DOCUMENT RESUME

ED 288 296 EC 200 767

AUTHOR Greenwood, Charles R., Ed.; And OthersTITLE An Eco-Behavioral Approach to Research in Special

Education. The Juniper Gardens Children's Project.[Volume 1].

INSTITUTION Kansas Univ., Lawrence. Bureau of Child Research.;Kansas Univ., Lawrence. Dept. of Human Services andFamily Life.; Kansas Univ., Lawrence. Dept. ofSpecial Education.

SPON'a AGENCY Office of Special Education and RehabilitativeServices (ED), Washington, DC. Div. of PersonnelPreparation.

PUB DATE 30 Apr 85GRANT DH-30066NOTE 94p.; For volume 2 of this monograph series, see EC

200 768.PUB TYPE Reports - Descriptive (141) -- Information Analyses

(070) -- Collected Works - General (020)

EDRS PRICE MF01/PC04 Plus Postage.DESCRIPTORS *Behavioral Science Research; Cultural Differences;

*Disabilities; Early Childhood Education; *EcologicalFactors; *Educational Psychology; ElementarySecondary Education; Evaluation Methods;Instructional Development; Language Role; Outcomes ofEducation; Program Evaluation; *Special Education;Student Behavior; Time on Task

IDENTIFIERS Behavioral Psychology; *Ecobehavioral Analysis;*Ecological Psychology

ABSTRACTThis monograph contains five papers that discuss an

eco-behavioral approach to psychology, special education, and appliedbehavior analysis. The papers point out the advantages of assessingecological factors (such as natural stimuli and special educationprocedures) in a quantitative fashion and in a temporal relationshipwith student behavior. An introductory chapter by Charles R.Greenwood defines eco-behavioral research and indicates its potentialbenefits. The first paper, a literature review by Donald Dorsey,finds that frequently contradictory research on the effective use ofinstructional time should be made more precise and broader in scope,by pairing behavior analysis and ecological psychology. Next,Greenwood assesses co-behavioral relationships in special educationoutcome research and argues for the development of a powerful, widelypracticed technology of instruction. Carmen Arreaga-Mayer examinesthe literature dealing with ecological and teaching variables thataffect the academic performance of culturally and linguisticallydifferent learners. Judith Carta proposes an eco-behavioral approachto evaluation of preschool programs for the handicapped, which allowsfor determination of early intervention effectiveness andspecification of the factors responsible. Gary Verna examinestheoretical and empirical issues related to current ecologicaldevelopments within the field of applied behavior analysis. (JDD)

AN ECO-BEHAVIORAL APPROACH TORESEARCH IN SPECIAL EDUCATION

Edited by

Charles R. Greenwood, Ph.D.Project Director,

Juniper Gardens Children's ProjectUniversity of Kansas

Carmen Arreaga-Mayer, Ph.D.Project Coordinator

Juniper Gardens Children's ProjectUnversity of Kansas

Dorothy Clark-Preston, Ph.D.Therapist,

Adult/Child Center for Behavioral DevelopmentKansas City, Missouri

Juniper Gardens Children's Project,

Bureau of Child Research and Departments ofSpecial Education and Human Development and Family Life,

University of Kansas

AN ECO4BEHAVIORAL APRROACH TO RESEARCH IN SPECIAL EDUCATION

A product of the Juniper Gardens Children's Project, Bureau ofChild Research and Departments of Special Education and HumanDevelopment and Family Life

This publication and the work reported herein was supported byGrant Nix DH 30066 from the Personnel Preparation Division,Special Education Program, U. S. Department of Education. Theviews expressed herein are those of the authors and no officialendorsement of the Division should be implied.

Contributors

CONTENTS

Preface vii

Acknowledgements ix

Chapter I: introduction 1

Chapter An Ecological-Behavioral View Ofinstruction Time 9

Chapter ill: Assessing Eco-Behavioral Relationships InSpecial Education Outcome Research Or "WhereIs The Independent Variable?" 19

Chapter IV: Environmental Variables Affecting sheSchool Achievement Of Culturally andLinguistically Different Learners: Aninstructional Perspective 35

Chapter V: Eco-Behavioral Assessment: A MethodologyFor Expanding The Evaluation of Earlyintervention Programs 57

Chapter VI: Eco4ehavioral Analysis In The Field OfApplied Behavior Analysis 73

Chapter VII: Concluding Remarks 87

ill

CONTRIBUTORS

Carmen Arreaga-Mayer, Ph.D., 1983-84Project Coordinator, Juniper Gardens Children's Project

University of Kansas

Judith Carta, Ph.D. 1983-84Project Director, Junipeir Gardens Children's Project

University of Kansas

Donald Dorsey, Ph.D. 1983-85

Post-Doctoral Fellow, Juniper Gardens Children's ProjectUniversity of Kansas

Charles R. Greenwood, Ph.D.Project Director, Juniper Gardens Children'Project

University of Kansas

Gary Verna, Ph.D. 1983-84Research Fellow

Vanderbilt UniversityNashville, Tennesee

v

6

PREFACE

From 1983 to 1986 the Special Education Program of theDepartment of Education supported a post-doctoral fellowshipprogram In research at the Juniper Gardens Children's Project.

The Children's Project has been a 20 year program in research anddevelopment concerning the problems of low income and minoritygroup children and their families. Within the last seven years,research based upon an eco-behavloral interaction approach hasevoived. Part of each post-doctoral fellow's requirements duringtheir fellowship year was the development of a review paper Inwhich each fellow examined an area of their interest from theperspective of an eco-behavioral approach. The results of theseindividual scholarly efforis are contained In this volume. Thechapters span a range from the highly conceptual (i.e., thechapter by Verne), to a presentation at a conference on specialeducation research (1.1e, the paper by Greenwood), to reviews ofthe research literature (i.e., the chapters by Carta, Arreaga-Mayer, and Dorsey).

The major underlying theme to these works Is an eco-behavioral interaction view of the educational process. Aneffort is made in the Introduction to define eco-behavioralinteraction research. However, since this research, as a contentarea and as a methodology Is rapidly developing, the adequacy ofthis definition must be considered In light of continuingdevelopments.

This monograph Is intended for researchers In ecological

psychology, special education, and applied behavior analysis whoare interested in exanding their methodology to include thequantitative assessment of ecological and behavioral factors Intheir work. The benefits, significance, and dimensions of thisapproach will hopefully be revealed in this monograph.

vii

Charles R. Greenwood

Carmen Arreaga-Mayer

Dorothy Preston

Kansas City, Kansas

April 30, 1985

7

ACKNOWLEDGEMENT

We gratefOly acknowledge the invaluable help of Mary Todd,Bernadine Roberts, and Kathy Banks for their help preparing thismanuscript in Its many phases. For their careful reading of thechapters and their thoughtful comments, special thanks areexpressed to Don Dorsey, Judith J. Carta, and Gary Verna.

Ix

CHAPTER I

INTRv:'2CTION

CHARLES R. GREENWOOD

An emerging trend In research is analysis of ecology -behavior interaction. Tne assessment of eco-behavloralinteraction is based upon recordings of the momentaryInteractions of environmental stimuli and a person's behavior.Examination of the behavioral assessment literature, however,

reveals that traditional forms of assessment have focused almostexclusively upon the characteristics and behaviors of the childor adult :McReynolds, 1979). Assessment of child status alone,however, does not provide important information about either theImmediate or historic conditions within the child's ecology,which may contribute to child growth and development (Bijou,1981). Thus, important information concerning the dynamicInterrelationship between child behavior and the environmentremains largely unknown (Bronfenbrenner, 1979; Brophy, 1979).

Natural science approaches to behavior analysis have beenbased upon theoretical conceptualizations of eco-behavioralInteraction for years (Barker, 1961; Kantor, 1954; Skinner,1953), yet ernirical data in this area has been forthcoming onlyin the last 10 years. In applied research particularly, this hasbeen the direct result of technology advances, particularly incomputer-assisted observation systems. These systems haveenabled researchers to gather and analyze complex information onthe structure, sequence, and function of eco-behavloralphenomena. A premiere example of this research is that ofPatterson and his colleagues (Patterson, 1982). Their coercivetheory of aggression is based upon sequential analysis of child -adult interaction records. This procedure allows theinvestigator to examine the stimulus controls operating withinnatural settings. The approach is based upon conditionalprobability relationships between the behaviors of interactingpersons. Thus, in Patterson's work it is possible to identify afollowed six seconds later in time by child W Since thereare no theoretical limitations on the time interval betweenevents (only the length of the data record imposes limitations),it is possible to study the distal impact of particular stimulior setting events upon specific interaction patterns, like theone between parent and.child just mentioned. For example,Karpowitz and Johnson (1982) examined the relationship of social

This Introduction is based upon Greenwood, C. R., Schulte, D.,Kohler, F., Dinw idd le, G. 8 Carta, J. (In press). Assessment andana I ys I s of eco- behavioral interaction in school settings. I n

R. Prinz (Ed.), Mums j behavioral assessment ztill dren andfamilies, (Vol. II), Greenwich, CT: JAI Press.

9

2

stimuli to child behavior at 10 seconds versus 30 seconds apart.They concluded that the immediately preceding stimulus (10seconds earlier) was most predictive of child response. Otherresearchers are currently applying these methods to appliedproblems in fern! l Um Biglan, et al., 1984; Hops et al., 1984;family theroy, & Graves, 1983; language development,Hart & Risley. 1984; aeer_sociei interaction, Greenwood, et al.,1982; Kohler, 1984; and classroom instruction; Brophy, 1979;Greenwood, at al., in press.

Def inition Qi gco-Behavioral Easgi sh

Eco-behavioral research implies assessment and interventiondesigned to reveal sequential and concurrent interrelationshipsbetween environmental stimuli and organism response. A goal ofeco-behavioral assessment is to assess both the physical andsocial stimuli temporally associated with behavior. This goaldifferentiates eco-behavioral assessment to some degree fromother forms of sequential assessment commonly referred to as"social interact ion" (Cairns, 1979; Lamb, Suomi, & Stephenson,1979; Patterson, 1982). Social interaction research hasemphasized assessment of reciprocal social stimuli within dyadicinteraction. It has not broadly assessed physical stimuli inrelationship to subject behavior.

We have used the term "eco-behavioral" in our work inclassrooms: (a) to refer to the measurement of a broaderconstellaflon of stimulus events than Just person-personinteraction (lie., the curriculum, physical arrangements, andteacher behavior) and (b) to denote the non - social character ofthese stimulus events. Our interest has been in discovering themomentary instructional forces that effect student's academicperformance in the classroom. Thus, in our work we have includedthe subject matter, instructional materials, physical grouping,teacher location, and teacher behavior, which are the generai andspecific contexts for student's classroom behavior.

In eco-behavioral assessment, ecological and behavioralvariables are sampled in close temporal relationship (10 secondintervals). By systematically alternating sampling of ecologicalfollowed by behavioral variables, changes In environmentalstimuli and behavior are recorded in sequence. Thus, anobserver records the teacher's behavior In an Interval justpreceding the recording of the student's behavior. In thisfashion, the sequence teacher instruct, followed by student readaloud may be recorded. By alternately sampling the teacher, thenthe student, the contextual basis for student behavior isincluded within the record for later analysis.

Limitations Ieposes1 b4. Traditional Observational Assessment

The rather voluminous observational literature in education,for example, is devoted to complex assessments of classrocmclimate, school ecology, and teacher-student interaction. In

fact, Flanders (1970) in his studies of teacher-studentinteraction, was perhaps the first to use an interaction matrixto relate teacher and student events. The rather pervasiveproblems with these observational studies, however, are that: (a)the variables measured are not defined as observeable events, (b)interobserver reliability is either not conducted or is low andinadequate, (c) the data are simply descriptive and static, thatis, sequential relationships cannot be expressed in the data, and(d) the theoretical assumptions upon which the codes are based donot permit scientific investigations of direct environment-behavior influence (for a notable exception see Brophy, 1979) .

In applied behavior analysis and social learning theory,however, we have a tradition of observational assessment thatmeets rather stringent requirements of reliability and that hasbeen productively used in behavior change studies. However, thisassessment tradition has focused strictly upon measuring thesubject's behavior and has rarely included measurement of theenvironment. While behavioral assessment has been very effectivewithin the context of specific behavior change experiments, ithas not produced quantified descriptions of the settings andstimuli within which these behavior changes take place. Neitherhas behavioral assessment provided information concerning theputative natural conditions which create developmental deficitsand behavior problems which behavior analysts are called upon tocorrect. As Patterson, 1982, pointed out, "a science can be onlyas good as its assessment methodology". For many of thebehavioral phenomena we would like to predict and control,assessment of behavioral events, in the absence of ecologicalvariables, will be insufficent. Without information on naturalcontrolling relationships and subsequent tests of their causalstatus, the field will continue to confront problems, such as thefailure of behavior changes to maintain or of sane children toachieve in school. As Mahler and Fox, 1981 pointed out, therehat not been sufficent research attention paid to the structureand function of setting events or stimulus controls in appliedbehavior analysis. We would argue that this is largely theresult of new methodological problems related to use ofinteractive assessment in applied settings. However, otherreasons are apparent.

Our tradition of experimental research by necessity haslimited wide expression of setting variables due to the need +omaintain experimental control. As a result, setting factors havemore often been considered troublesome confounds, rather than asindependent variables in much of behavior analytic research(Foster & Cone, 1980; Rogers-Warren & Warren, 1977). Thus, onemay argue that many of our most prized functional relationshipshave been established within narrowly deflped setting conditions.Setting conditions or contexts are the establishing factorswithin which functional relationships will operate (Larsen, &Morris, 1983; Larsen, Morris, & Todd, 1984; Leigiand, 1984;Michael, 1982). Reinforcement, for example, can only operateto strengthen behavior under specific conditions of deprivationand subject history. Change these contexts ane you change the

11

4

functional relationship. Unfortunately, we often do not know howfunctional relationships hold up within extremes of settingvariation as occurs in the natural setting. The classic exampleof this point, for which we have yet to supply an adequateanswer, is, "How do schedules of reinforcement, (revealed withincontrolled research), operate to strengthen and maintain behaviorIn natural settings"?

It also can be argued that much of our current knowledge ofbehavior change is based upon data in which setting variation wassimply ignored. Reflected as variability in our baseline andtreatment data, is the operation of many setting variables forwhich we have no explanation or information since we choose notto assess them. Setting variations very likely operate toenhance or retard subjects performance.

As a result of both conceptual and methodological problems,we may find that we are unable to generalize our results tospecific contexts in which our target behavior occurs. Thisproblem is particularly evident within studies of generalizationand maintenance of behavior change wherein the setting dimensionsimportant to these phenomena were not included in the Initialanalysis. One means for representing both the immediate anddelayed setting factors in applied research is throughinte active eco- behavioral assessment.

agatiAL &mutt a Aft EgnBah v lora I Interaction Approseth

The solution to many of our current social problems (i.e.,academic retardation, effective special education treatments),may require a more fundamental understanding of eco-behavioralprocess. The benefits of an eco-behavioral interaction approachIle in several important areas. First, the recording ofecological variables, describes natural stimuli, their normativerates of occurrence, and their probability relationship tobehavior. As noted by Barker (1963), and more recently byBronfenbrenner (1979), there is no definitive data base on humanbehavior and its settings. Second, use of an eco- behavioralapproach as a process variable enables research on resultingdevelopmental impact or outcome (e.g. academic achievement). It

Is possible to develop base rates and base proabilities on eco-behavioral variables against which the effectiveness ofinterventions can be assessed. Third, examination of theconditional relationships between ecological variables andbehavior, may enable identification of particular ecologicalarrangements correlated with high levels of criterion behaviors.By identifying these arrangements In high performing subjects, itmay be possible to test these natural arrangements for functionwhen introduced to the environments of lower performers. Fourth,experiments testing these specific eco-behavioral hypotheses withlow performers could yield information concerning their causalstatus. Fifth, structural information from an eco-behavioralcode, defining standard and effective treatment formats, could beused to study the implementation of treatment by teachers overtime. Thus, it might be possible to identify and analyze factors

12ti

that interfere with standard implementation or that facilitatemaintenance of it.

The importance of an eco-behavioral interaction approachlies in its potential as a tool for exploring the effects ofnatural stimuli in the natural environment. Using thismethodology, it may be possible to develop precisioninterventions based on eco-behavioral data that will be effectiveboth in the initial stages of treatmont and in generalizing andmaintaining behavior change. In 1967, Patterson (1967) calledfor reprogramming the natural agents within environments toaddress the problem of maintenance. This involved training ofnatural agents in the use of contingency management procedures.Yet, this problem of limited generalization and maintance remainswith us today. While we have been successful training naturalagents, perhaps, we have armed them with procedures that simplycannot survive unaided in the natural setting. We need tosufficentiy understand the function of natural stimuli in theseenvironments. As with other approaches to assessment, theimportance of the eco-behavioral approach will be itscontribution to our ability to predict and control behavior inapplied settings.

13

6

&farm=Barker, R. R. (1961). IcologIcai pughcasgmf Stanford, CA:

Sianford University Press.

Biglan, A., Hops, H., Sherman, L., Friedman, L. S., Arthur, J.& Osteen V. (1984). Erdal= salving latacactions.slorLisik 11211111J1 And t spouses. Eugene, OR: OregonResearch Institute.

Bijou, S. W. (1981). The prevention of retarded development indisadvantaged children. In M. J. Begab, H. C. Haywood,& H.L. Garber, (Eds.), Psycheseclal influences 111. taficisaumuma ALUM And thiamin la development (Vol. 1).(pp.29 -46). Baltimore, MD: University Park Press.

Bronfenbrenner, U. (1979). Context_ of child rearing: Problemsand prospects. American psychologist, 14, 844 -850.

Brophy, J. (1979). Teacher behavior and its effects. Jour al ajEducational Psychology, B., 733-750.

Cairns, R. B. (1979). Ihe, analysis, a social interactIonstmAthigAL Issues. And III ustrat tons. Hi I Isdale, NJ: LawrenceErlbaum Associates.

Flanders, N. (1970). &Aiding ±AnchlAA, hatusylim. Reading, MA:Addison-Wesley, 1970.

Foster, S. L., & Cone, J. D. (1980). Current Issues In directobservation. fishalgrAd &numb 2, 313-338.

Greenwood, C. R., Delquadri, J., Stanley, S. 0., Terry, B., &Hall, R. V. (In press). Assessment of eco-behavioralInteraction In school settirgs. BAhgatarAlAAAAAAAAnt.

Greenwood, C. R., Todd, N. M., Hops, H.) and Walker, H. M.(1982). Global and specific behavior change targets In theassessment and behavior modification of soc;a1 ly withdrawnpreschool children. EtitinacallAssfissagat,Ar273 -298.

Hart, B., & Risley, T. (1984, May). The 'natural' conditionsfor learning to talk. In S. Warren (Chair), Contemporarybehavioral agnipardlyel IAAga Alaluign. Symposiumpresented at the 10th Annual Convention of the Associationfor Behavior Analysis, Nashville, TN.

Hops, H., Biglan, A., Sherman, L., Arthur, J., Friedman, L. &Osteen, V. (1984). H a family interactionsgf.Asarizsguixamon. Eugrne, OR: Oregon Research Institute.

Kantor, J. R. (1959). Interlehimaariamahaggx. Granville,OH: Principle Press.

14

Karpowitz, D. H., & Johnson, S. M. (1981). Stimulus control in

child-family interaction. aahavloral., 161-172.

Kohler, F. (1984). Social interaction raga LK, Intl Juniperiilagns. gem tutoring gm& Kansas City, KS: JuniperGardens Children's Project, Bureau of Child Research,University of Kansas.

Lamb, M. E., Suomi, S. J., & Stephenson, G. R. (1979). Socialinteraction Analysis. Madison, WI: University of WisconsinPress.

Larsen, S. E., & Morris, E. K. (1983). QiitmAsefu

salting taintrancsatla behavior znalals, Paper read atthe Ninth Annual Convention of the Association for BehaviorAnalysis, Milwaukee, WI.

Larsen, S. E., Morris, E. K., & Todd, J. T. (1984). Comments QadetermlnimADA, :nasality. Paper read at the Tenth AnnualMeeting of the Association for Behavior :41alysis, Nashville,TN.

Leigiand, S. (1984). On 'setting events' and related concepts.Iba Behavior Analysts 2_, 41-45.

McReynolds, P. (1979). The case for interartional assessment.Behavioral Assonant, 237-247.

Patterson, G. R. (1982). A microsocial analysis of structure andprocess. (pp. 169-198). In G. R. Patterson, Coercive familyprocess. Eugene, OR: Castalla Publishing Co.

Patterson, G. R., McNeal, S., Hawkins, N., & Phelps, R. (1967).Reprogramming the social environment. Journal pi QUA.Psychology And. Psychiatry, 1, 181-195.

Rogers-Warren, A., b Marren, S. (1977). Ecological pfirspfiableSighithatigc analysis. Baltimore, MD: University Park Press,

Skinner, B. F. (1953). /aura and human aanayhx. New York:MacMillan.

Wahl er, R. G. & Fox, J. J. (1981). Setting events In app I ledbehavior analysis: Toward a conceptual and methodologicalexpansion. dal Qf Appl led Behavior Anal ys ism, 14,327-338.

Wahler, R. G. & Graves, M. G. (1983). Setting events in socialnetworks: Ally or enemy In child behavior therapy?Behavior Therapy 14, 19-36.

15

9

CHAPTER II

AN ECOLOGICAL-BEHAVIORAL VIEW OF INSTRUCTION TIME

DON DORSEY

'Abstract

The effective use of instruction time has emerged as a majorconcern of parents, educators, and even of the United StatesGovernment. Tinkering with the length of the school day or year,along with school board mandates for more required instruction,are popular solutions for reversing the slide from academicexcellence that began in the 1960's. Since Carroll (1963)introduced instructional time as a component of pupil progress,interest in the area grew gradually, until now there is a fairlylarge body of research. However, educators have found littleguidance or comfort from this welter of data, because the resultsare frequently contradictory. Two steps need to be taken toreduce the research conflicts that have hobbled educators. Thefirst step Is precision. Precise language, precise measurementof both independent and dependent variables, and preciseapplication of appropriate experimental designs will result indata that are understood and believed. The second step is: Takeone step back. The view from one step back Is broader, and willallow researchers to place instructional time in the context ofthe classroom and school day. This two-step process pairs thesomewhat reluctant partners of behavior analysis and ecologicalpsychology. Whether a two-step to the music of instructionaltime will lead to a liaison or a brief twirl, neither partnerknows for sure. However, if an ecological -behavior! approach canrevolve some of the contradictions in instructional timeresearch, it will have been worth the f I i ng.

Introduction.

Instructional time, the current hot property in the Ed Biz,has come to public attention through the President's Commissionon Education in America. The Commission (1983) reported thatEuropean students get more of it than American children do, andwe should increase Instructional time if America is to retain itsslim competitive edge in trade and technology. Since Carroll's(1963) theoretical analysis of the relationship of instructionaltime to teaching and learning, interest in the topic has builtslowly, until now there is a fairly substantial body ofliterature. Yet that literature is diverse and contradictory,and basic questions such as the optimal amount of instructional

INNS 1011111.1111111.11111111111111NDIMINIOM

This Chapter is based upon a paper presented by the author in1984 at the Tenth Annual Meeting of the Association for BehaviorBehavior Analysis, Nashville, TN.

