DOCUMENT RESUME ED:234 573 EC 160 446 · DOCUMENT RESUME ED:234 573 EC 160 446 AUTHOR Haring, Norris; And Others. TITLE Investigating the Problem of Skill Generalization. INSTITUTION

DOCUMENT RESUME

ED:234 573 EC 160 446

AUTHOR Haring, Norris; And OthersTITLE Investigating the Problem of Skill Generalization.INSTITUTION Washingtbn Univ., Sedttle. Washington Research

Organization.SPONS AGENCY Special Education Programs (ED/OSERS), Washington,

DC.PUB DATE Jun 83CONTRACT 300=82-0365.NOTE 50p.PUB TYPE Reportt --Descriptive (141)

EDRS PRICE MF03/PCO2 Plus Postage.DESCRIPTORS Cognitive Processes; *Generalization; *Research

Methodology; Self Control; *Severe Disabilities;Teaching Methods; *Transfer of ,Training

IDENTIFIERS Washington Research Organization

ABSTRACTThe booklet describes the 'efforts planned by the

Washington Research Organization (UWRO) to study ways in which,generalization of skills may be fgcilitated by severely retardedindividuals.,It is explained that activities. of uwkw, are organizedaround four major tasks: descriptive and laboratory research;research in natural educational settings;. evaluation; andcommunication about ongoing studies, procedures, ,and products. Thecircular administrative structure.of UWROrs diagramed and makeup ofthe advisory committee described. A section of basic researchconcepts focuses on such topics as skill generalization,instructional piogramming for generalization, and methodologicalaspects. Four approaches to generalization are proposed: anecological approach to describing and then altering conditions withinecological

educational °environment; a performance pattern approach todescribing and matching individual learning characteristics andinstructional techniques; a self-control approachin which 'severely,handicapped students learn to manage their own behavior; and a'cognitive strategy approach that focuses on helping students usehigher-order levels of thinking. Each of the approaches is describedin terms'of background and purpose of studies, design of studies, andexpected outcomes and products. (CL)

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

* from the original document. *

***********************************************************************

.4 '

WashingtonResearch

Organization

Research in Education ofthe Severely Handicapped

Univ iit.Washingti.rn25A Parringtt,n Hall, DC-65

Seattle, WA 98195

t'3. DEPARTMENTOF EDUCATION

NATIONAL INVITUTEOF EDUCATIOLIED.J7:ATIONAL RESOURCES INFORMATION

ri.NT:ER/This dhzurnent has been reproduced as

',volved from the person or organizationmiginating it -

Minor changes have boon made to improver' ninciuchon quality.

Points of view or opinions stated in this docuinent do not necesserav renresent oilicial NIEposition or polar;.

INVESTIGATINGTHE PROBLEMOF SKILLGENERALIZATION

"PERMISSION TO REPRODUCE THISMATERIAL HAS BEEN GRANTED BY

<-.s7

TO THE EDUCATIONAL RESOURCESINFORMATION CENTER (ERIC)."

Norris HaringKathleen LibertyFelix BillingsleyEarl Butterfieldand Owen White

Norris Harir,g, Principal InvestigatorKathleen Liberty. Project Coordinator

Michael Boer. Editor

The activity which is the subject of this report was supported inwhole or in part by the U. S_ Department of Education (Cohtract No.300-82-0365)_. However, the opinions expressed herein do notnecessarily reflect the position of policy of the U.S. Department of

. Education, and no official endorsement by the Department shouldbe inferred.

Printed in the United States of America.

June 1983

Foreword

It is only within the short span of the last two decadesthat teaching models for the severely handicapped havebeen successfully demonstrated., As a result of system-atic instruction, the severely handicapped have demon-grated that they have We ability to acquire self:help,social, and vocational skills.

Haying attained some success, educators of the se relyhandicapped must face another problem: acquisition ofskills rarely guarantees that the individual can applythose skills in natural settings. The process of adap:ingskills from instructional to natural settings is calledgeneralization.

The generalization phenomenon has been recognizedand, studied in research laboratories for many years; butthe need for strategies to facilitate generalization hasnever been more critical than It is now As we face theintegration of severely handicapped persons into allfacets of society, we realize that the main stumblingblock is the difficulty these individuals.encounte7 ingeneralization.

As we examine the natural setting, weave the multitudeof unpredictable variables that stand in the way of adap-tation for the severely handicapped. Each new st;mulusdemands generalization within a class of responses to fitthe particular situation in the environment.

Up to now, educators have relied upOn two strategies forgeneralization. The first, what Trevor Stokes and DonBaer have called "train and hope," involves thoroughtraining during the acqUisition phase and hope for later

`-x

generalization. The second, identified by Lou Brown,utilizes the "criterion of ultimate 'functioning" to assurethat all the skills taught have an ultimately usefulpurpose or functiom

EdiicatOrs, following the course of least resistance; havetried to conduct their instruction in the natural. setting:This "practical" approach is simplisticr'and difficult toachieve. Educators must search for appropriate_ naturalenvironments; but effective teaching involves unavoid-able adaptation of those environments.

A further problem is that the natural setting .may notprovide sufficiently_ predictable opportunities fofseverely handicapped persons to practice particular newskills. Thus It is inefficient to develop large-')umbers-ofne \A" skills in this way. The challenge raised by thecurrent difficulties is to determine whether or .not. SkillSc. an he developed fo an acceptable rate within aninstructional 'environment and then transferred to thenatural environment.

Educators and advocates all over the United States haveexpressed the need. for solutions related to facilitatinggeneralization. As a result, the U.S. Department ofEducation's Special Education Programs (SEP) has setaside funding for study and analysis zif generarizationprocesses with the severely ha.ndicapped: The Wash-ington Research Organization, is pleased to participate intheSe efforts.

Norris . HaringPrincipal InvestigatorSeattle, 1983

vi

cy

Table of Contents

Foreword.

Table of Contents vii

ONE: The Washington Research Organization 2Research in Generalization 3

Evaluation 5

Communication 7

Advisory Committee 9Administration and Management 10

TWO Basic Research Concepts 12Skill Generatization 12instructional Programming for Generalization 15 .

Methodology. 17

THREE: UWRO'S Approaches to Generalization 22Studies in Ecological Variables 25.StUdieS in Performance Patterns . 29Studies in Self-Control ' '33Studies in Cognitive Strategies 37

FOUR: Putting it All Together 40

Ms. Cindy Burchan is pleased with her new job at the.S(,attte howl. She ha' loaded the industrial dishwasherfor the first all of the plates on the bottom in neatlows and all of the glasses on the top: It was easy tofigure out where they went: She c loses the door withsatistaction. But where are the buttons to start themar hint?? l bey 're not on the front cif the machine_; noron the side: Behind the dishwasher; on the. wall; Ciii0sees a row 01 buttons; switches, dials; and lights. Someof the lights are dark; while others are_ glowing red orgreen. She stands bewildered beforeJhe display. Themanager et the kitchen rushes over glares at Cindy; and

- rapidly puSh-es some buttons; sets a dial; and fliCkSswitch. He barks; "Start on the next load," wonderingwhy he ever agreed to give a retarded person a chance,anyway.

_Mr. White gazes at the assessment data for Richard. Heis deprr..csserf This is the third year he has had Richard inhis class, the third year he has conducted assessment,and the third year he must prepare instructional objec-tives for Richard:s IEP. Last year he taught Richard to say;"My name is Richard Clark," when asked, "What's yourname?" or ':Who are you?" This year; he only answerswith, "Richard.:' "That really won't help if he gets lost;"sighs Mr. White. He ruefully writes' the objective for"says own name" for the second year in a 'row. He looksat some More assessment data; collected over the first sixweeks of school: It is taking Richard even longer to learnto say his address than it did to say his name; and it lookslike there is no guarantee that he will remember that nextyear. Mr. White considersi just getting him an ,.d.bracelet; but remembers wh 4 Richard's parents said. Hewrites an objective for "sa own address" and shakeshis head.

\

lody is screaming, so loudly that his face is eggplantpurple-. Mrs. Loomis stares helplessly at him. She goesover and picks up the tennis shoes from the cornerwhere he threw them moments ago. She knows thatlady's teacher to!cl her that loch was able to put on thesevery same shoes without any help. The screaming is nowbroken 1-iy gasps; lolly winds up to an even higherpitch. Mr: Loomis veils up the stairs; "Where are you?We're all in the car waiting!" Mrs: Loomis quickly picksup /ody; puts his shoes on him; and carries him downstairs. loch writs screaming when the go out the door:

fhink goodness,- she say herseh. "-

=ONETHE WASHINGTON RESEARCH

ORGANIZATION

The problem for Cindy, Richar&and Jody is generalization; or thelack of it, tb be more precise. The setting changes, time passes,and it is somehow as if they had never learned what to do in the firstplace. ThiS is brie of the most important problems we have had toface since we began educating severely handicapped students, Itis brie that must be solved if education is to be an effective prepara-tion for life in society.

The Washington Research Org_anization LUWRO), located on thecampus of the University of Washington in7Seattle, was awarded dfive-year contract by the U.S. Department of Education't SpecialEducation Projects _(3EP); in October of 1982, The mission ofUWRO is to identify instructional strategies through. empiricalinvestigations that enable practioners to promote generalizedresponding by severely handicapped persons. If the research weconduct is productive; 'we will develop practical _instructionalmethods that ensure that severely handicapped individuals areable to use learned skills in environments outside of the trainingSetting, and that those skills remain useful ongafter formal educa-tion has ceased. We will also take steps to see that the methods wedevelop are available to all practioners:

r:eJ

2

A sequence of objectives will need to be met if we are to besuccessful in accomplishing our mission. First, UWRO investiga-tors will conduct descriptive and intervention _research to _identifyspecific environmental, instructional, and individual characteristicsthat affect the probability of generalized responding. Although rari-ables so identified may increase our understanding Of ger)eraliZa-non. little of practical value is accomplished by identification alone.Our second objective 18 to conduct research designed to test theeffects. of manipulating or`-changing environmental, instructional,and individual performance variables. Third, UWRO investigatorswill_conducUntervention research in controlled and_natural settingsto deterinine instructional and curricular Strategies that increase.theprobability of generalizeci _responding. This lead.; us to our fourthobjective. Teachers and other practioners will be trained in instruc-tional_ methods and curricula identified by UWRO research; in orderto determine whether procedures may be implemented effectivelyand practioally, within operating and budgetary constraints ofeducational settings:

.The success of meeting each objective will be determined bychanges in pupil p_erformance data, and by determining the overallpractical impact of such changes. The effects of interventions willbe evaluated according to the change in frequency, quality, andquantity of generalized responding __from pre_-interventign_le.vels.The results of VW, 10's studies will be evaluated according topsychological and educational_ research sten erds for reliability,Validity, methodological considerations, and alytic techniques:The extent_to which severely handicapped inci viduals successfullydeMonStrate generalized behavior will be the extent of our successin meeting these objectives.

