DOCUMENT RESUME

ED 375 227 UD 030 130

AUTHOR Jordan, Will J.; And Others

TITLE Exploring the Complexity of Early Dropout Causal

Structures.

INSTITUTION Center for Research on Effective Schooling forDisadvantaged Students, Baltimore, MD.

SPONS AGENCY Office of Educational Research and Improvement (ED),Washington, DC.

REPORT NO CDS-R-48PUB DATE Aug 94CONTRACT R117R90002NOTE 38p.

PUB TYPE Reports Evaluative/Feasibility (142)

EDRS PRICE MF01/PCO2 Plus Postage.DESCRIPTORS *Academic Aspiration; Black Students; *Causal Models;

Dropout Research; *Dropouts; *Educational Attainment;

Educational Research; Expulsion; Females; HispanicAmericans; Intervention; Males; *NontraditionalEducation; Prediction; Prevention; RacialDifferences; Reentry Students; Secondary Education;Sex Differences

IDENTIFIERS *African Americans; *National Education Longitudinal

Study 1988

ABSTRACTThis study analyzes NELS:88 data from a dropout

sample of students who were enrolled in the eighth grade in 1988 but

who were not enrolled in school in 1990. The data for this analysis

were collected in Spring 1990 to examine reasons for dropping out and

?lams for dropouts to resume their education. In both areas,

differences were found on race-ethnicity and gender. Concerning

reasons for dropping out, a larger percentage of white and Hispanic

dropouts cited school-related factors as a cause than did African

Americans; African American dropouts cited suspension and expulsion

more often than any other group. A significantly larger percentage of

male than female dropouts cited job-related factors; females cited

family-related reasons more often than did white females. The

overwhelming majority of dropouts did have plans for resuming their

education, but these plans differed by race-ethnicity and gender.

Both male and female white dropouts more frequently planned to take

equivalency tests; Hispanic adolescents favored attending alternative

high schools, and African American adolescents more often planned to

return to a regular high school to earn their diplomas. Six tables.

(Contains 18 references.) (Author/BJ)

***********************************************************************Reproductions supplied by EDRS are the best that can be made

from the original document.***********************************************************************

NuZ

U S DEPARTMENT OF EDUCATION

n nt EduratoonalResev,t, andehOh>vemere

MICATIONAL RE SOURCESINFORMATION

CI NUR IERiCi

10,s documenthen teen

reproduced as

chwed horn the De/sunof o,ganualto,,

hgmatmgchanges have

beer, made to enwove

reO,OdUr hOhdualthe

P,notS of wee ot op,hohs stated,nthsCor

men! do not neress.wyrepresent otbc,a'

OE RI Dos.100Or pilot r

Exploring the Complexity of

Early Dropout Causal Structures

Will J. Jordan

Julia Lara

James M. McPartland

Report No. 48

August 1994

C'ENTER OR "RESEARCH QN `EFF.ECTIVE:Silf604.I.ING.FOR. DISADVANTAGED

2BEST COPY AVAILABLE

National Advisory Panel

Beatriz Arias, College of Education, Arizona State University

Mary Frances Berry, Department of History, University of Pennsylvania

Anthony S. Bryk, Department of Education, University of Chicago

Edmund Gordon (Chair) Department of Psychology, Yale University

Ronald D. Henderson, National Education Association

Vinetta Jones, Executive Director, EQUITY 2000, College Board

Arturo Madrid, Director, Tomas Rivera Center

Heenan LaFontaine, Administration and Supervision,Southern Connecticut State University

William Julius Wilson, Department of Sociology, University of Chicago

Center Liaison

Harold Himrnelfarb, Office of Educational Research and Improvement

BEST COPY AVAILABLE

Exploring the Complexity of

Early Dropout Causal Structures

Will J. JordanJohns Hopkins University

Julia LaraCouncil of Chief State School Officers

James M. McPartlandJohns Hopkins University

Report No. 48

August 1994

Published by the Center for Research on Effective Schooling for Disadvantaged Students, supported as

a national research and development center by funds from the Office of Educational Research and

Improvement (OERI), United States Department of Education (R117R90002. The opinions expressed in

this publication do not reflect the position or policy of OERI, and no official endorsement should be

inferred.

Center for Research on Effective Schooling for Disadvantaged StudentsThe Johns Hopkins University

3505 North Charles StreetBaltimore, Maryland 21218

4

BEST COPY AVAILABLE

The Center

The mission of the Center for Research on Effective Schooling for DisadvantagedStudents (CDS) is to significantly improve the education of disadvantaged students at each

level of schooling through new knowledge and practices produced by thorough scientificstudy and evaluation. The Center conducts its research in four program areas: The Early andElementary Education Program, The Middle Grades and High Schools Program. theLanguage Minority Program, and the School, Family, and Community Connections Program.

The Early and Elementary Education Program

This program is working to develop, evaluate, and disseminate instructional programscapable of bringing disadvantaged students to high levels of achievement, particularly in thefundamental areas of reading, writing, and mathematics. The goal is to expand the range ofeffective alternatives which schools may use under Chapter 1 and other compensatoryeducation funding and to study issues of direct relevance to federal, state, and local policyon education of disadvantaged students.

The Middle Grades and High Schools Program

This program is conducting research syntheses, survey analyses, and field studies in

middle and high schools. The three types of projects move from basic research to useful

practice. Syntheses compile and analyze existing knowledge about effective education ofdisadvantaged students. Survey analyses identify and describe current programs, practices,and trends in middle and high schools, and allow studies of their effects. Field studies areconducted in collaboration with school staffs to develop and evaluate effective programs and

practices.

The Language Minority Program

This program represents a collaborative effort. The University of California at Santa

Barbara and the University of Texas at El Paso are focusing.on the education of Mexican-A-merican students in California and Texas; studies of dropout among children of recentimmigrants have been conducted in San Diego and Miami by Johns Hopkins, and evaluationsof learning strategies in schools serving Navajo Indians have been conducted by theUniversity of Northern Arizona. The goal of the program is to identify, develop, andevaluate effective programs for disadvantaged Hispanic, American Indian, Southeast Asian,

and other language minority children.

The School, Family, and Community Connections Program

This program is focusing on the key connections between schools and families andbetween schools and communities to build bet' rr educational programs for disadvantaged

children and youth. Initial work is seeking to provide a research base coNcerning the mosteffective ways for schools to interact with and assist parents of disadvantaged students andinteract with the community to produce effective community involvement.

Abstract

This study analyzes NELS:88 data from the dropout sample collected inSpring 1990 to examine reasons for dropout and plans for dropouts to resume their

education. In both areas, differences were found on race/ethnicity and gender.Concerning reasons for dropping out, a larger percentage of white and Hispanicdropouts cited school-related factors as a cause than did African Americans; AfricanAmerican dropouts cited suspension/expulsion more often than any other group. Asignificantly larger percentage of male dropouts than female dropouts cited job-related factors; females cited family-related reasons more frequently than did males.Hispanic and African American females cited family-related reasons more often than

did white females. The overwhelming majority of dropouts did have plans forresuming their education, but these plans differed by race/ethnicity and gender. Bothmale and female white dropouts more frequently planned to take equivalency tests;Hispanic adolescents favored attending alternative high schools, and AfricanAmerican adolescents more often planned to return to a regular high school to earn

their diplomas.

iii

6

Introduction

There has been an outgrowth of concern aboutthe dropout crisis among middle and high schooladolescents. At a time when most new jobopportunities demand not only the Three Rs, butalso an ever-increasing level of technicalcompetencies and communication skills, alongwith educational credentials which attest to these,dropping out of school prior to earning a diplomahas serious repercussions. Recent research hascentered around contrasting dropout rates acrossracial/ethnic and gender groups, and discoveringpossible determinants of dropouts (Alsalam et al.,1993; Fine, 1986). Typically, these researchefforts have found that adolescents who areminorities and poor drop out more frequentlythan their middle class, white counterparts(General Accounting Office, 1986; Mann, 1986).