16

10

time for different subjects at different ages, or how to allocateinstructional time, remain unanswered. Where research hasaddressed basic questions, the answers are sometimescontradictory. For instance, Husen (1967) conducted a massiveinternational study of mathematics, and concluded thatinstructional time contributed perhaps 3% of the assignablevariation on total math scores. Yet other researchers(Greenwood, Delquadri, Stanley, Sasso, Whorton b Schulte, 1983;Rosenshine, 1978) assert that instructional time, particularly ifit is well structured, is a major contributer to pupil progress.

How is it that despite a history of more than twenty yearsresearch, educators are unable to agree on the nature ofinstructional time, its effects on achievement, or even how toincrease it? From the beginning, instructional time research hasbeen approached from several theoretical perspectives, has used astartling number of experimental designs, has sometimes usedpeculiar vocabulary as if it were generally understood, and mostgrievously, has often failed to specify the independent anddependent variables under study. It is not surprising, then,that teachers and educators are still waiting for clearprinciples and useful teaching practices to emerge frominstructional time research. The President's Commission hastaken the common sense approach that instructional time is anecessary ingredient if children are to learn skills. And theAmerican public seems to agree. Teachers, however, must havemore than a common sense notion of instructional time if they areto use the school day most effectively. One way to bring orderto this area is to observe precisely in schools, to useexperimental designs chosen specifically to control for thecomplex nature of real c i assrooms, to use a vocabulary ofcareful ly defined terms, to specify both the independent anddependent variables cleanly, and to make some allowances for theschool setting itself. Such an approach can be calledecological-behavioral, and it can be used with many educationalquestions, though it seems to have a special potential forbringing order to the instructional time question. A briefconsideration of these steps may illustrate the usefulness of theecological-behavioral approach to educational research.

Evardsnix Mined lams.

One source of confusion in instructional time research hasbeen the cavalier tendency to use some terms as if they were bothInterchangeable and mutually understood. Thus, time, teachingtime, instructional time, allocated time, scheduled time, on -task, engaged time, and other terms have been tossed into aresearch stew. Results, not surprisingly, have been mixed.Karweit and Slavin (1981) have addressed this issue directly.They found that scheduled instructional time and teacher reportsof daily schedules, had no relation to pupil progress in grades2-3, and a weak relationship to pupil progress in grades 4-5 onthe California Test of Basic Skills (CTBS). In fact, they foundthat scheduled instructional time did not correspond closely to

.17

11

actual instructional time. Teachers' instructional schedules

tend to the optimistic, something noted by Smith (1979), as wellas Hook and Rosenshine (1979). Despite the unreliability ofteachers' reporting their own scheduled instructional time as theindependent variable.

Smith (1979) correlated unverified teacher reports of

instructional time use with the STEP II social studies test, andfound that instructional time was weekly correlated withachievement (riff.23). She found that measures of staticvariables, IQ for example, were more highly correlated withachievement on the STEP II test (rs.63). Jarvis (1962) conducteda massive study of Texas Gulf Coast schools and concluded thatreading should not be scheduled for more than 50 irinutes per daysbecause schools scheduling more than that learned less! Hismeasure of instructional time was even more removed from theclassroom than was Smith's. Jarvis got his independent measure,allocated time, by writing to schools, asking them what theschool district guidelines were for instructional time, and howwell they were complying with the district policy. No attemptwas made to observe instructional time directly. It is naive toassume that school personnel can accurately fill out aquestionnaire about how well they comply with district policy.One does not even know if the questionnaires were completed byprincipals, counselors, teachers, or secretaries.

Kiesling (1975) studied 5,800 children to test for theeffects f,f "instructional time" on criterion - referenced and norm-

referenced tests. What Kiesling called "instructional time" wasactual ly scheduled time taken from teacher reports andinterviews. No direct observation was made, and Kieslingreported that scheduled time had a slight effect on criterion-referenced tests, and no effect on norm-referenced lasts.

Smith (1979) and Kiesling (1975) studied scheduledinstructional time, whereas Jarvis (1962) studied allocatedinstructional time. None of them defined the terms they wereusing, except to take what information was offered to them byteachers, school officials, and school board policy. In no casewere children or teachers observed to see if reports of use oftime were accurate. Given what Smith herself and others haveobserved about the accuracy of teacher reports, it is doubtfulthat these data are reliable or valid. In addition, Smith (1979)said the a panel of experts found that the STEP II test she usedas a dependent variable was an inappropriate measure, becausethere was so little correspondence between what the children weretaught and the test items that the results could not be offeredwith confidence.

Perhaps the best illustration of how precisely defined termscan sort the significant from the muddy was provided by Karweitand Slavin (1981). They looked at the different effects thatinstructional time had on achievement depending on the kind ofmeasurement used. Their research spoke directly to the confusiongenerated by researchers using vaguely defined independent

18

12

measures. In the Karweit and Slavin work, pupil progress wascorrelated with four kinds of instructional times teacher -scheduled time; actual minutes allocated; engaged time; andengaged rate. The measure of teacher scheduled time was takenfrom lesson plans. Actual minutes allocated was tallied fromdirect observation of the number of minutes spent teaching. adistinct departure from many earlier studies. Engaged time wasdefined as that part of the instructional period in which astudent is on-task, or interacting with the materials. Engagedrate was defined as engaged time per actual minutes allocated.Except for teacher-scheduled time, these measures were obtainedthrough direct observation. Karweit and Slavin found distinctlydifferent relationships between these four measures ofinstructional time and achievement on the California Test ofBasic Skills. "The detection of effects of time measures issensitive to the proximity of the measurement used, with the

measures most accurately reflecting individual students use oftime showing the strongest effects" (p. 1651. That is, the moreprecisely the independent variable is measured, the more likelyit is to show an effect.

In their emphasis on measuring the effects of instructional

time on individual students, Karweit and Slavin are following aline of research that began with Bloom in 1956. Though that linehas been productive, it has not gone unchallenged in the clash oftheoretical viewpoints.

Divergent ,Theor Tess Different Procedures, Mean Different Results

In a Walt Kelly comic strip, Pogo was elucidating apolitical question for an insect. It was all a matter of viewpoint, explained Pogo. The insect looked up and replied. "Fromwhere I stand, there's only one point of view".

Two distinct points of view guide instructional timeresearch, each proposing different methods, and producingdifferent results. Bloom (1956, 1967) argued that learning takesplace in individuals, thus each student is the proper unit ofinterest. Smith (1979) criticized Bloom for this, preferring touse the classroom as the unit of analysis so that data could moreeasily be gathered and submitted to statistical analysis. Thoseopposing points of view underline most of the instructional timeresearch, leading experimenters to ask different questions, usedifferent methodologies, and to produce results that aresometimes in disagreement. Before Bloom (1956) instructionaltime researchers were concerned with how variables affected theaverage gain of a class, a school, or a district. Naturally,demonstrating changes in the mean level of pupil progress tooklarge numbers of students and required a statistical analysis toInterpret the results. Representatives of this line of researchgenerally correlated results of standard tests with reports ofinstructional time use and found that instructional time was aweak contributor to achievement. Under these circumstancesstatic variables such IQ, income level, and fathers, Jobs weremore highly correlated with achievement than instructional time

19

13

variables. For instance, Husen (1967) engineered a massive andinfluential study that included the role of instructional time inachievement.

Husen's (1967) international study of mathematics was alarge scale effort to describe the Instructional and familialcomponents of mathematics achievement in 12 countries. More than132,000 students and some 13,300 teacLars were involved In thiseffort. From their data Husen concluded that instructional timeand homework variables together accounted for only 3% of theassignable variation in the total mathematics score, a trivialcontribution. Far stronger, in Husen's view, was thecontribution of static variables--a point that was to be echoedby Smith in 1979. How could this study, involving distinguishedresearchers from around the globe, come to the astonishingconclusion that homework and instructional time contributedvirtually nothing to mathematics achievement scores? Thestrength of this study is that its data are uninterpretable.Husen admitted that he had little control over his independentand dependent measures. The independent variables, assignedhomework, family Income, instructional time, etc., were col lecteddifferently from one country to the next, and in no case wasdirect observation used. Various kinds of surveys or estimatesfrom headmasters and government officials provided the data. Thedependent measures inspired a similar lack of confidence in theirreliability. The mathematics tosts were constructed andtranslated into the 8 languages used by children in this study.Difficulties in translation and testing conditions led Husen towarn the reader that, "even as regards certain basic statisticalinformation, such as per pupil expenditures or enrollmentfigures, there is a lack of uniformity in data reporting" (p.287). Husen (1967) went on to say, "the difficulties indicatedabove should be kept in mind when interpreting someinconsistencies that appear in the findings" (p.287.). One ofthe most glaring inconsistencies is the assertion thatinstructional time and homework account for negligible effects onmathematics performances. While Smith (1979) Husen (1976) andother exponents of research that locked at mean differencescontinued to publish, a parallel course of development wascharted by researchers who were interested In the individual.These two lines of research go on with little crossover or impacton one mothers' work. Carroll presented his "Model of SchoolLearning" in 1963, and in one stroke showed that time could betreated as an independent variable, and that individual studentswere the most important unit of interest. The model says: timeto learn a task is a function of aptitude in a child's repertory,minus time saved by previous learning.

Carroll realized that the quality of instructional time, aswell as the history of each student, influenced the rate oflearning. Bloom (1974) used this time-based model as thefoundation for mastery learning. He claimed that most studentscould learn most skills if enough instructional time wereprovided. in summarizing the strengths of a time-based model,Bloom (1974) noted that: time can be measured with precision that

20

14

Is rare for educational variables; time can be measured on aninterval scale, with each unit of measurement equal, and with anabsolute zero; time can be used as a base to measure costs ofinstruction; perhaps mosi importantly time can be a measure ofinstructional effectiveness.

Among the researchers who have followed the Carroll-Bloom

line are Bennett (1976), who was forced to look at the classroomecology to account for unexpected results; Rosenshine andBerliner (1978), who focused the concept of instructional timeinto academic engaged time; the Beginning Teacher EvaluationStudy group (BTES), who sharpened the concept of on-task; theJuniper Gardens research group, who collectively developed andtested the concept of opportunity to respond. The work of thissomewhat divergent group shares several characteristics. Theseresearchers assume that individual students are the unit ofInterest, because learning takes place with the child, not withthe classroom or school. This group of researchers has evinced adetermination to pursue instructional time in its simplest form,pealing away extraneous constructs, and measuring time as it isactually used by children and teachers. One other common threadmay be the most important. This group of researchers directlyobserves the phenomena of interest, and measures them precisely.Questionnaire research -- hypotheses by mail - -wil I not be found inthis body of work.

Bennett's work (1976) Is particularly interesting because itshows the persistence of a researcher trying to account forunexpected results. This persistence took Bennett from hisoriginal line of inquiry and led him to examine instructionaltime. Bennett set out to compare the effects of teaching styles

on cognitive and social development of elementary students in theUnited Kingdom. His independent variable was continuum of twelveteaching styles. from formal to informal, with various mixturesin between. Formal classes were teacher directed and curriculumoriented. Informal classes had flexible physical structures,allowed students wide ranges of choice, and encouragedindependent work. Mixed classes had various proportions of thesecharacteristics. and others such as homework assignments, freedomto move in the classroom, and other descriptive variables.

Bennett found that children in informal classes did significantlypoorer in reading than children from formal or mixed classes. In

mathematics children from formal classes gained skills at abouttwice the expected rated, whereas children from mixed or informalclasses gained skills at about 'half of the expected rate. Therate differences in mathematics culminated in achievementdiffences of 3-4 months of pupils progress per year. Thesepuzzling differences in mathematics culminated in achievementsent Bennett back to his data and the classrooms. His post -doc

analysis showed that the critical differences between formal andother classes was the amount of structured learning timeavailable.

Rosenshine and Berliner have used Carroll's framework toextend and develop some of the ideas in the Model of School

21

15

Learning. In a series of papers (Rosenshine, 1976, 1978;Rosenshine & Berliner, 1978) they developed the related conceptsof academic engaged time and direct instruction. They saw theirwork as significant partly because it focused attention ondynamic variables (socio-economic status, birth order, IQ) thatheld little significance for policy or for teaching. Academicengaged time, as developed by Berliner and his colleaguesBerliner, Fisher, Filby & Marl iave, (1976) refined the definitionof on-task, and made it a more active concept. Academic engagedtime was defined as content covered in combination with studentattention to materials. Of course, these kinds of data can onlybe collected by going into classrooms and watching children.Rosenshine (1976, 1978) made an additional point: children learnthrough direct instruction, not by accident or through self -directed discovery. Bennett's (1976) work confirmed this point.In child-centered classrooms children failed to learn academicsubjects, mainly because they were never exposed to thecurriculum. Left to choose, most children choose math, spelling,and greaser. By contrast, children in structured classrooms dowell in academics precisely because they are taught a curriculum,given homework, expected to master specific skills, and havelarge blocks of time devoted to practice and mastery.

Perhaps no group of researchers has pursued the concept ofinstructional time as extensively as the Juniper Gardens group.In a series of papers the concept of opportunity to respond wasdeveloped (Delquadri, Greenwood, Stratton, & Hall, 1933;Greenwood, Dequadri, Stanley, Sasso, Whorton & Schulte, 1981;Hall, Delquadri, Greenwood & Thurston, (1982). By opportunity torespond they meant the amount of time available for activepractice of the skill of interest. In spelling, for instance,opportunity to respond refers to practicing the writing or oralspelling of the words to be mastered. It does not mean using aspelling word while tolling a story to the class, or drawing aline from the spelling word In one column to a list ofdefinitions In another column. If the skill of interest islearning French vocabulary, then seeing a film about France doesnot qualify as opportunity to respond. Active responding isessential.

The concepts of academic engaged time, direct instruction,and opportunity to respond, share several characteristics.First, they are direct measurements of how children and teachersspend their time. These data cannot be collected any other way.Secondly, each has gone beyond the more passive concept of on-task, which by itself says nothing about how a child spends time,merely that the child is non-disruptive. Thirdly, each of theseconcepts is tied directly to student achievement. When academicengaged time, or direct instruction, or opportunity to respondare increased, reliable increases In student skills follow. thatIs not too surprising. That a child must practice themultiplication tables before they can be mastered may be selfevident, yet such a simple concept Is only gradually beingaccepted by education researchers.

22

16

In addition to refining instructional time into its

components, practice, reinforcement, and error correction, theJuniper Gardens group have taken another tack. They have takenthe one step back and watched the instructional activities withthe eye of an ecological psychologist. They have begun to seekcontexts for teaching and learning so that instructionalactivities that contribute most directly to learning can be used.This effort has involved collecting extensive ecological data,and associating various activities and techniques with specificstudent outcomes. The Code for Instructional Structure and

Student Academic Response (CISSAR) Is a system of observationthat simultaneously tracks classroom structure, teacher activity,and child responding. It is sensitive enough to show, forinstance, that using educational media Is less effective thanpaper and pencil for teaching skills. By focusing on theseecological variables the Juniper Gardens group has extended thetraditional preoccupation of behavior analystswith conseviences.

II 1 Lout d SAme. lima Mama 112 g fisa fromHera?

Two lines of instructional time research have beencontrasted, and it seems evident that the most productive workhas come from researchers who have hewed to precise observation,precise definit'on of terms, and assiduous selection of design.Further, some researchers have added the dimension of ecologicalobservation to the powerful tools of behavior analysis. Theresulting science shows promise because it holds the secret ofself correction: the ability to frame mutually exclusivehypotheses. Research that is descriptive or correlational may benecessary in the early stages of delimiting research questions.But as Platt (1964) stated so forcefully, only research that canreject one of two mutually exclusive hypotheses is capable ofadvancement. The rest is a Sargasso Sea of aimless hypothesesand endless circular debate. A behavior analytic approach toinstructional time can provide the structure for freeing thosehypotheses and the ecological perspective can insure that suchresearch Is conducted in a meaningful context.

23

17

References

Bennett, N. (1976). Teach I ng Ltd= and Aiwa Driss,Harvard University Press.

Berl iner, D. C., Fisher, C. W., Filby, N., & Marl lave, R. (1976).Proposal for Phase III of Bloginnina Teacher EvaluatioaStudy. San Francisco: Far West Laboratory for EducationalResearch and Development.

B 1 oan, B.S. (1956). Taxonomy j education ob jecti yes: theclassification jat hducatioR goats: liandbj cognitive

an. New York: McKay.

B I oan, B. S. (1967). alma characteristics And school l earning,

New York: McGraw-Hill.

B 1 °cm. B. S. (1974). T I me and 1 earn 1 ng. &ark.= Euchsalogat22, 682-688.

Carroll, J. B. (1963). A model of school learning. TeacherCol 1 age, fumed, fib 723-732.

Del quadri, J., Greenwood, C. R., Streeton, K., & Hall, R. V.(1983). The peer tutoring spelling gone: A classroomprocedure for increasing opportunity to respond and spellingperformance. Education And treatment LA Children, fa" 225-239.

Greenwood, C. R., Del !quadri, J. C., Stanley, S. 0., Terry, B., &Hall, R. V. (in press). Assessment of eco-behaviorinteractions in school settings. Behavioral Assessment

Greenwood, C. R., Del quadri, J., Stanley, S., Sasso, G., Whorton,D., & Schulte, D. (1981). Al locating opportunity to learn asa basis for academic remediation: A developing model forteaching. ligoagraRh in Maid= Disorders, /MOOD 22-33.

Hal I, R. V., Del quadri, J., Greenwood, C. R., & Thurston, L.(1982). The importance of opportunity to respond inchi I drenos academic success. In E. B. Edgar, N. G. HerJ. R. Jenkins, & C. G. Pious (EO.), Manta I ly hand !cappedchildren: Education And training. (pp. 107-140).Baltimore: University Park Press.

Hook, C. M., & Rosenshine, B. V. (I 979). Accuracy of teacherreports of their classroom behavior. Bev law a EducationalResearch, .49, 1-12.

Husen, T. (Ed.) (1967). International study achievement lamathematics: A ccanar i son a haw Countries (Vol s. I &II). New York: John Wiley & Sons.

Jarvis, 0. T. (1962). Time I !plaints and pupil achievement IA

24

18

the intermediate niamentary grades, A Texas Gulf Coast Study.(ERIC Document Reproduction Service No. ED 035 063)

Karwe it, N., & S I av in, R. E. (1981). Measurement and mode I ingchoices in studies of time and learning. American EducationalResearch ADurnig., 1,8, 157-17 I .

K les 1 ing, H. (1975). Ma relationship time saaat, mg reelinginstructiort 1 reading gel= MI measured, ty porm-referenced andgritarigrzte renced huts, (ERIC Document ReproductionService No. ED 116 135)

National Commission on Excellence in Education. (1983. A patloprisk 1111Peratimm educational reform.

Washington, D. C. U. S. Government Printing Of ice.

Platt, J. R. (1964). Strong inference. $c ience, Iral, 347-353.

Rosenshine, B. V. (1976). Classroom instruction. In N. L. Gage(Ed.), Lift aLychiaggx tairaing. mittids.. Seventy -fifthy.aar.haQk the National Sw.111.14. Inc the. Stud, Education.Ch icago: University of Ch icago Press.

Rosenshine, B. V. (1978). Academic engaged time, contentcovered, and direct instruct ion. Journa I a gducation, INA38-66.

Rosenshine, B. V., & Berl iner, D. C. (1978). Academic engagedtime. British ,Journal saf. Imam. Education, Al 3-16.

Smith, N. M. (1979). Al location of time and achievement inelementary social studies. Journal gi.auratignAL Research,12, 231-236.

25

19

CHAPTER III

ASSESSING FCO-BEHAVIORAL RELATIONSHIPS IN SPECIAL EDUCATIONOUTCOME RESEARCH OR "WHERE IS THE INDEPENDENT VARIABLE"?

CHARLES R. GREENWOOD

Abstract

Much of special education is concerned wJh issuesassociated with the identification and placement of children intospecial education programs. Risking oversimplification, myobservation Is this, a =hat A =Atha scis nc. mid brawl=at Instruction thart fact, cad. It XeILLI1 saw that Iliaissues a student, !dent 1 if [cation And Are warner& Egag,allailantad spacial education QL st least drastically reducedIna sharp focus a= the praluama at taachlag. and instruction,Specifically, I intend to address In this chapter issues related tothe eco-behavioral process assessment of instruction, processachievement relationships and their implications for research oninstruction.

Iha Eaklas

The problem we all share, of course, is the fact that underour current educational practices, there remains wldo variance instudents' rates of academic development over their time inschool. Historically, we have looked "inside" the child toaccount for this variance. I am sure you will agree that amajority of our research literature in special education is ofthis type. This literature has looked for brain damage, delaysin development, loss of sensory and perceptual functions, strongand weak modalities, loss of cognitive processes, etc. Thisliterature has given us instructional strategies that haveattempted remediation of delayed developmental stages prior toteaching academic skills or which have sought matchinginstruction to student strengths in the central nervous system,in cognition, in sensation, in perception, and/or in modalities.More recently in our history, however, we have had moments whenwe have locked at the environment and at environmental deficitsto account for this variance. We have assessed the physicalsetting, looked at student's home environments for differences inculture, language, and econamica, assessed teacher skills andtraining, etc. Based upon this literature, we have modified thephysical environment within the classroom, purchased newcurricula, increased or decreased the length of the school day,developed new service delivery patterns, provided students withmeals and health services, increased teacher certificationrequirements, created new instructional environments Ce.g,

MMII411.0011.11411114111.1Address Made by the Author to the First Annual PitsburghSymposium on Research with the Handicapped, May, 1984, Pittsburgh, PA.

26

20

resource rooms), introduced the most modern electronictechnologies, reduced or increased stimuli, etc. And yet, theproblem remains with us; the variance In academic developmentpersists In the face of these substantial, costly, and certainlydramatic efforts. Based upon the most recent research, however,we are now Just realizing that it may be the interaction betweeneach student's instructional environment and behavior thataccounts for this variance.

Eco og I cal -Balm/Iona interact1an Approach

An eco-behavioral interaction approach to the problem ofvariance in students development is not new in theory (Bijou,1979; Bijou, Peterson, & Ault, 1968; Bronfenbrenner, 1979;Brophy, 1979; Dunkin & Biddle, 1974; and Lindsley, 1964) orinstructional practice (cf. Becker, 1977, 1978). In fact,instructional practices based upon the general operant model

anteceder t stimulus, student response, consequent

stimulus), an environment-behavior interaction theory, have hadsubstantial impact upon the development of effectiveinstructional systems (Leinardt, Zigmond, & Cooley, 1981).Direct instruction, precision teaching, and personalized systemsof instruction, are all specific examples of effectiveinstructional systems based upon an environment-behaviorinteraction approach. What is new, however, are directassessment measures that can provide an analysis of this on-going interaction within a classroom setting.

An eco-behavioral assessment of instruction captures boththe important ecological variables In the classroom (e.g, teacherbehavior, grouping arrangements, tasks, activities ofinstruction, etc.), and specific student behaviors (GradenThurlow, d Ysseldyke, 1983; Stanley & Greenwood, 1981). sincethese ecological variables are represented within theobservational record frequently and systematically, the student'sbehavior can be analysed in relationship to these importantvariables. One of the fundamental problems in past educationaloutcome research has been the inability to obtain such animportant measure of the educational experience (Brophy, 1979;Good, 1979; Strain & Kerr, 1982). As a result, we have not beenable to explain the results of most outcome/efficacy studies,even in well-controlled outcome research. When outcome researchfocuses soley upon assessment of outcome variables (i.e., tests,grades, etc.), we simply cannot meaningfully relate outcomes toevents that transpired within the program. Frankly, because ofthis limitation, we can no longer afford to conduct outcomeresearch In the absence of high fidelity process measures.