The activities of the Washington Research Organization aredesigned to meet these objectives_ and are organized around fourmajor activity categories ('tasks''): descriptive and laboratoryresearch; research in natural educational settings, evaluation, andcommunication. These tasks will be supported by the activities ofthe Advisory Committee and by project management. This over-view describeS the activities of UWRO, basic concepts in general-ization research. and our four different approaches to the problemsencountered by Cindy, Richard, and Jody:

Research in Generalization

Research in _g_eneralizatibn constitutes the major activity of theInstitute. These activities are divided into two categories, "Descrip-

4

tive and Controlled Laboratorytudies" (Task 1) and "Research inNatural Settings" (Task 2):-.Task 1 activities are designed to identifyprecise variables that affect generalization and to test specific strat-egies under tight experimental control: They will be conductedprimarily during the first two years. These studies are designed to.,provide the background information needed so desperately in ourunderstanding of generalization.

Taek 2 research Will investigate the effeCts of interventions innatural _educational settings. In the fourth project year,_ Task 2research activities will include investigations of the efficacy ofguidelines developed for practioners from_ UWRO research. Theguidelines will define how procedures are to be applied in naturalsettings. These investigations will seek to determine how applyingthe guidelines affects the generalization demonstrated by severelyhandicapped individuals, and also how guidelines might beimproved for more accurate and effective implementation:

We are fortunate to have established cooperative arrangementsWith three local school districts to provide the settings and subjectsfor our research: These three local educational agencies-are _LakeWashington School District No. 414, North Shore school DistrictNo. 417,° and Issaquah _School_ District No. 411. Personnel fyor0these districts will work closely with 'he Senior Investigators,providing the first contact with parents; guardians, and teachersand arranging for research settings. -

Representatives of tfie districts wil; meet with UWRO staff asmembers of the Direct Service Congortium. Ralph Bohannon;Director of Special Services for Lake Washington, ;s an experi-enced researcher and has cooperated in previous University ofWashington research projects: This _large _district is also- repre-sented by Nancy Wilson, Principal of the Gordon Hauck Center;Ruth Hayes, Special Education Administrator; and Joyce VandenHOOrn, Administrative Assistant. Fred Rowe, Director of SpecialEducation, represents North Shore._Lorna Tripp Wilturner, Directorof Special Education, will represent Issaquah as well as the UWROAdvisory Comtnittee, of which she is a member. Joining the' repre7sentatives of the cooperating districts is Gregory Kirsch, Director- ofSpecial Edgcation for the state educational agency.

Research may also be conducted in the Eicperimental EducafionUnit (EEU) of the .Child Development and Mental RetardationCenter, located in the University Affiliated Facility at the Universityof Washington: Karen Morris _is the principal of the EEU, whichcurrently serves severely handicapped pupils from seven

4

surrounding school districts. Pupils are:placed at the EEU whenadministrators, parents; and teachers determine they would_tebetter served at the EEU rather than in their local programs. All ofthe pupils live with parents; guardians, or in group homes awayfrom school. The EEU is in session all twelve months of the year,which will permit UWRO to conduct research during the summer.

Evaludtion

Eath of the objectives is a necessary step in Achieving our mission:Evaluation of our progress in meeting the objeaRies is an_ongoin_gprocess and constitutes Task 3 of the Institute: Three general clas-sifications of data will be collected for overall Institute evaluation:intervention, formative, and summative:

intervention data, which will serve as the primary basis for eValu-tion: are measures of the adualperformance of the subjects duringinstruction and in nOfitraining settings; collected 'during theresearch activities. UWRO is conducting a wide range of carefullydeSigned studies in an attempt to identify and develop procedureswhich will help severely handicapped persons to generalize and ,effectively use the skills they learn. The data collected on theperformance of severely handicapped pupils during the researchstudies will be analyzed by a variety of procedures, includin_g_,yisualinspection Of graphed data, trend analyses, and time series anal,yses for repeated measurement of single subjects._ Analyses _ofgrdup data will Utilize correlational and standard tests of statisticalsignificance. Standard analytic practices will determine if an inter-vention has an effect on subject performance; and the extent ofsuch effects; Data will also be compared to the performance ofStudents participating in the long term study'of educational environ-ments, described in the first section of Chapter 3.

Many studies will begin in very special settings where the greatestcontrol over conditions can be exerted. It should be noted,however, that all research Studies include specific time lines formoving into aped settings 7- schools, homes, and -the community-- and eValUating`the impact of findings in the "real world:" Eachline of study is designed to culminate in a_material product, such asa manual or Set of materials, which describes exactly how theparent; teacher, or. other practioner can use UWRO's findings tofadilitato Skill generalization: Since things which are possible arenot necessarily easy or efficient, cost stbdies will be undertaken toeValuate the time, energy; and resources required for implementingthe procedures recommended by URWO. If necessary, recow-

5 1

mended procedures will be modified and retested to make themmore easily understood and implemented within the typical appliedsetting.

Some research designs disre_gard the fact that statistically_ signifi-cant effects obtained in controlled_studies may not have any prac-tical value in normal situations. Therefore, the -evaluation Of theUWRO will rest very heavily on demonstrating that its work is actu-ally having an impact on the lives of severely handicapped persons,not in special laboratories or experimental programs, but in theirregular classrooms, homes, and places of work.

The purpose of formative evaluation is to demonstrate the extentthat research and communication activities contribute to thedevelopment of UWRO's. research and attainment of our overallmission. Formative evaluation will incorporate data and descrip-tions documenting existing research and demonstrationprocedures reviewed. applications and adaptations of existingprocedures; and research activities utilizing existing procedures.Products which result from UWRO's contacts with other institutesand agencies will be reported; Any products_ disseminated as aresult of either individual stpdies within UWRO or-through contactand collaboration with researchers outside UWRO will .aiso bedocumented.

Summative evaluation contributes to assessment of the lastingimpact of UWRO's various activities. We will collect and analyzedata on (1) the extent to which UWRO's research and interventionprocedures and materials are adopted by locaOeducational agen-cies, (2)' evidence of the quality of research;ikills acquired byUWRO Research Assistants, (3) evidence of cost savings resultingfrom implementation of UWRO-developed procedures; (4) theadoption of UWRO-developed procedures in teacher preparationand inservice training courses; (5) changes in peer interactionsresulting from UWRO. research and intervention procedures and

'products; and (6) the overalt_attainment of UWRO's goals. _Thesedata will be disseminated to the other institutes and researchers inthe field on an annual basis; for their information and to stimulateinteraction with UWRO. This documentation also VI serve asevidence of any "ripple effects" produced by UWRO erts.

Finally; the overall impact of UWRO's efforts must be measured interms of how much generalization is facilitated. This will be evalu-ated by conducting extensive inservice training seminars for'teachers; parents; administrators; and Other training professionals.and determining, through folloW-up evaluation, the extent to Which

6

the participants adopted the procedures and whether there wasani demonstrable beneficial impact on the lives of their children orclients_ by increasing_the nature or extent of generalization from-Dre-UWRO" levels. That will represent the most meaningful evalu-?tion of UWRO's success.

Communication

Communication about ongoing studies; training; and disseminationof procedures and products derived from our research is an impor-tant component of the UWRO mission: Cooperation with _otherresearchers exploring issues related to skill generalization is thenecessary first step in maximizing the potential benefits of UWROactivities. We will be working closely with the three other. Institutesfor Research in Education of the Severely Handicapped.

Two of these institutes will focus on methods of facilitating integra-tion from restrictive to least restrictive educational environments.The University of Minnesota's "Consortium Institute for Educationof Severely Handicapped Children" directed by Luanne Voeltz; andSan Francisco State University s "California Research Institute on'Transition of Severely Handicapped Students to the Least Restric-tive Environment" directed by Wayne Sailor, will study.the waysand means of integrating severely handicapped students with theirnonhandicapped peers. Another institute will also study_generaliza-

: tion. "Extending Competent Performance: An Institute for the Studyof Generalization with Severely Handicapped Students" is underthe direction of Robert -Horner at the University of Oregon. Sincethe process of integration is likely to involve the necessity ofgeneralized responding in "new;" integrated environments, thework of each institute Will relate directly to the work of the others.

The four institutes are committed to maintaining active interactionwith one another. Researchers may assist each other by replicatingvarious procedures or interventions: Conclusions drawn at oneinstitute may be incorporated into designs for studies at other insti-tutes. We will also be able to share our failures -- importantinforma-tion that is seldom published -- to_prevent investigation of ineffec-tive procedures: MethodOlogicat problems and solutions can beshared, preventing duplication of mistakes. Under normal circum-stances; new data are seen by other researchers only after theyappear in a professional journal. Since the publication processoften takes as long as two years; relevant data May not be avail-able when needed. This problem will be circumvented by monthlycommunication and inter-institute meetingS. Data from the other

three institutes will affect the direction and content of our reseach,and stimulate creative approaches to our work.

second important communication activity is training. Thisactivityvilf commence with the training of Research Assistants bySenior Investigators. Efforts will be made to employ ResearchAssistants who are students enrolled in graduate programs inSpecial Education and related fields. These potential researchersand professionals will acquire training in those skills required toconduct different types of research in laboratory and appliedsettings. At the same time, they will acquire experience inpromoting generalization in educational settings. We expecttraining of Research Assistants to be conducted continuouslyduring the project.

The results of individual studies in generalization conducted byUWRO; as well as the results of studies conducted within otherresearch institutes, will be disseminated via courses taught bySenior Investigators who are also teaching facultyet the Universityof Washington. Consultations and workshops given by SeniorInvestigators at other agencies, colleges, and universities mayinclude results to date.

During the third project year training in instructional and curricularprocedures will involve personnel from Direct Service Consortiumschools. Training will be conducted by UWRO staff. The nature ofthe training will depend on 7he requirements of the local educationalagency and will focus directly on the application of procedures.innatural settings. 0_pportunities for training will be extended topersonnel from all local educational agencies in Washington duringthe fourth projectyear in cooperation with the Offide of the Wash-ington State Superintendent of Public Instruction.. As informationfrom the "guideline" studies is collected, training content will bemodified. During the fourth project year, it is expected that thetraining will emphasize the guidelines for each area and practicalmethods of integrating the approaches in educational settings.

