ED 249 738

AUTHORTITLE

INSTITUTION

SPONS AGENCY

PUB DATECONTRAC:NOTEPUB TYPE

EDRS PRICEDESCRIPTORS

DOCUMENT RESUME

EC 170 850

Haring, Norris; And OthersInvestigating the Problem of Skill Generalization:Literature V 'iew I.Washingto v., Seattle. Vashington ResearchOrganization.Special Education Programs (ED/OSERS), Washington,DC.Jul 84300-82-036459p.; For related document, see EC 170 849.Information Analyses (070)

MF01/PC03 Plus Postage.*Basic Skills; Elementary Secondary Education;*General:zation; Literature Reviews; ResearchMethodology; *Self Evaluation (Individuals); *SevereDisabilities; Training

ABSTRACTAs I,rt of a 5-year investigation of skill

generalization in severely handicapped students, a literature reviewwas undertaken within each of three general approaches: performancepattern studies, ecological studies, and self-control studies. Thefirst paper, "Descriptive Analysis of Extant Research Literatureconcerning Skill Generalization and the Severely/ProfoundlyHandicapped" by Owen R. White, reviews 30 studies from the WashingtonResearch Organization (UUIRO) and concludes that only 13 of thestudies provided information useful in systematically refining aninstructional technology for promoting skill generalization withseverely handicapped persons. Ann K. Berman and Cheryl L. Opaiski("The Impact of Functional Trial Sequencing upon Generalization")review effects on skill generalization of the nature of the task, theenvironment, and the task's impact. In the final paper,"Self-Monitoring and Skill Generalization: A Review of CurrentResearch," Kathleen A. Liberty focuses on studies of self-monitoring,one of three most common self-control techniques. (CL)

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

* from the original document. *

*************************************ft*********************************

U 8 otrammore Of EDUCATICIIN)NISTrf49 f0. foucArcm

A .

Pr\

cr,WashingtonReseart hOrganiration

ULJ

w

4949

PERMISSION TO REPRODUCE THISMATE RIAL HAS BEEN GRANTED BY

di

TO THE EDUCATIONAL RESOURCESINFORMATION CENTER (ERIC)

INVESTIGATINGTHE PROBLEM

OF SKILLGENERALIZATION.

LITERATUREREVIEW I

, 4 1

111

aSdi PG /P

Norris Haling. Principal investigatorKai Neen Ube', Protect Coordinidor

fAichael Boer, Ed tor

The activity %which is the subject of this report was supported Inwhole or In part by the U. S. Department of Education pct No.30042-0354). licatever. the oPinlilds wiPidlidad homed do notwawa* reflect the aosiNon or policy al the U.S. Department ofEducation. and no official endoreemert by the Departmerd shark'be Warred.

Printed In the UnNed Stake of America.

y1984

fi

4

4

Preface

The Washington Research Organization (UWR()), anInstitute for Researc h in Education of tile Severely Hand-l( apped, is conducting a five-year investigation of theproblem of skill generalization. During its first projectyear. UWR() investigated three approaches todeveloping strategies for facilitating skill generalization:

I Prtormant e Patti rn Studies began with a retros-pet five analysis of existing data sets, and will pro-eed to the' «illection of descriptive data in public

st hexil c lassrooms. These data will be used to de-termine a set of experimental ck..cision rules formatt hing spec 'tic instructional methth;s to individ-MS I learners.

lthIgif studies initiated a four-year longitudinal(lest rip study of factors in educational settingswhit la may influence generalization. Interventionstudies are also int ludvd within this approach.starting with a pilot study of massed vs. distributedmstrut tumid trials on generalization.

sett -Itintroi Studies mrnmenced with three studiesof the cite( is of self-monitoring proc Mures On skillgeneralwation, This area of studies will also in-

lude investigations of self-reinforcement and self-wising hon.

During the first year of resarc h, UWRO investigatorshave «intim to d reviews of applicable literature withineat h of these general approac hes:

I )r t men R. White present', in "Descriptive Analysis ofKa,irc h i itrature Concerning Skill Generaliza-

5

tion and the Severe ly/Profoundly Handicapped," adescriptive analysis of 30 published studies fromUWRO's retrospective analysis of existing data sets.

Ann K. Berman and Cheryl L. Opalski's "The Impact ofFunctional Trial Sequencing Upon Generalization"provides a review of the literature that pertains to oneaspect of UWRO's ecological studies.

Dr. Kathleen A. Liberty's review of the literature, "Self-Monitoring and Skill Generalization: A Review ofCurrent Research," is related to the subject of her firstyear of UWRO studies in self-control.

UWRC) will produce a total of four literature reviewproducts, of which this is the first. Each of the currentresearch approaches will continue through the next fouryears, with the ultimate objective of producing a unifiedset of guidelines for practitioners to use in facilitatingskill generalization.

Norris G. HaringPrincipal InvestigatorSeattle, 1%4

f;

Table of Contents

Trigs of Contents

White. Owen lt. "Descriptive analysis of extant researchliterature concerning skill generalization and the se-verelyiprofoundly handicapped.

vii

Berman. Ann K. and Opakid, Cheryl L "The impact offunctional trial !sequencing upon generalization." 21

Liberty. Kathie"-, ,tilf-monitoring and skill genendiza-tiisn: A re .9.w of current research." 37

vii 7

DESCRIPTIVE ANALYSIS OF EXTANTRESEARCH LITERATURECONCERNING SKILLGENERALIZATION AND THESEVERELY/PROFOUNDLYHANDICAPPED

Owen R. White

The billowing analyses are based on 30 studies of skill general-flat i acid severely, profoundly handicapped persons. Thegeneral procedures for this retrospective analysis of publisheddata included the development of a form for coding informationabout the studies. development of coder rellabi;ity on theontent and format of the coding form, entry of coded informa-

tion in a computer. computerized summaries of the datacollected, and analysis.

Articles were selected for coding on the basis of the followingmerit'. 111 Data on generalization must be included in the

article. 121 The data must show responscs of at least one indi-vidual subject (group data alone are not sufficient). (3) The studymust concern at least one behavioral target that is an acceiera-t ion target. 141 The study must have been published since Stokesand Haer's 114771 article summarizing generalization studies todate.