However, little educational research has investi-gated explicitly whether minority adolescents,particularly African Americans and Hispanics,exper:ence greater complexity in causal structurewhen they drop out. That is, although we havegaincl some knowledge of broad patterns ofdropout behavior, the research literature has notafforded much guidance in understanding howthe various reasons for dropping out differbetween groups. To effectively provide interven-tions for adolescents at risk of dropping out aswell as to establish dropout recovery programsfor those who have already interrupted theirschooling, social science must better informpolicymakers and practitioners on the complexpattern of reasons surrounding the process ofdropping out.

In the present study, preliminary evidencesuggests that Hispanic and African American

1

adolescents indeed not only have higher dropoutrates, but the causes of their dropping out aremore complex than for whites. Moreover,African Americans, Hispanics and whites, aswell as males and females, seem to differdrastically from one another regarding theunderlying reasons for dropping out.

Although researchers have long engaged inidentifying causes for dropping out (Cervantes,1965; Rumberger, 1983), systematic insights arenot readily available on whether and howdropouts in different raciaUethnic groups aredriven by different causal processes. The lack ofattention to the causal complexity issue is

understandable partly because the prevailingtheoretical orientation generally favors oneparsimonious model that applies to all groupsand partly because the usual methodologicalapproach relies exclusively on indirect inferenceof separate factors. Evidence on the complexityof dropout causal structures drawn from thisstudy will fill the knowledge gap created by thesedeficiencies.

In addition, there are practical implications. IfAfrican American and Hispanic adolescents areexperiencing not only higher dropout rates butalso significantly greater degrees of causalcomplexity, then the actual difficulties involvedin dealing with both the quantitative and qualita-tive aspects of the problem may have beenseriously underestimated by researchers andpolicymakers alike. In this case, the currentstrategies for ameliorating the dropout crisisshould certainly be reassessed and reevaluated.

The Data

The data for this study are drawn from theNational Education Longitudinal Study of 1988(NELS:88), administered by the National Centerfor Educational Statistics. NELS:88 provides arich source of information on the process ofdropping out of school as it takes place from theeighth grade onward. These data allow us toexamine school context, student engagement.aspirations, achievement and future plans, andthe broader contextual factors affecting students.The NELS:88 design employed a two-stage,stratified, random sample of 25,000 eighthgraders in some 1,(M schools, who are followedup at two-year intervals. In addition to thestudent data file, two of the respondent's teach-ers, the school principal, and parents of therespondents were administered detailed question-naires. A separate dropout component wascreated for students who participated in the baseyear but were not enrolled in school in subse-quent waves.

The present study analyzes data from the dropoutsample collected in Spring 1990. We refer torespondents in this sample as the early dropoutsbecause they left school at some point betweenthe eighth and tenth grades as opposed to duringlater secondary grades. Because adolescentscannot legally drop out in many states until age16, most of the existing research literaturefocuses primarily on older adolescents, thosenearing the end of high school. Although mostdropouts do, in fact, leave school from highschool, important evidence for understanding thecumulative process of dropping out is missedwhen we omit younger adolescents. Theseyounger adolescents, or early dropouts, aretherefore the focus of our investigation.

The NELS:88 dropout data file contains informa-tion on 1,000 students who were sampled in the

2

eighth-grade cohort, but were not enrolled inschool in 1990. According to the user's manual,75c/( of the respondents completed the full, orslightly modified, version of the dropout surveyinstrument, while the remaining 25% completedthe abbreviated version. Only critical questionswere asked in the abbreviated versions andadditional weights were created to compensatefor nonresponse bias.

Early Dropouts

We view dropping out of school as a cumulativeprocess arising out of a series of events andexperiences which impact on youth in theirtransition into adolescence. These socialinfluences cause the adolescent to becomealienated and disengaged from school, andinevitably some drop out. Thus dropping out isnot an episodic event, occurring as result ofshort-term or whimsical displeasure with treat-ment in school. Nor is dropping out simply anirrational decision made as result of a combina-tion of economic hardship and prior schoolfailure. Although this may explain a part of thephenomenon, the process of dropping out is morecomplex.

Educational research has identified at least twocategories of influences of dropping out -- bothpush and pull effects (Rumberger, 1987;Wehlage & Rutter, 1986). Push effects arefactors located within the school itself, whichnegatively impact the connection adolescentsmake with the school's environment and causethem to reject the context of schooling. Thisrejection, on the part of students, may manifest indisruptive behavior, absenteeism, or a cessationof academic effort. In sanctioning this kind ofbehavior by suspending, failing, or issuing poorgrades to students, schools sometime producecontinued failure, rather than getting students oncourse to success as they intend.

In other words, when troubled adolescents cometo realize that they have fallen into a hole thatlooks too deep to climb out of, many stop tryingaltogether. For example, if a student reaches themaximum number of absences that automaticallymeans a failing grade, he or she may easily loseany incentive for continuing to come to schoolfor the remainder of the term. Further, long-termabsences may evolve into actual dropout througha natural progression if fail-safe mechanisms arenot put in place to guard against it. In addition,students who are often suspended begin to seethemselves as incapable of succeeding in school-- and the school affirms this with each newsuspension. For all intents and purposes, in theworst cases, adolescents are evicted from schoolin that they have slipped so far behind anddeveloped such an adversarial view of schoolingthat the probability of getting hack on track isquite small.

Thus, internal factors, which may he structural,contextual, climate-related, and/or individual-ized, cause certain at-risk students to view schoolas an unwelcoming place, and they becomealienated. Over time the degree of disengage-ment increases for these students, they put outless and less academic effort, and they eventuallydrop out.

Pull effects take into account that school ismerely one aspect of the complex adolescentsocial milieux. Although compulsory educationoccupies a great deal of time and takes consider-able effort on the part of the adolescent, morecompelling external factors may usurp anemphasis on schooling. The family, neighbor-hood peer groups, community organizations, andreligious, legal, and health institutions often playa critical role in the stability and development ofthe adolescent. Therefore, conflict between theeducational objectives of schools and socialforces external to the school may interfere with

3

student success -- for example, having to care forfamily members, needing to hold down a job, orbecoming pregnant may have a negative impacton school performance and in some instanceslead to dropout.

Although the seeds of dropping Out may be sownearly in one's educational career, the problem ismanifested in high school. NELS:88 data showthat about 7% of dropouts leave school whilethey are in the eighth grade, which is typicallythe final year of middle school. However, in theninth grade -- usually year one of high. school --38% of the dropouts leave school. Severalresearchers have posited that transition years arecritical points in the lives of children andadolescents (Wheelock & Dorman, 1988), andthese data add empirical support. But thedifficulties caused by transition and adjustmentdo not necessarily have an immediate impact onadolescents. As discussed above, dropping out isa process which stems from an embroidery ofadolescent experiences -- as a consequence ofseveral conditions, some adolescents become leftbehind in the transition from the middle grades tohigh school and gradually drop out. Themajority of the dropouts in the present study,some 55%, departed in the tenth gradegenerally their second high school year.However, many of these adolescents may havefelt alienated, disengaged, or simply unable tokeep up with increasing demands during theirfirst year and prior, but it took time for them toreach the decision to dropout.

It is difficult to say whether academic failure issymptomatic of losing interest in school anddropping out, or causes adolescents to loseinterest and drop out. Educational research hasnot sufficiently explained the cause-and-effectrelationship surrounding dropping out and schoolperformance (Pallas, 1984). How best can wedescribe the sequence of events? For example,

9

do students first lose interest in school, becomeless motivated to do academic work, receivepunishment or few rewards for their poorperformance, receive no intervention from theschool, family or community, and then drop outafter falling too far behind? Or, conversely, dothey first try hard to succeed in school, becomediscouraged because they cannot keep upacademically, lose interest after they have fallentoo far behind, and then drop out in despair?Perhaps both of these occur to some degree.However, we do know that most dropouts are notmaking the grade when they leave school. (Orr,1987; Peng, 1983). Eighty-one percent of thedropouts in our study reported that they did notpass the last grade they attended in school.