Training is perhaps the most active communication process, but itwill math only a small percentage_of interested professionals. Inorder to increase the potential benefits of UWRO procedures. tech-nical information and the guidelines for practical application will bedisseminated through publications and direct mailings.

Technical information will include precise and detailed descriptionsof research methodology, analytic procedures, the relationshipbetween the research conducted by UWRO and the existing body

of research infOrmation; and _presentation and discussion of theresults of individual studies. Technical information will be in theform of individual articles prepared for journals, in proceedings fromthe inter-institute conferences, in annual "Review of the Literature"publications_, in Annual Reports_ from UWRO, in:the Final Report;and in the Research Monograph to be produced during the fourthand fifth project years. -This information will also be disseminatedthrough discussions with researchers at inter-institute meetings; atnational conferences, and at quarterly professional seminarsconducted at the University of Washington..

UWRO will produce several publications of !'best practices'_'_ guide-lines. These materials will be assembled for specifiC audiences,including teachers, teacher trainers, parents; supervisors; adminis-trators, curriculum specialists, and related professionals. A widevariety_ of persons interested in the research will_ receive this-prac- -

Qtical information, which will be disseminated via training, presenta-tions at conferences, and mailing of project products. Nationaldissemination targets will be identified, but persons interested inreceiving project information will_ be able to contact UWRO directlyand obtain any product at a small cost:

UWRO's activities will be of little Ultimate value if the results are not, available to those who need them. Communication' activities willinclude cooperation; training; and product dissemination.

Advisory Committee

To ensure that research Will have practical application to a widevariety of potential consumers and to provide advice from profes-,stonal perspectives, adminis s, parents, researchers andothers met during the for ation of the UWRO proposal. Now

, meeting as the Advisory Committee, they provide advice onpngoing activities and assist the project in maintaining a practicalapproach to the education of the severely handicapped. Themembers represent the full range of professional activities andservice deliVery systems in the State of Washington.

Local educational agencies are represented bx, _three individuals:Genevieve Fisher, Chairwoman of the Advisory Committee, Coor-dinator of Child Find and Staff Development for Tacoma SchoolDistrict No 10; Lorna Tripp Wilturner, Director of Special Educationfor Issaquah School District No. 411; and Bill Tilley, Director ofSpecial Education for Seattle School District No: t_ Intermediateeducational agencies are represented by Donald Whitney, Director

.

of Special Services for Educational Service District No. 121 whichserves 36 local education agencies: Judy Schrag is the AssistantSuperintendent for the Division of Special Services. ir) the Office ofthe Washington State Superintendent of Public Ins uction and willbe our liaison with the state educational agency. I Bauer repreBents the 49th Legislative District in the Washingto State Senateand sits on the Rples, Wayg and Means, Financial I titutions, andLocal Government Committees. Joseph Jenkins _is D actor of theExperimental Education Unit and a member of the fa ulty of theCollege of Education of the University of Washington. A notedresearcher, he will contribute information from the perspective of aresearcher_ and as a representative of an institution of higherlearning. Margo Thornley is Executive Director' of the Wiser Voca-tional Institute, which provides vocational evaluation and training toseverely handicapped individuals: She represents other serviceagencies on the Advisory- Committee. Kathleen Knowlan is astudent in Speech and Hearing Sciences at the University of Wash,ington. She has completed _a B.A. in Communication Disorders andplans to complete a graduate program in Clinical_ SpeechPatholo_gy, She is the parent of a handicapped child. Together,these individuals'will bring_ a wide background of experience, avariety of perspectives. and a incere interest in the education ofthe severely _ handicapped 'to assist the Washington ResearchOrganilation in meeting its goals.

Administration and Management

Administrative activities support the_ research evaluation; andcommunication tasks of the Institute. General administrative tasksrelating to employment personnel_ management; purchasing;budgeting, and federal reporting requirement8 are covered by thiStask,

While such administrative_ tasks are conducted in every organize-tiOn, the structure of UWRO is designed to facilitate our uniqueactivities. Rather _than an hierarchical system where responsibilityand information flows from "top to bottom," we have a circularstructure. Information flow is both circular, within the rings, andlinear, to and from each circle; In addition; most personnel willParticipate in more than one group, thus increasing the nonhier-archical structure of communication:

The overall responsibility for UWRO activities_rests with the Prin-cipal Investigator and the Project Coordinator, but decision making

10

Cl

is shared_by all groups. Individuals will make deCisions related toactivities fOr Whith they are responsible. Decisions for group andintergroup activities will be reached by consensus. The model forcommunication at UWRO is shown in the following chart

UWRO Organizational,Structure

11

TWOBASIC RESEARCH CONCEPTS

Skill Generalization

e

Sometimes we want generalization to occur and other times we donot. For example; if we are successful in decelerating or eliminatingmaladaptive behaviors during training, such as spitting and hitting,we want those behaviors to not occur in other environments._ Theaim of such programs is generalization of nonresponding. Sincespecial conditions and circumstances surround this kind of training,and since instances of maladaptive behaviors may actuallydecrease as skill and competence increase, very little of UWRO'sresearch will be _ concerned with the generalizaticrh of _"noresponse." Most of the time, as educators, we do want generaliza-tion to occur: If we train feting at home; we want to see_toileting atschool. Generally, the behatIors we train may be called "skillS" or"skilled behaviors." because they proifide the' student with thecompetencies needed (Or normal living; these will be the ones ofmost interest in our investigations.

Broadly speaking; skill generalization is appropriate responding inthe absence of programmed training procedures. Severely handi-capped individOals are taught specific es_ponses under specialconditions, involving' instructional techniques developed throughexperimentation. These techniquel involve variety of elements,including the events that immediately precede the response; suchas verbal directions (e.g., "Get dressed." "Put on your shoes. ").these antecedents may come to control the response and are thenk,alled discriminative stimuli for responding. The student responds'When discriminative_stimuli are present; and does not respondwhen they are not. Other discriminative stimuli may include specificmaterials_ or objects (T-shirt; shoes) used during jastruction, thesetting of the instruction (the desk; the room) and the trainersinvolved. Instructional techniques also involve events that followone or more responses; like praise or:candy ("Yes; that's the way toget dressed."), or feedback on incorrect responses ("No, that goeson your other foot."). These events are called' consequences.Consequences are usually arranged to follow the response; theiroccurence is contingent upon the response. Contingencies are

12

scheduled during training, and may vary from one consequence foreach response to one consequence for several responses.

Generalization is concerned with the performance of the response.outside of training settings. When the specific events that occurredduring training are not available. different stimuli may- serve toSignal the response. Outside the training setting; contingencies forresponding are different; consequences may or may not follow the ,

response: In analyzing why generalization does or does not occur;investigators have found it useful to examine separately each of theareas where differences exist: io stimuli; people; consequences,settings, and ever time.

When the student responds appropriately to untrained instances,objects, or cues, "generalization across stimuli" is said to occur.For example, instruction in putting on shoes may have includedonly loafers; if the student is able to put on a slipper, using the samemotor skills in her response; generalization across stimuli hasoccurred. In cases where _generalization does not occur, it hasbeen hypothesized that discrimination training has been sosuccessful that the student will respond only to stimuli that areidentical to the training stimuli. When the stimuli changer thestudent -recognizes- the change, and thus does not respond. If thestudent does respond to stimuli that are similar to the trainedstimuli, then generalization has occurred.

Ancither problem area in generalization appears to involve thetrainers. Often trained responses occur only in the _presence of thepeople who trained the response; even if the same antecedentsand consequences are involved. When the student respondsappropriately to people who did - not train him, "generalizationacross people" is said, to occur. For example, if the student hasbeen taught to say; "Hi; my name is Charles," and is able torespond to a stranger's introduction witb those appropriate words,generalization across people has occurred.

Many instructional situations, especially during skill acquisition,involve consequences for each response. One to one contingen-cies are unusual outside of acquisition programs. Also the conse-quences available during instruction, such as candy or hugs; maynot be as available after instruction ceases. When the studentresponds appropriately in the absence of the consequences avail-able in the training environment or to different contingencies ofcorsequation "generalization across- equences7 is said tooccur. For example. training proced es ma ave included candyfor eacn correct response; If the st dent responds apps priately

13

A

and continues to respond with only intermittent praise, generalize-tion across consequences has occurred:

"Generalization_ across settings"_is a broad descriptor which_ incor-porates each of the types described abOve and generally definesthe incredible variety of changes that occur when the student isexpected to respond in new settings. FOr example, training a pupilto identify buSes by number; to enter the bus; to pay for his fare,and to exit from a bus at his destination may all occur within a class-room setting. However, the__student must be able to apply thislearning to actual travel. If successful; generalization acrosssettings I,as occurred. The differences between the training settingand the actual use of city buses are so many_and so varied that thiscategory is used to describe the collective differences.

We include another category of generalization, "generalizationacross time."_ If the response continues to be performed appropri-ately after training ceases, generalization across time hasoccurred. This is also called "maintenance" or "retention," butsince training has ceased, the conditions have changed (i.e.,antecedents and consequences may be different or presentedirregularly), and thus a response that maintains may be appropri-ately described as generalization.

So far, generalization has been described as occurring when thetrained response is performed under untrained situations.However, the true purpose of teaching generalized responding is toprovide the individual with means of -adapting to new situations,solving problems, and living in different settings. Each responseshould be appropriate. "Hi, my name is Charles," may be saidperfectly in a new setting, but if it follows the stimulus; "Put on yourjacket," it is entirely Wrong.

If the true aim is getting along in new environments, then theresponse must also he moaified; or physically adapted, to fit thesetting. Many instances of generalization involve changes in thephysical actions that constitute the response. For example, puttingon a T-shirt with long sleeves requires slightly different physicalmovements. than putting on aoshort-steeved T-shirt. In other casesvery different physical responses will be required to achieve thesame effect as that achieved by performing the trained response. -

For example, training a student to put on a shoe ar hieves the effectof covering and_protecting the feet. Putting on a pair of rubber bootsachieves the same effect; but physically different responses areusually involved.

2,14

Other problems must also be SOIVed if the student is to respondsuccessfully in new environments. One method of solving proptemsin new environments is to combine two or more responses thatwere learned separately. For :example; the student may learn howto reach for.and grasp something an a shelf above his head: Inanother training situation. he may be taught how to Stand on achair. If he were to successfully use both of these skills to get tilsi

lunch_ from a high closet shelf without training or prompting, hewould have solved a typical problem situation that may occur when-ever he is in a.different environment. When decisions are required,a response adapted. or a problem solved,. generalization involvesmuch more than simple application of a learned skill; it involves-adaptation. UWRO investigators will study both types of general-ization: application and adaptation.