In addition. since there were a wide variety of professional jour-nals from which to choose, journals that had a high probabilityof containing studies that met our criteria were searched first11- g, Amine! of Applied Rehovior Analysis was searched beforeAmerican Journal of Mental Deficiency). An effort was made toinclude articles where at least one of the subjects was severely.profoundly. or multiply handicapped. but other studies werealso im bided. since studies with more severely handicappedsubjects that met our criteria were difficult to identify. Overall.median reliability for article coding was MM.

1

The analyses provided below are descriptive in nature.providing infurmation concerning the frequency with whicheach coded condition ix.curred in the articles reviewed.tbierved conditions are evaluated in terms of their implicationsfor the study of generalization per se and the utility of availabledata in resolving certain questions being posed by UWRO. Noattempt is made to draw conclusions concerning the relation-ships among variables (e.g.. the relationship between varioustypes of generalization strategies and eviaitual success) due tothe limited number of studies thus far reviewed and coded.Certain relationships do appear to be emerging, however. andwill be reported if confirmed by further review.

Subjects

Number of Subjects

The number of subjects in each study or series of related studieswas generally small. The actual range of subject sample sizesextended to a maximum of 12, but a boxlike analysis iTukey.19771 revealed that the typical distribution ended at a maximumcif tt subjects per study. and that the 3 studies employing 12subjects should he considered exceptions to the general case.

As all studies reviewed employed some form of single-subjectanalytic strategy. the small sample sizes per se did not neces-sarily threaten internal validity. Oven the types of handicap-ping conditions represented by that smell number of subjects,however, the external validity of encoded studies with respect tothe population of "severely handicapped persons" as a wholemust he questioned.

Table 1 1

Divabibtito Itsiffessabed by %Medi

Dnatutisv type Nutntiorr at %Ate,

Normal 11143 sittaibilitlest if idItt1Mental Ketwistion

Mmirt4te, Monit4114efartfittnots !..11,111PIG Met LISewn/ Monstai Wag ittilatiom 7,1, re 1611illriAlvlIsrohnutel Mental flailerclatton Laid LITIVILILIAMultiply ilanditapped If leiPhystr ally f fandetappedt'oatolly impaired

usittot siv Impairede Pnonurni.aikern Impiturd&hovels Dukabelity ADeaf Madafter Hand mappens' earn

9

Subject Types

The distribution of disability types repr "stinted in the 30 studiesis illustrated in Table 1-1. The total number of cases noted (38)exceeds the number of actual studies reviewed (30) due to casesin which studies involved subjects with more than one disabilitytype.

MI studies reviewed involved at least one subject who could beconsidered severely handicapped in some manner. Disabilitytypes which could not in themselves constitute a severehandicap txxurred in studies employing control or contrastsubjects for other, more severely handicapped subjects.

t!WRO reviewers attempted to concentrate on studies dealingwith the severely mentally retarded. It is not possible. therefore.to make general statr.nentr concerning the representation ofvarious disability types in the entire body of research literatureas 4 whole. Given that a high proportion of severely mentallyretarded persons display one or more other handicaps as well.however, the pour representation of subjects with physical.visual. and auditory handicaps must be considered a seriousthreat to the ovriall ',sternal validity of the research reviewed todate

Subject Ana

StIbl,P4 is ranged in age from less than 2 years to in excess of 50sears The maturity of studies involved subjects ranging froms in all years of age to 21 years of age (b7" of the studies). Gener-ally. theretare. the studies reviewed appeared to provideadequate representation of the school population with respect toage.

Summary

Stint! ail attempt was uncle to review amides studying skillgeneralization with otverely mentally retarded subjects. nostatements tan be made concerning the degree to which all avail-able research represents the total 'emulation of severely,profoundly handicapped. Within that constraint, however. moststudies deal with a small number of subjects (Fs or less; in the ti-to-21 year age range. Notably, no studies were reviewed whichincluded sublet:is with physical. visual, or auditory handicaps.so fax the most part, conclusions based upon the UWRO reviewmust be limited to what might be called a sampling of -physi-cally intact" mentally retarded individuals.

30

Behaviors Studied

Behavior Chula

Most behaviors studied fell into one of six motor classifications:social. communication. vocational/prevocatiOnal. self-help/independent living. cognitive,academic skills, and cognitivestiategies. The majority of studies investigated the developmentof communication skills (liac, expression of need). self-help orindependent living skills (e.g.. use of local transportationsystems). social skills (e.g., engaging in cooperstive play behavtors), academiucomviptual behavior (e.g., the discrimination

"survival" worth in the environment). Somewhat surpris-ingly. 1:1 percent of the studies investigated general cognitivestrategim (e.g.. strategies for eliciting reinforcement from otherpersons in the environment( while only circa 10% of the studiesinvestigated vocational or prevocational skills le.g.. janitorialskillsl.

Famtionelity of Behavior

The great majority lgali( of behaviors selected for study wereperceived by both the original investigators and the reviewers astieing of immediate functional utility to the subject. That is inconformance with recent instructional trends. and should facili-tate generalization by increasing the probability of naturalconsequatitm outside the instructional situation.

Complexity of the Behavior

Training targets varied considerably in complexity. ranging fromsimple one-step behaviors le.g.. "responds appropriately to vesno questions") to tasks in which UM separate subtasks wereidentified lie g., "cleans bathroom"). Howe/P.1r. nearly two-thirdsid the investigated tasks contained fewer than ten separatesubtasks.

Pretraining Performance Level

Most studies i741%i involved the development of new skillswhere pretraining performance levels were demonstrated orassumed to be zero. At times the study of generalization of suchskills was incidental to the cic-moestration of skill acquisition,although the development of new skills was most commonly ariche le fur controlled studies of eventual skill generalization.

11

tkcasionally investigators continued to refine skills after initialacquisition or work with behaviors that had already beenacguinal at some level in an attempt to produce generalizedresponding, or to work with a behavior which had already met"unterion- in one setting or situation but which had not yetgeneralized appropriately.