The data in Table 1 Illustrate the reults of an ecc,behavioral process assessment of one student's instructional dayat school. The Table is organized by instructional activity (thesubject of instruction on the left side) and specific ecologicalarrangements in association with the student's academic behaviorduring that arrangement (displayed on the right of the Table).

27

21

Thus. for example. we can see that reading covered 36% of the

observation (50 minutes). During al 1 of reading the student'sacademic behavior was as follows: 10.7% writing, 4.3% silentreading. 1.7% academic to and .7% answering questions. Thestudent's composite academic responding was 17.4% overall, thesum of the separate categories. These data constitute a molardescription of reading and student behavior. A molecularanalysis of eco-behavioral interaction is displayed within thecenter of the Table section devoted to reading. Here we SO4 thatthe most frequent arrangement of stimlus events used by theteacher (32.3% of reading time) was R TSD SG AS T. (1.0., reading

Table 1

An Ecom-Behavioral Analysis of a Student's Instructional Day

Arrangements p(Al/An)

Activity AC TA ST TP T8

IMPOIMMStudent Academic Response p(R/Ai)

COMPO-W AGP RA RS TA ANQ ASK SITE

Reading R TSD SG AS T 32.7 8.2 3.1 1.0 12.3(36%) R TSD SG S T 10.3 0.050 min. R LL SG AS T 9.7 0.0

R *WS SG AS T 8.7 42.3 3.8 46.1R W8 SG AS T 5.7 11.8 35.3 47.1R PP SG AS T 5.7 35.3 35.3

Reading All (Baserate) 100.0 10.7 4.3 1.7 0.7 17.4

Math M PP EG S T 32.2 44.6 1.8 46.4(21 %) M PP EG AS T 8.6 46.7 46.729 min. M PP EG 0 NR 8.0 35.7 35.7

M PP EG S NR 6.3 45.5 45.5M OM EG IF T 13.2 13.0 4.3 17.3M TSD EG IF T 4.6 12.5 12.5

Math All (Baserate) 100.0 33.3 4.0 0.6 37.9

Spelling S PP EG IF T 36.1 46.2 46.2(1%) S PP EG S T 22.2 50.0 50.06 min. S PP EG B T 19.4 71.4 71.4

S PP EG AS T 19.4 0.0 14.3 14.3

Spelling All (Baserate) 100.0 44.4 2.8 47.2

Language I. PP EG AS T 34.6 17.8 2.2 20.0(16%) I. PP EG AD T 12.2 23.2 6.3 29.522 min. I. PP EG AD NR 7.7 28.9 10.0 38.9

I. OM EG SG T 13.1 13.1 5.9 19.0

Language All (emirate) 100.0 6.2 3.8 10.0

28

22

Soc/Study SS TSD EG AS T 83.3(75) SS TSD EG AS NR 13.09 min. SS TSD EG AS D 1.9

SS TSD EG AS OT 1.9

33.342.9

33.342.9

0.00.0

Soc/Study All (Baserate) 100.00 33.3 33.3

Free/ FT RR EG AS T 47.1

Time FT RR EG AS NR 29.4(15) FT RR EG S OT 8.86 min.

12.5 18.8 31.3

0.00.0

Free All (Baserate) 100.0 8.8 8.8 17.6

Trans,- TN FP EG IF OT 16.8tion TN FP EG AS T 14.0(13S) TN FP EG 0 NR 9.318 min. TN FP EG AS OT 6.5

TN FP SG AS T 9.3

10.0

0.00.010.00.00.0

Trans,- All (Baserate) 100.0tion

4.7 0.9 0.9 6.5

Note. This table contains classroom ecology/student behavior

interaction data for one student over one day at school basedupon observation with the CISSAR Code (Stanley & Greenwood.1981). The percentage occurrence of specific ecologicalarrangements is reported with response conditional probabilities(e.g.. writing, reading aloud, composite, etc.). Thearrangements and their percentage occurrence are found to theleft in each panel, while the conditional probability valLes arefound to the right. To make inspection of the Table clear. anumber of conventions were used to organize the informationwithin it. First, arrangements are organized withininstructional activities. Second, the arrangements in each grouphave been sorted by task and structure variables to aid in theevaluation of different but somewhat similar arrangements. Thus,all paper/pencil. entire group arrangements are listed before thepaper/pencil. smell group arrangements, etc. Within the Table.comparisons can be made: (a) between arrangements within eachactivity. (b) between common arrangements occurring in differentactivities. and (d) between individual arrangements and thebaserate for each activity.

Table abbreviations are as follows: Category Codes--Activities (AC): R = Reading. M = Math. S = Spelling. L =Language. SS = Social Studies. FT = Free Time. TN = Transition.

29

23

Task (TA): RR = Reader, PP = Paper/Pencil, TSD = Teacher-StudentDiscussion, WB = Workbook, WS = Worksheet, LL = Lecture, OM =Media, FP = Fetch/putaway. Structure (ST): EG = Entire Group. SG= Small Group, I = Individual. Teacher Position (TP): AS = AmongStudents, S = To the Side, AD = At Desk, IF = In Front, B =Behind, 0 = Out of Roan. Teacher Behavior (TB): T = Teaching, NR= No Response, 0 = Disapproval, OT = Other Talk. StudentBehavior Codes: W = Writing, AGP = Academic Game Play, RA = ReadAloud, RS = Read Silent, TA = Talk Academic, #NQ = AnswerQuestion, ASK = Ask Question, AT u Attention.

teacher + student discussion + small groups + teacher located amongstudents + teacher engaged in teaching behavior). Studentbehavior during this single reading arrangement was 8.2% writing,3.1% academic talk, 145 answering questions for a compositescore of 12.3%.

These molar and molecular data a low one to ask and answerseveral types of questions about the student's instruction andbehavior. First, (and most molar), the time devoted to separatesubject matt.r over the day can be seen. Reading was taughtduring this observation for 365 of the time or 50 minutes.Spelling and Free Time occurred least at 1% or 6 minutes duringthe observation. Second, the student's academic behavior can becompared across activities. For example, spelling instructionresulted in the highest levels of academic behavior, 47.2%, whileas one might predict, transition activities produced the least at6.5%. It was also evident that major activities such as reading,17.42, and language, 10.0%, resulted in low levels of studentacademic behavior compared to math, 37.9%, spelling, 47.3, andsocial studies, 33.3%. Third, (and most molecular of all),within activities one can examine the relationship of specificarrangements to student behavior. In reading, it was evidentthat the most used arrangements involving teacher-studentdiscussion or lecture, (RTSD SG AST, RTSD SG ST, and R LL SGAS 1), were least related to academic behavior (12.3, 0.0, and0.0, respectively). However, the remaining arrangements in whichworksheet, workbook, or paper/pencil tasks were used resulted Incomposite scores ranging from 35.3 to 46.1%, two to four timeshigher. Thus, the molecular analysts of eco-behavioralinteraction reveals information about the configuration of theinstructional session and their behavioral relationships. Italso suggests instructional changes, the reconfiguring ofarrangements, which might be tested in order to increase studentacademic behavior. Facilitator arrangements, those with highbehavior relationships (above 30% for example), might be usedmore often while the use of non-faciliating arrangements might bereduced or modified. Process assessment data such as this can beused as a feedback device for teachers interested in increasingstudent academic responding.

30

24

naCormlatIon fatween,gco-Behavioral Interaction gadAchieve ant

Between 1978 and 1981, our group at the Juniper GardensChildren's Project, reported a relationship between low readingachievement, low IQ, and fourth grade students daily levels ofacademic behavior (Greenwood, et al., 1981; Hall, Delquadri,Greenwood, d Thurston, 1982). This appeared at about the sametime that other researchers were reporting a relationship betweenstudents' academic learning time or engaged time and theirachievement (e.g., Rosenshine, 1979). What was significant aboutthese data is that these researchers were basing their findingsupon direct observation of what teachers and students did in theclassroom, as opposed to what teacher's said they did or whatteacher's schedules indicated would be done during classroom time(Brophy, 1979).

Our data suggested that low SES students, well known to beat risk for academic retardation, were being educated within

classroom environments and instructional methods associated withlowered levels of academic responding (i.e., writing, academicplay, reading aloud, reading silently, academic talk, askingquestions, and answering questions,. This daily difference incomposite academic behavior averaged 5% or about 11-13 minutesper day between Title 1 and non-Title 1 students (Stanley I.Greenwood, 1981). The data In Table 2 illustrate this differencefor both a fourth grade and a first grade sample.

Table 2

Percent of Day Spent in Academic Behavior by Groups

Group Academic Behavior CodesW AGP RA RS TA ANS ASK COMPOSITE

Fourth Grade Sample (:980'

Title 1 15.7 1.2 0.5 5.0 1.5 0.8 0.4 25.1 48* *

Kritle 1 17.2 1.3 0.6 8.2 .2.3 0.7 0.3 30.6 45

First Grade Sample (1984)

Chapter1 18.0 0.7 2.4 8.4 1.9 0.9 0.3 32.5 25* * *

NChapter1 24.0 0.8 1.3 9.8 1.1 0.1 0.2 37.6 23

Table 2 Continues (pg. 2)mommemmommodim14.1.

25

*p< .05

Note. Academic Behavior Abbreviations are: W = Writing, AGP =Academic Game Play, RA se Reading Aloud, RS = Reading Silent, TA =Talk Academic, ANS = Answer Question, ASK = Ask Question, N =Number of Students Observed. Group Abbreviations are: NTitle 1 =Non-Title 1, NChapter 1 = Non-Chapter 1. The Composite score isthe sum of the seven code scores.WFurthermore, the range in academic behavior across individualstudents in inner-city schools was large, from 10 to 109 minutesspent in academic behavior per day. A secondary analysis ofthese data confirmed that students in the sample attending Title1 schools (now Chapter 1) had fewer academic interactions in eachof their major instructional activities during the day (i.e.,

reading, math, language, spelling, etc.). We also have recentlyreplicated the fact that students may be well-behaved at schoolbut not be actively engaged in academic behavior. These data aredisplayed in Table 3. Students' means for total appropriatebehavio-, which includes both academic and task managementresponses, are compared to students' composite academicresponding. These data demonstrate that the correlation is notperfect. In both samples, some schools high in total appropriatebehavior were not high in academic behavior (See for example,Welborn and Lindbergh).

Table 3

Appropriate Behavior vs. Academic Responding Summary by Schools

School Appro. Behavior Acad. Response NM Rank Students

Fourth Grade (1980 Sample)

Kennedy 90.4 31.7 1 78Quindaro 90.2 26.2 3 65Welborn 87.9 29.5 2 69Banneker 81.6 23.6 4 59

Overall M 87.5 27.8SD 4.1 3.6

32

26

Table 3 Continues (pg. 2)MPIMIKINMOOMOMM1!

First Grade (1983 Sample)

Kennedy 94.8 41.9 1 11

Lindbergh 89.9 23.8 9 14

Parker 87.9 41.0 2 11

Eugene Were 86.9 30.0 8 13

Chelsea 85.1 36.8 3 12

Fairfax 83.5 30.4 7 10Qeindaro 78.7 35.2 4 12

Hawthorne 73.1 33.1 6 12

White Church 69.5 33.8 5 12

Overall M 83.3 33.7 107

SD 12.3 11.0

Note. These data are based upon the CISSAR code (Stanley &Greenwood, 1981). Appropriate Behavior is defined by the sum oftwelve codes devoted to academic responding and task management.These are: (Writing, Academic Game Play, Reading Aloud, ReadingSilent, Talk Academic, Ask Question, Answer Question, Attend,Raise Hand, Look for Materials, Move, and Play Appropriate). TheAcademic aspansa score is defined by the sum of the first sevenbehaviors Just listed, Writing through Answer Question. Thus,the difference between these two composite scores is accountedfor by the five task management codes, Attend through PlayAppropriate.MIMINMENwnwalMormimlialwiNIa=mminisalmo

Because our code was developed to broadly represent ecologicalfactors in the observational record, in addition to studentbehavior, it became possible for us to also examine the contextsof instruction in which student behaviors occurred. Thisrevealed that Title 1 teachers In comparison to non-Title 1teachers, used more media, and more discussion, which were nothighly related to studentslactive academic behavior (e.g.,reading aloud or academic talk). Conversely, the non-Title 1teachers used more reader, and paper/pencil tasks in theirlessons which yielded a greater, relationship with students'academic behavior (Greenwood, Delquadri, Stanley, Terry, & Hall,in press).

In subsequent experimental research using classwide peertutoring in reading, spelling, mathematics, and vocabularyinstruction, we have reported findings that have supported theearlier results based upon descriptive designs. Increasedacademic responding is related to increased academic outcomes.Compared to instructional methods often used by teachers(baseline phase), peer tutoring increased the presence of keystimuli during the session readers, paper/pencil,

33

27

worksheets,), increased active student responding (e.g., readingaloud, academic talk, writing), decreased task managementresponding (e.g., attending, looking for materials, raisinghand), decreased competing student behaviors (e.g., lookingaround, disrupt, etc.), and increased gains on Friday tests andreading rate checks (Delquadri, Greenwood, Stretton, & Hall,1983; Greenwood, Delquadri, & Hall, 1984; Greenwood, et al., In

press).

Thus, not only can we point to wide variance in academicachievement, but also a related wide variation In students'academic behavior measured either as molar time spent engaged orIn terms of specific academic responses, and in teachers'arrangement of the classroom environment during instruction.This variance has been noted across schools, classrooms, andindividual students. However, this variance can be reduced andcontrolled with implementation of an effective instructionalprocedure. These data suggest that one causal variable withrespect to student achievement will be each teacher's ability togenerate high levels of academic responding within specificinstructional activities and over the school day.

Icetions ism Instruction

I have now discussed the problem of academic delay,described process assessment based upon eco-behavioralinteraction, and have argued a correlational and functionalrelationship between students' achievement and eco-behavioralvariables at school. The next topics, then must be devoted tothe implications for instruction. The relevant questions are:"What are the features of instructional methods required toreduce this variance in student achievement"? and "How can thesevariables be addressed within special education outcomeresearch"?

Iffectlite Instructional =sheds, The research literature onteaching which has included process measurement tells us thatinstructional methods that: (a) provide freeyeat apportunItits, fQresponds Shl imam that ALL students roma correctly. And laArstylAn rainfarrament j student raspandina are mosteffective in terms of traditional academic test measures (i.e.,tests, grades, etc.). Given that these are the basic goals ofany instructional method, the question then is "How do weimplement these events within the classroom?" Certainly, otherconsiderations are necessary for a sufficient analysis ofinstruction, (e.g., educittional objectives for each lesson, taskanalysis, programming/sequencing of tasks,introducing/practicing/reviewing, etc.), but these, I wouldargue, are not as fundamental as the response opportunityvariables. In other words, all students can realize achievementgains in any curriculum, given frequent opportunity to respond,correct responding, and positive reinforcement. To affectoutcome measures, we will have to actively engage students in theacademic behavior, rather than simply expose them to the academicmaterial. Since students who spend their school time In

28

unresponsive instruction will realize lower outcome gains overthe same amount of time, as will students in responsiveinstruction, we must create instructional environments thatproduce high rates of academic responding for all students. Weshould also recognize that instructional procedures vary in theirability to create and maintain student responding. Thus, theselection of appropriate procedures will necessarily be basedupon research. Procedures must also meet a reasonable criterionof efficiency in terms of teacher implementation. Proceduresthat work but are difficult or effortful to implement will not bemaintained.

frayliumagpaadjulitt ROMA, In order toactively respond, students must be presented with a task stimulusand a teacher command stimulus occasioning their response. Boththe form and rate of these opportunities are important to therate of students' academic Interactions. The common form ofthese opportunities are teachers' questions, task presentations,and/or commends to respond. In naturalistic instruction, thesecommands to individuals have been estimated to be as low as 1.40per minute (Grinstead, 1982). We have also reported occasionswhen low reading groups, scheduled last, were not held at all dueto extra time spent with more advanced reading groups.Obviously, students have no opportunity to learn in these illmanaged circumstances. in Direct Instruction reading groups,where a frequent task presentation rate is emphasised, theopportunity rate has been estimated at 12.0 task presentationsper minute (Canine, 1976; Carnine & Silbert, 1979, p. 26). InClasswide Peer Tutoring methods, the rate that first grade tutorspresent new spelling words to the tutoring has averaged 6.0 perminute (Kohler, 1984).

The second form of opportunity is present when students usecurriculum materials and the students typically determine theirown task presentation rate. When the pace of task presentationand responding is selfmanaged there Is obviously a widevariablity in rate and content coverage during a period ofinstruction. This variability can be reduced by individualizingmaterials or by using contingencies that reinforce higher ratesof task completion. Timings, in which students complete as manyitems as possible with In several minutes, are another procedurefor accelerating task presentation rate within curriculummaterials.

111) LU. Students Bassaid.. Instruction in which allstudents are required to respond, either to the sameopportunities or to different opportunities, is more effectivethan those that allow variation in who responds. Instruction inwhich teachers lecture or discuss with students seldom requiresall students to respond in an active way. Rather, individualstudents are called on or asked to volunteer a response. InDirect Instruction reading groups however, teachers require groupresponding to their response prompts. All children respond tothe same prompt signal. Students who are late in responding, whohesitate, or who are wrong, are then drilled individually in

35

29

order to ensure that they can emit the response. In PrecisionTeaching timings, all students start and stop at the same time.In Classwide Peer Tutoring, all students are simultaneouslyresponding to individual task presentations of their tutors.

(111) positive Reinforcement, To the extent that studentsare provided reinforcing consequences for their academicbehavior, the probability of their continuing to respond isincreased. The social reinforcement that students receive innaturalisitic instruction is known to be very infrequent (White,1975). It is also well-accepted that other forms ofreinforcement (e.g., performance feedback, token reinforcementwith backup consequences, etc.) also are used infrequently innaturalistic instruction. Moreover, the rules of reinforcereffectiveness in terms of immediacy, schedules, and avoidence ofsatiation are either unknown or simply not widely practiced byteachers.

Effective instruction takes full advantage of the positivereinforcement principle. Teachers in Direct Instruction praisecontingent upon correct student response or for improved testscores. Point earning is used for tutee responding In ClasswidePer Tutoring, as are interdependent group-orientedcontingencies, as is performance feedback through the posting ofstudent's point totals and Friday test scores. Similarprocedures are used in Precision Teaching to reinforce studentgains in correct response rates.

ResearchDulsaja, Sonja Education Outcome Research

Since the mission of special education is student change andgrowth, outcome research must focus directly upon student change.Outcome research must by definition be the study of dynamic andnot static phenomena. Research of this type requires repeatedassessments of outcome variables over time in order to examinegains or change from occasion 1 to occasion 2. Procedures thatare effective producing gains are those needed by the field.Static research designs (e.g., posttest only designs, includingmost evaluation designs), are simply not able to provide preciseinformation on child change before, during, and after treatmentand are therefore, inappropriate for examining the causal effectsof special education. Simillarly, designs based upon multipleregression models that interrelate variables assessed at onepoint in time will not qualify as a useful outcome researchdesign. Designs Involving experimental manipulations of specialeducation variables (e.g., teaching variables), and includingrepeated measurement of student change over time are needed.Designs that meet these criteria include single-subject researchdesigns (e.g., reversal, multiple baseline, or alternatingtreatment manipulations) and experimental- control group designswith repeated measures. Single-subject designs are particularuseful for discovering functional relationships between specialeducation treatments and outcome variables. Experimental-controlgroup designs with repeated measures are useful in establishingthe generality of these initial discoveries across units of the

'6

30

service delivery system (i.e., students, teachers, schools,districts).

Research that is particularly needed are studies that alsocombine repeated measurement of outcome with educational processvariables (e.g., eco-behavioral interaction and studentachievement). As previously discussed, the interpretation ofattained outcomes is enhanced by the inclusion of processmeasures. While one may conclude in experimental outcome studiesthat the treatment was in fact causally related to gains inoutcome variables, without process assessment, It is difficult toknow little more. Why did the treatment work? Were theexperimemal variables implemented as planned or were they onlypartially implemented? What aspects of the treatment related tochange in the outcome variables and which did not? Was studentbehavior in relation to the treatment as expected or otherwise?If there was no change in outcomes was it due to a treatment thatapparently is not effective and should be discarded or was it dueto a failure of the staff to implement as planned?

Within this framework of process-outcome research,investigators must continue to build into their investigationsthe means to account for special education as the independentvariable. Eco-behavioral interaction is one means for doing thissince specific variations of the instructional ecology can beexpressed in terms of student behavior. This is a dynamicanalysis since change in student behavior can be noted inrelationship to change in instructional variables. Thismetholodolgy allows one to address questions concerning studentbehavior that transcend the well- established findings concerningstudent's levels of academic engagement and/or learning time.Future studies must examine the important topographies of studentbehavior in relationship to instruction with the intent ofrevealing the logic and sequence of these interactions, if we areto provide teachers with precision advice on how to teach mosteffectively.

Research must continue the search for effectiveinstructional procedures. However, research is more urgentlyneeded to identify the variables that account for teacher'simplementation of effective classroom procedures. If increasedopportunity to respond is related to increased studentperformance, what affects teacher use and generalization of thisprocedure over time? Given effective Instructional variableswith clear histories of outcome'research support, what are thefactors that ensure teacher selection and use of these proceduresin the classroom? What are the factors that account for teachersdrifting from established procedures and hov can this beprevented or modified?

Similarly, outcome research must address questionsconcerning the effects of partial or incomplete implementation.Although teachers often seek to create new and interestinginstructional procedures, they must also understand that variancein outcome performance is related to deviation from effective

37

31

teaching procedures. Creativity must be channeled within thecontext of the use of effective standard practices If we are toever approach the GA1 of equal outcomes for children in specialeducation. As we continue the discovery of effectiveInstructional procedures, discovery of the means to ensurequality implementation of those procedures will become the mostpressing issue within special education outcome research.

Summary And elisions

This Chapter began with the supposition that the majorpreoccupation in special education with the identification andplacement of students was in fact a direct result of a failure todevelop a powerful, widely practiced technology of instruction.It was further argued that low academic achievement and variancein students rate of academic responding in the classroom settingis related to instructional practices and the daily productionof student academic behavior. Will it ever be possible toprovide students equal outcomes in special education? Researchon teaching has revealed that instructional methods which: (a)provide frequent opportunities to respond, (b) allow all studentsto respond, and (c) provide reinforcing consequences forresponding are most effective increasing student gains inacademic outcome measures (e.g., tests, grades, etc.). The newgeneration of outcome research in special education must focusupon the assessment of student change in terms of both outcomesand instructional processes. This research will enableresearchers to: (a) tease out the important ecological variablesand stimulus arrangements related to optimal student performanceduring instruction, (b) precisely control and compare alternateinstructional procedures, and (c) identify Important factors inthe implementation of effective instructional procedures.Research of this nature, based upon experimentation, will beimportant to the development of a powerful and widely practicedtechnology of instruction in special education.

32

References

Backer, W. C. (1977). Teaching reading and language to thedisadvantaged: What we have learned from field research.}larvard Educational Review, 42,, 518-543.

Becker, W. C. (1978). The national evaluation of follow-through: Behavior-theory based programs come out on top.Education AMAdAm Society, 1Q, 431-458.

Bijou, S. W. (1981). The prevention of retarded development indisadvantaged c h i l d r e n. (pp. 29-46). In M. J. Begab, H. C.Haywood, & H. L. Garber, (Eds.), Paychosocial Influences inLabiate aarlarmances lasmas and theories slayalapmant LYskl.11. Baltimore, MD: University Park Press.

Bijou, S. W., Peterson, R. F., & Ault, M. H. (1968). A methodto integrate descriptive and experimental field studies atthe level of data and empirical concepts. Journal a

APPlilidashadatAnalYsis,e1J 175-191.

