756www.luminafoundation.org

Lumina Foundation for EducationNew Agenda Series™

Susan O. ConnerExecutive Editor

David S. PowellDirector of Publications

Gloria Ackerson-SeatsAmy SchrammEditorial assistance

Natasha SwingleyDesign and production

Printed by SPG Graphics Inc.

Lumina Foundation for EducationNew Agenda Series™ is published periodically byLumina Foundation for EducationP.O. Box 1806Indianapolis, IN 46206-1806

First-class postage paid at Indianapolis, Ind.Copyright © 2005Lumina Foundation for Education, Inc.All rights reserved.

Send all name and address changes to:Lumina Foundation for EducationAttn: Amy SchrammP.O. Box 1806Indianapolis, IN [email protected]

NEW AGENDA SERIES™

January 2005

®

®

Documents

Paths to Persistence - ERIC · Paths to Persistence: An Analysis of Research on Program Effectiveness at Community Colleges NEW AGENDA SERIES™ Thomas R. Bailey and Mariana Alfonso