Outline of Analyses

Below we examine two aspects of the dropoutcrisis, as it has affected those adolescents whoalready left. school at some point between theireighth and tenth year. First, we investigate thevarious reasons different adolescents give fordropping out. We explore patterns of differencesacross adolescent subgroups defined by their

race/ethnicity and gender. The race/ethnicitygroups compared are Hispanics, AfricanAmericans, and non-Hispanic whites. There areseveral steps to our analysis of reasons fordropping out. We begin the investigation bydeveloping composite scales, or summaryfactors, of the different types of reasons adoles-cents give for leaving school. Next, we tabulatethese scales by race/ethnicity and gender sub-groups in order to assess differential patterns ofcorrelates. Next, SES background is controlledin order to examine the degree to which the zero-order patterns are influenced by social classdifferences. Finally, we consider possibleinteraction effects among race/ethnicity, genderand SES.

Perhaps, the least that can be said about droppingout is that it interrupts one's schooling; however,it is not necessarily permanent, nor is it the endof the road. The second part of this studyexamines ethnic/racial and gender differences inthe plans for further education of early dropouts.Here, we are primarily concerned with intentionsof dropouts to return to school, take equivalencytests, attend alternative schools, or to notcontinue their formal education.

I. Popular Reasons for Leaving School

Overview of Dropout Reasons

Early dropouts were given a list of 21 reasonstypically cited for leaving school and asked toselect all that applied to them. These reasonscovered a variety of areas, includingschool-related variables, peer influences, familyconstraints, and accelerated or adult role transi-tions (such as having children and/or needing to

4

work). The reasons reflected both push and pulleffects, the two notions on dropping out (whichare not mutually exclusive) that have arisen toaddress the locus of factors influencing anadolescent's decision to quit school.

On the one hand, it is believed that the overridingcauses of dropping out are located outside of theschool's environment -- i.e., within the family,

10

community, and/or among peer groups(Rumberger, 1983). By definition, these externalfactors are difficult for the school to manipulate,and sometimes they are completely beyond theschool's control. Thus proponents of this viewtend to believe that the schools can do little tostem the tide of pervasive student departurewithout help from outside agencies.

On the other hand, some argue that the realcauses of dropping out are inherent in the schoolcontext itself (Wehlage & Rutter, 1986). Theseresearchers assert that a broad range of structuraland contextual conditions within the school oftenmake it difficult for some students to succeed; asa consequence, schools generally do not serve allstudents well. These structural and contextualfactors include, for example, administrativesupport for teachers, teacher effectiveness,morale and expectancies, student morale andacademic investment, class size, parental in-volvement, fiscal inequities, physical appearanceof facilities, and the like. Because of thedynamics of internal school conditions, manyyoung students become alienated and disengagedfrom the learning process and lose interest inschool. Furthermore, for these adolescents,school becomes viewed as a place where theirshortcomings and failures become increasinglyreinforced, both by teachers and fellow class-mates. As a result of the disconnection, contin-ual failure, and bad feelings and images ofschool, many students simply quit.

While both external (pull) and internal (push)factors play a role in dropping out, the verdict onwhich is most critical is still out. However, inthe present study, within-school factors (as

reported by the dropouts themselves) appear tobe most important for predicting their decision todrop out. This pattern is consistent across allracial/ethnic and gender groups. Presented inTable 1 are weighted tabulations of adolescents'

5

reasons for dropping out. Respondents were toldto mark all- that apply. We recoded each of thesevariables so that "1" indicates an item waspicked, and "0" represents that it was not.Hence, the means are equivalent to weightedpercentages. As shown in Table 1, the mostfrequently cited reasons offered by dropouts forleaving school were related to contextual factorswithin the school itself, as opposed to externalinfluences. For example, the majority of students(51%) indicated that one of the reasons theydropped out was because they simply did not likeschool. Forty-four percent said "I was failing inschool," while 34% indicated "I couldn't getalong with teachers." In addition, aboutone-third of the respondents reported they couldnot keep up with their schoolwork, andone-quarter of them reported that they did notfeel they belong in school.

In contrast, these data show that external basedreasons for dropping out were reported much lessoften. Only 7% of the respondents indicated thatthey needed to help support their families. Sixpercent wanted to have a family of their own.Seventeen percent and 18%, respectively, "had toget a job" and/or "found a job." Twelve percentof the adolescent females indicated that theydropped out because of pregnancy, and 12% ofall teenage dropouts cited parenthood as a reasonfor leaving school.

Insert Table 1 about here

Although descriptive of the various reasonsadolescents give for dropping out, these percent-ages alone do not identify causal processes acrossrace/ethnicity and gender. However, usingvarious comparison techniques, which areelaborated below, our analyses suggest thatcausal structures that affect dropping out differ

ii

across groups. For example, factors that influ-ence adolescent Hispanic males to drop out areconfigured differently than those that influenceAfrican American males or Hispanic females. Insome cases, there are severe gaps betweengroups; in others the differences are subtle. Inaddition, as alluded to earlier, the relationshipsbetween the reasons themselves are at timesconfounding. For example, adolescents whoindicated that they could not keep up with theirschoolwork conceivably grew to that point as aresult of not liking school and not being able toget along with their teachers. Thus theirdisinterest and disengagement led to an unwill-ingness to put forth academic effort which, inturn, caused them to fall behind in schoolwork.This two-way process can lead to a cycle offailure which sometimes ends in total disengage-ment from school -- dropping out.

Some adolescents, particularly African Americanand Hispanic males, dropped out because theirschools lacked the capacity and resources tohandle their disruptive conduct. When a studentis suspended or expelled, the school is sending astrong message that his or her behavior isunacceptable and has no place in school(Wheelocl & Dorman, 1988). However,sanctions arc sometimes internalized by adoles-cents to the degree that their self-concept isaffected. Over time, certain students disciplinedtoo often cannot separate the behavior beingsanctioned from their core identity. Theseadolescents may come to believe that who theyare and how they conduct themselves is incom-patible with how things work in school. As aresult, they grow to believe that they have noplace in school. Among the NELS:88 respon-dents, 16% reported that one of the reasons theydropped out was because of too frequent suspen-sions, and 13% said that they were expelled.

6

Facilitating the Analyses: Scale Construction

We used the factor analysis procedure tocategorize the various dropout reasons intological measurement scales. Combining the 21items into scales was done for various method-ological and intuitive reasons -- one of which isthat scales generally provide stronger, moreaccurate measures than single dichotomousvariables. Moreover, putting together multipleitems eliminates the problem of having limiteddependent variables. Using varimax rotation,seven composite scales were extracted. Thesescales ranged in size from two items to fiveitems. Two-item scales were kept and includedin the analysis primarily to explore the full rangeof reasons for dropping out. The reliabilities ofthe scales also varied, but the mean alpha(excluding Factor 7) was roughly .60. We didnot, however, discard any factors even thoughsome reliability alpha levels were smaller thandesired. The rationale for keeping all of thescales is that in most cases, the individual itemsfit together intuitively. Again, our primary aimhere was to condense the variety of commonreasons for dropping out int( a smaller numberof useful, stronger measures.

A summai of the factor analysis results isdisplayed hi Table IA. The labels we attached tothe extracted factors once we created thecorresponding scales were: FAMILY 1; FAM-ILY 2; SCHOOL; JOB; SAFETY; EXPELLED;and MOBILITY. Once we identified theindividual items, a simple summative formulawas used to compute the scales. The followingsections describe each of these in some detail.

Scale,. (%Item, + % Itemh +...+ %Itcmk)N of Items

Family-Related Reasons. A confounding set ofvariables was extracted within the category offamily-related reasons for dropping out. Basedon the factor analysis statistics, it was difficult toseparate reasons related to the respondent'simmediate family (parents and siblings) fromreasons related to their own or potential family(children and spouses). To resolve this, wecombined both items having to do with therespondent's own family and his or her immedi-ate family into a single scale. A gender-specificitem was an added wrinkle in that, obviously,males were not asked whether they werepregnant. To address how this item would playout in the analysis we combined and deletedsome items in order to create two new scales.The end result was the creation of two scales, onefor males, the other for females, which werecomposites of family-related reasons for drop-ping out; we labeled these FAM and FAMFEM.