Bronfenbrenner, U. (1979). Contexts of child rearing: Problemsand prospects. Amedranpsychoinots.t, 844-850.

Brophy, J. (1979). Teacher behavior and its effects. Journal satEducational Psycho 11141L, 11., 733-750.

Carnine, D. W. (1976). Effects of two teacher-presentationrates on off-task behavior, answering correctly, andpartIcIpation. Journal Qf. Appl led Behavior Analysis,

199-206.

Carnine, D. W., & Silbert, J. (1979). Direct instructinmaReading. Columbus, OH: Charles E. Merrill.

Dunkin, M., & Biddle, B. (1974). Ile study a A.New York: Holt, Rinehart & Winston.

Good, T. L. (1979). Teacher effectiveness in the elementaryschool. Journal a luau Umatilla, 30., 52-64.

Graden, J., Thurlow, M. L., & Yssel dyke, J. (1983).Instructional ecology and academic responding time forstudents at three levels of teacher-perceived behavioralcompetence. Agurnald, Experimental Child Enychnilwy,36A 241-256.

Greenwood, C. R., Delquadri, J., & Hall, R. V. (1984).Opportunity to respond and student academic performance.(pp. 55-88). In W. L. Howard, T. E. Heron, J. Trap-Porter,& D. S. Hill, (Eds.), Focus DA. MAXI= analysis LLaducatlon. Columbus, OH: Charles Merrill.

Greenwood, C. R., Uel quadri, J., Stanley, S., Sasso, G., Whorton,

J(9

33

D., & Schulte, D. (1981). Allocating opportunity to learn asa basis for academic remediation. Monograph fgt Childrenlath Seam klahAldot filsoalact, Summer, 22-33

Greenwood, C. R., Delquadri, J., Stanley, S., Terry, B., & Hal 1,R. V. (in press). Assessment of eco-behavioral interactionin school settings. Behavioral Assiisfimsat,

Greenwood, C. R., D inwiddie G., Terry, B., Wade, L., Stanley,S., Thibadeau, 5., b Del quadri, J. (1984). Teacher- versuspeer - mediated instruction: An eco-behaviora I analysis ofachievement outcomes. journal gf. App I led, Dewey tor Analysts.II, 521-538.

Grinstead, J. (1982). IAA iamts at law= signals., groupparticipation And error correction procedures, ix student's

ua math responses, Doctoral Dissertation. Departmentof Human Development and Family Life, University of Kansas.

Hal I, R. V., Del quadri, J., Greenwood, C. R., 8, Thurston, L.(1982). The importance of opportunity to respond Inchi I dren's academic success. (pp. 107-140). In E. B. Edgar,N. G. Haring, J. R. Jenkins, & C1 C. Pious (Eds.), Mental I yhAndlaamed children: Education pod training, Baltimore,MD: University Park Press.

Kohler, F. (1984). Process Assessment age= tutoring.Personal Communication.

Leinhardt, G., Zigmond, N., & Cooley, W. W. (1981). Readinginstruction and its effects. American Educational ResearchApurnal, 11, 343-361.

Lindsley, 0. R. (1964). Direct measurement and prosthesis ofretarded behavior. Journal af. Ethic/Wan, lib 62-81.

Rosenshine, B. (1979). Content, time, and direct instruction.In P. L. Peterson & H. J. Walberg (Eds.), Research gateachings CopuptAL findings, And implications. (pp 28-56).Berkley, CA: McCutchan Publishing.

Stanley, S. O., & Greenwood, C. R. (1980). CISSAII: do fir,Instructional structure And student =dada responsesObserver's Manual,. Kansas City, KS: Juniper GardensChildren's Project, Bureau'of Child Research, Universityof Kansas.

Stanley, S. 0., & Greenwood, C. R. (1983). How much"opportunity to respond" does the minority disadvantagedstudent receive in school? Childreq,41, 370-373.

Strain, P. S., & Kerr, M. M. (1981). Mainstrelminga childrenIMLAChools. New York: Academic Press.

40

34

White, M. A. (1975). Natural rates and teacher approval anddisapproval in the classroom. Journal sit Applied BehaviorAnalysis, lb 367 -372.

41

35

CHAPTER I V

2NVIRONMENTAL VARIABLES AFFECTING THE SCHOOL ACHIEVEMENT OFCULTURALLY AND LINGUISTICALLY DIFFERENT LEARNERS:

AN INSTRUCTIONAL PERSPECTIVE

CARMEN ARREAGA-MAYER

,Abstract

The literature offers many explanations for the differentialeducational attainment of culturally and linguistical ly differentlearners. However, research on effective methods Is relativelyscarce. This review of the literature was designed to examinethe research basis for the school achievement of culturally andlinguistically different learners from the standpoint ofinstructional effectiveness and the opportunity to learn. Thisreview concentrated upon research which directly assessed

specific ecological/instructional processes in relationship toachievement outcome measures. An important contribution of thepresent review Is the focus upon functional ecological andteaching variables affecting the academic performance ofcultural ly and 1 inguistical ly different learners.

Introduction

Despite apparent gains in educational equity in America,especially in the past two decades, differential educationalattainment remains between many minority and majority groups.This phenomena is also noted in other countries wherein the

Inequity remains greatest for refugees and children from cultureslacking modern school experiences (Rolider & Greenwood, 1982).Many reasons have been put forth to explain this pervasive andcontinuing inequity. Some researchers have viewed children fromdifferent cultural and linguistic backgrounds as inherentlydeficient and inferior, thus making a pessimistic appraisal oftheir chances of success in school even with curriculum reform(Miller, 1978). Other researchers have argued that differentialsuccess in school is best explained by differences in language,values, customs, attitudes, and norms which arecharacteristically associated with certain ethnic, racial,national orig!n, gender, and socioeconomic groupings (Dixon,1977; Genova, 1981; Marjoribanks, 1974; Ramirez, 1974; Stodolsky& Lessor, 1967). Still another group of researchers hasemphasized the ef, cts of learning process variables on minoritystudents' responding and academic achievement (Brophy, 1979;Brophy & Evertson, 1981; Greenwood, Deiquadri, & Hall, 1984;11=01011M1110

This Chapter is based upon an earlier draft of Arreaga-Mayer, C.,& Greenwood, C. R. (In press). Environmental variables affectingthe school achievement of culturally and linguistically differentlearners: An Instructional perspective. IslurnaLatthatintinul

ASSOC i Itt toll a BIUMULa Education.

42

36

Hall, Delquadri, Greenwood & Thurston, 1982; Harris, 1977;Henderson, 1969; Stanley & Greenwood, 1983; Trueba, Guthrie, &Au, 1981). Lavints (1965) extensive review of the predictorsof academic performance among students of varied culturalbackgrounds concluded that child intelligence accounts for lessthan half of the variance in school achievement. This findinghas been supported by many studies, including the research ofBared (1977), Genova (1981), Greenwood et al., (1981), Stallings(1977), and Walberg and Marjoribanks (1976). These findingshighlight the importance of studying the environmental correlatesof academic achievement: the ecological - behavioral variables inhome and school environments.

Cultural ly and linguistically different learners includePuerto Rican Americans, American Indians, Mexican Americans,Black Americans, Portueguese Americans, Jewish Americans, IrishAmericans., Asian Americans, and all the diverse racial, cultural,and linguistic groups that constitute the school population inthe United States. However, a recent review of 178 studiesinvestigating an ecological-instructional hypothesis of academicperformance revealed that 90% of studies dealt with Black andMexican American populations (Arreaga-Mayer, 1984).

The aim of this paper is to identify those variables thatare relevant to the academic learning of culturally andlinguistically different students, with a major emphasis upon theBlack and Mexican American students. This review coversinvestigations of the ecological-behavioral variables in home andschool environments. Following discussion of research examiningthese two areas, recommendations and implications for futureresearch are presented.

Ecp logical-Behavioral Yjithstafts.

Gordon (1)76) criticized research completed before 1976 forroutinely focusing upon broad demographic variables as theenvironmental basis of student's academic ability andperformance. According to Gordon, educational research has beentied to the loosely defined notions of ethnicity, socioeconomicstatus, language, gender, geographic origin, and level ofsocialization. It seems reasonable to conclude that as widelyrepresented as these indicators (demographic categories) are inthe literature, they are too vague and ambiguous to be thefunctional determinants of achievement (Coleman et al., 1966;Lavin, 1965; Ornstein, 1982; Weinberg, 1977). What is missing inthis literature are direct studies within home and classroomenvironments of multicultural children that empirical ly assessthe learning process; variables that are the putative causes orlimiting factors upon child achievement at school.

Delquadri (1978), in a comprehensive review of 51reading-learning studies, noted that over 505 of the participantsin these studies had been described as "minority" children. Hereported, in comparison to children from the dominant culture,that "minority" children displayed identry learning patterns on

37

many learning tasks which were represented in those studies.Under structured learning conditions, minority children learnedIn the same manner as their majority group peers In terms ofresponse frequency, the slope of the learning curves, learningtrials, use of discriminative stimuli, and reinforcementprocedures. Yet, research indicates that minority group childrenfrom low socioeconomic backgrounds frequently enter school atkindergarten or first grade with significantly fewer academicskills based upon standardized tests (e.g., Lelnhart, 1980).

Similar lower performance findings have been noted at latergrades, even after children have been within the school programfor several years (Becker, 1977). In fact, low socioeconomic andminority group students have not been expected to perform higherthan the 20th percentile on academic tests throughout theirschool years. Additionally, mincrity group students have beenoverrepresented in special education programs for the retarded(Dunn, (1968) and learning disabled (Maheady, Towne, Algozzine,Mercer, 8. Ysseldyke, 1983). What is the environmental basis andwhat may be the solution for this phenomena?

82111.01X11 INIMIBLICh11112211±Al variable& Dave (1963)and Wolf (1964, 1966) demonstrated that home environmentalvariables, focusing on behaviorally defined events in the naturalenvironment, displayed substantial relationships to concurrentmeasures of intelligence and academic achievement. Their workfocused on what parents actually did in their interactions withtheir children at home, rather than what parents were in terms ofstatus level, source of income, type of dwelling, or some otherdemographic variable. The environmental variables identified bythe authors as likely to relate to academic achievement were:ta) the climate created for achievement motivation (e.g.,parental asper-tions for the chi Id's education, the rewardsaccorded 8Ca' accomplishments, etc.); (b) opportunitiesprovided fc: I development; (c) nature and amount ofassistance p. Ad in overcoming academic difficulties; (d) theactivity of si4..ificant individuals in environment; (e) level ofintellectuality in the environment; and (f) kind of work habitsexpected of the individual. Environmental varibales identifiedas likely to be related to general intelligence were: (a)

stimulation provided for intellectual growth; (b) opportunitiesprovided for, and emphasis on, verbal development; and (c)provisions for general types of learning in a variety ofsituations. Their instruments included a focused interviewschedule and rating scales which were intended to define andmeasure variables selected from' the literature in learning, childdevelopment, and and related areas. In the later study (Wolf,1966), the correlation between the directly observed homeenvironment rating and school achievement was +.80, indicatingthat measures of what parents do with their children can relatehighly to the child's academic success in school.

Henderson and colleagues conducted a series of studies toevaluate the predictive and concurrent validity of homeenvironmental variables with respect to the academic performanceof low achieving minority group children (Henderson, 1966, 1969;

k,

44

38

Henderson & Merritt, 1968; Valencia, Henderson, & Rankin, 1981).For example, Henderson and Merritt (1968) and Henderson (1969)used nine directly assessed environmental variables (i.e.,achievement press, language models, academic guidance, activenessof family, Intellectuality of the home, work habits,identification with models, range of social interaction, andperception of practical value education), to predict the academicachievement of six year old Mexican American children from aneconomically depressed area. The children were divided into highand low contrast groups based upon measures believed to predictschool performance Goodenoug5 -Harris Drawing Test and VanAlstyne Picture Vocabulary Test). These findings demonstratedthP.t children in the high potential group came from homeenvironments and family backgrounds that offered a greatervariety of stimulating experiences than were available to thosechildren in the low potential group.

Matuszek and Raskin (1978) conducted a parental survey of asample of 533 parents of Anglo, Black, and Mexican Americans inAustin, Texas. Their highest correlation was located between thevariable of student reading in the home and academic achievementat the fifth grade level. While it cannot be inferred from thesecorrelational data that a child who reads more at home will thenimprove his reading achievement, these data do suggest a possiblefocus for experimentation with specific instructionalintervention and parental involvement programs.

Valencia, Henderson, and Rankin (1981) studied therelationship of 13 family constellation and socioculturalvariables (i.e., age of child, sex of child, number of children,birth order, language of test, father present, schooling offather, schooling of mother, country of mother's schooling,country of father's schooling, language of the home, socialposition score, and social class level), to the intellectualperformance of 190 Mexican American preschool children from lowincome families. They concluded that the most academicallycompetent children were those who came from homes in which: (a)the dominant language was English, (b) who were tested in Englishrather than Spanish, (c) whose parents were educated in theUnited States rather than Mexico, and (d) whose parents hadattained the highest levels of formal education among thoserepresented In the sample. The authors further stated that theeffects of schooling are very important, and that skills andconcepts implicit in the school.culture are passed on by parentsto their children. The assessment and analysis of this actualprocess, however, (e.g., the functional effects of parentseducated in Mexico upon students learning at home), remains amajor challenge for future research.

Another group of researchers pursuing various issuesrelated to family interaction have directly assessed naturallyoccurring name processes using direct observational codingsystems (Hart, 1983; Patterson, 1982; Wahier, Berland, Coe, &Leske,'1977). These researchers have established clearrelationships between language development rd family interaction

45

39

variables (Hart, 1983), aggressive child behavior and familyinteraction (Patterson, 1983), and mother's depression as afunction of home interaction (Wahler, et al., 1977). Theseapproaches. based upon an eco-behavioral interaction perspective,expand the focus of assessment to account for the interactionsoccurring between individuals within and outside of the homesetting. For example, Wahler et al., (1977) discussed learning asthe product of a system in which behaviors are determined byecological subsystems. At the first level is the covaryingsystem of behaviors within a child's repertoire of responses. Atthe second level is the interacting behavior system of thechild's primary group, such as the family. At the third level,the family subsystem serves as a component of the communitysystems, such as the school. This system model and its relevandata directly assessed, describes the functionalinterrelationships among the child's behaviors, those of fa&members. and those of the school in terms of correlations amothem in their day-to-day occurrence. This approach howeveryet to be applied to the evaluation of educational programsculturally and linguistically different learners.

yng

hasfor

t

Studies have shown that children learn language and academicskills (e.g., reading), through both incidental and formal

teaching interactions with their parents and/or older siblings.For example, parents may engage in informal tutoring by askingquestions or having the children identify letters or words in astory while reading to them. Becker (1977) reported thatlanguage, including specific vocabulary, is learned within theinformal context of family interaction. Thus, the languagespoken at home and the level of vocabulary used can be expectedto exert a powerful ceiling influence upon ch.ld language skillsand development.

In formai studies of home instruction, Thurston (cited inHall, Delquadri, Greenwood, & Thurston, 1982) reported anexperimental study in which a Black mother was trained to tutorher student reading sight words. The student was tested on theseitems by the teacher at school. Results indicated that childincreased her mastery of the content due to tutoring at home,while during reversal conditions (in the absence of hometutoring), she made little progress. Delquadri, Whorton,Elliott, & Greenwood, 1983 cited in Greenwood, Delquadri, & Hall,1984 trained 19 inner-city Black parents to tutor their childrenin oral reading at home using a.ten minute procedure involvingpractice and a modeling error correction procedure. When thesechildren were assessed at school by the teacher these studentsincreased their rates of correctly read words from 60 correctwords per minute before parent tutoring to 70 during the tutoringprogram. Reading errors were cut in half, from 3.0 per minuteprior to the program to 1.5 per minute during the program.Similar findings were reported by Wedel & Fowler, 1984, with astudent for whom English was a second :anguage. This studentmastered 26 sight words in 14 weeks as a result of the parentpointing to a word in a reader and asking the child to read it.When correct, the parent confirmed the response by saying "yes,

46

40

that is correct" and praising the child occasionally for gettingIt right. If the child was incorrect the parent modeled thecorrect pronounciation, then asked the child to say It. Thesestudies are but a few of the experimental studies in which parenttutoring at home has been functionally related to improvedstudent academic performance at school.

In summary, the research reviewed In this section hasrevealed important information concerning environmental processesand methods of assessment necessary to the functional study ofthe academic development of minority group children in the homesetting. Compared to prior studies of poorly defined demographicvariables, these studies directly measured and/or manipulatedecological variables (e.g., instructional interactions) withinthe home setting. Moreover, the inclusion of specific ecologicalvariables within the designs of some of these studies (e.g.,Valencia, et al., 1981) helped to more clearly identify theenvironmental variables ftsociated with student's academicperformance and ability. In the case of the home tutoringstudies, experimental designs were used and functionalrelationships were established. These studies, takencollectively, suggest that an observational analysis ofecological process variables and their function Is needed toreally understand the operation of a home environment on schoollearning.

The research discussed in this section clearly bringsthe traditional research (e.g., Coleman et al., 1966; Fink, 1962;Hernandez, 1973; Ornstein, 1982; Weinberg, 1977), bated uponbroad demographic variables, such as socioeconomic status,parental educational level, aspirations, and emphasis oneducation, etc. into serious question. This research also bringsthe field a giant step towards the empirical validation ofimportant causal relationships and a technology of effectiveinterventions.

Research) instructional Variables

Typically, achievement studies in the United Stateshave not considered the opportunity to learn within the classroomsetting as an achievement factor. Weinberg (1977) stated thatthe assumption in past achievement research with minoritystudents has been that all learning is created equal.

"Variations of teacher quality, differentialopportunities arising from other school functions,and the duration of the school day and the school

year are all of possible significance inconditioning achievement (p. 88)",

but have not been typically accounted for in the literature.Hall and his colleagues, (Greenwood, Delquadri, d Hall, 1984),however, have focused upon what they call the opportunity torespond during instruction. Their work has demonstrated thatstudents, and classes vary in the opportunities presented by the

47

1 16 ki 111 I I I I I I I I I I I I 111=101111111i111111MIII I I I I I I NEM no milmiiii

41

teacher and the lesson, which have a powerful effect upon theamount and quality of student academic responding duringinstruction. They defined opportunity to respond as aninteraction between the stimulus dimensions of the lesson (e.g.,teacher, curricu I a, and method), and student behavior.

The lack of information on learning opportunity asenvironment-student behavior interaction has been widelyrecognized in education (Brophy, 1979; Dunkin & Biddle, 1974;Hoge & Luce, 1979), psychology (Barker, 1968; Bronfenbrenner,

1979), and behavioral analysis (Bijou, 1978; Foster & Cone, 1980;Patterson, 1979; Rogers-Warren & Warren, 1977; Mahler & Graves,1983). Yet, there have been relatively few empirical studies ofeco-behavioral events (i.e., environment-behavior interactions)that have applied experimental design to eco-behavioralhypotheses of learning and even fewer have addressed the culturaland linguistically different population. These studies requiredirect assessment of ecological events, student behavior, and theconjunction and sequence of these events over time.

Dunkin and Biddle (1974), Borich (1977), Brophy (1979),and Foster and Cons (1980), have all presented the case for thestudy of the effects of instructional interventions above andbeyond well-controlled contextual conditions of past experimentalstudies, in order to better understand how classrooms andinstructional procedures contribute to achievement andsocialization. These researchers emphasize the need forexamination of instructional practices within the framework ofbehavior change studies. Only then, it seems, will it bepossible to design classrooms that maximize academic achievementgains over time.

failighalaacialtessarch, With the advent of recentimprovements In observational systems and design methodology,researchers are beginning to establish clear relationshipsbetween classroom ecological variables (e.g., teacher behaviors,commands or instructions, materials used, tasks, etc.), andstudent outcomes (e.g., Improved academic achievement scores).For example, Trueba, Guthrie, and Au (1981) c monstrated that thedramatic progress shown by students in programs that recognizetheir linguistic and cultural differences is not the result ofchanges in materials or curriculum, but stems from complex andsubtle changes in the teacher-student relationship, in theorganization of instructional tasks, and in the role studentsplay as primary agents of their learning.

Rosenshine's review (Rosenshine, 1977) disclosed someof the instructional processes found to be positively andsignificantly related to student school achievement. These were:

1. Observed time spent directly on instruction as opposedto non-academic activities (Stallings, & Kaskowitz, 1974).

2. Student attention or on-task behavior as opposed todisruptive off-task behavior (Brophy, Evertson, 1974; Stallings,

48

42

& Kaskowitz, 1974).

3. Frequency of direct factual single answer questionsposed by teachers instead of complex divergent questions (BrophyEvertson, 1974; Soar, 1973; Stallings & Kaskowitz, 1974).

4. Student involvement in large group Instruction ratherthan unsupervised independent study (Soar, 1973; Stallings &Kaskowitz, 1974).

Similar research focusing on students' learninginteractions during classroom instruction and their effect onacademic achievement have been conducted by Hail, Delquadrl,Greenwood, and Thurston (1982). Their work has been directed atcomparing instructional arrangements (i.e., process or contextualvariables) in inner-city classrooms with those in suburbanschools serving minority students and the students' responsesassociated with these particular instructional arrangements.They have developed a comprehensive classroom observation system,the Code for Instructional Structure and Student AcademicResponse (CISSAR) to sample sequentially the ecology ofinstruction; that Is, activities (the subject of instruction);curriculum task type; structure (grouping); teacher position withrespect to target student, and teacher behavior; and thestudent's behavior (Stanley & Greenwood, 1981).

Among their findings based on an eco-behavioral perspective,were the following:

1. Inner-city minority group students emitted significantlyless active academic responding than suburban students even whenIQ and socioeconomic status were statistically controlled(Greenwood, Stanley, Delquadri, & Hall, 1981).

2. While 74% of the day in inner-city classrooms wasdevoted to academic subjects, only 25% of the day was spent Inactive academic responding writing, talking, reading,asking or answer questions, or reciting). The majority of theclassroom day (45%) was devoted to passive attention (Hall,Delquadri, Greenwood, & Thurston, 1982).

3. Instructional arrangements were different in inner-cityand suburban schools. In inner-city schools, teachers were morelikely to assign Eeatwork and allow students to workindependently (Greewood, Oelquadri, & Hall, 1982; Stanley, &Greenwood, 1981).

4. Specific instructional arrangements were found to bemost related to academic responding (e.g., generally those codedas including paper-pencil or reader tasks); others were found tobe least related (usually those that included teacher-studentdiscussion (Greenwood, Delquadri, Stanley, Terry, & Hall, 1982).

5. Academic responding was significantly correlated toreading and mathematics achievement. Of all response categories

49

43

on the CISSAR code, academic behavior was most related to studentachievement (Greenwood, Delquadri, & Hall, 1982).

6. Effects of intervention changes in instructionalcontexts, and student responding, co-vary with achievement gains(Delquadri, Greenwood, Stretton, & Hall, 1983; Greenwood,Delquadri, & Hall, 1984; Greenwood, Dinwiddie, et al., in press).

These researchers have demonstrated that an eco-behavioralinteraction approach (i.e., process-outcome approach) can teaseout the ways thstructional variables affect students' academicbehavior and achievement, Independently of socioeconomic level orminority group status. This eco-behavioral research paradigm isproviding some practical answers regarding which classroominstructional variables critical to promote more effectiveteaching interventions for all children, and especially for theculturally and linguistically different learner.