'rhe relatively low incidence of studies involved in the system-atic. :efinempot of existing skills is a dis'emoititment ten tIWROinvestigators. It is hypothesized that th, fluency with which aperson is dhle to engage in a behavior during instruction will berelated to the probability of eventual generalization. sa) it isemiorturicite that few investigators studied skill refinement or.indeed, even collected the type of performance data whichwould enable the iSSIlti of flUMIEV to be studied at all (see thesection on "Assessment Data." below

General Settings and ConditionsDuring Liitial Training

Settings

Niel," studies involved initial training within public schoolsettings ;37".,,), but the most common training setting was aspec cal laboratory or toom 147%I. albeit. thet room may havebeen located within a public school. Very few studies conduttedInitial training within the cimirminity or home (1(3% each). A.higher pert entage 1503 %) did study remeralizat:on within thenatural setting" after initial trair.:ng [see "GeneralizationAmilitions and Strategies. below,. but due to the nature of the

skills being studied. those natural settings timded to be school,special vocational, or residential facility environments. ratherthan the community nr home. That would seem cmitrery to theintent emphasis on community integration and school4mw

c tioperative programs.

Trainer Type

The maturity of studies (77%) employed a member of there sean h staff (experimenter or research assistant) to implementinitial skill training programs. Only 1" of the studiesemployed the suhlect's regular teacher. therapist. or counselor

5

12

during initial skill training4d only circa 7% employed thesubject's parent or guardian. Dispite the apparent. simplicity ofmost instructional procedures employed. therefore. little directinformation exists in the reviewed literature cone ring theability of regular practitioners or parents to implement theprocedures under study.

Trainert'Sublect Eat%

All but two studies conducted initial training sessions in a one-to-one setting (one trainer with one subject). Only one studyemployed a group situation (one trainer to more than fivesubjects). and one study employed several peer trainers withindividual handicapped subjects. While effective small or largegroup training alternatives would be desirable. most training inclasses for the severely handicapped is still conducted on a one-to-one basis. so the use of such procedures in the studiesreviewed cannot be considered a serious impediment to even-tual adoption within applied settings.

AratecedentsiCuestPrempts

A wide range of instructional cues and prompts were usedduring initial skill training. Most. however. were "task specific,"designed primarily to elicit a particular response. Only circa2316 of the cues and prompts could be considered specificallydesigned to pramite a fly7-,, 4..t generalized response class (e.g..generalized imitation in no:. se to the cue. "do this"). More-over. most cues wee "art. .al" (i.e.. would not occur in thenatural environment), with only 3 studies employing somemixture of what might be considered "natural" and "artifice)"

and only a single study employing what might be consid-ered "all natural" cues (i.e.. cues which could occur in thenatural environment).

ConsequencestReinlintement

Most studies employed some mixture of social consequation.correction procedures. amiltor repeated trials (for errors) duringInitial skill development. *lore studies used at least a mixture ofnatural and athficial consequent es than was the case with cuesor jitompts, but only a single study used whet might be rAinsid-wed "all natural" consequences during initial skill training.

WA,

Anrangsnuent/Contingeocyffldseduie

Approximately two-thirds of the studies employed continuousschedules of consequence' (1:1) during some part of initialtraining. Such contingencies are not noted for their resistance toextinction and would, for the moat part, reduce the probabilitycalf successful generalization. One-third of the studies did eventu-ally employ some systematic method for adjusting or fading thefrequency of consequation. but somewhat more studiesemployed fixed schedules of cansequation rather than variableschedules (the latter would generally be considered more resis-tant to extinction). Finally, only one study attempted to baseconsequation (at least in part) on a temporal feature of thebehavior (latency). and in that sandy the subject was allowed 10seconds to respond before correction procedures were under-taken. It would seem, therefore, that fluency of response in moststudies was not I mnsidered of sufficient concern to warrantcontingencies which might explicitly facilitate its development.

Initial 'indult,* Ellactivineare

Since each rewewed study used idiosyncratic trainingprocedures. and few studies conducted direct comparativeanalyses concerning different approaches to training, virtuallynothing can he said concerning the relative efficacy of initialtraining procedures per se. Unlike the literature of a decade ago,however. the great maturity (87%) of studies were able to teachalt subjects the initial training task, and all but one of thesubiects who did not complete the initial training procedurewere dropped due to conditions beyond the control of the inves-tigators (e g., sublet:I illness or transfer away from the trainingsite!. It might seem that skill acquisition is less of a problem thanit once was, and that a focus on the development of procedureswhich facilitate generalization after initial skill acquisition isappropriate. One should be careful in accepting that conclusion,however, inasmuch as many studies carefully screened subjectstwfore training, at ierist in part to ensure a subject sample thatwould be likely to acquire the task in question.

'14

Genera lia don Condftieus and Strategies

Strategies Used to Pawnee Generalization

Virtually all of the training strategies identified by Stokes andBaer (1977) which might lead to generalization were representedin the studies reviewed (see Table 1-2). As found by Stokes andBaer. the most common "strategy" (actually. a lack of strategy)was the simple "train and hope" model. A somewhat greaterproportion of the recent literature employed loose training.""sufficient exemplars." and "natural maintaining contingen-cies" than Stokes and Baer obsinved. but no studies werereviewed that employed the use of "indiscriminable contingen-cies" and only two studies used procedures which involvedspecific contingencies designed to foster the development ofskill veriatiorsigeneralizatitm per se (i.e., "train to granulize").One additional strategy. not originally mentioned by Stokes andBaer. appears to have gained considerabhi popularity: "trainingin the natural ienviroranent A few studies conducted compara-tive analyses of at low, two strategies. but in virtually all of thosestudies the comparison was limited to "train and hope" versussome more formal procedure for promoting generalization. Verylittle information is available. therefra, concerning the relativeefficacy of alternative strategies specifically designed to prorategeneralization. Also. for obvious reasons, the "train and hope"procedure was always attempted first, so the effect of the alterna-tive procedure was always confounded with intervention-sag uence.