Insert Table IA about here

As shown in Table 1A, FAM I is a four-itemscale comprised of F1D6G, "(For females only)I was pregnant;" F1D6H, "I became a

mother/father of a baby;" F1D6I, "I had tosupport my family;' and F1D6N, "I had to carefor a member of my family." The first two itemsrefer strictly to the respondent's own or potentialfamily, while the latter two can go eith.t. way.FAM2 has two items, F1D6F "I wanted to havea family," and F1D6S "I got married or plannedto get married." These scales (FAM1 andFAM2) were combined and modified in order tocompute two new measures of family influencessensitive to gender.

School-Related Reasons. The school scaleconsists of items related to the degree to whichadolescents dropped out as result of feelingdiscontented, alienated, disengaged, or incompe-tent in school. Five items loaded into this scale:F1D6C, "I didn't like school;" F1D6E, "I couldn'tget along with teachers;" Fl D6P, "I felt I didn'tbelong at school;" FID6Q, "I couldn't keep upwith my schoolwork;" and F1D6R, "I was failingschool." We labeled this scale SCH. Two ofthese items, getting along with teachers and senseof belonging, loaded less strongly into this scalethan the other three. However, combined, thereliability alpha was .609.

Work-Related Reasons. In addition to family-related responsibilities, the perceived need tofind work is among the key accelerated roletransitions which determine dropping out ofschool (Pallas, 1984). The JOB scale wascomprised of three items: F1D6A, "I had to geta job;" FID6B, "I found a job;" and F1D6U, "Icouldn't work and go to school at the same time."The reliability alpha was .628.

Safety, Suspensions, Mobility and FriendshipReasons. After the initial factors were extracted(i.e., family, school and job-related reasons), thereliability of subsequent scales weakenedsomewhat with each iteration, as is usually thecase in factor analysis.

The final three scales, SAFETY, EXPELLEDand MOBILITY, contained only two items eachand possess borderline to small reliabilities. Thesize of any reliability, however, is affected by thenumber of items in the scale, so these results arenot surprising. Notwithstanding the limitedreliability alpha levels, as presented in Table 1 A,the factor loadings suggest that there is indeedinterrelatedness among items in the scales.

7

13

The items which make up SAFETY arc: F1D6E,"I couldn't get along with other students" andF I D6K, "I did not feel safe at school." Althoughthis scale measures respondents' reasons relatedto within-school peer context and climate, thesetwo variables did not load on the school-relatedscale outlined earlier. The factor loadings for thecomponents of the SAFETY scale, F1D6E andFID6K, were .7730 and .7538, respectively.These results suggest that an adolescent's senseof security, within the context of school, can beviewed as a function of an interaction betweentheir general sense of security and getting alongwith other students.

As discussed above, a number of adolescentsdrop out, either directly or indirectly, as result ofadministrative sanctions against their behavior.The most severe disciplinary measures which canbe taken by a school are to suspend or expel thestudent. EXPELLED is an important scale,which is also related to school context. F2D6J,"I was suspended too often" and F1D6O, "I wasexpelled from school" are the items whichcomprise EXPELLED. The factor loadings forthis scale are .7670 and .8214, respectively.

The final scale, MOBILITY, is a two-item scalewhich combines Fl D6L, "I wanted to travel" andF1D6T, "I switched schools and didn't like mynew school." While we know that the term"mobility" may not capture what the sum of theseitems is about, they each have to do with wantingor having to make physical transitions. Thefactor loadings are .6855 and .6608.

One item, F1D6M, "my friends had dropped outof school" did not load on any of the abovescales. However, this variable is importantbecause it is assessing the role of peer andfriendship influences in the process of droppingout. Thus, in this case, we used this item as an

8

individual dichotomous variable in the analysisand labelled it FRIENDS.

Analyses of Race/Ethnicity and GenderDifferences in Reasons for Dropping Out

After constructing the above scales, we con-ducted a breakdown of them by race/ethnicityand gender. Table 2 displays the means andsample sizes for the various groups by each ofthe scales. For males and females, the scales arelisted in descending order, from the mostfrequently cited category of reasons. One clearpattern is that across each of the raciallethnic aswell as gender groups, school-related factors arethe most cited reasons for early dropout. Thetotal mean for males is .429 and for females is.312. However, white adolescents appear morelikely than African Americans or Hispanics toreport school-related reasons. The mean SCHfor white males was .474, as compared to .384and .315 for Hispanics and African Americans,respectively. Females showed a similar pattern

white female adolescents were more likely toreport school-related reasons than both Hispanicand African American females.

Insert Table 2 about here

In this initial breakdown, family-related reasonsfor dropping out showed sizable gender differ-ences. Females reported family reasons as adistant second to school-related reasons. Its totalmean was .149. However, for males, family-related causes were cited least often. This is trueparticularly among white males, for whom themean was .038. Among all groups, AfricanAmerican females reported family-relatedreasons most often, with a mean of .181.

14

Hispanic and white males reported job-relatedreasons for dropping out second to school-relatedreasons, with means of .241 and .233, respec-tively. However, for African American males,the second biggest reason was being suspendedtoo often or expelled from school. In fact,African American males cited suspen-sion/expulsion as one of the reasons why theydropped out of school far more often than anyother group in the study. The mean for AfricanAmerican males was .274, as compared toHispanics and whites, whose means were .152and .190, respectively, and all females, whosetotal mean was .079. In addition, AfricanAmerican females ?, more likely than otherfemales to cite suspension/expulsion as a reasonfor dropping out. The mean for AfricanAmerican females on EXPELLED, .146 ascompared to .072 and .064 for Hispanic andwhite females, respectively, is surpassed only byAfrican American and white males.

In summary, a rank ordering of the variousreasons for leaving school shows that the biggestdifferences are gender differences. However,there are several ethnic/racial differences as well.The primary three reasons adolescent males citefor dropping out are school-related, job-related,and suspension/expulsion. For females the orderis school-related factors, followed by fam-ily-related and job-related factors. There are alsoimportant ethnic/racial differences within andacross each gender group. One of the morestriking results is that the mean for AfricanAmerican males on EXPELLED was consider-ably higher than that of both Hispanic and whitemales. In addition, African American males werealso more likely to report having friends whohave dropped out as one of the reasons why theydecided to do likewise. Similarly, amongfemales, African American females' mean valuesfor EXPELLED and FRIENDS were notablyhigher than the means of Hispanic and white

9

females. Also, white females had higher meansthan Hispanics and African Americans on job-related reasons for dropping out.

Significance Tests. We conducted a series of t-tests to examine the significance of emergentpatterns of the initial phase of the analy.iis.Again, male and female dropouts were separatedand, within gender groups, we compared each ofthe race/ethnic groups. A summary of the t-testresults are shown in Table 2A.

Insert Table 2A about here

The results of the t-tests largely affirm our earlierobservations. For example, family-relatedreasons, for males, were cited significantly moreoften by Hispanics and African Americans thanby whites. For the Hispanic/white comparison,t=2.25 and p<.05; for the African Ameri-can/white comparison, t=2.80 and p<.01. Forfemales, the sole significant difference onfamily-related influences existed betweenHispanics and whites, t=1.97; p<.05 -- that is,Hispanic girls were much more likely than whitegirls to cite family-related reasons. Although themean for family-related reasons for AfricanAmerican females was higher than for bothwhites and Hispanics, it was of borderlinesignificance, p=.066.

Also, the t-test supported the earlier results thatwhites were more likely than African Americansand Hispanics to report school-related factors fordropping out. The largest difference in

school-related reasons was between AfricanAmerican and white males (t=-4.43; p<.01). Inaddition, the Hispanic/white t statistic was -2.66,and also significant. Hispanic and AfricanAmerican males, however, did not diff,r signifi-candy on school-related reasons for dropping

15

out. There was a similar pattern of results amongfemale dropouts. Hispanic females weresignificantly less likely than whites to citeschool-related reasons, t=-2.84 and p<.01, butthere was no real statistical difference betweenHispanics and African Americans. In thecomparison between African American and whitefemales, the results were t=1.94; p<.05.