Effective I nstructiona 1 Mode I s

The bi I ingual -bicultural intervention efficacy literaturethat currently exist Is mostly characterized by studies focusingupon child outcome data (e.g., changes In IQ, achievement gains)as indicators of program success. There Is a paucity ofliterature on effective alternative instructional strategies thatcan be used with minority students at risk for academic failure(process data). Unfortunately when thinking of multi-culturaleducation, the difficulties associated with non- or limited -English speaker tend to surface or many of the variables on whichwe have traditionally blamed the under achievement of minoritystudents such as race, working mothers, SES, and single-parentfamilies tend to be put forth as explanations. The problems thatminority culture children face In schools are by no means simple(Phillips, 1983; Spindler, 1982). The complexity of theseproblems demands a closer look, a more molecular analysis ofcontributing factors In the child's school environment toacademic achievement. This approach, however, has seldom beenused In the evaluation of instructional program for culturallyand linguistically different learners (Maheady, Towne, Algozzine,Mercer, & Ysse I dyke, 1983).

The instructional strategies and techniques that have provena greater impact on student achievement outcome include: Directinstruction; Ciasswide Peer Tutoring; Precision Teaching; andPersonalized System of Instruction.

Ilat Instruction, The most powerful demonstration todate that 'nstructional practices lead to powerful gains instudents' academic outcomes, including those of culturally andI inguistically different populations, was made by Directinstruction (DI) In the Follow Through program evaluations and inrecent follow-up studies (Becker, 1977, 1978; Becker, & Gersten,1982; Stallings, 1975, 1977). The features of the Di programsincluded scripted and field tested lessons for teachers(Instructional quality control), task analysis and programming of

44

Instruction, unit mastery, error correction, practice and groupresponding, signals for student response, small group instructionwith emphasis on oral communication, and use of positivereinforcement.

The evaluation of DI Included eight thousand low-incomestudents from 20 communities across the United States. Thesample included rural and inner-city Blacks, rural Whites,

Spanish Americans in New Mexico, Mexicao Americans in Texas,American Indians in South Dakota and North Carolina, and otherstudents from a variety of ethnically mixed communities. Thestudents received DI for three years. Results for the DI groupindicated that these children at risk for Icy academicachievement (low income, disadvantaged and culturally-linguistically diverse children), who scored at the 18thpercentile in reading at grade 1, were found to be at the 63rdpercentile on the third grade reading achievement posttest. In

grade equivalents, these students were more than one year abovethe national norms. Similar results were found for DI inmathematics (from 19th percentile to 54th percentile). A recentfollow-up study of this population by Becker and Gersten (1982),at the fifth and sixth grades where they are receivingtraditional instruction, indicated that these sizable benefitsfrom the earlier interventions were maintaining. On readingachievement, these students were still above the 50th percentilerelative to test norms. Major findings from the independentFollow Through programs evaluations, have been that a specificeducational procedure (DI), was demonstrated to (a) contribute asmuch or more to the variance in student's academic achievement,as did their entering IQ scores and (b) that DI resulted In moretime devoted to instruction (Stallings, 1975, 1977). Rosenshine(1977, 1979), and Rosenshine and Berliner (1978), explained theseresults as a function of: (a) regular exposure to instruction,and (b) increased student engagement with the academic task oracademic learning time. Becker (1977, 1978), Englemann, Grazin,and Severson (1979), and Gersten, Carnine, and White (1984) alsooffered the specifics of the DI methods as the most Importantfactors accounting for the Follow Through students' achievementgains (e.g., progammed lessons, signals to respond, etc.).

Duran (1980) studied the effects of DI in the teaching 117first grade bilingual Hispanic children to read. The studentscame from two rural schools In the Southwest near the Mexicanborder. Both schools employed bilingual educational programs.The experimental manipulation consisted of teaching reading forone hour per day for eight consecutive weeks in each school,using either the DI method of bilingual education developed bythe author, or the regular bilingual method already in effect inthe schools. The author concluded that: (a) instruction basedon the principles of DI can improve beginning bilingual

children's achievement significantly more than that of regularbilingual instruction, and (b) that curriculum materials can bedeveloped to teach reading to disadvantaged Hispanic childrenbased upon DI. Results of this investigation provide furthersupport to previous DI studies in which DI clearly takes

51

45

advantageous use of Instruction time and frequent practiceopportunity to effect academic achievement gains.

ClasswIde jegu. Tutoring, A second instructionalprocedure that enhances active responding and achievement gain inminority group, inner-city students Is Classwide Peer Tutoring(Delquadri, Whorton, Elliott, Greenwood, 8, Hall, 1981; Greenwood,Delquadri, b Hall, (1982). Delquadri, Whorton, et al., (1981)devised a classwide peer tutoring program whereby entireclassrooms are divided intc tutoring dyads. The tutoring pairsalternate tutor and tutee roles for 10 minutes each on a dailybasis. Tutors are required to: (a) monitor ongoing tuteeresponses, (b) identify and correct errors, and (c) give pointsfor correct performance. The teacher monitors the classwideprocess, answers tutees questions, and awards points to tutorsfor correct tutoring. Additional features of the classwide peertutoring program include: (a) weekly competing teams. (b) publicposting of game point charts, (c) modeling error correctionprocedure, and (d) teacher assessment of individual studentprogress on a weekly basis (Greenwood, Delquadri, b Hall, 1984).

The classwide peer tutoring program, In contrast totraditional forms of tutoring is a system that allows allstudents in a classroom the opportunity for sustained practice ona task for 10 minutes each day. Procedures have been reportedfor oral reading (Delquadri et al., 1981), spelling (Delquadri etal., 1983), and sight words (Heron, Hewer:lob Cooke, in press).Delquadri, Whorton Elliott, Greenwood, (1983) conducted a studyInvolving a total of 65 students attending inner-city schools.fifty-two percent of the students belonged to minority groups andall attended grades three through six. Four teachers and twentyparents also participated in the study. The students wereassigned randomly to four groups based initially uponstandardized reading test scores. The experimental school groupreceived the classwide peer tutoring procedure for oral reading.The experimental home group received treatment at home, as theirparents or an older family member was taught to use a hometutoring procedure (Whorton it al., 1982). The control andaverage peer groups received only the regular instructionalprogram at school and no home instructions Following threemonths of school and home interventions in the two experimentalgroups, results Indicated that both home and school tutoringgroups achieved statistically lower error rates during theirpassage reading to the teacher.. The two experimental groups alsoread with significantly fewer errors than did either the averagepeer or control groups, dramatically increasing their readingfluency. Delquadri, Greenwood it al., (1983) assessed theeffects of classwide peer tutoring on the weekly spelling testscores In a third grade classroom consisting of six learningdisabled (ID) children and 18 average peers. The participatingschool was located In an inner-city, low Imams, minorityneighborhood. Three of the children were at or above thenational average for spelling achievement on the MetropolitanAchievement Test. The remainder were below national norms withsix (ID) displaing a beginning first grade level of achievement.

52

46

Results of this study demonstrated that the class-wide peertutoring technique dramatically improved spelling performance,particularly by the lower functioning children (from 9.0 errorsat baseline to only 2.5 during tutoring). These lower errorrates were equal to the levels being achieved by the averagepeers during baseline. The individual data for the "lowachievers" showed a dramatic increase in perfect scores (100%) onFriday spelling tests when compared to baseline condition (from arange of 8.0 -11.0 errors to .5 - 3.5 during first phase oftutoring, up to 6.0 - 15.0 during second baseline, to a range of0 - 3.0 during the last tutoring phase). The average peers alsobenefited from the spelling tutoring game (from 3.0 errors perweek before to .5 errors during tutoring).

In summary, these researchers demonstrated that, academicresponding can be contextually controlled, that the arrangementof this control via instructional design (e.g., classwide or hometutoring) are of prime importance to the educational gains of thechildren, and that increased opportunity to respond is one causalcomponent of academic achievement.

Personalized Satem a instruction, Personalized Systemcf Instruction (Keller, 1968) is a procedure for engineeringpersonalized instruction in the classroom. It is an alternativeinstructional procedure that can be effectively implementedacross a broad range of academic skills, instructional materials,and settings. It has most commonly been used at the level ofuniversity teaching but has also been used with students at lowergrade levels (Kirigin, Braukman, Atwater, & Wolf, 1982). Keller(1968, pp. 83) summarized features which distinguished PSI fromconventional teaching procedures as follows:

1. Self-pace feature (Born & Herbert, 1971; Fawcett &Fletcher, 1977).

2. Unit - perfection requirements for advance on thematerial (Bitgood & Seagrave, 1975; Born & Herbert, 1971; Fawcett& Fletcher, 1977).

3. Stress upon the written word in teacher /studentcommnunications (Semb, 1974; Williams & Lawrence, 1975).

4. Use of lectures and demonstrations as vehicles ofmotivation (Keller, 1968).

5. The use of proctors (a student chosen to serve asteaching staff), which permits repeated testing, immediatescoring, tutoring, and a marked enhancement of the personal -social aspect of the educational process (Bostow & Blumenfeld,1972; Klrigin, et al., 1982).

Keller stresses as especially important, in a course taughtby such methods, that any differences in social, economic,cultural, and ethnic backgrounds are completely subordinated to afriendly intellectual relationship between two human beings

53

47

(student-proctor) through the period of the course. Studiescomparing PSI to traditional lecture methods of instruction havedemonstrated increased grade point averages and test performancewith PSI (Hess, 1977; McMichael & Corey, 1969; Schimpfhauser &Richardson, 1977; Sheppard & MacDermot, 1970).

Precision TeachingL, Precision Teaching (Beck, 1979,White & Liberty, 1976), is a set of measurement procedures toassist and guide educational decision makers in making betterinstructional decisions. Unlike the previously discussededucational procedures, Precision Teaching (PT) is not a methodof instruction in the sense that it does not specifyinstructional procedures. However, it is a means for powerfullyassessing the progress children make attaining instructionalgoals via what ever instructional methods are used.

Precision Teaching equips the classroom teacher with: (a)

precise, direct, and daily measures describing student academicperformance, (b) daily accuracy and error rates, (c) formativeeducational decision from charted data; and (d) a bank ofpractice sheets designed to assess student performance uponstandard classroom curriculum object!va-. I n conclusion,although PT is not a complete teaching method per se, it sharesmany of the characteristics of the effective instructionalprocedures previously discussed ifl this section. First, the needfor direct measurement of student academic performance. Second,high performance standards (accuracy and spelling). Third, theneed for frequent, intense student practice sessions. A fineexample of PT application and outcome data is the chapter byLovitt, 1976.

In summary, the successof the four instructional strategiespreviously discussed rests on the molecular analysis andmanipulation of contributing factors in the students' schoolenvironment to academic achievement ( i.e., eco-behavioralanalyses).

Discussion

This article has reviewed the environmental basis forthe school learning of culturally and linguistically differentchildren. Much of the traditional literature in this area hasbeen devoted to the use of indirect and molar measures of theenvironment (e.g., socioeconomic status, ethnic-culturebackground, etc.). Moreover, these studies often treat thesemeasures as unitary and homogeneous, when in fact the events thattranspire in one's !Ife that Impact school learning (one'shistory) are neither unitary or homogeneous in nature.

Most promising, Is the identification and empiricalvalidation of classroom teaching variables (ecological variablesor process variables) by a variety of investigators as importantcausal factors affecting the academic performance of students.As evident in the literature, achievement can be negativelyeffected when: (a) the opportunity to learn and time on-task is

48

low, and (b) when the structure of the instructional program isnot oriented toward monitoring, controlling, and coordinating theamounts of active academic responding for culturally andlinguistically different learners.

Some recent research has improved the measures used forevaluating home and school environments. The use of directassessments of the environment in interviews, ratings, andobservations is offering a new look at both the quality andfrequency of interactions students have in relationship to theiracademic development. As long as we continue to treatinstructional procedures as a singular entity and disregard thecomplex array of contextual variables that control the occurrenceand topography of learning behaviors in natural environments(1.e. home and school), we will continue to merely "expose"students to the curriculum and all students will fail to learn.Hence the need not only for an eco-behavioral method of analysingthe interactions that occur in homes and classrooms forculturally and linguistically different learners, but also for amethod for recording its pragmatic consequences, its actualeffects on the learner.

In conclusion, the eco-behavioral approach to research(process-product) equips the education of the culturally andlinguistically different learner with new dimensions andmethodological alternatives as means of guiding educationalreform. It can provide a fine-grain analysis of what is going onIn the lives of the culturally different learner and how theseexperiences affect academic achievement. Thus, program outcomecan be evaluated in terms of the instructional procedurereceived. Because the findings of eco-behavioral are concrete

CISSAR data previously described), not abstract, they canmotivate educators and communities to act in ways that make adifference.

It appears that potentially effective alternatives tocurrent educational practices are available for consideration anddevelopment as researchers assist school districts move towardthe improvement of the programs. Culturally and linguisticallydifferent learners will succeed only to the extent that theinstructional technology implemented with students remains thefundamental factor in education services.

55

49

References

Arreaga-Mayer, C. (1983). Assess I ng env I ronmenta I and instructionalvariables affecting, j school schLeyemsa p. culturally sadl I ngu I st I cell y satsrsa learners: Aft instructionala I ternat Imo, Juniper Gardens Chi I drenls Project, Bureau ofChild Research, Lawerence, KS: University of Kansas.

Bakeman, R. (1978). Untangling streams of behavior: Sequentialanalysis of observation data. in G. P. Sackett (Ed.),

ing behavior: Oda co l lection And ana ys I s methods.(pp. 63-78). Baltimore, MD: University Park Press.

Baral, D. P. (1977). Achlayiepent Amu foreign-born aad.pat' ve-horn Mexican-Amer icsa students. (ERIC DocumentReproducing Services No. Ed 167-310).

Barker, R.G. (1968). Ecological Psychology. Stanford, CA:Stanford Un 1 vers 1 ty Press.

Beck, R. (1979). Report I= .±hi Office a Education - JointDissemination final. Great Pal Is, MT.

Becker, W. C. (1977). Teaching reading and language to thedisadvantaged: What we have learned from field research.& Educational &Limo AZ, 518-543.

Becker, W. C. (1978). The national evaluation of fol low-through: Behavioral-theory based programs come out on top.Education Asa Ilan Society, j, 431-458.

Becker, W. Co! & Gersten, R. (1982). A fol low-up of fol low-through: The later effects of the direct instruction model onchildren in fifth and sixth grades. Amer scan, Edurati.Research ilaurnali 12, 75-92.

Bijou, S. W. (1978). Some comments on trends in behavioralresearch with children. In T. D. TJossem (Ed.), Interventionstrategies LIE jugh rills. lama Add young ch 1 dren (p. 235).Baltimore, MD: University Park Press.

Bitgood, S. C., & Seagraves K. A. (1975). A comparison ofgraduated and final point systems of contingency managedinstruction. In J. M. Johnston (Ed.), Behavior, research sadtechnology In high= education. Springfield, IL: Charles C.Thomas.

Bronfenbrenner, U. (1979). Context of child rearing: Problemsand prospects. American psychologist, no 844-849.

Borich, G. D. (1977). lha appraisal ianchlool. cdrcepts sadprocess. Reading, MA: Add 1 son-Wes I ey.

Born, D. G., & Herbert, W. A. (1971). A further study of

50

personalized instruction in large university classes,Journal at Experimental Education, 42, 6-11.

Bostow, C. E., & Blumenfeld, G. J. (1972). The effect of twotest-retest procedures on the classroom performance ofundergraduate col lege students. In G.Semb (Eds.), Behav f orAnalysis ma education. Lawrence, KS: University of Kansas.

Brophy, J. E. (1979). Teacher behavior and it effects. Journalfducat Iona I Psychology, 21, 733-75C.

Brophy, J. E., & Evertson, C.14. (1974). Teacher behav for andstudent learning in second and third grades. In G. D. Borich(Ed.), Lift appraisal cf. hanchlAgl. Concepts And process(pp. 79-95). Reading, MS: Add ison-Wes i ey.

Brophy, J. E., d Evertson, C. (1981). Student, characteristicsAnd inching, New York: Longman.

Coleman, J., Campbe I IP E., Hobson, C., McPart I and, J., Mood, A.,Weinfield, F., & York, R. (1966). gojral lty aducatIonalopportunity. Washington, DC: U. S. Government Printing Office.

Dave, R. H. (1963). Ina identification and measurement id,environmental Arm= 3=111121es that are rat Ed aducAtinnal.Achievement, Unpublished doctoral dissertation, University ofCh icago, Chicago, IL.

Delquadri, J. (1978). Reading: StAte tha ALI AS, perceivedkm An ApALLAA behavior anal yALI paradigm. Unpub I ishedmanuscript, Juniper Gardens Chi I drents Project, University ofKansas, Lawrence, KS.

Del quadri, J C., Greenwood, C. Re, Stretton, K., & Hal IP R. V.(1983). The peer tutoring spelling game: A classroom procedurefor increasing opportunity to respond and spelling performance.Educatioq and Treatment a Children, A(3), 225-239.

Del quadri, J., Whorton, D., El I loft, 14., & Greenwood, C. R.(1983). Training Achcci consultants,

.ingaittgrs, nnalft and

school tutoring AccgrAAAL gulag= And affects, Kansas City,KS: Juniper Gardens Children's Project, University of Kansas.

Del quadri, J., Whorton, D., El ! lot+, M., Greenwood, C. R., &Ha 11, R. V. (1981). Ihft Juniper, card= clAssrpon aimtutoring program ism, oral reading. Kansas City, KS: JuniperGardens Children's Project, University of Kansas.

Dixon, C. N. (1977). Matching reading, instruction ± cognitivestyle Icr, MexthAntlimAricna Children, (ERIC Document ReproductionSery ices No. ED 158 269).

Dunkin, Md & BI dd 1 e, B. (1974). MA study j21. IA=111o. NewYork: Ho I 'PP Rinehart & Winston.

57

51

Dunn, L. 14. (1968). Special education for the mildly retarded--Is much of it justified? Exceptional Chi [siren, 25, 5-22.

Duran, E. (1980). Dina reading tQ asavaned Hispanicchildren Wald gg direct instruction. New Mexico StateUniversity, Las Cruces. (ERIC Document No. ED 191 624).

Eng I emann, S., Granz in, A., & Severson, H. (1979). D I agnos I ng

instruction. Tha Journal a special Educat ion, a,355-363.

Fawcett, S. B., & F I etcher, K. R. (1977). Communityapplications of instructional technology: Training writers ofinstructional packages. Journal QL App I led BehaviorAnalysis, 111, 739 -746.

Fink, 14. B. (1962). Self-concept as it relates to academicachievement. Callforni g Journal, a Educat fort Research,57-62.

Foster, S. L., & Cone, J. D. (1980). Current issues in directobservation. aenaglgral Assessment, Z, 313-338.

Genova, W. J. (1981). A study,p. Interaction Affects a schoolAnd jam environments ga students a varying race/ethnicity,clAsAL And gender. Newton, MS: TDR Assoc iates, Inc. (ERICDocument Reproducing Services No. ED 221 622).

Gersten, R., Carnine, D., & White, W. A. T. (1984). The pursuitof clarity: Direct Instruction and appl led behavior. In W.L. Howard, T. E. Heron, D. S., H i l l , & J. Trap-Porter (Eds.),Ems. ga Behavior Analysis Ift Education (pp. 38-57),Columbus, OH: Charles E. Merrill Publishing Company.

Gordon, E. W. (1976). Toward an understanding of educationalequity. Equal DOADE±MaittridAILL ERIC clAaring. house 211ur education, 1-3.

Greenwood, C. R., Del quadri, J., & Hal I, R. V. (1982). Theopportunity to respond and student academic performance inschool. Paper presented at the conference on BehaviorAnalysts jg Education, Ohio State University, Col umbus, OH.

Greenwood, C. R., Del quadri, J., & Hal I, R. V. (1984).240:builix pond And student eaedamia performance.Kansas City, KS: Juniper Gardens Children's Project,University of Kansas, Lawrence, KS.

Greenwood, C. R., Del quadri, J., Stanley, S., Sasso, G., Whorton,D., & Schulte, D. (1981). Al locating opportunity to learnas a basis for academic remediation. Monograph j Children

Severe Behavior Disorders, Summer, 22-23.

Greenwood, C. R., Del quadri, J., Stanley,S. 0., Terry, B., &Hal I, R. V. (1982). Relationships wog instructional

58

52

contexts, student 11Ahay.lczA. And academic achievement. KansasCity, KS: Juniper Gardens Children's Project, Universityof Kansas.

Greenwood, C. R., Dinwiddle, G., Terry, B., Wade, L., Stanley, S.0., Thibadeau, S., & Delquadri, J. (1984). Teacher- versuspeer-mediated instruction: An eco-behavioral analysis ofachievement outcones.lournAi Agpl led BehaviorAnalysis, 12, 521-538.

Greenwood, C. R., Stanley, S. 0., Del quadri, J., & Hal l, R. V.(1981, February). AMILIAIAgelmulus control a =Weak.tospandLaga Ihmt _op Ice Ions 2.t. observation methodology /2 the.design a instruction, Paper presented at the SimposiumInternacional de Analisis Conductal Aplicado en Educacion,Universidad Nacional de Mejico, Ciudad de Mejico.

Hail, R. V., Del quadri, J., Greenwood, C. R., & Thurston, L.(1982). The importance of opportunity to respond inchi I drents academic success. In E. B. Edgar, N. G. Haring,J. R. Jenkins, & C. Pious (Eds.), Mentally handicappedchildren: Education And training (no. 107-140). Baltimore,MD: University Park Press.

Hart, B. (1983). Assessing spontaneous speech. BitinxInral.Assessment, 1.(1), 71-82.

Henderson, R. W. (1966). Environmental stlmu I at Ion AndIntel I actual dnyAidgmi a Mexican-Amer ice, gh I 1 drent Aaexploratory study. Unpublished doctoral dissertation,University of Arizona, Tucson, AZ.

Henderson, R. W. (1969). Environmental variables =predictorsof, Academic performance, Tucson, AZ: Research andDevelopment Center. (ERIC Document Reproduction Services No.036 592).

Henderson, R. W., & Merritt, C. B. (1968). Environmentalbackground of Mexican-American children with differentpotentials for school success. "Mt Journal giBiacilsjPsycholtav, LI, 101-106.

Hernandez, N. G. (1973). Variables affecting achievement ofmiddle school Mexican - American students. Built df.Educational Research, A(1), 1-39.

Heron, T. E., Howard, W. L, Cooke, N. L.. & Hill, D. S. (Inpress). Evaluation of a classwide peer tutoring system:First graders teach each other sight words. Education AndTreatment 2f, Miriam

Hess, J. H. (1977). Large-scale individualized instructionalsystems: A comparison of three models. Educational Technoloay.a, 37-44.

53

Hoge, R. Do! & Luce, S. (1979). Predicting academic achievementfrom classroom behav lor. Rev few a Esurat1121111 Research, .42)479-496.

Kel ler, F. S. (1968). "Good-bye teacher....". Journal aApp 1 i Q Behav (era I Mal Ws, 1(1), 79-89.

Kirigin, K. A., Braukmann, C. J., Atwater, J. D., & Wolf, M. M.(1982). An evaluation of teaching-family (AchievementPlace) group homes for Juvenile offenders. Journal aApplied Behavior Analysis, 1,1, 1-16.

Lavin, D. E. (1965). Da prediction a academic arismaanaeltheoretical Analysis And rev iew a research, New York:Vintage Books.

Leinhardt, G. (1980). Transition rooms: Promoting maturation orreducing education? Journal a Educational elyriha55-61.

Lov itt. T. C. (1976). Applied behavior analysis techniques andcurriculum research: Implications for Instruction (pp.112-156). In N.G. Herring & R. L. Schiefelbusch (Eds.),Teaching, Special Children, New York, NY: McGraw-HI I I Co.