Instructional Programming for Generalization

recently: many people expected generalization to occur spon-taneously after training; a "passive" approach to instructing forgeneralization has been common. We know now that the "train andhope" method does not result in much generalised responding bythe majority of severely handicapped students. Trevor Stokes, ofthe University of Manitoba; and Don Baer; of the University ofKansas, publiShed a major analysis and summary of research ingeneralization in 1977: This - article; and the discussion it provoked,had a major impact on shaping SUba-ettiJent resefirch in generaliza-tion: They argued that it is-better-to view generalization as an activeprocess and to try to develop instructional methods-that ensure thatgeneralization does occur.

Stokes and Baer identified methods in addition to "Train and Hope"that had been reported in published research. in "sequential modifi-cation," the behavior is trained in one setting and then, if general-ization does not occur in the next setting, training is programmedfor that setting, and so On foe each setting. This is actually not avery practical solution to the problem by itself, since it would meanthat training would have to occur in every setting and each time theindividual moved to a new setting.

A similar technique, which Stokes and Baer fee' is more promising,requires including Many different types of similar antecedents intoine training situation. By "training sufficient exemplars," the indi-vidual is thought to-learn a general category of items or objects to

15

. -which to respond. For example; instead of teaching "putting on asweater!' with just long-sleeved-crew-necked sweaters, V-neckedsweaters, short - sleeved sweaters and so on are trained. With morevaried instructional antecedents, generalization to untrainedsweaters (e.g.; turtlenecks) may occur.

An extension of this technique was.classified by Stokes and Baeras. 'train loosely," in which many different antecedent events areintroduced' during training. For example, instead of prefacing eachtrial, "Put on your shoes," the student mayitiear, "Put it on," or "It'stime to go outside, shoes on," or even bVgiven the shoes withoutany verbal direction: The more specific "program common stimuli"technique would be to identify stimuli commonly found in differentenvironments and include those in the training setting.

In addition to problems associated with antecedent stimuli; it hasbeen hypothesized that generalized responding does nbt occur ormaintain because the consequences available in natural settingseither are not reinforcing to the individual or do not occur asfrequently as they did during training. Research data has alreadyshown us that if -a response that has been frequently reinforced isperformed under infrequent reinfcement, that response is likely todisappear --L% to be extinguished: The technique "use indiscrimin-able contingencies" involves gradually replacing trainin_g conse-quences and schedules with those available in natural settings. Inthis manner, naturally available consequences acquire reinforcingpowers through pairing with programmed consequences; beforetraining consequences are discontinued. Similarly, the schedule ofone consequence for each response commonly used- duringtraining is gradually replaced with-a schedule of intermittent conse-quences, so that the student is unable to discriminate when atesporiSe is likely to be reinforced and when it is not; This method isdesigned to ensure that generalized responding (Will occur andendure with the infrequent natural consequences available outsideof training settings.

.

Another method identified by Stokes and Baer is toradually intro-duce the individual to "natural maintaining contingencies;" This canbe achieved most easily by teaching behaviors that are functionalin nontraining Settings. For example, teaching appropriate eatingbehavior would introduce the child to the contingencies that occurnaturally for such behavior; such as compliments; access todifferent foods; opportunities to eat at restaurants, or outings withfamily and friends. The natural consequences would then reinforce"good eating behavior" after training ceased; The student is intro-duced to natural consequences by teaching her a response that will

216

be naturally reinforced in nbrmal settings.

Another technique that haS been used to "train to generalize"seems to be at odds with most established instructional methodsdirected at acquisition. In this method, consequation occurs only forgeneralized responding. In such situations, the learner would notbe reinforced for learning a new skill, but only for using it appropri-ately outside of training situations.

A final category of research involves processes that Stokes andBaer identified as "mediate generalization:" Teaching the individualnew methods of thinkin_g_ and acting or to use cognitive strategies isan example of teaching' "mediated generalization" skills; ratherthan directing programming at generalization Of specific .skilledresponses:

Each method shows some promise but to date no approach hasdemonstrated consistently good effects in controlled settings, andlittle research has been conducted in classrooms and homes withteachers and parents implementing the procedures: Our researchwill seek to extend and develop these and other approaches to theproblem. using the methodology discussed in the next section.

Methodology

Subjects and Settings

The subjects will be students attending _the Experimental EducationCUnit and SthoOIS of the Direct Service Consortium who meet local;

state, and, federal classifications as severely handicapped,profoundly handicapped, severely behaviorally disordered; autistic;childhood schizophrenic, deaf-blind, or multiply handicapped. InOtter to facilitate the identification of subjects while respecting theRights of Human Subjects Guidelines, these districts will writeletters to parents of students explaining the research. Parents orguardians will be given the opportunity to voluntarily consent totheir child's participation in a specific study. Teachers of studentsfor whom consent is obtained may also consent to participate inresearch studies.

Since we are investigating skill generalization, measurement ofgeneralization will occur in a wide variety of natural educationalsettings, including classroom, school, home, community, and voca7 .

tional environments: Some studies will involve subjects working

17

directly with a researcher in _a separate room or in a part of thetraining or classroom setting. Results_of research in such controlledsettings will be applied to more normal settings. When this occurs,the setting is called an "applied setting:" Other studies will involvestudents working individually or in a group with their regular class-room teacher: Studies in nonschool environments will involveparents, supervisors, peers, neighbors, or others who interact withthe subject during normal daily routines.

Subject Responses

In each study. the performance of the student will be measured.Such measurements are used to determine the effects of differenttypes of training, the effects of changing trainers, and the effects ofchanging settings. Performance data will provide the informationwe need to better understand the phenomenon of generalizationand practical methods of achieving it for many different individualsand many different skills:

The selection of skills or 'behaviors to be measured will be deter-mined by methodological factors, but will also involve educationalconsiderations _ of _ functionality and age-appropriateness.Increasing emphasis is beingplaced on teaching_ severely handi-capped individuals functional and age-apppropriate skills: It may beappropriate to teach playing with blocks to a preschool child as aleisure_ skill; but bowling is a far more aae-appropriate leisure skillfor a teenager. This concept also extends to the selection ofinstructional materials. While beads and blocks may be appropriatematerials to teach a youngster to discriminate objects by shape,spoons and forks are more appropriate for teaching the same skillto a teenager. Furthermore, rather than teaching skills with limiteduse in most daily environments (i.e., making holiday ornaments);teachers are now concentrating instruction in areas more relevantto daily living and vocational success and ones which introduce thepupil to natural maintaining contingencies:

Some of our research will involve collecting data on behaviorstargeted on IEPs, but other factors most be considered. Forexample, if there are differences between skills or skill dusters,only certain skills will be selected. For examplei generalizationmight better be achieved fur the group _of dressing skills if trainingincluded practice with a wide vanety of different types of clothingitems_ (e.g., sweaters trained: cardigans, pullovers; zipbacksweaters, v-necked sweaters, etc.), than with repeated practice onitems of a single type (crew-necked sweaters). Conversely,

182 4

generalization may be hindered for the class of grooming behaviorsby training in a wide variety of items; but facilitated by teaching thestudent to check his own appearance. 'In some studies, therefore,grooming might be selected, while in others, dressing: As informa-tion about_g_eneralization accurriulates, factors such as these mayinfluence the selection of student responses for study:

In the selection of subject responses it is necessary, especially inthe early stages of research, to make sure that any observedchanges are the result of the intervention being_ tested._ Theresearcher may need to ensure that experimental instruction is the

{ only training affecting the performance of _the subject. This control`svery difficult to achieve if common functional skills are selected.How can the researcher wttio ._selects dressing skills be sure thatinstruction is not being conducted in the home orschool, even inti-dentally? In order to eliminate such effects, tasks which are rele-vant only to the study may be_ selected: Effective strategies iden-tiled by studies measuring artificial exQerimental responses will beapplied to functional tasks later on:

Measurement of Generalization

The basic concerns of the collection of data on generalizationinclude, in addition to standard research concerns of reliability andvalidity, the scheduling of generalization "probes" or measures, thefrequency of measurement, and the quality of the generalizedresponse. In most published research, generalization is measuredby one or mote "probes" or "tests" following the conclUsion oftraining or after the subject has met a predefined criterion perfor-mance level on the trained skill. These data can provide us withevidence that generalization did or did not occur. However; ifmeasured "after" only,.we really can't determine when generaliza-tion began or compare performance with preintervention levels.

Measuring generalization both_ before and after provides informa7tion din the net impact of the training, but leaves other importantquestions unanswered. Does generalization begin to occur gradu-ally, paralleling acquisition of the skill; or only as some level ofmastery is reached? Do different methods of teachinggeneralizedresponding promote generalization earlier than others? Doesgeneralizat ,tcur soon after training begins or only toward theend of training th such information; we can begin to understandthe relationship between skill acquisition, fluency-building, applica-tion. and adaptation., These questions can only be answered bymeasuring generalization during training and repeatedly over time,

2L,19

as PPR° Will do. Repeated measures or opportunities to performthe generalized response will provide information on the progressOf generalization as an ongoing; active process, rather than as ansingle spontaneous event. Not only will data be colleCted atdifferent times in relation to an intervention, but multiple probes willbe scheduled at each time.

IRepeated measures of generalization will also provide 'informationon another aspect of generalization; one that has received_ littleattention -- "training savings." An individual who has been success-fully instructed in one skill may. learn- another skill very quickly as aresult of the previous instruction. For example; a student maycomplete all of the steps required to boil an egg accurately (withoutbreakage or overcooking) and fluerttly (in the time it takes anaverage adult to boil an egg) in seven training sessions of 15Minutes each. Following egg training; thestudent may need onlyone session to master broccoli cookery. This May bq comparedwith another student who was taught to boil broccoli without eggtraining, and who took eight sessions to achieVe the level of

.mastery the first student achieved in one. This "savings" of _timespent in instruction is another important dimension in buildinggeneralization skills and is of practical significante to educators.

Measures of generalization often include only "yes/no" data onwhether g_enerali±atibh occurs ands a statement of the accuracy ofperformance (e.g., 80% correct), These factdrs alone are_probablyinsufficient tbr a thbroUgh understanding of the quality of ,thegerralized respon'se, We. know that the "time's of the response,expressed as either rate, latency; or duration; is required inadditionto accuracy data in order to understand how severely handicappedindividuals acquire and build fluency in skills. The length ofwaiting" time before responding (latency), the rate of responding,

and the duration of the response itself each provide important infor-mation on the quality of the response.