Evidence from the t-tests also supports thefinding that African Americans, generally, aremore likely than Hispanics and whites to reportsuspension/expulsion as a reason for earlydropout. This is true among both sexes, but thelargest gap exists between African American andwhite girls. The t value in this comparison was2.77; p<.01. The mean for African Americanfemales was also higher than for Hispanicfemales on EXPELLED, t=-1.88, and thisrelationship was close to significance, p=.061. Asimilar pattern was discovered among males inthe study. The mean for African American maleson EXPELLED was larger than that of bothwhites and Hispanics (.274, as compared to .190and .152), and these differences, as shown inTable 2A, were statistically significant.

The African American/white females compari-son showed only between-group differences inthe influence of friends. African Americanfemales reported at a significantly higher ratethan white females (t=2.03; p<.05) that one ofthe reasons they dropped out is because theirfriends had dropped out. Although AfricanAmerican females reported this reason moreoften than Hispanics as well, the differencebetween means was not statistically significant(t=1.73; .085).

Comparisons Controlling on SES. Athoughthe descriptive analyses discussed above cast

10

some light on differences in reasons for droppingout across racial/ethnic and gender groups,critical background factors were not taken intoaccount. The data used in our analyses are basedon a sample of adolescents who have alreadydropped out of school. Previous studies havereported a connection between SES and droppingout (Rumberger, 1983, 1987) -- more specifi-cally, that lower social class adolescents aredisproportionately represented among dropouts.This finding is undergirded in the present studyusing NELS:88 data. We found that 82% of allearly dropouts, those who left school betweengrades eight and ten, belong to families belowthe mean in SES. Table 3 depicts a breakdownof SES means by race/ethnicity and gender. TheSES measure is a composite variable comprisedof several items, including family income,parents' education, and parents' occupation.

Insert Table 3 about here

Not surprisingly, each racial/ethnic group ofdropouts was between two-thirds to one fullstandard deviation below the mean SES.However, white dropouts, both males andfemales, were higher SES than Hispanics andAfrican Americans. Our t-test results examiningSES differences across groups are at the bottomof Table 3. Among females, no significantdifferences between Hispanics and AfricanAmericans were apparent. However, generally,there were significant differences in the His-panic/white and African American/whitecomparisons of SES -- in each case, whites hadhigher means. However, given the smallvariances in SES between racial/ethnic groups, asdiscussed below, the introduction of SES into theanalysis has only modest importance on thepattern of relationships outlined earlier.

The Core Analyses of Dropout Reasons

in the core analyses we used multiple regressionto further examine relationships and interactionsbetwem race/ethnicity and gender on the variousreasons adolescents report for dropping out. Abattery of ;egression equations were run investi-gating each of the composite reasons scales,together with SES, a dichotomous variable sex,and two interaction factors (i.e., the product ofrace/ethnicity and sex, and the product ofrace/ethnicity and SES).

The regression results showed the differencesindependently due to race/ethnicity, gender andSES, controlling on each of the factors collec-tively, as well as interactions between them(Table 4). The coefficients for race/ethnicity,gender and SES were drawn from regressionequations that included these three measures aspredictor variables for each of the dropoutreasons scales. The interaction coefficientsinvolved running separate analyses.

Insert Table 4 about here

Summary of Results. The final results supportseveral of the patterns discussed above and pointto others that were undetected previously. As inthe descriptive analysis, family-related reasonsfor dropping out were much more frequentlycited by females than males (Table 4). Inaddition, African American and Hispanic maleswere somewhat more likely than their whitecounterparts to report family-related reasons fordropping out. Although the standardizedregression coefficients (betas) were small (.078and .092), both were at p<.05. There was nodiscernible difference, however, betweenHispanic and African American dropouts inreporting family-related reasons.

Job-related reasons for dropping out also varied.The source of variation was primarily betweengender and not so much between racial/ethnicgroups. In each of the analyses involving JOB,males and females showed significant differ-ences. Males were, in fact, more likely thanfemales to report job-related reasons, even aftercontrolling on SES. In the African Ameri-can/white comparison, the beta for sex was .179,p<.01, and for the Hispanic/white and His-panic/African American analyses the betacoefficients were .189 and .214 respectively,both significant.

In regard to school-related factors for droppingout (e.g., not liking school, not getting alongwith teachers), the patterns that emerged out ofthe initial descriptive analysis largely remainedafter background controls were added to theequation. Underscoring the evidence of theabove t-tests, regression results indicated thatwhite males cited school-related factors moreoften than any other group. The results compar-ing African Americans and Hispanics to whites(beta=-.161 and beta.-.103) both indicate thatschool-related factors influenced white adoles-cents more heavily (Table 4). In addition, theregression coefficients for sex in these sameequations, .190 and .209, are each significant,indicating gender differences as well -- males,more often than females, cite school-relatedreasons for dropping out.

Mann (1986) found that African Americanadolescents were suspended three times moreoften than their white counterparts. Similarly,according to Garibaldi (1992), African Americanmales are disproportionately represented amongstudents who have poor academic performance,are retained in grade, and are suspended andexpelled from school -- almost every categc ry of"academic failure." The present study alsosupports the findings of these researchers. Our

results suggest that African Americans and malesare significantly more likely than whites andfemales to report frequent suspension or expul-sion as one of the reasons for dropping out. Thisfinding, too, holds after SES is controlled (Table4). However, although African Americans werealso more likely than Hispanics to cite suspen-sion/expulsion, the difference hetwee.n these twogroups was not significant.

Last, the effect of friendship circles and peergroups on the dropout problem is also a criticalissue (Fine, 1986). The pattern of results showedthat African Americans are slightly more likelythan whites and Hispanics to report dropping outbecause their friends had dropped out (Table 4).Again, the regression equation for AfricanAmerican v. white adolescents on FRIENDSresulted in a .073 beta, along with mild signifi-cance, p=.056. The gap between Hispanics andAfrican Americans on FRIENDS was muchlarger than the African American v. whitedifference. Here, the results were beta=-.116;p<.05. There was no significant differencebetween Hispanics and whites on this factor.

The Effect of SES. Because dropouts generallywere concentrated well below the mean in SES,we did not expect SES to explain much of thevariability in early dropout reasons beyond whatis already explained by race/ethnicity and gender.It is useful, however, to observe that controllingon SES in these sets of equations does not alterthe pattern of relationships discussed above(Table 4). Both the family and job-relatedcomposite variables have negative signs on SES-- meaning that dropouts who cited these reasons

12

are among the lower SES levels. However,statistical significance is only evident for the SESdifferences on family-related reasons fordropping out (Table 4).

Interactions. There were few interaction effectsfound in our analyses. As presented in Table 4,the only significant interactions involvedrace/ethnicity-by-SES (for school-related reasonsacross AA/W and H/AA groups, on family-related reasons in the H/AA comparisons, andon suspension/expulsion in the AA/W compari-sons). In each of these equations, SES was notitself significant as a main predictor variable. Inview of these interaction results, we conductedan examination of cross-tabulations of SES byrace/ethnicity for each of the dropout reasonswhich did not produce consistent and interpret-able patterns. The sample sizes in many cells ofthis cross-tabulation were too small to giveconfidence to further interpretations (Table 5).We therefore conclude that the principal findingsof our analyses are the ones described above forthe main effects of race/ethnicity differences andsex differences on dropout reasons. Our evi-dence suggests that SES background and interac-tion effects appear to matter little in predictingand explaining the proximate reasons whyadolescents drop out. However, these back-ground and interaction variables, such as socialclass and the product of social class andrace/ethnicity. may go a long way in making usaware of the characteristics of adolescents placedat risk of dropping out.

18

Insert Table 5 about here

Educational Plans of Early Dropouts

The decision to drop out of school is notnecessarily the end of the road for adolescents.Often it is only after dropping out that manyadolescents realize the economic, social andpsychological consequences of their actions. Inthis case, some adolescents actually "stop outand return to school to complete their educationthrough the regular process. Others enroll inalternative education programs designed to meettheir individual needs, while still others opt totake a high school equivalency test.