Maheady, L., Towne, R., AI gozzine B., Mercer, J., & Yssel dyke,J. (1983). Minority overrepresentation: A case foralternative practices prior to referral. Learning DisabilityQuarterly, 1(4), 448-456.

Merjor !banks, K. (1974). Env ironments icE issralsg. At I anticHighlands, NJ: Humanities Press.

Matuszek, P., & Hoskin, C. (1978). 11312 are ±he, disadvantagedand 'hat shouli Jig dg, Inc them/ J re I ationshlo Qf .thmLLyvariables fQ achievement and some Implications Inceducational programming. (ERIC Document ReproductionSery ices No. ED 171 438).

McMichael, J. S., & Corey, J. R. (1969). Contingency managementin an introductory psychology course produces betterI earn 1 ng. Journal Qj Applied Behavior Analysis, 2,79-83.

Miller, H. L. (1978). Social foundations a education. NewYork: Holt, Rinehart, and Winston.

Ornstein, A. C. (1982). The education of the disadvantaged: A20 year review. Educational Research, 24(3), 197-211.

Patterson, G. R. (1979). A performance theory for coercivefamily interaction. In R. B. Cairns (Ed.), Tha analysis asocial Interactions: Methods, Issues, and i I lustrations (pp.

119-162). Hillsdale, NJ: Lawrence Erlbaum Associates.

60

54

Patterson, G. R. (1982). Coercive family process (Vol. 3, pp.41-65). Eugene, OR: Caste I la Publishing Company.

Phillips, S. (1983). Ita Inv is lb le culture. New York:Longman.

Ramirez III, M., Castaneda, A., & Herald, P. C. (1974). Therelationship of acculturation to cognitive style among Mexican-Amer !cans. Journal gf. Cross Cultural Psychology, 1(4),424-433.

Rogers-Warren, A., & Warren, S. F. (1977). Ecologicalperspect I vas la behavior Dna! ys is. Ba I t I more, MD:

University Park Press.

Ro I I der, A., & Greenwood, C. R. (1982). Research. with childrenfrom depr tv I ng environments In Israel, Proposal submitted tothe United States-Israel Binational Science Foundation,Jerusalem, Israel.

Rosenshine, B. (1977). Review of teaching variables and studentachievement. In G. D. Borich (Ed.), the appraisal sit teaching:Concepts And process. Reading, MA: Add 1 son-Wes I ey.

Rosenshine, B. V., & Berliner, D. C. (1978). Academic engagedtime. British Journal a Teacher, Education, I 3-16.

Schimpfhauser, F., & Richardson, K. (1977). Medical educationand personalized instruction. Educational Lichnologyz.212., 31-36.

Semb, G. (1974). Personal ized instruction: The effects of gradingcriteria and assignment length on col lege student testperformance. Journa l AO I tad Behavior Ana I ys is,

61-69.

Sheppard, W. C., & MacDermot, H. G. (1970). Design andevaluation of a programmed course in introductory psychology.Journal al Applied Behavior Angayslul 5-11.

Soar, R. S. (1973). fallsatrihmugh claws= asnmsmeasurement and Anal growth (1970-1971): M =mt.Gainesville, FL: Col lege of Education, University ofFlorida.

Sp Ind I er, G. (Ed.). (1983). Doing the ethnography a schoo I ing.

New Yorks Hoit, Rinehart & Winston.

Stallings, J. (1975). Implementation and child effects ofteaching practices in follow through classrooms. Monographs

j.;:q Society .f Research J Chill Dave I opment, An, (7-8,Serial No. 163).

Stallings, J. (1977). How instructional processes relate to tochild outcomes. In G. D. Borich (Ed.), nil pppraisal

61

55

t Doncepts And process (pp. 104-113). Reading, MA:Add 1 son-Wes I ey.

Stal I ings, J., & Kaskowitz, D. (1974). Fol tow - through lomcLiery at! on Evaluation, 1972-1973. Mer I o Park, CA: StanfordResearch Institute.

Stan I ey, S. 0., & Greenwood, C. R. (1981). Code jor,instructional structure Ala student, Academic response(CI SSAR): Observer's manual. Kansas City, KS: JuniperGardens Children's Project, University of Kansas.

Stodol sky, S. S., & Lesser, G. S. (1967). Learning patterns inthe disadvantaged. Harimg Educational &LEAK, up,

Trueka, Ho, Guthrie, G., Au, K. (1981). Claim= nag the.bilingual classroom. Row I ey,MA: Newbury House.Valencia, R. Re, Henderson, R. W., & Rank in, R. (1981).Relationship of family constel lotion and school ing to

Intellectual performance of Mexican American children. Journala Educe tonal Psycho 1 ogy, 7,1(4), 524-532.

Wahl er, R. Ge, Ber land, R. M., Coe, T. D., & Leske, G. (1977).Social system analysis: Implementing an a I ternat I ve behaviormodel. In A. Rogers-Warren, & S. F. Warren (Eds.), fcologicalDerspecti_ves ja heav for Analysis. Baltimore,l MD: UniversityPark Press.

Wahl er, R. G., & Graves, M. G. (1983). Setting events in socialnetworks: Ally or enemy in child behavior therapy? BehaviorIhereay, 14, 19-36.

Walberg, H. J., & Marjoribanks, K. (1976). Family environmentand cognitive development: Twelve analytical models. Review

Educational Research* 41, 527-551.

Weinberg, M. (1977). Minority students: A research appraisal.Washington, DC: U. S. Government Printing Off ice.

White, 0. R., & Liberty, K. A. (1976). Behavioral assessmentand precise educational measurement. (pp. 31-71). In

N.G. Haring & R. L. Schiefel busch (Eds.), Teaching SpecialChi I dren, New York, NY: McGrew-Hi I I Co.

Whorton, D., Sasso, Go, El I iott, M., Hughes, V., Greenwood, C.R., & Del quadri! J. (1982). Teaching othi thatmaximize the apartanax IQ luau Parent and pew, .tutor in°.=grams, Kansas City, KS: Juniper Gardens Chi I dren'sProject, University of Kansas.

W i l l i a m s , R. L., & Lawrence, J. (1975). The effects of quizzingIn a lecture course. In J. M. Johnston (Ed.), Behaviorresearch And technology jn h gher educat ion.. Springfield,IL: Charles C. Thomas.

62

56

Wolf, R. M. (1964). liaft identification And Measurement aenvironmental process variables related /2 intelligence.Unpublished doctoral dissertation, University of Chicago,Chicago, IL.

Wolf, R. M. (1966). The management of environments. In A.Anastasi (Ed.), Testing problems, In perspective (pp. 491-503). Washington, DC: American Council on Education.

C3

CHAPTER V

ECO-BEHAVIORAL ASSESSMENT: A METHODOLOGY FOR EXPANDINGTHE EVALUATION OF EARLY INTERVENTION PROGRAMS

JUDITH J. CARTA

Abstract

57

The effectiveness of early Intervention programs has been anarea of controversy since the early 1960's. Unfortunately, manyin the academic community and the general public remain

unconvinced that spending government funds for the education ofyoung handicapped children is an effective use of publicresources. Research on this effectiveness issue has been limitedto the measurement of singular outcomes (e.g., changes in thebehavior and/or adjustment of the handicapped students or parentsthat have been the targets of intervention). This paper proposesa methodological expansion in evaluation research, an eco-behavioral approach, that has been used recently in teachereffectiveness studies within elementary grade classrooms. Thepaper discusses the implications of adding the eco-behavioralapproach to the evaluation of preschool programs for thehandicapped. It argues that this methodology will allow both (a)determination of early intervention effectiveness and (b)specification of factors that are responsible for thiseffectiveness.

Introduction

For almost twenty years, the effectiveness of earlyintervention programs has been the subject of public and academicdebate. Does educational intervention in the lives of youngchildren with special learning problems produce important andlasting gains? From the Inception of programs for disadvantagedpreschoolers in the early 19601s, to the more recentestablishment of educational services for preschoolers withhandicapping conditions, researchers have sought empiricalanswers to this question.

Although many of the earliest studies reported no effect ofearly childhood programs (Cicirelli, 1969; Bronfenbrenner, 1975),the most recent and convincing research studies indicate thatearly intervention programs are successful in producing lastinggains in the development of disadvantaged and handicappedchildren (Lazar d Darlington, 1979, Schweinhart d Weikart, 1981;Stock et al., 1976).

MIIMMM.This Chapter is based upon an early draft of Carta, J. J., &Greenwood, C. R. (in press). Eco-behavioral assessment: Amethodology for expending the evaluation of early interventionprograms. Topics In, Early Childhood Special Education.

64

58

arzidars slt he Current Efficacy Literature

The early intervention efficacy literature that currentlyexists focuses almost exclusively upon child outcome data.Review of the literature indicates that a variety of outcomemeasures have been employed as Indicators of the effectiveness ofearly intervention programs. As indicated in Table 4, by far themost ccaaon outcome variable used In this research has beencriterion-referenced tests. Criterion-referenced measures areuseful for evaluating preschools because they serve two purposes:they point to gains made by participants on tests of preschoolskills, and they assist program managers identify skilldeficiencies and specifying behavioral goals for Intervention(Bricker & Dow, 1980).

Table 4

Outcome Measures Emp 1 oyed 1II Past Efficacy ResearchQ far I y, i ntervent ion

4110 4111 41

Number of StudiesOutcome Measures Employing Measure

IMO

Percent of StudiesEmploying Measure

Number of Skills Acquired as 21 70Measured by Criterion -

Referenced Tests

Improvement on Norm 9 30Referenced Tests

Measures of Transitions 9 30Into Subsequent SchoolEnvironments

Social Competency Measures 8 27

Children's Attitudes and 4 13Values

Changes in Parents' 4 13Behaviors and Attitudes

Total 55 183

ON,

A second class of outcome variables is Improvement or changein developmental quotients derived from norm-referenced

65

59

Instruments of intelligence. An individual's quotient measuredat two points In time is often compared to the rate of changeexpected In the non-handicapped population as indicated by testnorms. These types of measures are useful because they allowcomparison of mild to moderately handicapped progri.n participants"against a set of developmental expectations generated from theperformance of a substantial number of normal children" (Kopp,1979, p. 35).

A third class focuses upon participants' successful

transitions to school environments subsequent to preschool. Avariety of these indicators have been used to measure programgraduates' success In adapting to the elementary school setting.The most prominent among these include whether or not the programgraduate was placed In a regular education classroom, whether ornot the child was academically retained, and whether or not thechild required special services (Karnes, Schwedel, Lewis, Ratts,& Esry, 1981; Lazar, Darlington, Murray, Royce, & Snipper, 1982;Schweinhart & Weikart, 1981). Programs using such measuresprovide evaluation data that parents, administrators,policymakers, and the scientific community can all understand.More important, however, they indicate whether or not a programhas met one of its foremost goals: to enhance its participants'probability of success In the compulsory education system(Takanishi & Feshback, 1982).

A fourth class of outcome measures Is improvement Inchildren's social competence or behavioral adjustment. Thesemeasures vary in both content and instrumentation. They rarge,for example, from ratings by teachers and parents of children'sproblem behaviors (Strain, 1981); to direct observation of socialand nonsocial behaviors towards adults and peers (Greenwood,Todd, Hops, & Walker, 1982); to self-reports of delinquentbehavior In long term follow-up studies (Schweinhart & Weikart,1981). These measures have more relevance to social adaptationthan do norm-referenced measures such as IQ tests (Zigler &Balla, 1982).

A fifth class is motivational, emotional, and self-conceptmeasures. A number of authors have argued the importance ofincluding attitude variables because these variables may modifythe effects of a child's educational environment In either apositive or negative direction (Bell, 1968; Sameroff, 1975). Theadvantage of including measures of attitude and values whenevaluating early intervention programs Is that they sample anImportant domain and may relate to other outcome measures.

The sixth class of outcome measures Is assessments ofparents' behavior and attitudes as a result of earlyIntervention. Parent measures provide an important Indication ofwhether training activities directed at them have beensuccessful. Additionally, changes are sometimes noted Inparental behaviors when parents were not themselves tar;ets ofthe intervention procedures (Gray, 1977).

66

60

Ijig Problem with, Current Evaluation Research

Evident in the literature reviewed were six general classesof outcome measures which have been used to evaluate theeffectivensss of early Intervention programs. While thesemeasures have been useful in documenting that important changeshave occurred In the lives of children who have been involved inearly intervention programs, they have provided littleinformation regarding why those changes have occurred. This isbecause the entire focus of assessment in these earlyintervention studies has been upon child-centered outcomes,.thedependent variables. Little, if any, attention has been paid tothe independent variable, the preschool treatment that has beenprovided. Past evaluation studies have treated preschoolinterventions as unitary variables that were either "on" or"off". There has been little specification and little assessmentof the many components that differentiate one program fromanother such ass the behaviors engaged in by the teachers theobjectives taught, and the manner in which objectives wereimplemented as instruction. Past evaluation studies have treatedearly interventions as singular entities and have disregarded thecomplex array of dynamic variables that determinc programoutcomes on a day-by-day and even moment-by-moment basis.

The knowledge that a program has brought aboutIncreases in participants' skills or improvement in IQ scores orsuccessful transitions for Its graduates into the elementarygrades, may indicate that the program was successful. However,this design provides no information about the specific componentsthat made the program successful. The science of earlyintervention service delivery can only move forward whensuccessful programs can be replicated, and those programs canonly be replicated when the independent variables have beenclearly delineated, assessed, and experimentally examined(Sidman, 1960; Strain & Kerr, 1981). A new set of controlledefficacy studies must be undertaken to examine the specificprogrammatic factors responsible for the positive outcomesresulting from early intervention.

Forerunners lat its goo-Behavioral mixort-b.

An eco-behavioral approach to program evaluation is a meansof assessing program variables through systematic observation andmeasuring the moment-to-moment effects of an array of variablesupon student behavior. The momentary Interactions betweenimmediate program variables as ecological stimuli and studentbehaviors are the units of analysis for predicting or otherwiseinvestigating program outcomes (e.g., developmental gain or long-term achievement). Thus, this approach differs dramatically fromother well-known interactional approaches, namely aptitudetreatment interaction or ATI (Cronbach b Snow, 1977), in whichinteraction refers to a two-way factoral interaction effect uponoutcome variables as a result of divergent levels of aptitudes(,4., learning modalities) and instructional treatments.

C7

61

The basis of an eco-behavioral approach to programevaluation derives from environmental determinism and can betraced to three different fields: behavioral ecology, (Barker &Wright, 1968; Bronfenbrenner, 1979); applied behavior analysis(Bijou, Peterson, & Ault, 1968; Patterson, 1982; Rogers-Warren &Warren, 1977; Wahler & Fox, 1981); and process-product researchin education (Brophy, 1979; Dunkin & Biddle, 1974).

Behavioral ecollsy, Behavioral ecology (Barker & Wright,1968) calls for the recording of events in the life of anindividual in terms of behavioral settings; descriptions of theenvironment on par with descriptions of the subject's behavior.These behavior settings and their standing patterns of behaviorare termed synomorphic relationships and their identification anddescription Is the goal of ecological research. The ecologyobserved in a particular study may be defined quite broadly, asin the analysis of setting variables and their impact onchildren's learning and play behaviors (Gump, 1969). On theother hand, only specific aspects of the environment may bestudied, such as group size (Barker & Gump, 1965), density(McGrew, 1970), and spatial organization (Prescott, 1981). In

any event, the focus is upon the study of stimulus elements thatprovide the occasion for persons to behave. These events areoften described through verbatim accounts, narrative records, andother less than quantitative methods.

Appl led B0havior Analysis The field of applied behavioranalysis is founded upon the operant model of behavior, or theinteraction of a setting event, discriminative stimulus,response, and reinforcing consequent stimulus. Assessment Inapplied behavior analysis emphasizes recording behavioral andenvironmental events in observable, quantifiable terms (Bijou,1968, p. 175). Measures describing these events are coded intoquantifiable units of behavior such as frequencies or durationsand these ar3 then combined across observation periods, analyzed,and interpreted. Descriptive studies in applied behavioranalysis that have focused upon naturalistic observations ofinteractions between setting events, discriminative stimuli,behaviors, and consequences have been used to determine thestimulus control of deviant behavior (Karpowitz & Johnso,, 1981;Patterson, 1982). Similarly, Wahler and his colleagues ( Wahler &Graves, 1983) have examined the setting events for familyinteractions as a way of predicting successful or unsuccessfulfamily behavior therapy.

t Studies a Teaching, Effectiveness. A thirdperspective that has influenced the eco-behavioral approach toprogram evaluation has been process-product studies of teachereffectiveness (Brophy, 1979). Process measurement, the assessmentof teacher behaviors, student behaviors, and other classroomstimuli (e.g., grouping, physical structure, curriculum, etc.)has been successfully used to explain academic products (outcomegains) of students resulting from instruction. Process measuresare typically gathered through systeinat!c observations of teacherand student behaviors (Flanders, 1970; Medley & Mitzel, 1963).

68

62

In fact, Brophy (1979) refers to process-product stuaies (teacherbehavior-academic products) and process- process - product studies(teacher behavior-student behavior-academic products) in whichwhat teachers and students do in the classroom is quantitativelydescribed and examined in terms of academic gains in controlledstudies.

Ths. Anal ys Is Eco-Behav [oral Processes

Eco-behavioral processes can be described In two basic ways.First, observations of specific variables can be staticallyscored, that is, the frequency of each coded event can be totaledand expressed in terms of the grand total of all coded events, asan unconditional proportion or percentage score. Theseproportions act as molar descriptions or session estimates of therelative rates of occurrence of each coded classroom event.Alternatively, classroom observation systems can be dynamicallyscored, that is, events that occur contiguously (those that co-occur in the same time interval or those that follow each otherin subsequent Intervals) can be combined to form conditionalproportion scores. Summaries of these Jointly occurring eventscan then be combined to form molecular descriptions, thwconditional relationship between ecological and behavioralevents. These classroom processes achieve added significancewhen they are related to product measures (gains in academicachievement).

Several studies have been conducted in regular classroomsettings to determine the molar classroom processes that arerelated to student achIeveme:. A review of process-productresearch by Rosenshine (1977) pointed to the following processesas significantly and positively related to school achievementgains:

1. Observed time spent directly on instruction as opposedto non-academic activities (Stallings & Kaskowitz, 1974).

2. Frequency of direct factual single answer questions byteachers instead of complex divergent questions (Brophy &Evertson, 1974; Soar, 1973; Stallings & Kaskowitz, 1974).

3. Student attention or on-task behavior as opposed todisruptive off-task behavior (Brophy & Evertson, 1974; Stal I ings,& Kaskowitz, 1974).

4. Student involvement in large group instruction ratherthan unsupervised independent study (Soar, 1973; Stallings &Kaskowitz, 1974).

A ram In

In our own work, we investigate molar and molecularprocesses using a sequential classroom observation system, theCode for Instructional Structure and Student Academic Response( CISSAR). The CISSAR (Hall, Delquadri, Greenwood, & Thurston,

69

63

1982; Stanley & Greenwood, 1981) focuses upon a single studentfor an entire school day and allows observers to record fivemajor categories of ecological variables: activity, task,teaching structure, teacher position, and teacher activity.Observers also record three categories of student responsevariables (i.e., academic responding, task oanagement, andcompeting, inappropriate behavior).

Data derived from the CISSAR code have been used for threemajor purposes:

1. To provide molar descriptions of contrasting educationalsettings and instructional approaches.

2. To provide molecular descriptions of co-occurringclasssroom events and related student responses.

3. To conduct process-product analyses of academicachievement gain.

Molar, raftia Results, Most of the molar descriptions usingthe CISSAR have been used to contrast inner-city and suburbanschools. For example, fourth-grade teachers in inner-cityschools were found to use different instructional tasks thanteachers in suburban schools. In inner -city schools, fourth-grade teachers were more likely to use media (overheadprojectors, films, etc.). In suburban schools, teachers weremore likely to assign seatwork and allow students to workindependently (Greenwood, Deiquadri, b Hall, 1984). When thebehaviors of students In these groups were compared, inner-citystudents were found to emit significantly less active academicresponding than suburban students, even when iQ and socioeconomicstatus were statistically control led (Greenwood et al., 1981).These differential findings are molar descIptions of classroomecology and student behavior that are helpful In making globalcomparisons between programs.

Molecular ragaResults, Molecular descriptions ofclassrooms have revealed the diversity of ecological stimulusarrangements students experience within single lessons and thevariability In specific academic behaviors temporally associatedwith each particular arrangement (Greenwood, Delquadri, Stanley,Terry, d Hall, in press). Analysis of the most frequentlyoccurring ecological arrangements resulted in the identificationof arrangements that accelerated or decelerated student'sacademic responding over a baserate, the molar average proportionof academic responding for an entire observational session. Forexample, the baserate probability for academic responding duringspelling was .33 for the inner-city sample. The probability ofacademic responding Increased to .62 when the followingaccelerator arrangement of variables occurred: Task:paper/Pencil, Structure: fntlre group, Teacher Location: .Desk,Teacher Response: ha, Teaching, In contrast, the probability ofacademic responding decreased to J04 when the followingarrangement occurred: Task: Teacher/Student Discussion,

70

64

Structure: Entire, prow, Teacher Location: In Front, TeacherResponse: Teaching,

process-Product CISSAR Results. Process- producr analysesusing the CISSAR have been used to find the best predictors ofstudent achievement among the student response variables.Greenwood et al., (1984) reported that a composite of sevenstudent behavior variables, termed academic responding, was mostpredictive of achievement (.42). The individual behaviors ofwriting and silent reading were also significantly correlated tostudent achievement. Attentive behavior defined as looking atthe teacher, the lesson, or a peer answering a question was not asignificant correlate of achievement.

In a second line of research, functional analyses ofinstructional interventions have shown that important changes ineco-behavioral processes co-vary with gains in studentachievement (Greenwood, et al., 1984; Greenwood, Dinwiddie, etal., in press). These changes included: increased use ofpaper/pencil and reduction in teacher-student discussion,location of the teacher among and to the side of students, ratherthan In front or at desk, increased teaching behavior, etc.Thus, it has been demonstrated that an eco-behavioral approachcan define the ways classroom variables effect student behaviorsand subsegently, student achievement. However, the questionremains: "Can this methodology work In evaluating earlyintervention programs?"

A Prototype Eco-Behav I ore I Approach I= gva I uat Ina Early,Intervention Programs

The real need at present is the development of an eco-behavioral code that captures important instructionalinteractions that occur in classrooms for handicappedpreschoolers. This instrument must be complex enough to assessboth the diversity and sequence of ecological variables, teacherbehaviors, and student responses in the preschool classroom. Asample prototype observational code being developed by theauthors appears in Table 5. An expanded version of thisinstrument will be used to assess

eco-behavioral processes for a single student observed duringindividual sessions or spanning an entire preschool day.Students' observation records will then be summarized to providethe following descriptions of classrooms:

1. Molar descriptions of preschool programs derived bycomputing the percentage occurrence of each variable on the code.These descriptions will be statements about the classroomecology, such as the percentage of the preschool day spent inspecific activities, and in what types of structure, grouping,and composition of handicapped/nonhandicapped students in whichactivities occurred. Similar descriptions can be made about theproportion of the day that the teacher engaged in variousbehaviors and likewise the types of behaviors the students

71

emitted.