A good example of the importance of the temporal quality ofgeneralization is dressing: If a child is taught -to dress herself accu:rately and to finish within 10 minutes duringtraining; it is importantto know not only Aetheror not dressing occurs at home lyes/nodata), how many items of clothing are put on correctly (accuracy,data); but also,how long it takes her- luency data). Presumably theten - minute training criterion is sego allow :the. child to completedressing within a time limit that is functional for her hbme environ-ment (e.g.. a morning schedule which does not allow formore than10 minutes dressing). Thus, if training in dressing prciduces accu-rate an speedy dressing- in the hone; the training may be

.

20

regarded as entirely successful. However, what is the quality ofgeneralization if the child dresses accurately, but takes,35 minutesto do so? Obviously, the significance of the generalizatiOn achievedis less than in Cie former case. Thirty-five minute dressing mayeven have more serious consequences for behavior maintenance.The parents. anxious to see the child dressed and breakfastedbefore the school bus arrives; may decide to "help" the child dressor even dress her themselves. Over time, the opportunity to dressis withdrawn and we would expect that the skill of dressing mayeven be lost. In order to measure all of the important dimensions ingeneralized responding; UWRO researchers will collect_ yes/no,accuracy, and flUency data as measures of skill generalization.

Procedures for Descriptive Studies

UWRO's research activities will begin with studies designed to__ provide additional information about variables already identified,

such as the stimuli, contingencies, consequences, settings, andconditions in environments where generalized responding is desir-able. DescOtive studies will also include examination of other vari-ables that may affect skill generalization; such as the scheduling ofinstruction or the learning characteristics of the individual.

The collection of descriptive data will involve three different types ofanalyses. A ilyses of data collected previously may be used togenerate hypotheses, since it is unlikely that the experimenter'sbias could affect the data. Similarly; analyses of published researchusing statistical summaries across studies and discriminate anal-ysis techniques may provide additional information. Descriptivestudies will also include data collection in educational settings,without any intervention.

Procedures for Intervention Studies

UWRO's intervention _studies include both controlled laboratorystudies and investigations in applied settings. The interventionresearch_ will utilize two distinct methodological approaches ininvestigating generalization in severely handicapped individuals:"single subj_ect" and "'group" designs In each methodology, ourinterest_ is in determining the effects of .various interventions ongeneralization of the subjects involved in the study.

Single subject research designs irnlude repeated measurement ofthe target behavior, and thus provide information on the process of

21

2,

change of tne behavior. Data are collected on the target behaviorover a period of time before an intervention is introduced: Theeffect of the intervention is determined by comparing performancebefbre, dUhrig, and after the intervention.lhe relative strength ofan intervention_ is tested__ by _withdrawing the intervention andanalyzing any Changes: If the intervention cannot_be withdrawn,. aswhen an intervention has taught a new way of responding, theintervention is implemented with other behaviors and with -othersubjects. The data collected on each subject are studied inclividu-

.ally and analyzed to determine the process of change involved:Replicating the stuc!,:-S will provide information on the generality ofthe results:

In group designs: subjects are selected to -be representative of alatge population and then randomly divided into two or moregroups. Sometimes a ;measure_ of performance Of the targetbehavior is used as a pre-measure or pre-test: One group ischosen as the control group and another as the intervention group.There may be several different types of interventions tested, butusually only one per experimental group. Following the interven-tion, a post-test or measure-of performance is taken. The effects ofthe intervention are determined by comparing the performalce ofthe experimental group with that of the control group. The data oneach group are studied as a single unit to determine the product ornet effect of the intervention. The performanceof a single individualis important only as an indication of individual differences within thegroup. Inferences and results obtained by studing a group may leadto ihforMatiOri about how procedures may be likely to affect thepop, 'ation from which the group was originally draWn.

THREEUWRO'S APPROACHES TO

GENERALIZATION

UWRO's research will involve four different but interrelated lines ofinquiry to approach the fundamental questions about generaliza-tion: Why do some students generalize and others not? What canwe educators do to see that all students are able to generalize?

9

22

These approaches are distinguished by their basic assumptionsand by the types of the intervention strategies investigated. Thefour approaches to these questions are:

(1) An "ecological" approach to describing and thenchanging conditions within the educational environ-ment.

(2) A "performance pattern" approach to describing andmatching individual learning characteristics andinstructional techniques.

(3) A "self-control" approach to teaching severely handi-capped individuals to manage their own behavior:

(4) A "cognitive strategy" approach that cbricentraes onteaching severely handicapped students to use 'higherorder levels of thinking:

The time line for the research activities of the Institute proceedsgenerally from descriptive studies and tightly controlled laboratoryintervention studies to intervention studies in natural environments.The longitudinal descriptive study of existing conditions in trainingand nontraining settings will continue throughout the project. Thehypotheses of the studies designed to intervene in existing ecolog-ical conditions are, of all of the areas, most firmly rooted in existingresearch. Therefore, intervention studies will begin initially inapplied settings (e.g., public school classrooms).

The performance pattern research will begin with descriptivestudies involving analyses of existing data sets; and then proceedto the collection of descriptive data in public school classrooms.The descriptive information will be used to determine a set ofexperimental decision rules for matching specific instructionalmethods to individual performance which will be tested in interven-tion studies in applied settings during the third and fourth projectyears.

Without existing data sets or even very much applicable literature,studies in self-control and cognitive strategies will begin with tightlycontrolled intervention studies under laboratory conditions. Eachcognitive strategy and each self-control skill will be investigated inthe laboratory before intervention research begins in appliedsettings. Studies in cognitive strategies will also include investiga-tion of methods of teaching cognitive strategies in applied settings:

During the fourth year of the planned research; guidelines for theapplication of each method of facilitating skill generalization will betested in natural settings. According to the contract plan, the

223

research activities of the Institute will be concluded by the first partof the fifth project year. Our final activities will emphasize dissemi-nation of t4 research findings. The background, design; andexpected results of each of these approaches will be discussed inthis chapter.

RESEARCH TIME LINES--,

Ecology of Performance Self-Control CognitiveTraining Pattern Studies StrategiesSettings Studies Studies

Year 1 Deadriptive Descriptive Intervention82-83 Environment X Eteskno (Laboratory)

Ind,/ Chwactet Data Sets SollMonitoringisticst Group 1)

intervention(Applied)Massed y Dist.

ributed Trials

Year .2 Descriptive Descriptive Intervention Intervention83-84 Env X Charac Existing (Applied) (Laboratory)

(Croup 1 & 21 Instruction Self-Monitoring Skill ri e co m .

X _Pen_ binationIntervention Patterns Intervention(Appsed) (Laboratory)Competing Seli.Rnin- Intervention

Behaviors, lorcement (Applied)Instructional Sell-Inst- Teach

Schedules 'ruction CognitiveEnvironmental Strategies

Simulation

Year 3 Descriptive Descriptive Intervention Intervention84-85 Env, X Charac Exiating inst X (Applied) (taboratory)

(Groups 1. 2. Pert. Patterns Self -Rein- Self-& 3) lorcement Evahation

intervention Self-MonitoringIntervention (40.0) Intervention(Applied) Decision Intervention (Applied)Instructional Rules (Laboratory) Teach _ _

Schedules_ . Sell- CognitiveEnvironmental Instruction Strategies

Simulation CombinedStrategies

Year 4 Descriptive Intervention Intervention InterventkM85-86 Env. X Charac (Applied) (Applied) (Laboratory)

( Groups 1, 2, Decision Sell- Other3.8 4) Rules Instruction Cognitive

Guidelines Corribitied__ . Strategies_ _Intervention Strategies ilintenrention(Applied) Guidelines (Applied)Environmental Teach -

Simulation CognitiveGuidelines Strategies

Guidelines

24

Studies in Ecological Variables

Background and Purpose

While educators frequently lament the failure of pupils with severelearning handicaps to gerieraiize; research has suggested a varietyof instructional strategies which could potentially be applied ineducational programs to increase the probability of obtaininggeneralization: Broad categories of modifying instruction forgeneralization were discussed in the preceding chapter. However,the degree to which these praCtices have been incorporated intoeducation is unknown, as are factors in educational settings thatmay limit the effectiveness of these strategies..We will use the word"ecOlogy" to refer to the total of all observable factors and condi-tions which_comprise the educational setting._ The purpose ofresearch in this area is to explore the current educational ecologyof severely handicapped pupils and selected ways in which ecolog-ical conditions might be modified to enhance generalization.

Ecology studies will begin with a four year descriptive study: Thelongitudinal descriptive study will serve two purposes. The datacollected will be used as a general baseline for all of UWROresearch, to determine the extent to which our procedures areadopted; as a basis for cost comparisons, and as a general _indica-ton of the level of generalized responding with and without UWROprocedures, However, the primary purpose is toexplorea variety ofthe factors in educational settings that may influence generaliza-tion. Factors so identified will be investigated in a series of interven-tion studies.

Design of Longitudinal Descriptive Study

This four year study will explore five major issues:

(1) The nature and number of pupil goals and objectivesthat include the intent to promote g_eneralization orinclude behaviors which require generalization in orderto be of functional value.

(2) The extent that Pupil performance data indicate attain-ment of goals/objectives related to generalization.

(3) The degree to which formal instructional programmingis designed to facilitate generalization, the nature ofprogramming provided, and the extent to whichinformal practices are employed which seem likely topromote generalizalion.

3,_

25

(4) The degree to which generalization occurs as a resultof formal or informal programming.

(5) General ecological conditions which might facilitategeneralization.

Ecological conditions which may be examined include-the numberof managers adminiStering fOrmal or informal programs throughout'the day: the ,number of intra - school environments_ in whichprogramming occurs, the perCf.int of the school day in communityenvironments; the degree of interaction or opportunity for interac=tion with non- handicapped or with lesser handicapped peers; andthe number of school-community cooperative programs adminiS-Wed iollowing school hours (e.g:. programs administered, by-parents). Other conditions may be identified through observationand factor analysis.

The descriptive study will include two _types of activities. _First, areview of existing recbitit (e.g., IEPs, formative date, and lessonplans) for the original sample_ of approximately 25_severely haridi=_capped pupiliS Selected frOM Direct Service Consortium schools willbe undertaken to collect information on- objectives and educatiOnalplans. Sedbrid, interviews with teachers; parents, andr other care-givers, as well as data Collection_ in_classrooms_ anq other envirbri-Merit§ by members of the project staff; will be necessary to gatherinformation related to current conditions, implementation ofprocedUres, and pupil performance:

Selected members of the original pupil sample will_be followed upeach year for three additional years in order that a longitudinalrecord may be developed of generalized skills as they are acquired.This study will elk) record the changing nature of conditions towhich pupils are exposed over time. In addition, new pupils will beselected and added to the sample each year with similar informa-tion collected on each.