The second part of this study is a descriptiveinvestigation of the various education plans ofdropouts. NELS:88 respondents were askedwhether they planned to earn a high schooldiploma, or GED, and to chose one of amongfive responses: 1) "yes, I plan to go back toschool to get a diploma;" 2) "yes, I am enrolledor plan to enroll in an alternative school or get aGED;" 3) "yes, I plan to take an equivalency testsuch as the GED;" 4) "no, I do not plan to get adiploma or GED., and 5) "I already have as GEDor equivalent" (Tables 6 and 6A).

Insert Tables 6 & 6A about here

Table 6 presents tabulations of the racial /ethnicand gender groups on the educational plans ofdropouts. Several patterns emerged in thisanalysis. African Americans, both males andfemales, disproportionately indicated that theyplan to return to regular school to earn a diploma.Forty-two percent of the African American malesand 39% percent of the African Americanfemales, a larger rate than any other group,reported an intention to return to school.However, we are somewhat skeptical that these

13

intentions might actually be carried out.Dropouts often fall so far behind their ex.classmates that catching up, or progressing at anormal rate, seldom occurs when they return.The extra help they need to he brought up tospeed is often unavailable. Moreover, it is acommon scenario that dropouts were strugglingacademically even before quitting school, asdiscussed above. These influences make theprobability of dropouts earning high schooldiplomas by returning to school quite small,unless special provisions are made for them uponreturning.

In contrast to regular high schools, alternativeschools attempt to foster environments bothsupportive and sympathetic to the needs ofadolescents who may have dropped out, orplaced at risk of dropping out. Alternativeschool teachers are often given additional supportand training they need in order to better serve achallenging student body. In the present study,African American females, Hispanic males, andwhite females reported that they plan to enroll inalternative schools, or schools designed to

prepare them for the GED, at a slightly higherrate than other groups (34%, 33% and 32%,respectively, as compared to a 26.3% averageamong the other groups).

Perhaps the quickest and least arduous route to ahigh school credential for adolescents who haveinterrupted educational histories is via the GED.However, preliminary results suggest reluctanceamong some adolescents, African Americans andHispanics in particular, to take this route. Three-quarters of the African American males andHispanic males, 71% of Hispanic females, and80% of African American females had no plansto take an equivalency test. However, thiseducational option was somewhat more popular

19

among white adolescents. Thirty-nine percent ofthe white males and 34% of white femalesplanned to take an equivalency test.

Across all groups, only a handful of adolescentshad no plans to complete a high school diplomaor earn an equivalent alternative credential.Among Hispanics, 8% of the males and 7% ofthe females; among African Americans, 0% ofthe males and 1% of the females; and amongwhites, 5% of the males and 4% of the femalesdid not intend to pursue further education.

In order to further examine the race/ethnicity andgender differences in educational plans we againconducted a series of t-tests. For the most part,the results discussed above were supported in thet-test analysis (Table 6A). Again, AfricanAmerican dropouts, generally, were found to bemore likely than whites to report planning toreturn to school (e.g., African American/whitecomparison: t=3.06; p<.01). The African American

males were also more likely than Hispanic malesto cite a desire to return to school (t=-1.65), butthis difference was not significant. Similarly,African American females were significantlymore likely than white females, but not Hispanicfemales, to report plans of returning to school(Table 6A).

Finally, these t-test results add to earlier evidencethat African American and Hispanic adolescentstend to avoid the equivalency test route to a highschool diploma at a significantly higher rate thantheir white counterparts. For males, AfricanAmerican and Hispanics, although not signifi-cantly different from each other (t=.080), weresignificantly less likely than whites to plan totake the GED test (t=-2.55, t=-2.24; both p<.01,respectively). However, for females the solesignificant difference in plans to take the GEDwas between African Americans and whites (t =-2.28; p< .05), where white females had a muchhigher rate reporting this option.

III. Conclusion and Implications

The purpose of this study was to cast light on thedegree to which various common reasons fordropping out of school and the educational plansof early dropouts differ across racial/ethnic andgender groups of adolescents. Above, we haveshown that there are multiple factors that operatesimultaneously to precipitate the decision ofracial/ethnic minority and female .adolescents todrop out. The following is a summary of majorfindings and some preliminary implications ofthe findings.

School-related reasons for quitting school werecited more frequently by early dropouts than any

14

other category of reasons, including family-related, job-related, safety, friendship influences,mobility and suspension/expulsion. Theschool-related scale included items such as notliking school, not liking teachers, and feelings ofnot belonging in school. In addition, theproportion of adolescents who cited school-related influences as contributors to dropping outdiffered across race/ethnicity. In general, alarger percentage of white and Hispanic dropoutscited school-related factors as reasons for leavingschool than did African Americans. Furtherdescriptive analysis of the NELS:88 datarevealed that African American dropouts cited

20

suspension/expulsion more often than any othergroup. This is indicative of the difficulties thatschools might be having with African Americanstudents' culturaUlinguistic styles and behaviors,and merits close qualitative analysis. In addition,these results underscore others studies (e.g.,Garibaldi, 1992; Mann 1986; and Willie, 1991)which have found that African Americans,particularly males, are disproportionatelydi.,-;ciplined and labelled as having behavioralproblems by schools.

Several other categories of reasons for leavingschool had a differential impact across groups ofdropouts. For example, a significantly largerpercentage of male dropouts than femaledropouts cited job-related reasons (e.g., findinga job, or keeping one) as one of the reasons forleaving school. In addition, females citedfamily-related reasons such as pregnancy,parenthood, or caring for a family membersignificantly more frequently than male adoles-cents. However, the family-related influences ondropping out did not affect raciallethnic groupsamong females equally. Many more Hispanicand African American females cited fam-ily-related reasons for dropping out than didwhite females. There arc perhaps a number ofways that these data can be interpreted. Oneexplanation is that Hispanic and AfricanAmerican female adolescents may have moreresponsibility in the home; another is that theymay become pregnant in their early teenage yearsmore frequently than white female adolescents.

The influence of friendship circles also had adifferential impact on dropping out. We foundthat dropping out because a friend had done so is

more prevalent among African Americanadolescents than among their white or Hispaniccounterparts. This was observed across gender.

In this study, we statistically controlled on SES.Although 80% of the dropouts in the samplewere low SES, small differences were revealedin the analysis. Generally, white dropouts werehigher SES than African American and Hispanic

students. In addition, the lowest SES levelsgenerated the most dropouts who cited familyand job-related reasons as a cause of their schooldeparture. We also found that, for Hispanics,school-related factors and suspension/expulsionwere reported most often by those abovemid-SES levels. We also examined possibleinteraction effects and found that none appear toalter the earlier descriptive patterns that wefound.

The final set of analyses focused on the futureeducational plans of early dropouts. AfricanAmericans, both males and females, were mostlikely to report planning to return to school andleast likely to take the GED test. The pattern ofresults for Hispanics was similar to AfricanAmericans, but of a lesser magnitude. However,previous research has shown that AfricanAmerican and Hispanic students are less likelythan whites to return to school and eventuallyearn diplomas once they drop out (GeneralAccounting Office, 1986). Nonetheless,dropouts often come to realize, some muchsooner than others, that their economic prospectsarc limited without a high school diploma -- i.e.,only a very small number of dropouts had nointentions to complete a high school diploma oran equivalent credential.

15

2 i

The overwhelming majority of dropouts did plan

to earn a high school level credential in thefuture, but they had different paths for meeting

this objective, and these paths differed byrace/ethnicity and gender. We found that bothmale and female white adolescents, as compared

to African Americans and Hispanics, plannedmore frequently to take equivalency tests as ameans of reaching their goal. Also, generally,Hispanic adolescents favored attending alterna-tive high schools, while African Americanadolescents favored going back to a regular highschool and earning their diplomas. It will beimportant for follow-up studies on these earlyadolescents to learn which of these reported plansare actually realized, and to consider ways ofincreasing the uses of the dropout recoveryroutes that are found to be most successful for allgroups.