Mangos in Canditirms Durk* Generatintigia Probes

Collectively. reviewed studies probed for generalization across awide range of conditions (see Table 1-3). Although some studiesinvolved only what might be considered "trivial" generalizationparameters (e.g., generalization across time of day or specificmanager), most made some effort to probe for generalizationwithin the setting or situation where the trained skill wouldmost naturally be employed. hi four studies (13%). responserequirements under generalization probe conditions were alsochanged. requiring some form of response adaptation or induc-tion in addition to simple stimulus generalization.

15

TAW 1

TiouswolisistSse TriagingMeth

Strom Number of Studies

from I tope A44401711.42P1i10[4061efiTi1

Naturist Settinsi 2141U13444,14440tirqueni sat%Wan Arion

fnuntwoomrh

%Latin tent1..***nolgit* VII. 140t Sluttish

NiAturni Muni.nnturrni licral

%timbalsultrsal (Awn

1 fmli h1Grua/411,r

atits i I

limps in Condition* duringConeratieution Probes

t it ion Number ut Stusisres

`u.tsitrio 7- 14.4.111; i4

I inn. 1, IT ri 1411;14, 144444. -41

ileum ',annuli . I. I44114.1144,11,144,1

Pr/44v 1-441;

Ant t .tirm r 1.:(444,161440M144,11:4,11Tit/714

Pqa ,Itle1.14444.44714iT/14.14711, PLIP11;11.1

9

16

Assessment Data

Trainiabarsta

Training data were provided in 25 (87%) of the studies reviewed:four studies reported only that "training was conducted success -fully." With those studies, of course, no analysis of the relation-ship between initial acquisition curves and eventual generaliza-tion is possible. All of the 26 studies reporting training data didreport the data for each individual sub*ct, but in only six (20%)of the cases was it possible to identify precisely how manyresponses (correct/error) were made during each trainingsession. The remaining studies reported data in "blocks ofsessions** or as some computed statistic (e.g., "percent ofpossible," where the number possible was not reported). Suchsummaries can seriously distort the actual nature of an acquisi-tion curve and obviate meaningful analysis of its relationship toeventual skill generalization. Finally, only two (7%) of thestiiies reported time-based data (e.g.. rate, latency) for initialtraining, obviating the possibility of evaluating the relationshipbetween initial response fluency and eventual generalization.The lack of time-based data during initial training also limits theevaluation of changes in response characteristics during general-ization probes. That is, while it might be possible to determinewhether subjects did or did not generalize at a3 on a given oc ca-sion. it was not possible with most studies to determine whetherthere was any decrease in response fluency under generalizationconditions.

Generalization Probes

in outer to form a complete picture of the relationship betweenintieil training and eventual skill generalization, systematicprobes fur generalization should probably occur before, during,and after each phase of skill training. Unfortunately, that was notalways the case.

Seven studies (23%) did not probe for skill generalization priorto the initiation of training. In most of those cases baselinetraining probes (i.e.. probes to determine pretraining level ofperformance in the training situation) indicated less than crite-rion performance. so one might assume that no satisfactorygeneralized responding would occur.

Six studies (20%) did not probe for generalization at intervalsduring the initial training process. With those studies it would

to7

not be possible to determine whether meaningful generalizationoccurred prior ie, the point when the submits reached the oftenarbitrary criterion for initial skill acquisition.

Eignt studies (27%) failed to probe for generalization after thetermination of training. Such shidies did probe for generaliza-tion at least once near or at the and of training, but would haveno measure of "maintenance" of effect, even over very briefperiods. Five studies (17%) probed for generalized respondingonly after training had been concluded. In those studies. noassurances are provided that generalized responding did notix.cur prior to training, and no estimate of when generalizedresponding might first have occurred (or been possible, given theopportunity) was possible.

It should be noted that there are potential iustifications forminimizing the number of generalization probes undertaken'luring a study. Aside from the potential difficulty mkt cost ofsuch probes. frequent generalization probes may constitute oneor more c onfounds. For example. if reinforcement (natural orotherwise) is provided fur appropriate responding in generaliza-tion situatkins. that reinfortement would constitute a potential"training strategy" for continued responding, obviating any"clean- assessment of further generalization per se. Conversely.if reinforcement is withheld in generalization probes, repeatedexposure to those conditions would constitute "extinctioncond:tions" and could lead to the formation of a discriminationbetween training and riontraining situations, the opposite ofwhat one .e.t ual I y hopes to accomplish. It is necessary. therefore.to consider the issue of repeated and frequent generalizationprobes very carefully. but the failure to probe for generalizationbefore. during. and after initial training should be avoidedwhenever possible.

Outcome: and Implications

Level of Generalization

in all but three of the 30 studies reviewed. authors chicacterizeiltheir sublects as generalizing "some" or "well." However.antliors characterized all of the subjects in any given study asgeneralizing "well" in only 11 (37%) of the cases. UWROreviewers tended to be somewhat more conservative, recordingthat in their opinion, only 2 (7 %) of the studies demonstrated

11

oinplete generalization," and that only 12 (40%) of the studiesappeared-to employ methods which facilitated some meaningfulgeneralization in most sublects. The discrepancy in opinionsmight be due to an unconscious predisposition to find successon the part of the authors, but in at least some cases authors wereactually more reserved in their opinions than the reviewers.More likely, therefore, the discrepancy is simply due to a lack ofLoncenSUS concerning the level of performance that constitutes"meaningful generalization."

Applicability of Findings

Reviewers felt that 16 (53%) of the studies employed proceduresthat could most likely he applied in a wide range of nonresearchsettings. In the remaining cases, however, one or more featuresof the methcallsi employed would likely serve as a seriousimpediment to its widespread adoption. in most of those cases,the maim priklem appeared to be a simple matter of cost, eitherbecause of the personnel and time required to implement theprocedure, or the special equipment/settings involved.