Table 5

Preschool Observation Systemigclorelinteractions- ---------------------------

EcologicalCategories Description Code ExamplesN OMMMNIDO.MOWOM

65

Designated The subject of instruc- Free play, Pre-Activity tion academics, Language,

F!ne motorMONONIMMMMOOMWAYMINOMD4.1.ONAMMM4110.11.MMONOM.Ii OWN

Structure The amount of struc-ture provided

Teacher-directed,Teacher-guided,Child-guided

NININ.M.M.MIN411.01NOM401110411.0.1111iM...ONMMaterials

=1Location

Objects which thestudent engages orattends to

Physical placementof the observedstudent

Manipulatives, Artmaterials, Largemotor equipment

....00.0011WIMMONNW101100041MM.0001.MMOMMINGOM

On floor, At tables,On equipment, Inchairs

MONNiMm.....MMINNONDINFIND MMMMM MOOMINIOMmM4101..mGrouping Size of group in

same activity asobserved student

Small group, Largegroup, Whole class

MilmNWIN.111MNISOMMIDANNOINNNOMMOODMOWMM.MINOSComposition Mix of handicapped

and non-handicappedstudents inInstructional group

Ail handicapped,Mixed, All non-handicapped

MM14/1400101.6DINIIM110~0000.00MOMOINN.INONDOMMINNNOMIM =IMO IMENM.M1=1MO.MFTeacher BehavCategories

for

DescriptionIMMEM00011.1111TeacherDefinition

Code ExamplesMONIONWIIMMANDOOMOOSOOMMMMMWMINIMNONNANOWIONIMMIN!Primary adult

interacting withobserved student

M..M.1.1110104.01ON01..MONNINNOMM410.Teacher TeachN. behavior

Behavior relative toobserved student

MM4111. MMMMM W111101WING,...D411.104110MINIMMMENIMO.

TeacherFocus

Direction ofteachers behavior

Teacher, Aide,

Student teacher.Ancillary staff

OwlmommiNNOwdmomm.,Verbal instruction,

Physical assisting,Approval, Disapproval

MMMMM emoimmobsomMmsmmommo....

Target child only,Target ch110 andentire group, Otherthan target child

MMOO MMMMMM .1114. MMMMM NIOMMSOM MMMMMM OOMMOOMPODMONIM

2

66

Table 5 Continues (pg. 2)

Student BehaviorCategories Description Code Examples

Appropriate Specific active on- Fine motor, GrossEngaged task responses motor, Pre-academics

Appropriate Prerequisite or Waiting, Looking forNon-Engaged enabling responses materials,

Transition.....MNNMINNallInappropriate Behaviors which Disruptive,

compete with Inappropriate play,appropriate, Off-taskengaged behaviors

Interaction Verbal or non-verbalengagement of a peer

or adult

Verbal interactionwith peer, VerbalInteraction with adult

GrMOW.iI.WMM........W.M.NIMWIMMOWNeOiMiAOOlp4OMIm.

2. Molecular descriptions of programs derived by computingthe conditional probabilities of various combinations ofvariables on the code. These descriptions will make possible thefollowing types of statements: "Given a specific type ofactivity or materials, in what types of behavior was the studentmost likely engaged?", "Given a specific classroom structure orinstructional grouping, in what types of behavior Is the teachermost likely to be engaged?", and "Given a specific teacherbehavior, In what type of behavior Is the student most likely tobe engaged?".

3. Process-product analyses will be conducted by correlatingspecific eco-behavioral processes with outcome measures such asgain scores on developmental tests or measures of successfultransitions to post-preschool environments.

These three types of descriptions can be used in several waysto evaluate early intervention programs. First, they can formthe basis for defining the program variables across differenttypes of preschools in a quantifiable manner. For example,programs embodying different philosophies (e.g. behavioral,Plagetian, or Montessori) can be qualitatively and empiricallydescribed and contrasted across variables like the activities andmaterials provided, and the behaviors engaged in by teachers andstudents. In a similar fashion, programs that reflect differentservice delivery models, such as mainstream programs versus self-contained, or half-day versus full-day programs, can becontrasted.

Second, molar and molecular descriptions can be used toexamine the fidelity of program replications. If an original

"I 3

67

program model can be quantifiably described across a variety ofdimensions, such as the content that is taught, the materialsused etc., then those quantified dimensions can become a templateagainst which replications can be compared.

Third, the molar and molecular descriptions can provide ameans of documenting specific changes in programs. Some examplesof programmatic shifts that could be monitored are: changesincurred by the institution of a new curriculum; changes broughtby a shift in the classroom population, such as the integrationof non-handicapped peers into a program; or changes brought aboutwhen a teacher decides to systematically alter some behavior,such as the rate of "approval" statements. If data are gatheredon students in the program both before and after changes are madein the classroom, molar and molecular descriptions derived fromobservational records can be analyzed in time-series fashion todocument shifts in the specific independe0 variable beingmanipulated (e.g. activity, composition of group, teachingbehavior). At the same time, changes in student behavior thatco-occur with the program variable alterations can be monitored.

Fourth, the molar descriptions of student behaviors can beused in process-product analyses to determine the specificclassroom behaviors that are most related to developmental gain.This type of evaluation will allow early intervention programs todetermine the skills that are most predictive of developmentalgain. This type of information will help program developers inchoosing the behaviors (e.g. pre-academics, free play, language)that will be the focus of the program.

Fifth, the molecular descriptions of eco-behavioralinteractions can tnen be used to make precision diagnoses ofinstruction, that is, a determination of the specific combinationof ecological and teacher variables that are most related to theclassroom skills and that are critical to the enhancement ofdevelopmental outcomes. These instructional variables can thenbe the variables targeted for improvement.

lummecy.

What has been described is a powerful methodologicalimprovement for use in the next generation of efficacy studies.Unlike many approaches in which the child is the center ofassessment (e.g., aptitude-treatment-interaction and much ofapplied behavior analysis), the eco-behavioral approach willexpand the focus of assessment to account for both theindependent and dependent variables to explain student outcomes.Obviously, the suggestions for using this methodology that arediscussed above are only a starting point for many new anddifferent applications. Indeed, a truly exciting feature of thismethodology is that measurement of the independent variableallows us to ask a whole new set of questions. These questionshave never been asked at the level of quantitative analysis aboutearly intervention program processes and outcomes.

74

68

Very simply, an eco-behavioral approach to the educationalevaluation of preschool programs will allow definition ofInteractions between programs and behavior in a precise empiricalfashion. Program outcomes may be examined and interpreted interms of the indices of treatment received. This willdramatically improve our ability to design, deliver, and supportearly intervention programs.

F.15

69

References

Barker, R. G., & Gump, P. V. (1965). &La school_, smell school .Stanford, CA: Stanford Unversity Press.

Barker, R. G., & Wright, P. E. (1968). Bcologi_cill psychology!Concepts And. l athgads. ice studying the env Ironment a humanbehavior. Stanford, CA: Stanford University Press.

Bel I, R. (1968). A reinterpretation of the direction of effectsin studies of social ization. Psychological Review, n,81-95.

Bijou, S. W., Patterson, R. F., & Ault, M. H. (1968). A methodto integrate descriptive and experimental f lel d studies atthe I eve I of data and empirical concept. Journal a Ala I led

Behavior Analysis. 1, 175-191.

Bricker, D., & Dow, M. (1980). Early education with the youngseverely handicapped child. Amaral at. the Assoc t at ion for.±be. Severely Handkjaped, 5, 130-142.

Bronfenbrenner, U. (1975). Is early intervention effective?Facts and principles of early intervention: A summary. In

A. M. CI arke, & A. D. B. C I arke (Eds.), garl y experience!

Myth and evidence, New York, NY: The Free Press.

Bronfenbrenner, U. (1979). Contexts of chi id rearing: Problemsand prospects. Awl= Psychologist, SA 844-850.

Brophy, J. E. (1979). Teacher behavior and its effects.Journal a Educational Psychology, n, 733-750.

Brophy. J. E., & Evertson, C. M. (1977). Teacher behav for andstudent learning In second and third grade. (pp. 79-95).In G, D. Borich (Ed.), 11lik appraisal tiech f nos raicepts.nit process. Reading, MA: Addison-Wes I ey.

Cicirel I I , V. (1969). pm impact af. kind sigith axiiiitho.the effect, a acad. Start jaa chi ldrents gogn it! va And

of development, Springfield, VA: U. S. Departmentof Commerce Clearinghouse, PB 184-328.

Cronbach, L. J., & Snow, R. E. (1977). Aptitudes Ansiinstructional methods, New York, NY: Irvington.

Dunkin, M., & Biddle, B. (1974). Ih study la teaching..Naw York, NY: Holt, Rinehart, & Winston.

F I enders, N. (1970). Anal yzing, teaching 'allay tor. Reading, MA:

Addison-Wes I ey, 1970.

Gray, S. W. (1977). Home-based program for mothers of youngchildren. In P. M itt ler (Ed.), EMU= .±Q practice inMOB! retardation Care And intervention (Vol. 1).

76

70

Baltimore, MD: University Park Press.

Greenwood, C. R., Del quadri, J., & Hal I, R. V. (1984).Opportunity to respond and student academic performance. InW. L. Heward, L. E. Heron, J. Trap-Porter, & D. S. Hi I I

(Eds.), Focus pa Behavior Analysis In Education.Columbus, OH: Charles Merrill.

Greenwood, C. R., Del quadri, J., Stanley, S., Sasso, G.,Whorton, D., & Schu I te, D. (1981). A I I ocat I ng opportunityto learn as a basis for academic remediation: A developingmode I for teach I ng. Monograph in bhav lora! D I sorders,

Sonar, 22-23.

Greenwood, C. R., Delquadri, J., Stanley, S., Terry, B., & Hal I,R. V. (in press). Assessment of eco-behavioral interactionIn school settings, Behavior Assessment,

Greenwood, C.R. Dinwiddie, G., Terry, B., Wade, L., Stanley, S.,Thlbadeau, S., & Delquadri, J. (1984). Teacher- versuspeer- mediated Instruction: An eco-behav lora I analysis ofachievement outcomes. Journal a App I led Behav for Anal ysls.11., 521-538.

Greenwood, C. R., Todd, N. M., Hops, H., & Walker, H. M. (1982).Behavior change targets in the assessment and treatment ofsocial ly withdrawn children. Behavioral Assessment,273-297.

Gump, P. V. (1969). Intra-setting analysis: The third gradeclassroom as a special but Instructive case. In E. WI I lems& H. Rausch (Eds.), Natural istic viewpoints jnpsychological research, New York, NY: Holt, Rinehart,& Winston.

Hal I, R. V., De I quadri, J., Greenwood, C. R., & Thurston, L.(1982). The Importance of opportunity to respond inchildren's academic success. In D. Edgar, N. Haring, J.Jenkins, & C. Pious (Eds.), Serving young handicappedchildren: issues ma research, Baltimore, MD: UniversityPark Press.

Karnes, M. B., Schwedel A. M., Lewis, G. F., Ratts, D. A., &Esry, D. R. (1981). Impact of early programming for thehandicapped. Journal gt tut Div is ion I= Earl y Chi I dhood,

62-79.

Karpow 1 tz, D. Ho? & Johnson, S. M. (1981). Stimulus controlin chi Id- fann y interaction. Behavioral Assessment,161-172.

Kopp, C. B. (1979). Mildly to moderately handicapped Infants.What should influence your approach to measurement? In T.Black (Ed.), Perspectives pa measurements A cal I action sitreadings fix. educators saf. rang handicapped children. TADS

77

71

Series Paper 11. Chapel Hill, NC: Technical AssistanceDevelopment System, Frank Porter Graham Ch I I d Development Center.

Lazar, I., & Dar I ington, R. (1979). Sumery, reports Last 1 ngg fleets After preschool. DHEW Publication No. (OHDS) 79-30179. Washington, DC: Government Printing Office.

Lazar, I., Dar I Ington, IL, Murray, Ho, Royce, J., & Snipper, A.(1982). Lasting effects of earl y education: A report Irogthe consortium for longitudinal studies. Monographs a the.

Soc !sty 12r, Research Child Devei opment, AZ, 2-3.

McGrew, P. L. (1970). Social and spatial density effects onspacing behavior in preschool ch I I dren. Journal at, Chad,Psychology end Psychiatry, 2, 197-205.

Med ley, D. M., & Mitzel, H. E. (1963). Measuring c I assro:mbehavior by systematic observation. In N. L. Gage 'Ed.),Handbook at research pD ±ftwhing. Chicago, IL: RandMcNally.

Patterson, G. R. (1982). Coercive faint I y process. Eugene, OR:Castel la Press.

Prescott, E. (1981). Relations between physical setting andadult /child behavior in day care. Adyences IA EarlyEducation mut On Cara, 2 129-158.

Rogers - Warren, A., & Warren, S. F. (1977). Eco 1 °glee Iperspactives_ bohav tor Dna I ys Is. Ba I t I more, MD:University Park Press.

Rosenshine, B. (1977). Review of .1-etching variables andstudent achievement (pp.114-120). In G. D. Bor ich (Ed.), Ihaappraisal Qj uuu rignaults. And process,. Reading, MA:Addison-Wesley.

Sameroff, A. J. (1975). Early inf I uences on development:Fact or fancy? Merr I I 1-Palmer Quarter 1 yo 21.267-294.

Schweinhart, L. J., & We kart. D. P. (1980). Young chi ldrengrow up: The effects of the Perry Preschool Program onyouths through age 15. Monographs pf /hit High/ScopeRe_ es Foundation,

Schweinhart, L. J., & Weikart, D. P. (1981). Effects of thePerry Preschool Program through age 15. Jour e1 a till

ion j Early, raUslhilado AA 29-39.

Sidman, M. (1960). Tactics j sci ent I fic research.New York, NY: Basic Socks.

Soar, R. (1973). Fol low- through Classroom groces measurementgag pupil growth (1970-197111 Final report. Ga I nsv I I le, FL:

78

72

Col lege of Education, University of Florida.

Stal i ings, J., & KaskowItz, D. (1974). Fol low-through calm=observation ovaImatIon. 1972-1973, Menlo Park, CA; StanfordResearch Institute.

Stanley, S. 0., & Greenwood, C. R. (1981). =SA& Code IS2r.Instructional structure And student academic response:Observer's Manual. Kansas City, KS: Juniper GardensChildren's Project, Bureau of Child Research, Universityof Kansas.

Stock, J. R., Wnek, L., Newborg, J., Schenck, E., Gabel, J.,Spurgeon, M., & Ray, H. (1976). gva I uation sat handicappedtchildren's early education program AcEEEk amt. ot(Contract No. OEC-0-74-0402). Columbus, OH: Battelle Centerfor Improved Education.

Strain, P. S. (1981). Conceptual and methodological issues inefficacy research with behavioral ly disordered children.Journal Qj ihs. Div ision jor, gar] y Chi ldhood, A, 110-124.

Strain, P. S., & Kerr, M. M. (1981). ilignittliallana.children In schools. New York, NY: Academic Press.

Taken I sh I, R., & Feshback, N. N. (1982). Ear I y ch i I dhood

special education programs, evaluation, and socialpolicies. Topics js,farly Chi ldhood Sasclal Education,1(4), 1-9.

Wahler, R. G., & Graves, M. G. (1983). Setting events in socialnetworks: Al ly or enemy in chl Id behavior therapy? BehaviorTherapy, IAA 19-36.

Zigler, E., & Bal la, D. (1982). Selecting outcome variables inevaluations of earl y childhood special education programs.Topics la Ear y Childhood Spec al Education, 1(4), 11-22.

79

73

CHAPTER VI

ECOBEHAVIORAL ANALYSIS WITHIN THE FIELD OF BEHAVIOR ANALYSIS

GARY B. VERNA

Abstract

In 1976 a conference held at the University of Kansasbrought together a group of ecological and behavioralpsychologists to discuss issues related to the work in eachdiscipline, concerns from across disciplines, and areas ofsubstantial agreement. The purpose of this paper Is to examinesome of concerns raised by ecological psychologists with thefield of applied behavior analysis. The intent of thisdiscussion is to determine areas of potential development withinapplied behavior analysis that may be necessary and profitable inthe development of an eco- behavioral science.

Introduction

In the mid 19701s a concept was coined and added to thecurrency of behavior analysis, ecobehavioral analysis,which like the Susan B. Anthony dollar appears to be stillseeking Its place in everyday commerce. At that time interestin, as well as concern over, the prospects of an ecobehavioralapproach to behavior analysis seemed high. The term had beenspurred by arguments from ecological psychologists that behavioris a part of a delicate system of interdependencies such that ifa single behavior were changed there would likely follow other,unanticipated changes.

The height of the discussion between ecologists and behavioranalysts was reached at a conference held specifically to niciressecological concerns with behavioral analysis at the University ofKansas in 1976. The proceedings from that conference werepublished in a volumne edited by Rogers-Warren and Warren (1977)and included papers presented by ecologists and behavioranalysts. Of particular concern to ecologists at that conferencewas whether the science of behavior analysis was capable ofaccounting for, let alone was sensitive to, the broaderecological network within which behavior was embedded. Thatconcern Is the focus of the present paper.

Most of us have forgotten the Susan B. Anthony dollar. It

was irregular in shape and, therefore, awkward or uncomfortableto have in one's possession. We were often, while thumbingthrough a hand full of change, at a loss as to what it

----- - - - ---

This paper was presented by the author in 1984 at the TenthAnnual Meeting of the Association for Behavior Analysis, Nashville.

80

74

represented. To make matters worse, the coin did not appear tohave an appropriate place in the natural community of currencyexhange. Others were often intolerant of receiving it as legaltender. It did not fit Into devices for which other coins hadalready been used with proven success. Indeed, the coin wasinconvienent to have around. Accordingly, it might be safelysuggested regarding the Susan B. Anthony dollar that, prior toits issue, there was not a natural community poised for itsentry. Is the same true of eco-behavior analysis?

The argument that I would like to make is that bothecologists and behavior analysts would benefit from a continuingdial ogue over ecological- behavior relationships, provided eachscience articulated eco-behavioral concerns from within, Its.framework. However, this dialogue awaits the commitment toeco-behavior analysis by the larger community of behavioranalysts. I shall argue also that, an eco-behavioral approachindeed has a place in the larger community of behavior analysis.And, although important work toward this end has already begun(Baer, 1982; Baer d Wolf, 1970; Greenwood, Delquadri d Hall,1984; Larson d Morris, 1983, Wahler et al., 1977, 1978, 1981),considerable work is yet to be done in order to provide for itsentry into this larger community of psyctologists. In

particular, a number of concerns both within and without behavioranalysis need to be addressed before either or both of the abovegoals can be reached.

Concerns Aca IQ Messed

Outside the field of behavior analysis, a number ofquestions have been raised regarding whether as a system it canadecAtely represent ecological-behavior phenomena. Withinbehavior analysis one factor in particular stands out. Thedevelopment of an eco-behavioral approach makes considerabledemands on a science of behavior. First, as Wahler and hiscolleagues (Wahler, Breland, Thomas, d Leske, 1977; Wahler & Fox,1981) have pointed out, an eco-behavior approach raisesconsiderable question regarding the relative emphasis to be givenspecific factors within the traditional three-term contingencyoperant paradigm (SD_R_Sr+), as well as raises questionconcerning which parameters of the stimulus-response elements arefunctional. In addition, to address eco-behavioral relationshipsrequires considerable empirical effort in the form of datacollection, analysis, and rellabiliq in order to identify validrelationships between the behavior of the organism and thebroader social and physical environment within which the behavioris immersed. Both of these activities may have considerablenegative stimulus function, that is, negative reinforcingconsequences, for the behavior analyst. In compensation for thiseffort, however, there are factors which operate in favor of anecobehavior approach. For example, the issue of side effects, asdiscussed by Willems (1973a, 1973b), and the related behavioralconcern with response classes, have yet to receive the systematicattention they deserve. Next, within behavior analysis there

81

75

attention they deserve. Next, within behavior analysis thereremains the need of developing more effective means forgeneralization and maintenance of behavior---little progresstoward this goal seems to have taken place since Stokes and Baer(1977) published their, now classic paper addressing the problem.However, one very vomising approach which they discuss focuseson the poised natural community of reinforcement (Baer & Wolf,1970), with already practicP., effective schedules ofreinforcement in operatior waiting only for the correctbehavior(s) to be emitted. In other words, they suggest that thebehaviorist take advantage of the broader ecology of socialstimuli in order to maintain and generalize behavioraltreatments.

Future investigators may well profit from approaches, suchas the natural community of reinforcement, which are moresensitive to the molar environment, that is, the broader naturalcommunity and which seek a more systematic accounting of thisenvironment's role in shaping and maintaining behavior. it 1:precisely this consideration which separates the eco- behaviorapproach from the traditional behavior analytic approach.Specifically, an accounting of the natural community and Itsfunctions does not easily follow from the current body ofresearch in behavior anlysis. Demonstrations of how theprinciples of behavior are used to manage behavior In a trainingsetting/or situation do not tell us how benavior is managed Inthe natural situation. Neither, as the evidence ongeneralization and maintenance suggest, does it tell us how togain entry into the natural community of reinforcement In orderto support new behaviors, nor how to poise the natural communityto receive new behavior when appropriate contingencies are notalready present. It can be argued with some success thatdevelopment of a technology of behavior over these past fifteenor so years has provided us witn considerable evidence of theefficacy of its principles in what might be, to colt; a term: "thebuilt environment"---specifically the training setting orsituation. It has also provided us with evidence of successfulmodification of behaviors that make less demand on the naturalcommunity, either for cuing or support of behavior. However, ithas provided less systematic research regarding how these sameprinciples operate in the natural community.

We benefit much from a closer examination of the ecologist'sconcerns with behavior analysis, since they represent importantsetting events for what may develop to be an eco-behaviorapproach within behavior analysis. The most important of theseconcerns were presented at the Kansas convention; accordingly,comments here will be limited to the arguments presented byecologists at that convention.

Specific Issues

Although ecologists (Gump, 1977; Willems, 1977) focused onsomewhat different Issues, a single and clear message wasprovided behavior analysts. The message was that behavior

82

76

analysis was in trouble, whether it was in relation to dealingwith side-effects or handling larger segments of environmental-behavior interactions (e.g., whole classrooms, schools,hospitals). For example, Willems (1973) pointed out thatbehavior analysis had not adequately accounted for the ecologicalimpact of its interventions to modify specific target behaviorsin the individual. In attempts to change specific targetbehaviors, they argued, there are often unplanned-forconsequences of direct intervention in the natural environment.According to Willem, behaviorist thinking about these theseunplanned treatment effects, as either incidential or as theresult of poorly controlled manipulation, denies the functionalrole played by the larger environmental context. Gump (1977)also pointed out that behavior analysis has chosen to deal withonly a narrow and unique portion of the environmental context in

which the person and his/her behavior are embedded. Gump'sconcern was not so much one of whether the behavior analystrecognized that the larger environmental context existed---hebelieved that they do, but whether the recognition of the largercontext will become a legimete concern of behavior analyticscience.

Similar concerns also have been voiced from within the fieldof behavior analysis. Wahler and his colleagues (Wahler, et al.,1977, Wahler b Fox, 1981) argued that the problem with behavioranalysis is its narrow scope, that is, its predilection towardthe measurement of only a few responses and their dyadicsettings.

Ecologists had other concerns besides what they consideredthe narrow or molecular temporal-spatial context within whichbehavior analysis operated. Gump coninued his argument bysuggesting that the field of behavior analysis may lacksufficient scientific resources for extrapolation to the largerecological situation. Willems sharpened this argument. Hepointed out that knowledge about the principles that characterizeand govern the systems into which the analyst must intrude toalleviate human suffering is necessary. This position raises ahost of metatheoretical as well as theoretical and methodologicalproblems. In elaborating his point he stated, "When operanttechnology is applied with a particular behavioral outcome inmind and the result is outright failure, marginal success, orsome vexing behavioral drift over time, it is easy to assert thatno larger, system-wide problem or no theoretical problem hasarlsen..1 submit that there is a theoretical issue here that hasto do with assumptions and predictions not borne out and with theoverall adequacy of the operant view of behavior to deal withbehavior-environment phenomena.