Analytic techniques applied to the data will be primarily descriptiveand exploratory in nature: Ultimately; the results will be of value indetermining which currently employed instructional strategies havea high probability of success; whether certain skills are more likelythan others to generalize in the absence of formal programming forgeneralization; and the basic conditions within community, hortie,and work environments which should be considered whendesigning; implementing; and evaluating programs to facilitategeneralization for severely handicapped persons The datacollected will be used in the selection of ecological intervention

26Yr-

studies and in the selection of variables and methodologies forother UWRO investigations.

Design of Intervention Studies

The second set of ecological investigations will study interactionsbetween factors identified in the descriptive study and the effect ofchanging one or more of the existing conditions on the occurrence,quality, and quantity of generalized responding. Factors that arelikely to be studied include pupil response variables, general taskand setting variables, stimulus variables; response demands; andreinforcement variables as they pertain to the development ofgeneralization:

Although it is impossible to predict alt factors that will be investi-gated, previous stOdies have identified likely areas. Past researchleads us to believe that when generalization does not occur, theindividual may have previously acquired a behavior that achievesthe same outcome as the behavior being trained. This behaviormay compete with the trained behavior in the critical effect of aresponse. For example, tantrumming to be fed may compete withfeeding,oneself or asking for food; if each gets the-same results. Atleast one study will be conducted to investig_ate methods of identi-tying and managing undesirable competing behaviors andassessing_the impact of the interventions on generalization of moredesirable responses.

Another intervention study wilt examine the relationship of general-ization to the scheduling of instructional trials. Instructional opportu-nities or trials are frequently grouped into a single block_, with onetrial immediately following the next. For example, 10 or 15 trials of"buttoning" instruction might be presented daily from 10:00 to10:30 a.m. An alternative to this practice would be to provideinstruction in buttoning at times when a natural need exiets tobutton one's clothing (e.g:, upon getting up in the marling, beforegoing outside, after_gym, or after usingthe toilet), This method forscheduling instruction would result in the provision of trials spacedor naturally distributed throughout the day.

Schedaing instruction at the times when the target behavior wouldoccur in natural environments could prove beneficial for severalreasons. (1) It would Increase the similarity between the instruc-tional situation and the conditions in which generalized baspondingis deSired. (2) It might increase the likelihood that unprogrammedreinforcers would be available in the natural, generalized setting.

327

(3) It may avcid problems often noted_ with the severely handi-capped, like "poor attention spans," fatigue, and reinforcer satia-tion; This series of investigations will provide data on the relativeimpact of different trial scheduling formats on generalization:

A third potential area for ecological intervention studies involvessimulating "natural" environments in instructional settings: Since itis not aWays possible or practical to teach in natural environments,systematic assessment of the differences between "artificial" and-natural" settings should allow us to identify those variables thatactually affect whether or not a generalized _response_ occurs.These critical factors will then be systematically introduced into the"artificial- training situations; producing an arrangement that maybe similar to the "natural maintaining contingencies," "commonstimuli," and "indiscriminable contingencies" methods discussedby Stokes and Baer. For example, since the natural environmentfor instruction in washing clothes would be a laundry room ratherthan a classroom, perhaps an instructional program including theuse of coin-ope.iated washing machine and dryer in the trainingsetting willproduce generalization to all laundromats: Generaliza-tion obtained during and atter training in the "artificial," "modified-artificial," and "natural" environments can then be compared:

The methodology of the intervention studies will be single subjectdesigns replicated across subjects. Repeated measurements ofthe accuracy and either the rate; duration, or latency of perfori-mance of the subjects durin_g_the instructional sessions and in thegeneralitation setting will serve as the primary independent vari-ables. In cases where training occurs in the "natural environment;"generalization will be measured in different but similar settings.Analytic techniques will include visual inspection of graphe.q data;time-series statistical tecnnigues; and an overall statisticalsummarof performance for comparison between studies and withthe data collected in the longitudinal descriptive study.

Expected Outcomes and Freducts

The studies in this area should result in the development of a "bestpractices" manual, incorporating at least five areas: .

(1) Guidelines which identify current best- practicesexisting in public_ school settings will result from thedescriptive studies.

(2) , Guidelines for identifying competing behaviors andmethods for counteracting their effects in nontrainingenvironments.

(3) Guidelines for how to schedule instructional trials andlearning opportunities for different classes of skills-.

(4) Guidelines for how to (a) identify critical elements in the"natural" environment and (b) introduce thoseelements in the training environment.

(5) Other possible guidelines may be developeddepending on information from the descriptive studiesand the nature and results of the intervention studies.

Studies in Performance Patterns

Background and Purpose

Most people agree that each pupil is an individual and that whatmight work with one student may not work with another. There is a

-need to individualize not only in the selection of behaviors to teach,but in how we teach. Individualization usually begins with an identi-fication of the skill areas and behaviors to be taught; Next; detailedinventories of the pupil's skill in each area of the curriculum areconducted to determine; for each behavior selected, the_ exact levelor curricular step at which instruction should begin. Major pupilcharacteristics which might indicate the need for a particularinstructional approach are also identified, The teaching procedureswhich Might prove most effective with mentally retarded children;for example, might be quite different from those which work bestwith the deaf-blind. Surveys of "learning channels" and "reinforce-ment preferences" could also be used to help in the development ofspecific instructional plans. Overall, there is much that can be doneto select and develop highly igdividualized approaches for meetinga pupil's needs.

For the most part, however, educators tend to think of the factorswhich might determine __the effectiveness of_ an instructionalapproach as being rather fixed and unchanging. The student is andalways will be deaf-blind; the student is "visually oriented," or"prefers juice-instead of hugs." In reality, instructional approacheswhich work quite well on one day may actually hinder furtherlearning on the next. Truly individualized instruction will involve the

'continuous assessment of daily_pu_pil progress to determine exactly .

when and how instructional procedures should be modified to keep _

pace with the changing needs of the pupil.

Fortunately, research over the -ast decade has identified patternsin the way pupils' learning changes from day to day. Each patterncan be related to specific instructional needs. For example, there is

3-29

1-

a surprisingly consistent' relationship between a _pupil's overallfluency in performing a task and the need for additional guidance. Ifa pupil is performing a task very slowly (even if overall accuracy isfairly good), strategies such as increased cues; prompts; and direc-tive feedback may facilitate futher progress. However, if the pupil isperforming the task fairly quickly, those same strategies may bequite ineffective or even hinder further progress. After noticing theineffectiveness of one strategy, teachers may need to try three orfour different approaottes before finding one that works. Of course,soon after finding one that works, the pupil's needs change oncemore and the process of trying to find effective instruction begins allover again.

By examining the performance characteristics of students whowere acquiring or building fluency in a skill, researchers _foundcertain . elements of performance to be very important. TheSeincluded the student's rate or frequency of performance under openconditions. or the latency in responding to a cue or the duration ofthe response, in addition to the accuracy of the response, theweekly trend or direction of change in the response, and the varia-bility of responding. Five constellations of these elements wereidentified as specific performance patterns. By examining thesecharacteristics, researchers were able to predict whether or not aspecifi&strategy would help or hinder the student's learning at thattime. To replace the guesswork in programs for acquisition andfluency-building, rules to help teachers match instructional strate-gies to pupil needs as performance changes were developed.Research shows that teachers who follow the rules are able tochoose an effective strategy ten times out of twelve.

As successful as the performance pattern rule research has been,to date it has only looked closely at the way in which pupils learnand master new skills in specific instructional situations. Very littleis known about the relationship between those patterns of learningand the chances that the new skill will generalize to other situa-tions. It will be the purpose of the performance pattern research atUWRO to investigate those same elements to discover their rela-tionship to generalization. We will then try to match specificpatterns with instances and noninstances of generalization. Ifnecessary, we may look at other elements in responding, but wehope that the same elements will prove predictive of generalization.We will attempt to identify the instructional procedures with thehighest _ probability of promoting generalization. If we aresuccessful, we will be able to match particular types of instructionwith student's individual needs in order to facilitate generalization.

3

30

Design of Descriptive Studies

A great deal of potentially useful information concerning the rela-tionship between patterns of learning and generalization alreadyexists. For example, a_ published research_ study; originallyconducted to determine the usefulness of feedback in promotinggeneralization, might be _evaluated to look at the relationshipbetween performance patterns and generalization. Similarly, thedata already being collected in many claSsrooms to monitor pupilprogress may yield certain clues: There are at least two Idvari-tages in using existing data it is far less expensive and there is nochance that our expectancies of what should happen mightsomehOw affect what does happen. The disadvantages in usingexisting data lie in the fact that they may not provide all of the infor-mation required fOr the study (*g., a researcher ma_y haveexpressed performances in terms of simple percentages; or accu-racy statements, rather than in both accuracy and fluency asdesired for the current research); and _some question might _existconcerning the reliability of the data. Once the existing data havebeen used to refine specific hypotheses; we will begin to collect ourown data:

Experience in special education research has shown that valuabledata can be obtained from scientific observation of what is alreadyhappening in the classroom; before making any changes: Duringthis phase of the research, the project will simply monttcr and docu-ment what is already going on and_ how those activities appear torelate to generalization. During the second project year, descriptivestudies will be conducted within the claSsrooms of the DirectService Consortium. SIverely handicapped )upils with a widerange of disabilities wile included in the stuo j. in previous perfor,mance pattern research on skill acquisition and fluency, basic pupilcharacteristicsii.e., type and level of handicap, age, sex, etc.) werenot related to the way in which performance patterns predicted thesuccess 'of various instructional approaches. Nevertheless,detailed records of pupil characteristics will be kept and evaluatedto determine whether those charaCtenstics do relate to the useful-ness of performance pattern rules for predicting when and howgeneralization might be facilitated.

Teachers volunteering for the studylwill be asked to collect (or toallow project staff to collect) specific information concerning dailypupil progress in a sampling of instructional programs. Concur-rently, project staff will monitor each pupil in a variety of other situa-tions to determine if, when, and how the pupil begins to demon-. Strate new SkillS outside of the instructional setting.

3,31

Design of Intervention Studies

During the third and fourth project _years specific stUdies will be'conducted to clarify the relationship between performance patternsin an instructional situation and the likelihood of generalization:. Forexample, if the noninterventional studies conducted during thesecond year suggest that pupils who achieve- a specific level offluency in tho instructional setting are more likely to generalize theirskills, then a study during the third year migfft test that relationshipby banging new pupils up to that level of fluency and noting whethergeneralization does actually occur. When a reasonably compre-hensive set 'of rule.: has been developed, the impact of those rulesWill be testbd by training new teachers in their use and evaluatingtheftect of role use on generalization.