Iteplicability of Studies

Surprisingly, considering that all reviewed studies werepublished in research journals. reviewers only felt that 21 (70%)of the studies were described in sufficient detail to allow precisereplication. Problems centered most often on descriptions oftraining procedures or the conditions under which generaliza-tion was assessed. For example, one study desrxibed trainingprocedures as "modeling, role-playing, and social consigna-tion." It should be noted, however, that precise replication of allprocedures might not he required in order to achieve a func-tional replication of outcomes, and additional informationt -incoming procedures is often available from authors uponrequest.

implications fm. Development ofa Technology of Generalization InstructionThe demonstration that a particular procedure is capable offacilitating skill generalizaticn does not necessarily further thedevelopment of a precise instructional technology. in order forsuch a technology to evolve. studies must be conducted whichsystematically analyze variation in approach in a manner whichallows specific controlling variables to be identified. Unfortu-nately, most of the reviewed studies did not lend themselves to

12

1;i

such analyses. As menthmed earlier, most studies merelycompared a given approach with the "train and hope" paradigm,demonstrating at best that some directed attempt to facilitategeneralization was better than no formal attempt. In thosestudies which did make some attempt to compare alternativestrategies fur the active promotion of generalization. the"favored" strategy almost always followed an attempt toproduce generalization with less favored strategies -- raising thepossibility of an inbmention sequence confound. Overall.reviewers were of the opinion that only 13 (43%) of the studiesprovided information useful in systematically refining aninstructional technology for promoting skill -generalization withseverely handicapped persons: and indeed. many of thosestudies still contained potentially serious experimentalconfounds. The remaining 17 (57%) of the studies may haveprovided an example of how one might facilitate skill general-ization. but did not provide sufficient information to enable thescientist to refine hypotheses concerning controlling variablee,or the practitioner to make an informed choke among viableinstructional alternatives.

References

- --* Andel son. S . b Smndlin. 1.11980f. The generalized effects of productiveLibeling training involving common object classes. journal of theAssisi-ninon for the Severely Handicapped. 5.143 -157.

Campbell. C.. a suemei-carepbell. K. (19621. Programming loosetraining" as a strategy to facilitate language generalization.hiornal of Applied Behavior Analysis. 15. 295-301.

Loon. M , Vogelsberg. R., & Williams. W. 119811. Effiicis of classroompublic transportation instruction on generaiiitation to the naturalenvironment. Animal of the Association for the Severely Handi-capped. 6.46 -53.

Italie. f.. Marshall. A., Spradlin, 1. (19-1j. Time delay: A technique toxiscremete language use and facilitate generalization in retardedi.hikiren. journal of Applied Behavior Analysis. 12.431-439.

%filbert. V.. Minor. I.. & McCoy. J. (19771. intensive feeding trainingwith retarded children. Be;tavior Modification. 1.517 -529.

Stokes. T . & Baer. D. (19771. An implicit technology of generalization.journal of Applied Behavior Analysis,. 10.349-367.

13 20

Stokes. T.. Fowler. S.. & Beer. EL (19751. Training pi-mama' children' torecruit natural communities of reinforcement. faunal of AppliedBehavior Analysis. 11. 255-303.

'Thompson. T.. Bream. S- 6 Fuqua. W. (1982). Training and generaliza-tion of laundry shah: A multiple probe evaluation with handi-capped persons loaned of Applied Behavior Analysis. 15. 177-182.

Tucker. D.. & Berry, G. (1980). Teaching severely retarded multlhandl-capped students to put on their own hearing 'ids. Journal ofApplied Behavior Anidysis.'13, 85-75.

Tukey. J.W. (1977). Exploratory data analysis. Nadir*, PA: Addison-Wesley.

van den Pal. K. Nana. B.. Iyancil, M.. Page, T.. Need. N.. & Whitley. F.(1981). Teaching the handicapped to eat in public places: Acquisi-tion. genendizatkin, & maintenance al restaurant skills. Journal ofApplied Behavior Analysis. 14.61-01

Welch, S.. 6 Pear. f (1980). Generalization of naming responses to°Merin in the natural environment as a function of Minion stim-ulus modality with retarded children. loaned of AppliedBehavior Analyslia. 13. 829-643.

eiblitignipby adArtidesCoded

Adkins. & Matson. I. L. (1980). Teaching institutionalized meniallyretarded adults socially appropriate leisure skills. Mental Boar-&Mon. 10, 249-252.

Bornstein. P. H.. Bach. P. I.. McFall. M. E. Friman. P. G.. & Lyons. P. I).1480). Application of a social skills training program in the modi-

fication of interpersonal deficits among retarded adults; A clinicalreplication. Journal of Applied Behavior Analysis. 13. 171-176.

Bianstein, M. K.. Bellack. A. S.. & Hera% M. (1977).training for unassertive children: A multiple-baseline analysis.Journal of Applied Behavior Analysis. 1o, 183-193.

0rowder, U M.. mods, W W., Snell. M. E (1981). Using time delay toteeth manual signs to a severely mentally retarded student,Education and Training of the Mentally Retarded. 18. 252-250.

Borgia, L. D., Whitman. T. L.. Johnson. M. R. (1980). A self-instructionalpackage for increasing attending behavior in educable mentallyretarded children. Journal of Applied Behavior Analysis. 13. 443-460

24 21

Rummy I i1 . Russel. B., Shares. K.E. (19771. Developing social neiptelsoiin two 1114401mill, retarded American A.ssaciatinn forthe haul untie of the Severrly'Profirundly tfandicapped Review.Z. 53-60

Campbell. C. K.. & Strettiel-t:ampixill. K. (198Z1. Programming loamtraiiiiria d Sirdiejef to facilitate language generalization.Alumni id .ippheel Behavior Analysis. 15. 295-.102.

Combs. M. L & (Airy. H. (1981). A cognitive social skills trainingprograin Brhuvior Modification. 5. 39-19

Cooke. T. P & Apo Boni. T (19761. Dente loping positive social-emotionalhelm% tors A study of training and generalization effects. Inertialof Applied Helen uw Analysi.. 9.65 -78.

t :ono M E . Vitgelstwrg. K T . & Williams. W. (19811. Effects of class-room public transportation inserts two an generalization to thenatural environment. knumul of the A.ssiriation for the Severely/tondo ripped. 6.46 -5.1

iitoA t.e.d. Id H . & liorakove. L. S. 119701. Teen.hing janitorial skillsto the mentally retarded: Acquisition. generalization and mainte-nance foninitzl at Applied ffehaviur Analysis. I I. :1415-355.