It Is clear that behavior analysis was being attacked on atleast two general levels. We would error greatly, however, if wedid not separate our response to these different concerns. Eachlevel of concern calls for a difterent type of response as wellas has different implications for the prospects of an eco-behaviorapproach. At the first or descriptive level of concern,

83

77

ecologists were arguing for research that: (a) admitted thepossibility of complex interdependencies between behavior and thebroader social ecology and that (n) added procedures that allowedtheir detection and measurement when they occur. The first ofthese concerns requires little effort on the part of behavioranalysis to satisfy. A number of activities can be cited whichare valid examples of the behavior analyst's interest in thebroader ecological context. This can be documented as far backas 1948 when Skinner published his novel Walden II. This utopiancommunity not only extolled the virtue of contingency managemerr,but also gave considerable space to environmental management. AsGump points out in his analysis of Walden II, 100 behavioralsettings were mentioned and over thirty of these were describedin some detail by Skinner. Through one of the characters,Frazer, we ara told that small communities are preferred tolarger ones. From the standpoint of contingency management,small communities would allow more opportunities for contact and,presumably more opportunity for better arrangements ofreinforcement contingencies. Another excellent example of thebehavior analyst's concern with the broader ecological context is

Baer and Wolf's (1973) discussion of how the natural communitycould be used to maintain behavior after treatment. They pointedout that for some problems the behavior modifier may discoverthat there exists already an effective community of peerssocially skilled behavior modifiers, practiced. effective, andwaiting only for an introduction of the subject. Baer and Wolfdid nut present a functional analysis of how the naturalcommunity was poised or operated to maintain the behavior of anindividual intruducted into it; instead they drew attention toan analogy - --what we theoretical behaviorists call a model. Itis worth our time to consider this model as it leads naturally tothe methodological concerns voiced by the ecologists. The modelwas one of a mouse trap---a rather grim analogy.

Never the less, the essence of a trap in behavioral terms,according to Baer and Wolf, is that only a relatively simpleresponse Is necessary to enter the trap, yet once entered thetrap cannot be resisted In creating general behavioral change.For tho mouse, the entry response is merely to smell the cheese.Once it enters, the trap accomplishes massive behavior change.Also the modification has thorough generality: the change inbehavior will be uniform across all environments, it will extendto all of the mouses behavior, and it will last indefinitelyInto the future. Finally. the trap affords a great amount ofbehavioral change by a relatively slight amount of interventionby the analyst. For those of us who are experienced mousetrappers, it doesn't take much effort to recall that there ismore operating in a mouse trap than simple contingency managementof the mouse's behavior. That is, there is certainly moreoperating here than providing an appropriate discriminativestimulus (bait) and powerful consequence. There is considerableenvironmental management as well. For example, it is no minortrick getting the trap poised to carry off its task. Traps haveIdlosyncracies which not only differ one trap to the next butwhich make it necessary to learn what those idiosyncracies are in

84

78

order to set them. There has been many a time when a trap wasset, operated appropriately, but where the mouse escaped.Sometimes we use too large a trap for the size mouse we seek tocatch or one which requires greater force on the part of themouse to set off the trap than it is capable of emitting.Finally, it is possible that some mice do not like the smell ofcheese.

The point of these 4.wo examples is simple, yet illustrativeof behavior analysts existing commitment to the ecologistsconcerns with the descriptively mclar environment. First, bothexamples Indicate a concern with contingency arrangements withinthe broader ecological context. Both emphasize three importantelements of any ec6logy from the standrint of contingencymanagement: (a) setting availability, (b) response opportunityand, (c) the network of naturally occuring reinforcers. Next,both provide for the recognition that more is involved in thesuccessful operation of a community or trap than simply theprovision of discriminative stimuli and reinforcing consequences.Human ecologies, like mouse traps, are not arbitrary settings. The)ettings in which people behave have an evolutionary history justas the humans who inhabit them. That is, their features, havefunction, It may be that their most important function Is tooccasion reinforcing consequences instrumental to the shaping andmaintenance of appropriate behavior, as well as occasion theextinction of inappropriate behavior. Yet, we know very littleabout ecologies for natural communities. Behavior analysts knowvery little about how the natural community performs itstunctions. We simply have not been collecting the type of datawhich documents these functions of the natural community. Whatwe need at the descriptively molar level is the same commitmentto systematic, functional analysis that has characterizedbehavior analysis at the molecular level. It is this fact, morethan anything else that has drawn the attention of ecologistsbehavior analysis. We now consider some additional points.

First, there has been considerable confusion regarding therelationship between particular methodological procedures, (i.e.,naturalistic /descriptive versusexperimental /manipulativeapproaches, and the pursuit of molar phenomena. Specifically, ithas been argued by behaviorists that, because ecologists usenaturalistic methodology, molar phenomena can only be studieddescriptively. Bijou, Ault, and Peterson (1968) form within thefield of behavior analysis, and Willems and Rusch (1969) fromwithin the field of ecological psychology, have similarlyaddressed this point indicating there is no isomorphicrelationship between the phenomena of interest and a particularresearch methodology (Bandura, 1981). Both groups have gone onto suggest that with respect. to methodology, a pluralisticapproach would be advantageous within their respective fields.

The second type of concern raised by ecologists regardsquestions about behavior analysis as a science. Does it havethe resources TO deal effectively with molar phenomena? Thereare three possible interpretations one can give to this concern.

85

79

Let me separate two of them. The scientific basis of radicalbehaviorism, as with methodological behaviorism: is what might betermed "theoretical behaviorism." Accordingly, we need to ask ifthe ecologists concern with behavior analysis is a concern withthe behavioral or theoretical aspectz of behaviorism (Spiker, 1973).

Behaviorism names a commitment to two major points. First,as originally proposed by Watson (1913), the primative, undefinedterms of the psychological language need not differ from those ofthe physical and biological sciences. Spiker (1973) referred tothis as the definitional tenet, and by Kendler and Spence as theoperational tenet, of behaviorism. Second, given this basictenet, the task of psychologists is to find process laws aboutbehavior. Admittedly, radical and methodological behavioristsdiffer as to how to organize laws into theories as well as whentheory construction ought to begin; nevertheless, they are agreedon the two points just mentioned.

An examination of ecological literature suggests that it isnot the behavioral features but rather the theoretical aspects ofbehaviorism with which they are concerned. For example, inpointing out some basic similarities between behavioral ecologyand behavior analysis, Willeas (1973a) sates that in general theecologists and the behavior analysts place a great deal ofemphasis upon empirical data. Both tend to focus upon whatorganisms do, defined quite physicalistically, in relation to theenvironment and teak to deemphasize the use of hypotheticalconstructs to represent what the organism feels and thinks.Similarly, with respect to representi,1 behavior-environmentInteractions, both seek lawful relationships.

Despite these similarities, wa ought not lose sight of thedifferences between these behavior sciences which need to berecognised and dealt with. For example, the focus of ecologicalpsychology has been primarily centered around finding out whatgoes on in the natural community. Accordingly, its methodologyhas consisted of quantitative and systematic observation ofnaturally occurring phenomena with as little intrusiveness on thepart of the investigator as possible. In contrast, behavioranalysis has focused on indentifying socially significantconcerns with behaving organisms and intervening to correctproblems identified in the natural community. Empirically,behavior analysis focuses on what was done (the social problem),how it was done (intervention) and what was the outcome (socialvalidity).

Ecological psychology on the other hand, takes a rather widespan or molar approach to studying behavior in terms of spatial-temporal dimensions. It Is also interested in the long termeffects of the interactions between broad environmental settingvariables and the behaving organism. Also, it is primarilyinterested in raising questions such as, what is the pattern ofinterdependencies between organism - behavior - environment and whatare the far reaching implications of changing either theresponses of the organism or the setting in which the organism is

86

80

Immersed?

Behavior analysis, however, might be said to be moreinterested in behavioral incidents, specifically defined asdiscrete responses. Accordingly, its focus Is more narrnw ormolecular, both spatially and temporally, dealing within themoment to moment framework of specific, discrete responses andthe immediate stimulus or setting events surrounding them. In

behavior analysis the environment is arranged carefully so as tooptimize the use of discriminative cues, which either occurnaturally or are produced by the investigator or hisconfederates. Through the process of operant strengthening inwhich responses are reinforced, discriminative stimuli come tocontrol the emission of responses. Thus, independent variablesare manipulated directly.

The advantage of this highly controlled operative procedurehas been that behavior analysts can then make strong Inferencesregarding causal relationships. The unobtrusive, non-manipulative approach of the ecologists neither allows for thislevel of control nor this level of inference. Instead,ecologists look for nature-given comparisons (e.g., big versussmall school differences). Rival hypotheses are tested orexamined by internal checks of the natural ocurring phenomenabeing tested.

However, like methodological behaviorists, ecologists differfrom behavior analysts in matters of theory construction. Thisleads us to a consideration of theoretical behaviorism, and tothe final two interpretations of the ecologists concerns. It ispossible that, for ecologists, molar phenomena cannot bepredicted, explained, and understood in terms of the molecularprinciples of conditioning; hence other, molar units andvariables become necessary. That is, when we turn from a focus onspecific, discrete responses operating in the well controlledcontext of intervention to systems of responses and/or numbers ofindividuals operating in multiple settings within the naturalcommunity, contingency management oecomes less feasible and otherunits and variables become important to the prediction andcontrol of behavior. There Is one error of logic and oneprescription being made here.

First, this argument confuses a preference for how toexplain and/or write laws to represent these interdependencieswith a preference to investigate molar versus molecular phenomena.Cairns (1979) made the same point when he proposed thatsocial acts are embedted in a lager social matrix. Thishas implications for how one goes about understanding the natureand determination of social patterns. Thus, it seems possible toagree on the phenomena of Interest general ly, without agreeing onhow to explain or write laws to represent it.

Next is the prescription. It may be that ecologists are notarguing against the use of a molecular language to representmolar phenomena, at least not in principle, but that it !s

67

81

specifically the language of contingency management that isinsufficient to account for molar phenomena. It is difficult tosay when a particular language is no longer adequate to handlethe phenomena of interest. Probably, the best criteria would bethat it has failed when another theory/or language was shown topredict and explain everything that the old one did, and inaddition, correctly predict new phenomena that the old languagedid not. However, it is questionable if such a new theory hasshown itself to meet these criteria. On the other hand, theability of a new language to predict should not be taken toimply that Skinnerian conditioning principles are sufficient toaccount for molar phenomena. That is, It might be that behavioranalysis would benefit, as Whaler and his colleagues havesuggested, from further conceptual and methodological expansion.

l'hese researchers have suggested that there has been toomuch emphasis on the response consequence side of the threetermoperant contingency and not enough attention paid to theantecedent side of the contingency or reinforcement paradigm.Wahier and his col leagues (Wahler 8. Fox, 1981) have presentedinteresting data suggesting that mother's extrafamily socialinteractions may influence her Interaction patterns at home.That Is, events temporally far removed from the ongoing moment tomoment interactions between a mother and her child may havesignificance for the nature of that interaction. Greenwood,Delquadri, and Hall (1984) have also shown that the antecedent/orsetting events in the classroom predict academic performanc viatheir effect on student's opportunity to respond. Similarly,Epling and others (EplIng, 19?? ) have pointed out that there maybe a problem in the pursuit of interventions based on a principlethat stipulates a relationship between behavior and its immediateconsequence. However, evidence is increasing which supports thatbehavior is not always maintained as a result of contiguitybetween the response and the reinforcer. In environments wheremany sources of reinforcement are operating, behavior may beacquired and maintained on the basis of correlation between rateof response and rate of reinforcement. Behavior analysts,faced with a behavior problem, have traditionally soughtImmediately present events that can be altered to change specificbehaviors. In some cases environmental events that directlyfollow behavior cannot be found. This last example not onlysupports the argument that behavior analysis may need torethink the relative emphasis It currently gives to specificaspects of its paradigm, but raises another point which isimportant here.

Often times we are so used to looking In specific directionsfor explanations that we miss opportunities to expand our scienceIn a direction that would yield more accurate predictions andbetter explanations. Specifically, it is suggested that behavioranalysis may have been trapped from within partly by its pastsuccess as an emerging applied science. While it may have beenstrategic for behavior analysis to developed in the direction ithas over the past fifteen or so years, it may be that Itsrelative emphases have now become more problematic than strategic

88

82

for its own expansion.

Id Toward An Exisrfinhnylcx Analysis?

There may be factors within behavior analysis that eithermilitate agrinst an eco-behavioral approach or at least are in partresponsible for the slow rate of acceptance among the largercommunity of behavior analysts. One thing stands out as weexamine the behavioral literature since Willems (1973a) publishedhis first attack on behavior analysis. That is, there appears tohave been little in the way of a direct response on the part ofbehavior analysts to the concerns he voiced. For example, Wahler(1980) in his review of the Journal a Appl:ed Behavior Apalysisfound that between the years 1967-1977 only three references werelisted under the discriptor "setting events." An examination ofthe same journal for the period 1978-1979 showed that noreferences to setting events had occured. The present authorreviewed five major journals in the field of applied behavioranalysis from 1977 to the present. Less than twenty articlesfrom these journals was devoted to what could be labeled eco-behavior analysis. Likewise, no indication of a trend inincreased publications was evident. It is perhaps not simplycoincidental that during the past years there has been atechnical drift in articles published in the Journal ad:AppliedBehavioral Analysis and Behavior Modification. For examplesHayes and his colleagues examined the publication of MA from1968-1977. They found that studies which were devoted to thedevelopment and extension of behavioral principles in appliedsettings (systematic applications) were fast disappearing fromthe pages of the journal (35% in 1968-72 to 10% in 1977).However, methodological and purely technical (how to do it)articles have dramatically increased over the same period of time.Hayes et al., 19?? used their analysis of attek to make two points.First, they argued that basic or conceptually oriented researchwas usuful to the applied behavior analyst. Next, there has beena decrease over the years in this type of research. At somepoint we need to ask when the technical drift experience has beensufficient to begin calling it a conceptual shift, that is, ashift toward a new agenda or priority for the field as a whole.It may be that time has come. 1 point ow.' again the work thathas yet to he completed regarding our understanding of responserelationstOps, as in the case of response classes and sideeffects, as well as the work to be completed regardinggeneralization and maintenance of behavior. It is possible thatthe attitudes such a drift occasion in the minds of behavioranalysts would make it difficult for the eco-behavioral approach toacquire the attention some of us expect it should receive. Addedto this, the fact that eco-behavioral analysis may requirecommitment to a number of considerations including the following:(a) relaxation of the definitional tenet, (b) relaxation ofexperimental demonstration of functional relations at least for atime, (c) introduction of correlational data, and (d)considerable effort in data collection and analysis, and it maybe that an ecotehavioral approach to behavior analysis may berelegated to the peculiar tew, rather like coin collectors, who

69

83

are attracted to oddities like the Susan B. Anthony dollar.Research will determine the outcome and In that regardworkability not agreement wil I be the acid test of an eco,behaviorapproach.

90

84

references

Baer, D. (1982). Applied behavior analysis. In G. T. T. Wilsonand C. M. Franks (Eds.), Contempory behavior therapy.New York: The Guilford Press.

Baer, D., & Wolf, M. M. (1970). The entry into naturalcommunities of reinforcement. In R. Ulrich, T. Stechmite,and J. Mabry (Eds.), ripbtrol Qf limn Behavior! Volume11. Glenview: Scott Foresman.

Bandura, A. (1981). In search of pure unidirectionaldeterminants. 13eltav tor Therapy, 12, 30-40.

Bijou, S. W., Peterson, R. F., & Ault, M. H. (1968). A methodto integrate descriptive and experimental field studies at thelevel of data and empirical concepts. journal a Appl led.

Behavior Analysis, 1, 175-191.

Cairns, R. B. (Ed.), (1979). lie analysis pf. =rte.'. interaction.:Method Issitess And I I Lustartions, HI I isdale: LawrenceErlbaum.

Epi ing, F. W., & Pierce, W. D. (1983). App I led behavioranalysis: New directions from the laboratory. The. Behav forAnal I, 27-37.

Gump, P. V. (1977). Ecological psychologists. In A. Rogers-Warren &S. F. Warren (Eds.), Ecological Perspectives In BehaviorApalysis. Baltimore: University Park Press.

Greenwood, C. R., De I quadri, J., & Ha I I , R. V. (1984). Opportun I tyto respond and student academic performance. In W. L.Heward,T. E. Heron, J. Trap-Porter, & D. S. Hill (Eds.),Focuson Behavior Analysis Io,Educatton. Columbus: CharlesMerrill.

Hart, B., & Rogers-Warren, A. (1978). M I I ieu I anguage training.In R. L. Schiefelbusch (Ed.), Language intervention strategies.Baltimore: University Park Press.

Hayes, S. C., Rincover, A., Solnick, J. V. (1980). The technicaldrift of applied behavior analysis. llama nt AppliedBehayior Analysis, 11, 275-286.

Kend ler, S. C., & Spence, E. K. (1971). Tenets cf neobehav lorism.In H. H. Kend I er & J. T. Spence (Eds.), Essays Qnneobehemlorismi A memorial yninme IQ Kenneth If, Spence. NewYork: Appleton-Century-Crofts.

Larson, S. E., & Morris, E. K. (19831. On the usefulness of thesetting event concept in behavior analysis. Paper presentedal- the Ninth Annual Meeting of the Association for BehaviorAnalysis, Milwaukee, WS.

Rogers-Warren, A., Warren, S. F. (1977). Ecological

91

85

Perspectives. In Behavior Analysts. Baltimore: UniversityPark Press.

Skinner, B. F. (1948). Walden Iwo. New York: Macmillan.

Spiker, C. C. (1973). Behaviorism, cognitive psychology and theactive organism. N. Daton, H. W. Reese (Eds.), Life-spandevelopmental psychology Dialectical Prespective 213.

erw_Lmstalresearell. New York: Academic Press.

Stokes, T. F., & Baer, D. (1977). An imp I is it technology ofgeneral ization. Journal a App I led behavior Analysis,

111,, 349-367.

Wah I er, R. G., Breland, R. M., Thomas. D. C., & Leske, G.(1977). Social systems analysis: Implementing an alternativebehavioral model. In Al. Rogers-Warren & S. F. Warren (Eds.),Ecological perspectives j.p.BehmapL Analysis, Baltimore:University Park Press.

Wah I er, R. G., & Fox, J. (1981). Setting events in app I ledbehavior analysis: Toward a conceptual and methodologicalexpansion. Journal al.ftlied ,Behavior Analysis, 14,327-338.

Watson, J. B. (1913). Psychology as the behaviorist views it.Psychological Review, 2Q., 158-177.

Wil)ems, E. P. (1973a). Behavioral technology and behavioralecology. Journal plApplied Behavior Analysis,151-165.

Willems, E. P. (1973b). Behavioral ecology and experimentalanalysis: Courtship is not enough. In J. R. Nesselroade &H. W. Reese (Eds.), Life Span Developmental PsychaloorMethodoloalcad issues.

Willems, E. P. (1977). Steps toward an ecobehavioraltechnology. In A. Rogers-Warren & S. r Warren (Eds.),Ecological. Perspectives Ip. Behavior Analysis, Baltimore:University Park Press.

WII lems, E. P., & Raush, H. L. (Eds.), (1969). NaturalisticViewpoints In, Psycholtcdcal Research. New York: Holt,Rinehart, and Winston.

9 2

87

CHAPTER VII

CONCLUDING REMARKS

CHARLES R. GREENWOOD AND CARMEN ARREAGA -MAYER

The purpose of this monograph has been to serve as a forumfor papers developed by post-doctoral scholars concerning thedeveloping trends within an e:o-behavioral approach to psychology,special education, and applied behavior analysis. These papershave demonstrated that only recently have researchers attemptedto develop and assess ecological factors natural stimuli,program variables, special education procedures) in aquantitative fashiol and In temporal relationship with studentbehavior. The papers have pointed out that a quantitativeapproach to ecological factors is necessary if one wishes to:(a) define program events that children actually receive in thecourse of their daily activities in special education and (b)relate these events in meaningful ways to outcome gains (e.g.,academic achievement) that result over time. in the past we havenot had high fidelity measurement of ecological factors withinour evaluation designs. Thus, we have problems monitoring thequality of educational programs, quantifying their content, andrelating them to student progress.

The paper by Dorsey reviewed the literature relating toacademic learning time as a quantified and directly assessedmeasure of educational process. The paper by Greenwood arguedthat variation in students academic progress is a function ofvariation in student's opportunity to respond and described aeco-behavioral day in the life of a student based upon data froma direct observation code. The paper by Arreaga-Mayer examinedthe literature concerning home and school eco-behavioral datathat would explain the generally lower school achievement gainsmade by culturally and I inguistical ly different students. Shereported only a few studies using high fidelity eco-behavioralmeasurement. Of those few studies attempting to assess ecco.behavioral variables, only low fidelity techniques (e.g.,involving surveys and ratings), of questionable validity wereused. The paper by Carta examined similar issues within theevaluation literature for early childhood special education. Shealso concluded that quantitative assessment of program factors iscurrently lacking. She proposed an ecobehavioral observationsystem for use in early child settings and evaluation research.The last paper by Verna examined theoretical and empirical issuesrelated to current ecological developments within the field ofapplied behavior analysis. The paper examined issues such asside effects reported in applied behavior analysis research.Side effects are considered to be a function of a broaderecological field of variables which are not currently assessed inapplied behavior analytic studies. The paper questions whetherthere Is a mingling of ecological and behavioral psychology assome people have suggested or whether the two disciplines aresimply persuing eco-behavloral phenomena from d parallel

93

88

perspective, each with its own metholodology and unique points ofview.

In contrast to traditional assessment approaches, it hasbeen demonstrated that eco-behavioral interaction enab!es one to(a) display the structure and pattern of momentary ecologyvariables, for example academic instrucflon, and (b) behavioralrelationships to this ecological structure. Eco-behavioralasssessment is dynamic, focusing upon changing situationalfactors and subject responses. As the Chapters in this volumehave suggested, this approach is increasingly evident in theliterature, however it is Just beginning to have an impact onwork in applied settings. The relative newness of +h:5 approachis due to the conceptual and metholodological issues reviewed inthis volume, in addition to practical issues.

Perhaps the greatest practical problems, are tho increasedresources and costs associated with assessments of this typecompared to traditional forms of assessment. For observationalstudies, large samples of data are required for individualsubjects In order to conduct extensive studies of eco-behavioralinteraction. These studies require computer assistance forrecording, storing, and analyzing the data. Moreover, thestatistical procedures for these data are relatively new andissues concerning serial correlation and violation of assumptionsof independence remain to he solved. These costs may inhibitmany applied researchers. In Just the area of school-basedassessment, we can point to no examples of eco-behaviorinteraction being used (e.g., for screening, placement, orprogress monitoring, etc). However, with the seemingly endlessdevelopment of electronic technology and Its lowering costs, thisapproach becomes increasingly more feasible.

Programmatic studies based upon analysis of eco-behavioralinteraction offer the ability to investigate many of the currentissues and problems facing the field. These include (a) settingevents and stimulus control, (b) the natural conditionssurrounding development of specific behavioral repertoires, (c)maintenance and generalization of behavioral repertoires, and (d)interventions based upon precision interpretations ofnaturalistic events. Thus, an eco-behavioral interactionapproach may prove to be an instructive development in behavioraland special education research.