Policy Implications

Although the results of this study are primarilydescriptive in nature and preliminary, severalpolicy implications are relevant. As we continueto investigate the causal complexity of droppingout, we simultaneously suggest several modifica-tions in the current programmatic efforts toreduce the incidence of dropping out amongadolescents in the middle grades and in highschool. We stress, however, that further researchinto the process and ramifications of droppingout need to continue in order to discoverworkable solutions to this pressing crisis inAmerican education. Below is a preliminary listof possible policy implications:

o Programs that build employment oppor-tunities for adolescents while they attendschool should be developed and ex-panded to allow a greater number ofAfrican American and Hispanic males toparticipate.

16

Schools should develop programs andstrategies to ease the burden that AfricanAmerican and Hispanic girls carry interms of family responsibility. In additionto counseling, this might involve identifi-cation of affordable and accessible childcare facilities for the children of adoles-cents. Another option is to identifysocial service agencies to help with careof members of the extended family. Veryoften poor minority females are forced toaccept the responsibility of caring foryounger siblings, parents that are ill, orfor other members of the immediate orextended family.

Schools might form after-school pro-grams that target specific populations ofthe school. For example, if AfricanAmerican females are more influencedby their peers than are other adolescents,the school might strengthen its holdingpower by designing activities of particu-lar interest to African American girls.

The finding in this study that a highernumber of African American students,particularly males, leave school becauseof suspension/expulsion is not new.School administrators must study thissituation carefully to ascertain the originsof this problem, and based on this analy-sis develop school-wide staff develop-ment efforts that might include strength-.ening the cross-cultural communicationsof the school's staff.

References

Alsalam, N., Fischer, G.E., Ogle, L.T., Thompson, G.R., & Smith, T.M. (1993). The Condition ofEducation (NCES 93-290). Washington, DC: National Center for Education Statistics.

Cervantes, L.F. (1965). The dropout. Ann Arbor, MI: University of Michigan Press.

Fine, M. (1986). Why urban adolescents drop into and out of public high schools. Teachers CollegeRecord, 87, 393-409.

Garibaldi, A.M. (1992). Educating and motivating African American males to succeed. Journal of Negro

Education. 61, 4-11.

Hahn A., & Danzberger J. (1987). Dropouts in America: Enough is known for action. Washington, DC:

Institute for Educational Leadership.

General Accounting Office (June, '..986). School dropouts: The extent and nature of the problem.Washington, DC: United States General Accounting Office.

Mann, D. (1986). Can we help dropouts? Thinking about the undoable. Teachers College Record, 83, 3,

307-323.

Orr, M. (1987). What to do about youth dropouts. New York, NY: Structured Employment/EconomicDevelopment Corporation.

Pallas, A.M. (1984). The determinants of high school dropouts. Doctoral Dissertation, The JohnsHopkins University, Baltimore, MD.

Peng, S.S. (1983). High school dropouts: Descriptive informationfrom High School dc Beyond (NCES83-2216). Washington, DC: National Center for Education Statistics.

Polite, V.C. (1993). If we only knew then what we know now: Foiled opportunities to learn in suburbia.

Journal of Negro Education, 62, 1-18.

Rumberger, R.W. (1983). Dropping out of high school: The influence of race, sex, and familybackground. American Educational Research Journal, 20, 199-220.

Rumberger, R.W. (1987). High school dropouts: A review of issues and evidences. Review of Education

Research, 57, 101-122.

Valdivieso, R. (1986). Must they want another generation ofHispanic and secondary school reform.New York, NY: Clearinghouse on Reform Evaluation.

17

23

Velez, W. (1989). High school attrition among Hispanic and non-Hispanic white youth. Sociology ofEducation, 62, 119-133.

Wehlage, G.G. & Rutter, R.A. (1986). Dropping out: How much do schools contribute to the problem.Teachers College Record, 87, 374-392.

Wheelock, A., & Dorman, G. (1988). Before it's too late: Dropout prevention in the middle grades.Boston, MA: Massachusetts Advocate Center.

Willie, C.V., Garibaldi, A.M., & Reed, W.L. (Eds.) (1991). The education of African-Americans. NewYork, NY: Auburn House.

18

Table 1Means and Standard Deviations of Dropout Students'

Reasons for Leaving School

Reasons n=1,000 Mean

.17 .38FID6A I had to get a job

FID6B I found a job .18 .38

FID6C I didn't like school .51 .50

FID6D I couldn't get along w/teachers .34 .48

FID6E I couldn't get along w/other students .16 .37

FID6F I wanted to have a family .06 .24

FID6G (For females only) I was pregnant .12 .32

FID6H I became the father/mother of a baby .12 .32

FID6I I had to support my family .07 .26

FID6J I was suspended too often .16 .36

FID6K I did not feel safe at school .09 .29

FID6L I wanted to travel .03 .18

FID6M My friends had dropped out of school .12 .33

FID6N I had to care for a member of my family .07 .26

FID6O I was expelled from school .13 .33

FID6P I felt I didn't belong at school .25 .43

FID6Q I couldn't keep up w/my schoolwork .31 .46

FID6R I was failing school .44 .50

FIDES I got married or planned to get married .12 .33

FID6T I changed schools and didn't like my new school .11 .31

FID6U I couldn't work and go to the school at the same time .18 .36

Tab

le lA

Sum

mar

y of

Fac

tor

Load

ing

on R

easo

ns S

cale

FID

6AF

ID6B

FID

6CF

ID6D

FA

CT

OR

IF

AC

TS

/112

FA

CT

OR

3F

AC

TO

R 4

FA

CT

OR

5F

AC

TO

R 6

EA

TO

R

.607

4

.321

9

.713

1

.712

3

FID

6E.7

330

FID

6F.8

312

FID

6G.6

503

FID

6H.7

946

FID

61.6

397

FID

6J.7

670

FID

6K.7

538

FID

6L.6

855

FID

6M

FID

6N.5

862

FID

6O.8

214

FID

6P.3

729

FID

6Q.7

352

FID

6R.7

835

FID

ES

.795

6

FID

6T.6

608

FID

6U.7

892

VA

RN

AM

EF

AM

1**

FA

M2

SC

HJO

BS

AF

ET

YE

XP

ELL

ED

MO

BIL

ITY

N O

F IT

EM

S4

25

32

22

RE

LIA

BIL

ITY

a =

.631

cx =

.569

a =

.609

a =

.628

a =

.472

a =

.562

a =

.147

= F

AM

1,c.

, + F

AM

2., (

a =

.656

)F

AM

FE

M =

FA

W11

,,,(

-FID

6G)

+ F

AM

2" (

a =

.601

)

Tab

le 2

Sex

and

Rac

e/E

thni

city

Diff

eren

ces

on R

easo

ns

for

Dro

ppin

gof

Sch

ool

H

MA

LES

AA

WT

otal

H

FE

MA

LES

AA

WT

otal

SC

H.3

84.3

15.4

74.4

29S

CH

.265

.272

.343

.312

JOB

.241

.199

.233

.229

FA

M F

EM

.174

.181

.129

.149

EX

PE

LLE

D.1

52.2

74.1

90.1

96JO

B.1

14.0

92.1

31.1

20

FR

IEN

DS

.124

.207

.163

.162

SA

FE

TY

.122

.125

.115

.118

SA

FE

TY

.129

.158

.134

.137

EX

PE

LLE

D.0

72.1

46.0

6407

9

MO

BIL

ITY

.077

.104

.065

.068

FR

IEN

DS

.063

.139

.065

.077

FA

M.0

72.0

85.0

38.0

53M

OB

ILIT

Y.0

59.0

83.0

64.0

66

SA

MP

LE S

IZE

105

8230

649

3S

AM

PLE

SIZ

E11

172

260

443

B

29

Tab

le 2

AT

-Tes

t of S

igni

fican

ce o

n R

ace/

Eth

nic

Diff

eren

ces

I

H/W

Q.

MA

LES

.AA

/W

iQ

I

H/A

A

Q.

i

H/W

Q.

FE

MA

LES

AA

/W

IQ

.

H/A

A

IQ

.