!blown. f P.. Clark. If & Kisely. T. K. (19771. Peer tutoring amongelementary students: Educational benefits to the tutor. journal ofApplied flehavior Analysis. 21), Z31-238.

Down( k. P W . & Dove. C. 09801. The use of self-modeling to improvethe swimming per of spina bifida children. journal ofApplied Behavior Analysis. 3. 51-56

ads. . & Thorne. 1 M. 119781. Improving the generalized mnemoniciirtormint of a Down's syruirome child. Journal of AppliedRehm .4nolvsis. 1 1. 41:1 -419.

Filter I . rk tiasari. t: 119911. Acquisition. maintenance & generalizationof parent-taught skills with two severely handicapped infants.homed of the Arun intern for the Severely ffundicapped. 6. io-ta

Ciangrei ri. M F (1983). Teaching basic photography skills hid severelyh,inilic.apped young adult journal of the Association hit theSeverely flondu tipped. 8.43 -49

Hall. C . Sheldim-Wildgen. & Sherman. J. A. (19801. Teaching 10hinterview skills to retarded clients. journal of Applied BehaviorAnalysis, I:CC13 -442.

2213

Hart R & Risley. T R. (1080). In vivo language intervention. Journal ofApplied *hay no Amapa& 13.407 -432.

J. W.. Wehmsa. P.. & thirst. C. (1982). Toward generalization ofappropriate leisure and social behaves in severely handicappedyouth. Pinball machine use. Journal of the Association for theSeverely Handicapped. 8. 38-44.

Hill. W Wehman. P.. Pentecost. J. (1980). Developing job interviewskills in mentally retarded adults. Education and Training of theMentally Retarded. 13. 179-188.

Hodges. P. M.. a Deich, K. F. (1978). Teaching an artificial language tonon-vinbal retardates. Behavior Modify:idiom 2.489.509.

laciuston. C. M. (1979), The use of visual orientation feedback to facilitateattention and task performance. Mental Retwdation. 8. 281-284.

Johnson. B. F.. & Cuvo. A. (1981). Training mentally retarded adults tocook. Behavior Modification. 5.187 -202.

Kissel, R. C Johnson. M. R.. & Whitman, T. L (1980). Training aretarded client's mother and teacher through sequenced instruc-tions In establish self-feeding. Journal of the Association for theSeverely Handicapped. 3.102-394.

:lase). R. C.. t# Whitman. T. I.. (1977). An examination of the direct andgeneralized effects of a play-training and overtarrection proce-dure upon the selfetimuletory behavior of a profoundly retardedbtu American Association for the Severely and Profoundlylandicapped Review. 2. 131-148.

knet,01. K. & Williams. J. A. (1980). Behavioral contrastand generalization across settings in the treatment of autistic thil-&sin. helmet of Experimental Child Psychology. 30. 422-437,

kohl F. L . Karlan. G. R.. sit Heal. I.. W. (1979). Effects of pairing manualsigns with verbal cues upon the acquisition of instructionfollowing behaviors and the generalization to impressive languagewith severely handicapped students. American Association forthe Severely and Profoundly Handicapped Review. 4.291 -302.

Latiev. H. lituuneyer. M. K.. (Mani. C.. & Bens. V. E. (1977). Treat-ment of severe perceptual-motor disorders in children diagnosedas learning disebled ilithovior Modificatnin. 1. 123.140.

Landon'. G. . & Caccarani. P. S. 119811. Teaching independent Writingwithin the normal daily program: Two studies with profoundlyretarded children. Behavior Research of Severe DevelopmentalDisabilities. 2. 70-06.

23 1.

Matson. I L. 119811. Assessment and treatment of clinical fears inmentally retarded t hildnen limning of Applied Behavior Anal.Ysis. 14.287.294

Matson. L. & Adkins. I. (1980). A self-lust/Winne! social Skillstraining pnigrani for mentally retarded persons. Mental Retarda-tem. 111. 245-244.

Matson. I L.. & Andrasik. P. (1982). Training leisure lime siociiil-inkfrac.non skills to. mentally retarded adults. American Puma/ ofMental Deficiency. M. 533-542.

Miller. S. M. a Sloane. H. N (19M). The generalization effects of parenttraining mums stimulus settings. journal of Applied BehaviorAnalysis. 9.355 -370.

11thai ,,. E. 11478). CAW study in training generalized instruction-following eisponses to preposition-noun combination's in aseverely nitarded voting adult. American Assn( Minn for theSeverely and Profoundly Hand ii ripped Review. 3.238-245.

Mithaiiii 11 E /4 Wolfe, M S. (19741. Fanploying task arrangenmints andverbal r wan:gem iris to promnte verbalizations between retarded4 hilsinin journal of Applied ItehavMr Antdysis. 9. 301-314.

Ntunlis k. (; Y.. (;,,tn.ta. E K. & Hardiness. M.L. (1977). Generalizing arti-culately's traiiimpi with trainable menially retarded suhiects.!mania of Applied Behuveir Analysis. 10. 717 -733.

-Nadler. . Ai Reid. U. ti. 11978). Teaching retarded women a clothingselection skill using community norms. 'annuli of AppliedBehavior Analysis. 11.475 -487.

Page, 1 !wale. K . 14 Neel. N. A. 41976). Teaching peclestrian skills toretarded persons i;eneraliration from the classroom to the naturalenvironnwnt journal of Applied Behavior Analysis. 9.433 -444

Parsonson. H S Haw. I) M. 119781. Training generalized improvisa-tion of nails by preschonl children hiurnul of Applied BelitiviiirAmtlysis. 1I. lb.1-3/10

Reese S C & Filipteak, 1 (1980). Assessment of skill mineralization:Measurement stns setting. behavior. and times in an educationalsetting. Behavior Modifentum. 4. ziso-223.