During the earlier performance pattern studir:::, he success of aprogram change was judged by the immediate impact on performance. the change produced in average weekly _progress; and thenet effect of those two factors on eventual skill mastery: ThoseSame variables_ will be used to monitor the basic effectiveness ofany changes made in the instructional situation to improve general-ization, but special probes of the pupil's behavior in a variety f-

other situations wit alSO have to be conducted to examine generalization: Initially: the degree_ of generalization at any point in time willbe described in terms of the number and type of noninstructionalsituations in which_ the behavior is observed to occur, and thedegree to whiCh performance characteristids in the noninstructionalsetting approximate those observed in the instructional sittjatiOn (interms of fluency, accuracy; and improvement over time).

Expected Outcomes and Products

If the proposed studies are as successful as earlier work; it shouldbe possible to develop a set of rules which teachers can use toevaluate individual pupil performance and decide if; when; and howthey might change instructional procedures to facilitate generatiza-lion. Rather than impose a single approach to developing general-ization; the rules would help teachers to choose, from among avariety of possible instructional procedures, the best method tomeet the individual needs of a pupil at a_given point in time. Withsuch rules, it will be possible to truly individualize instruction to takeinto account each pupil's changing needs.

3,732

In addition to a series of research-papers and monoeraphs docu-menting the progress of individual studies; the performance patternresearch -should result in the creation of a brief "user's manual"which explains how the rules can be used_to facilitate skill_general-ization with severely /handicapped pupils. The manual will bewritten in a manner which is easily understandable to teachers andother educational practitioners and will be as self-contained aspossible. The manual will not assume that the reader has any priorknowledge of the skills necessary to use the rules. The actualusefulness of the manual will be tested on a group of teacherstoward the end of the fourth project year. The feedback gained fromthat trial implementation will be used to make modifications duringthe fifth and final project year:

Studies in Self:Control

Background and Purpose

Typical instructional procedures for skill acquisition and fluency-building rely almost exclusively on a teacher or other trainer actingas the _focal point. In almost every research and/or curriculumreport, the handicapped person is seen as the one whose behavioris to be changed; rather than the individual who is to change herown behavior. This emphasis is evident when you consider that inmost training programs:

(1) The behaviors to be changed are selected by others.(2) The training materials and procedures are selected by

others.(3) The training procedures are implemented by others:(4) Changes in behavior caused by training are monitored

by others.(5) Decisions about changes in training procedures are

made by others.Although this instruction has been effective in teaching specificskills, the collective effect of many years of such training may be_toteach the handicapped individual total dependence on others forcontrol in each situation: Generalized responding may fail to occursimply be-cause the individual is waiting for someone to give step-by-step instructions in what to do.

Self-control procedures offer an alternative. In self-control training,the individual is taught NA/ to use different techniques to direct herown behavior. It is easy to find examples of self-control techniques

33

in everyday activities. One common self-management procedure isself - monitoring; or counting the occurrence of one's own behavior.A person who says, "This is only my third cigarette today is moni-toring her own behavior: We've probably all_ heard someone say;"I'm getting fat, I'll skip desSert." Such individuals are not onlymonitoring their own behavior; but they are making a decisienbased on the information as well.

Another technique we use to manage ourselves is called self-instruction, directing the sequence of activities we are performingOr are about to perform. People facing several different tasks or aparticularly complicated task will often -audibly list, to themselves,the sequence of things they are going to do. For example;: "I'll startthe water for the noodles, then cut up the asparagus, then I'll putthe noodles in; next start the asparagus; and hope that they arefinished- cooking-at the same time." A third typical procedure is Self-reinforcement; including selecting and delivering consequences foractivities. For example, a person may reward himself with time toread the newspaper after he has washed the dishes.

While instances of these activities abound in our deity -lives, ntilrecently, little research in self-control has been reported: it is knownthat many people do not learn to use self-control skills withoutdirect training in the skills: Research does show that self - controlskills are usually just as effective as external-control proceduresinchanging behavior: Moreover; self-control may be better at facili-tating maintenance and generalization, since the individual learnsindependence; rather than dependence:

Can we teach Self-control skills to the severely handicapped? Onlya feW researchers have worked with handicapped individuals; sothis question has not yet been answered. We can develop empiricalstudies to determine if precise skills, such as pushing a button on acounter following task completion (i.e., self-monitoring), can belearned. We can also determine if other self-control activities helpthe person to change her own behavior and if they are effective inchanging other behaviors in new settings: A second puzzle forresearch concerns the nature of the training. If the methOds used toteach self-control skills rely on an external agent; will that methodcounteract the deyelo_pment of independent control? What othertypes of training can be used?

It is the purpose of research in this area to investigate whether ornot severely handicapped people can be taught to use methods ofself-control. If so, what are the. best methods of nd_if theself-control skills can be used by individuals to change their ownbehavior, dosuch skills improve generalization?

Design of Studies

The variables that will be investigated in these studies will include:(1) The accuracy and fluency of the performance of the

self- control skill.(2) The length of time required for acquisition and fluency

of the self-control skill.(3) The instructional procedures used to teach self-control.(4) The effect (Le., generalization) of the self-control skill

on other behaviors.

Although it is difficult to predict the course of future research, wewill attempt to investigate each of the three primary self-controlskills: self-monitoring, self-reinforcement, and self - instruction:

Three other self-control techniques will be integrated into thestudies: self-determination of behaviors for change, self-determina-tion of consequences, and self-determination of the ratio of behav-iors to consequences. Subjects will be able to selett consequencesand behaviors to change in most instances, so that they can imme-diately begin participating in the behavior change process. Individ-uals who do not respond to questions (e.g.; What would you liketowork for? What will you do if you make'a mistake? How many doyou want to earn?) will be_presented with a multiple choice situationvia pictures, objects; or words during each training session.

Each study will invoke the subjects in several different phases:' Aminimum of three different behavlori will be identified for eachsubject. One behavior, called the "training behavior;" will_ beselected because it is a skill the individual already does well. Twoother behaviors; the "target behaviors," will be the ones we hope tochange or affect by the self-control skill, and will be seletted by thesubject. Repeated measurement data will be collected on thetraining and target behaviors; in training and nontraining settings,and on the acquisition and generalization cf the self-control skillData will be summarized and analyzed according to accuracy andfluency of the response, and by changes in the individual's leveland direction or trend of performance:

Following the collection of baseline data, the subject will be taughtto use a setf-control skill. It is very important to select a trainingbehavior that the student has already learned, so that instructionand learning may be concentrated on the self-control procedure.For example, if we are investigating_self-monitoring we might teachthe student to count the number of sit-ups completed each gymperiod. if sit-ups are something the student can already do well.

35

Once self-monitoring training has started, the subject wilt have theopportunity to count the target behaviors. The target behaviors.might be talking out in class; finishing work on time, or a widevariety of other behaviors which the student selects to change. ifthe student does not begin to count the target behaviors; she will betaught to do so, btTvinning_with one of the target behaviors and thencontinuing with the others. The diagram below indicates thegeneral progression of activities. We will examine generalization ofthe self-control skill across behaviors. If generalization does occurwith some of the subjects, later studieS will involve the collection ofdata in different settings as well.

TrainingBe;tavier

TargetBehavior

TartletBehtActr

2

Design of Self-Controf StudieS

Baseline Opportunity Train:ngto aPOIY uvseltsell-control controlSkill skill

Maintenance

Baseline Opportunity Training Maintenance

Baseline Opportunity .Training Maintenande

Expected Outcor, les and Products

Since there are so few precedents for teaching- self-f:ontrol_ toseverely handicapped students, it is diffitUlt to predict thesequence and nature of the studies. Findings from one study willprobably change the direction and methOdS of subsequent inveSti-gations: We will begin _ with only a few subjects. If result3 areencouraging, let& stUdies will involve more subjects: If initialstudies ar,9_ successful._ classroom teachers will he taught how toinclude self-control skills in their curricula: It is Wu, possible thatparents and others may participate.

36

.

Ideally, this research approach will yield information on which sett-control skills can be taught and how to teach them to severelyhandicapped individualsWe hope that the product of UWRO's research in self-control will bea package of materials for trainers to use in teaching self-controlSkills to severely handicapped individuals in school, hoMe, commu-nity, and vocational settings. Whatever the results, -we expect thatresearch in self-control will produce information vital to our under-standing of and training for generalization.

Studies in Cognitive Strategies

Bvkground and Purpose

There are at least two important kirids-of generalization of learning.One kind of_generalization is to apply learned behavior in any othersituation.. Social skills provide good examples. For instance;regardless of the situation in which one encovters an acquaint-ance, one can inquire about his well-being with thV: same question;"How are you?" Once one has learned how to inquire afteranOther's well- being; it is important to generalize that samebehavior to different situations. One may_ ask, "How are you?"whether encountering an acquaintance on the street; upon arrivingat the acquaintance's home, or anywhere.

The other important' kind of generalization is to recognize when a -

specific response_ must be adapted to meet the - demands of a newsetting. Although there are many strategies that can be involved inskill adaptation; perhaps the simplest is combining two behaviorslearned separately to successfully solve a problem: The act ofcombining changet both of the originally learned behaviors. If theyare combined appropriately; they can allow sofution of novel prob-lems. Academic and cognitive skills_proVide food examples.Students can learn to count things; to measure distances, and tocombine numbers arithmetically in elementary school. Then, whenthey encounter physics problems in junior high school, they cancombine counting, measuring, and arithmetic to solve the prob-lems: At a simpler level; children can learn to recognizOhapes, toput things away, and to look for what they want. Then, when theyencounter a game in which different shaped blocks must be put intheir proper holeS in order to get them all in the same box; they canlook for blocks of different shapes one at a time and put them in theappropriate holes.

-7

37

4 07

It is important to know how to promote generalized skill application.so that severely handicapped _students will not need to_be _taughtevery behaVidt in every context in which it is to be used. It is equallyimportant to knew how to promote generalized" skill adaptation sothat they Will not need to be taught every conluintiiiv orbenaviorsfor every problEim to which it applies._ Although many theorists haveexpressed doubts that individuals with low levels of generral intelli-gence can lear the "higher" order thinking that seeme"to underlieSuch generaliz Hon; like Stokes and Baer, we suppose that all weneed are bette ways to teach "mediated generalization." That is,we need betted -ways to teach individuals new ways of organizingtheir thoughts bout what they are doing.