SC

H-2

.66

(.00

8)-4

.43

(.00

0)1.

62(.

106)

SC

H-2

.48

(.01

4)-1

.94

(.05

4)-0

.18

(.86

0)

JOB

0.23

(.82

0)-0

.87

(.38

5)0.

94(.

347)

FA

M F

EM

1.97

(.04

9)1.

85(.

066)

-0.2

0(.

839)

EX

PE

LLE

D-1

.06

(.29

1)2.

30(.

043)

-2.4

5(.

015)

JOB

-0.5

8(.

560)

-1.1

7(.

243)

0.58

(.56

2)

FR

IEN

DS

-0.9

7(.

332)

0.93

(.35

1)-1

.55

(.12

4)S

AF

ET

Y0.

20(.

838)

0.28

(.78

2)-0

08

(.93

8)

SA

FE

TY

-0.1

8(.

860)

0.74

(.46

0)-0

.75

(.45

2)E

XP

ELL

ED

0.38

(.70

8)2.

77(.

006)

-1.8

8(.

061)

MO

BIL

ITY

-0.9

0(.

369)

1.61

(.10

7)-2

.09

(.03

8)F

RIE

ND

S-

.08

(.93

4)2.

03(.

044)

-1.7

3(.

085)

FA

M2.

25(.

025)

2.80

(.00

5)-0

.51

(.61

4)M

OB

ILIT

Y-0

.26

(.79

8)0.

81(.

419)

-0.8

7(.

384)

3i

Table 3Sex and Race/Ethnicity Differences in

Family Socioeconomic Status for Student Dropouts

MALES

Hisp Afr Amer White Hisp

FEMALES

Afr Amer White

mean 0.903 0.835 0.658 1.023 0.952 0.672std. dev. .598 .673 .668 .580 .686 .684

n 98 83 305 118 73 262

t statistic

t - test Hispanic vs. Afr Amer -0.74 .462

Female Hispanic vs. White 5.15 .000

Afr Amer vs. White 3.08 .003

Male Hispanic vs. Afr Amer 0.72 .474Hispanic vs. White 3.24 .001

Afr Amer vs. White 2.13 .035

32

Table 4Summary of Multiple Regression Analyses of Student Dropout Reasons

on Race/Ethnicity, Sex, and Socioeconomic Status

eta

A.ANV

I Berta

H/W

1 g Beta

H/AA

FAM Race .092 2.47 .012 .078 2.17 .030 -.012 -0.23 .817

SES -.038 . -1.01 .311 -.090 -2.49 .013 -.017 -0.31 .751

Sex -.207 -5.59 .000 -.222 -6.29 .000 -.193 -3.66 .000

RXSex .648 .937 .641

RXSES .117 .406 .045

R .233 .265 .195

FAM FEM Race .083 1.50 .135 .070 1.32 .188 -.015 -0.20 .845

SES -.104 -1.88 .061 -.131 -2.45 .015 .006 .07 .941

RXSES .180 .428 .501

R .145 .163 .016

JOB Race -.054 -1.44 .150 -.014 -0.39 .699 .059 1.13 .258

SES -.024 -0.64 .521 -.055 -1.51 .132 -.018 -0.35 .728

Sex .179 4.80 .000 .189 5.26 .000 .214 4.08 .000

RXSex .917 .603 .765

RXSES .318 .928 .223

R .187 .195 .218

SCH Race -.161 -4.36 .000 -.103 -2.87 .004 .061 1.15 .251

SES .031 0.84 .102 .094 2.64 .009 .042 .80 .423

Sex .190 5.18 .000 .209 5.95 .000 .149 2.82 .005

RXSex .086 .657 .226

RXSES .020 .263 .010

R .256 .269 .166

SAFETY Race .030 0 78 .438 .003 0 07 .944 -.028 -0.53 .597

SES .017 0.44 .659 .007 0.19 .850 .022 0.41 .685

Sex .041 1.08 .282 .028 0.77 .440 .030 0.55 .580

RXSex .773 .750 .659

RXSES .841 .625 .527

R .052 .029 .050

EXPELLED Race .123 3.31 .001 -.002 -0.42 .967 -.114 -1.52 .123

SES .054 1.45 .148 .091 2.50 .013 .006 .124 .902

Sex .211 5.73 .000 .206 5 77 .000 .206 2.54 .012

RXSex .981 .220 .450

RXSES 052 .716 .256

R .247 .229 .232

MOBILITY Race .072 1.87 .062 -.013 -0 35 .729 -.108 -2 03 .043

SES .049 1.28 .200 .083 -0.26 .026 .007 .124 .901

Sex .014 0.38 .706 -.008 2.23 .822 003 .053 .957

RXSex .629 .585 .442

RXSES .212 .939 .184

R .082 .087 .109

FRIENDS Race .073 1.92 056 -.017 -0.45 .653 -.116 -2.19 .029

SES .018 0.48 .632 .053 1.45 .148 .000 .004 .997

Sex .135 3.60 .000 .137 3.79 .000 .094 1.76 .079

RXSex .610 .404 .905

RXSES 096 .870 .281

R .153 .153 .154

3,3

Table 5Interaction Effects on

Selected Student Dropout Reasons

REASON H

MALE

AA W H

FEMALE

AA

SCH SES LO 349 .334 .499 .248 .285 .312(70) (55) (165) (84) (49) (143)

SES HI .469 .296 .461 .344 .248 .386(26) (23) (128) (25) (21) (113)

EXPELLED SES LO .150 .300 .164 .060 .133 .042(70) (55) (165) (84) (49) (143)

SES HI .154 .239 .195 .120 .190 .093(26) (23) (128) (25) (21) (113)

FAM SES LO .091 .073 .029 .171 .176 .133(70) (55) (165) (84) (49) (143)

SES HI .038 .130 .044 .096 .095 .085(26) (23) (128) (25) (21) (113)

Tab

le 6

Edu

catio

nal P

lans

of H

igh

Sch

ool D

ropo

uts

by E

thni

city

and

Gen

der

Fi

MA

LES

AA

WH

FE

MA

LES

AA

W

Bac

k to

Sch

ool

31%

(107

)42

%(8

5)25

%(3

17)

32%

(116

)39

% (

74)

26%

(261

)

Enr

oll i

n33

%(1

07)

25%

(85)

25%

(317

)28

%(1

16)

34%

(74)

32%

(261

)A

ltern

ate

Sch

ool/G

ED

Tak

e25

%(1

07)

25%

(85)

39%

(317

)29

%(1

16)

20%

(74)

34%

(261

)E

quiv

alen

cyT

est/G

ED

No

plan

s to

take

GE

D8%

(107

)0%

(85)

5%(3

17)

7%(1

16)

1%(7

4)4%

(261

)

Alre

ady

had

2%(1

07)

8%(8

5)4%

(317

)2%

(116

)5%

(74)

3%(2

61)

GE

D

3)36

Tab

le 6

AS

igni

fican

ce T

ests

of R

ace-

Eth

nici

ty D

iffer

ence

s

MA

LES

FE

MA

LES

H/W

AA

/WH

/AA

cc

H/W t

a

AA

/W ta

H/A

A

a

Bac

k to

Sch

ool

1.07

(.28

7)3.

06(.

002)

-1.6

5(.

100)

1.09

(.27

8)2.

14(.

033)

-1.0

3(.

306)

Enr

ol in

alt/G

ED

1.44

(.15

2)-.

16(.

874)

1.21

(.22

8)-.

65(.

516)

.32

(.74

8)-.

78(.

439)

Tak

e G

ED

test

-2.5

5(.

011)

-2.4

2(.

016)

.08

(.93

4)-.

91(,

361)

-2.2

8(.

C23

)1.

39(.

167)

No

plan

sfo

r G

ED

1.00

(.31

8)-2

.26

(.02

5)2.

78(.

006)

1.23

(.21

8)-1

.28

(.20

3)1.

94(.

054)

Hav

e G

ED

-0.7

3(.

463)

1.41

(.16

1)-1

.69

(.09

3)-.

25(.

799)

.95

(.34

0)-1

.00

(.31

7)

238