Beet I) H.. & Hurlbut. H 11477j 'few 'sing ;ammo! communicationskills to multihandicappisi retarded adults. journal of AppliedBetratior Analysis, 10. 591 -1103.

" 24

Richman. 1 . S.. Sonderby. T & Kahn. 1. V. 119801. Prerequisite vs. in vivaacquisition of aelfheiding skill. Behavior Research and Therapy.18.337-332.

R.. & Cave. A. 1. (19801. Training nunitally retarded adults tomake emergency telephone calk, Behavior Modification. 4. 513 -528

Ragers-Warren. A.. & Warren. S. F. (1900). Minds for verbalization:Faciliterisql the display of newly trained language in children.Behavior Modification, 4.361 -382.

Russo. D.. & Kriegel. R. L. (1977). A method for integrating an autisticchild into a normal publk school Glassman:. MIAMI 4 AppliedBehavior Analysis, 10. 579-590.

Rychterick. R. G. & Bornstein. P. H. (1979). Training conversational skillsin mentally retarded adults: A multiple baseline analysis. MentalHet". 17. 289-291

Sc:hleien. S. 1.. Ash. T.. Kieran. I.. & Witham. P. (1981). DeveloPlnitindependent cookirqt skills in a profoundly retarded women.Journal of the Association for the Severely Handiropped. 6. 23-29.

hriebman. L.. & Garr. E. G.(1978). Eliminstism of echolalic respondingto quesi.ons through the training of a generalized verbal response.Mental of Applied Behavior Analysis, 11.453 -483.

. 1 . & Bucher. B. (19801. DPvIii0PMeet of spontaneous mending in6 language deficient children. Axonal of Applied Behavior Anal-

ysis. 13.523 -528.

Smaege, M. E.. Page. T. 1.. twee. 0. A.. & Jvsrule. M.1. 11980). TeachingMiNISLWOMOIli skills to mentally retarded students: Training.gemerelization and follow-up. Education and Training of theMentally Retarrhrd. 15.224 -230-

Strain. P. S.. Shwas, K L. a Timm. M. A. (1977). Effects of peer socialinitiations on the behavior of withdrawn preschool childrenkeinial of Applied Behavior Analysis, 10.289 -298.

Tofte-Tipps. S.. Mandonus, P.. & Peach. R. V. (1942). Training andgeneralization of social skills: A study with two developmentallyhandicapped. socially babied children. Behavior htorhileigion.6.45-71.

Trap, f. 1.. P.. lam. S.. & Cooper. 1. 0.119781. The effectsof feedback and consequences on transitional cursive letter forma-tion Journal of Applied Behavior Analysis. 11.301-394.

le 25

Tux ker. D I . & Rem. C. W Teaching severely mullibandicaPPIKI%tailetits to put on their own hearing aids. four:tat of Appliedt veep Anufyslti, I 1. 0575

Vaal:lit:welly!. A.. Spangler. P. F.. I Menthol!, A. M An embalm-mail examination of simple strategy fur hamming mealtimelanguage in residential facilities. Journal of Applied BehaviorAnalysis. 14. Z95. wS.

Vrtowlithurs. K. 7.. Raul. F. K. 12479!. 'Training miveritiv klandic4APPodNooftmta 19 (JOSS t,twitrulled intersintions. Journal of theAssociation for the Severely Handicapped, 4.264.273.

Warren. S F.. Raster. D. K.. Anderson, S. R.. Marshal A.. & Baer. U M.I pail,. t;erieralixation of question-asking by severely retardedindividuals Journal at the. Association fa; the Seventh. Horecfi-I upped. 6, 1S-21

Wellman. P . Kearatclia. A . Nm. C.. 40,uftr. K.. oi Kaman. () 119781.fievelooing a leisure skill repertoire in severely and prehtundlybandit lipped twrsints American Association for the Munition ofthe St9-ery1; Profountilv flandit upped finless, 3. 162- 172

19 26

THE IMPACT OF FUNCTIONAL TRIALSEQUENCING UPONGENERALIZATION

Ann K. Berman andCheryl L. Opalsid

Parents and teachers are often and frustrated by chil-dren with developmental ** who tie Asir shoes athome yet not at school; who do not brush their teeth at home yetbrush their teeth ewey day at school; who can operate one typeof clothes washer but not other ty . Such examples regesent afailure to *Aida skill ligation. Traditionally. manyresearchers considered skill generalizaticm to be a passive ,phenomenon (Straws & Baer, 1977). In other words. the transferof a skill from a training to a nontraintre; WWI= was consid-ered to be automatic; no specific training techniques werethought to be necessary to promote this transfer.

However. it is now broadlyrecognized that the meanie in whicha skill is initially taught may contribute significantly tosuuxissful skill generalization. A feature of training which hasreceived considerable attention in relation to skill acquisition isthe manner in which the training trials are sequenced. It is thepurpose of this review to examine the influence of variousmethods of trial sequencing on the generalization process.Specifically, 39 studies were reviewed which addressed theissue of skill generalization andlor trial sequencing in themoderately to profoundly handicapped. Special attention waspaid to the type of trial sequencing used by experimenters andhow that sequencing might affect skill generalization.

Driinittriate

Research into trial sequencing requires establishing clear defini-tions of several concepts. Thal sequencing is the process ofpresenting a stimulus across time and situation. Trialsequencing is conaidered faudienni if it results in a responsewhich is trusted to oprwate in nodule) envinunnents (Stokes andBaer. loin Thus, the mechanism for programming generaliza-tion is called fenciknat trfnl sersencing.

21 27

This reviev explores the impact of the four metlmds of trialsequencing upon generalization. Since there is variation in howtrial sequencing is defined from one study to the next, thefollowing definitions of trial sequencing types are used in thisreview:

(1) Massed Trials: Trials which occur so closely togetherthat no other behavior is emittedbetween trials (Mulligan. Lacy, &Guess. 1982).

(2) Distributed Trials: Trials which occur in such a way thattrials from another program ourbetween two trials from the sameprogram (Mulligan, at aL, 1982).