Part of being all effective and independent person is to recognizewhen to do what one knows. Another part iszto recognize when onecan tOmbihe things that are already known to solve problems onehas never sol ed before. No research has_been done to determinehOW to teat such cognitive strategies to severely handicappedpeople, alt ugh progress has been made in teaching normal andMildly retarded children how to combine behaviors they havealready,-learned. in order to perform effectively in novel situations.Such /instruction represents a cognitive approach; since itsernpffasis is on changing strategies of thought. By teaching infor-mation proCessing strategies which provide means of reorganizing'available knowledge, the student should be able to generalize hisor her behavior skills: The purpose of the cognitive strategy studiesat UWRO is to detentino Whether the principles and practicesdeveloped to produce mediated generalization by normal_ andmildly handicappea people can be used effectively with severelyhandicapped individuals.

Design of Studies

The initial studies will be tightly controlled laboratory investigations.If we are succeSSIU1 in laboratory settings, we will then work directlyWith teachers in developing means of translating the methodologyinto_practical procedures. Data will be colletted in classroom envi-ronments to determine if procedures retain their effectivenessoutside of laboratory conditions.

Since this line of investigation has not been applied to severelyhandicapped individuals before, our plan is to select vanables forinvestigation after we observe how potential subjects behave inpre-investigation observation and testing. Although we are unableto predict the difficulties or the variables which will be involved, it is

38 4

tike y that data will be collected on the cognitive capabilities ofstudents with or without training in information processing strate-gies, the degree to which such strategies can be taught, and theeffects of strategy training on the occurence of generalization.

We wiji probably use two groups of subjects, one group receivingtrainigg anclthe other not in our laboratory studies. Measurementof performance during training will include measures of the accu-racy of performance and the number of trials required to achievecriterion. Criterion statements will be used to define levels of perfor-mance to be reached before training is terminated. Each criterionstatement will include quantifiable measures of how accurately thebehavior is performed; as well as the latency or duration of theresponse.

Since it is most im_portant to discover if _generalization occurs,generalization will be tested after training is concluded and take theform of providing "prciblems" which the subjett can solve correctlyonly by using the cognitive strategies induced by the training;These measurements will include not only a simple statement as towhether generalization occurred, but how quickly the individual isable to "solve" the problem and how Much savings of instruction isachieved. Individual and group statistical treatments will be appliedto the data.

In the studies in applied settings, similar data will be collected bythe experimenter, but procedures developed for the classroomteachers will include methods of mi4asurement that are appropriatefor use by practitioners in evaluating the effects of instruction.

, The first step in our experimental approach will be to determinewhether severely handicapped people spontaneouSly combine twosimple behaviors. The procedure is to teach simple behaviors andthen present the subjects with a situation in which they can securea desired object only by combining two of the behaviors (i.e., by firstdoing one behavior to secure an item necessary to conduct thesecond behavior, which will secure the desired object); We expectthat many severely handicapped subjects will not spontaneouslycombine two behaviors. If they do not; we will move to the secondphase of our design.

The second phase is to teach a large group of distinct behaviorsand then to,-select pairs of behaviors which the subjects will betaught to combine to achieve desired Objects. The focus of thesestudies will be upon the ways of teaching "how to combine" so thatit will quickly generalize to the recombination of other behaviors.

39.

If the studies in teaching skill recombination are successful, we willnext try to teach the higher order strategy of self-evaluation. We willtry to define methods of teaching -the stuFlents to predict, beforethey actually respond in a novel situation; what the effects of theiractions are likely to be. Predicting and judging te effects of one'sown actions, and then making decisions about hick behavior willbe the best one, requires sophisticated patterns of thought. Suchpatterns are essential to the development of generalized adapta-tion. The design of these studies will proteed from an investigationof how 'severely handicapped individtials behave in situationsrequiring YUd_gement and prediction, to the implementation of care -

-fully designed teaching strategies and the investigation of. theireffects.

Expected Results and Outcomes

Since this area of UWRO research has few precedents our expec-tations must be. cautious. We hope that we will be able to identifycognitive strategies, be inning with skill recombination andproceeding to self - evaluation; that will teach severely handicappedindividuals how to generalize. If successful, we hope to be able toidentify practical methods that will allow teachers to provide cogni-tive instruction in regular school environments. We will then be ableto produce not only _a series of research reports and publications,but also a manual Or teachers to introduce "training to generalize"into the educational environments of severely handicapped individ-uals.

FOURPUTTING IT ALL TOGETHER

Guidelines for identification and manipulation of a wide' range ofconditions within educational settings will result from the studies inthe ecology of training settings, The performance pattern studieswill contribute a set of guidelines specifically for instructionalmethods that educators can use to ensure generalization. Guide-lines from the ecolOgy studies will be directed at fairly globalmanagement of the instructional setting, while decision rules fromthe performance pattern studies will be directed at the selection ofprecise instructional methods used in individual programs.

4

40

LtWRO Research Schematic

HOWOo -CONOMONS IN THEENVIRONMENT AFFECT SKILLGENERAUZATIONI

_ECOLOGICAL STUDIES

HOW_ TO CHANGE_TRAINING SETTINGS

DESCRIPTIVE STUDY_OfTRAINING ENVIRONMENTS

HOW TO-TEACHSELF-CONTROL

7.SELF-CONTROL STUDIES

CAN SEVERELY HANDICAPPED_INDIVIDUALS/E _TAUGHT TO USESELF-CONTROL SOUS TO

GENEMIZTIONI

INTEGRATg',BEST PRACTGUIDELINES FORPROFESSIONALS

HOW DO INDIVICRJALDIFFERENCES-IN -LEARNING -AND

IN TRAINING TECHNIQUESAFFECT SKILL GENERALIZATION?

/PERFORMANCE -PATTERN

STUDIES

HOW TC/ MATCHINSTRUCTION TO

STUDENTS

DATA FROM. OTHER INSTITUTES

HOW_TD TEACH _ _

COGNITIVE STRATEGIES

COGNITIVE STRATEGY. STUDIES

-CAN SEVERELY - HANDICAPPEDINDIVIDUALS SEIAUCHT_ TO _USE

COGNITIVE STRATEGIES -TOMPROSIESKILLGENERALIZATION?

During the fourth prcject_year, methods will be developed tocombine the guidelines from the ecology and performance patternstudies with other empirical data, into an integrated set of !bestpractices for generalization." Such guidelines would probablyestablish a decision hierarchy for use at administrative, trainingsetting, small group, and individual pupil levels. For example, asequence of decisions might Include:

(1) Determine what skills should be programmed forgeneralization.

(2) Determine the appropriate instructional settings (i.e.,home, school, or community) for each skill.

441

(3) Determine the characteristics of the setting in whichgeneralization is desired.

(4) Determine for each student the percentage of eachschool day to be spent in each setting, or how _to integ-rate factors from the generalized setting into thetraining setting.

(5) Determine if instructional trials will be massed ordittrIbuted.

(6) 'Determine the specific instructional procedures foreach student.

The guidelines that may result' from the studies in self-control andcognitive strategies will affect the curricula of training settings bysuggesting changes in the skills that are currently taught. Reborn-mendations, such as the inclusion of self-monitoring in the curric-ulum; will be accompanied by precise directions as to whom toteach such skills and how such skills might be most effectivelytaught It is expected that information on other curricular changesthat affect generalization, produced by the work of other institutes,would be used to produce a set of integrated guidelines for curric-ulum content. If all conditions are ideal, perhaps the guidelineS forcurricula will be integrated with the guidelines for intervention in thesetting and included as an aspett of the decision rules, prdducing afully integrated single set of practices.

At this time it is difficult to predict the nature_ of the various guide-lines to be developed or if the guidelines will fit together, since theymust be based on empirical evidence that the strategies do; in fact;promote generalized responding. The Institute will be able to drawon the expMise of the Advisory Committee and the Direct ServiceConsortium the development of guidelines: We will also haveaccess to resu.. horn the other institutes. All of the informationavailable will be integrated into the guidelines eventually produced:It is our hope that the four approaches will provide solutionsconverging into an integrated set of :guidelineS for users; Theschematic, shown on the title page that precedes Chapter 1, illus-trates how UWRO hopes to increase interaction and irltegration ofthe results as research proceeds, to the development of an inte-grated set of guidelines for practitioners.

The Washington Research Organization con bines four differentand complimentary conceptual approaches to the problem of skillg_eneralization. We believe thatp ,prsuit of these four lines of inquiryrepresents a strategy with the highest probability of definingreplacements for the "train and hope" methods on which educatorsmust currently rely. Implementation of the concept of afree appro-

4G

42

Otiate public ethitatibh in the least restrictive environment for allstudents should not be undermined by ignorance. _Thecontribu-tions the Washington Research .Organization makes to thedevelopment of a technology -of skill generalization are contribu-tions to the:work of all who strive for the realization of our socialcommitment to an effective and laSting education for all severelyhandicapped individuals:

Cindy is apprehensive her first day an the job at th2Pacific Oyster Bar. She failed so badly at the Seattle

Intel. She looks carefully at the dishwasher, and loadSthe bawls and cups. She closes the door. She searchesand findS- the buttons on the side of the machine. Theyare strange, but the little stickers just below them are justlike the ones at Scifabl. She confidently pushes theseries, and smiles when the dishwasher horns intoaction. At the end of the day, the kitchen supervisor says,"Good work today, Ms. Burchatt." He smiles as Cindygets her coat and loaves. Still smiling, he looks again atthe little stickers the trainer from the Seattle TrainingCenter had put on each of the dishwashers. He thinks;"Well, you learn something new every day:"

Richard leaves the office of the head housekeeper. As hewheels himself toward the chain of pink cabins of theSunset Motel; he repeats to himself, "Knock. Then saY,'Housekeeping here:' " Over and over he says theseinstructions, just as Mr. White taught him to do when heWas teaching him to say his name and address, all thoseyears ago. He is pleased that he can practice by himself.At Cabin 1 he stops,- squares his shoulders, and knocks'briskly. "Housekeeping here." He unlocks the door andgoes in to earn his first wage.,

4.3

lody is screaming so loudly that his' face is eggplantpurple again. Mrs. Loomis smiles to herself, and walksout the door to join, the rest of the family waiting in thecar, leaving Jody'.s jacket on the floor where he threw it.She gets in the car,\"Now where's Jody?" asks Mr.Loomis. "lust wait," she replies. ln 30 seconds, Jodycomes flying out the door, zipping his jacket. "Don'tforget to shut the door," cries his mother. She thinks Withsatisfaction of lody's teac* -- she was right, after all!Jody does know how to put oh his jacket.