(3) Serial Trials: Trials in which one item was trainedto criterion. a second item was trainedto criterion. and a third item wastrained to criterion, etc. (Waldo.Guess, & Flanagan, 1982).

(4) Concurrent Trials: Trials which involve training severalitems simultaneously to a specifiedlevel of criterion (Waldo. at al., 1982).

Having established these definitions, it should be noted thatmethods of sequencing are generally combined to form fourbasic types:

(1) massed-serial.(2) massed-concurrent.(3) distributed-serial, and(4) distributed-concurrent.

Variabke

The current literature suggests that trial sequencing. in and ofitself. does not significantly impact skill generalization.However, three variables were identified which. whencombined with specific types of trial sequencing. do affect theacquisition and generalization of a skill. These variables are:

11 ) the functionality of the skill.(2) the purchase power of the skills, and(3) the environmental relevance of the skill.

No one factor was found to be solely responsible for success orfailure of generalization. Each of these three variables isaddressed separately in this review.

22 2 ei

Both the method of sequencing and the variables withinsap:fences, as they relate to generalization. have been examinedfor this review. It should be noted that the studies reviewedvaried along a number of important dimensions including age ofsubjects (i.e.. infant to adult). the type of skill being taught (e.g..use of nonsense syllables. riding a bus. etc.). the degree to whichthe generelization environment differed from the training envi-ronment. and the implied meaning of the term "generalization."Conclusions drawn on the basis of this review are, therefme, of ageneral and tentative nature.

The Functionality of the Skill

The literature indicates that functional skills generalize moresin cessfully than nonfunctional skills (see Table 2 -1). A skill islonsidered functional if it has utility for the subjects_ in theirnatural environments. According to the definitions used in thisreview. examples of functional skills include: boarding a publicbus it :ism. Vinotistittrg. & Williams. 19B11. playing pinball (Hill,Wellman, & thirst. 1982), asking curious questions (Hung. 1977),f * ,in .ounting (Lowe & Cuvo. 1976). and cleaning a restroom(Ciao, Leaf. & iktraknve. 19791. Nonfunctiunal skill examplesim tulle. touching head on cue ()Cowl & Rincovar, 1977). ream -to.. skills (Charlop. 1933), and some verbal skillsIt;arcia. 14174: McLean & McLean, 1974: Stevens-Long. Schwarz,et Bliss, 1976: and Calm & Hear, 1973).

Skills taught in a distributed format tend to be functionalwhereas about 50% of those skills taught in a massed format arerated nonfunctional (see Table Z -2). Thus. generalization tendsto he more successful with functional skills. and functionalskills tend to be related to distributed rather than massedsentient .ing.

Three factors within the general issue of functionality seem rele-vant: I I t whether the skill is initiative or retentive in nature, (2)whether the skill employs cues and/or stimuli already present inthe individual's natural environments, and (3) whether the skill

a form of the behavior already within the individual'srpertoii e..

Initiative tw Receptive Nature of Task

The individual contods where and when to use an initiativeskill (in the other hand. a receptive skill is one that always

"29

Table 2-1

numbhweal nessralisslim neanks Accairdhesto Ike Fusistlimalny at the MIN

Stradies addressinghiactional skins

Studies adrinissingnouhuicliasel

skins

arraps..

Number of skills'

Studies reporting atleast same rworelization

Studies reporting 100%

18

18 (100%)

10

9 (10%)

puendtaatian intimodiatuly 14(70%) 0(0%)

Studios owning 100%gsseralLuation attarfurther training 3 (27%) 3 (30%)

Studies reporting lam dm)100% genaudisation emsviler further Staining 1 V") 41810%)

Table 2.2

7 fie nalaltasship nebrass Fwactieuelny andTrial Salusuciag tits S lissimmed

Nusahmed Studlesa

NumberFunctional

PercentFunctional

Massed - Serial 13 7 47%

tAnieed-Concutrent 4 2 50%

Diteributed-Smail 1 1 100%

Distributed-Concurrent 8 8 100%

TOW 28 18 04%

1 Them include only the studies ft/owed which addressed lioneralhationusing one typo at trial saquencing.

depot& on cues hum another individual in order for thebehe.ior to occur appopriately. The majority of skills trainedina massed-serial and maseed-concormt sequence are receptivein nature while those ski& taught in a distributed-serial ordistributed-concurrent hornet tend to be initiative (see Table 2-3).

Skills which require independence or initiation of a behavior onthe part of the child are charectaistically more functional in thechild's natural environment, while receptive skills are lessuseful in any environment outside of the !minim; situation. Furexample passively responding to an experimenter's three ques-tions about cue cards (Clark & Sherman. 1975) requires receptiveskills. while a self-help skill such as washing hands (Zifferbbnt.Burton. Hamer. k White. 1977) requires initiative. Overall. skillsrequiring initiative on the part of the child generalize moresuccessfully than receptive skills (see Table 2.4). However, theresults are not conclusive since the initiativelrecinnive nature ofthe task was not an experimental variable within the studiesreviewed.

Use of Cues and Stimuli within the Natural Environment

Oxm. et at. (1981) trained one retarded person to board and ridea public bus by providing instruction in the classroom and inpublic places. The skill did not successfully generalize untilnatural environment training was impiemented. Marholin.O'Toole. Machete. Berger. and Doyle (1979) successfullytrained four retarded people to ride a bus to a specified destina-tion, order an item. and purchase the item by training in theclassroom. on a bus. and in a shopping center. Generalization onthese skills was succeeshil. Coon et-at.-used a massed-serialformat while Marholin at al. used a distributed-concurrentformat. Thus. successful generalization can not be attributed tothe trial sequencing. Rather. the results suggest that generaliza-tion may have been facilitated by incorporating into training theconcrete, dependable stimuli present in the natural environmentfor the skill.

Some studies probed for skill generalization in environments inwhich the child did not ordinarily operate (i.e., a wooded areawith a stranger. Koegel & Kim:over. 1977; Rincaver & Kassel.1975). In these studies, the children were placed in a nonnaturalsetting and asked to perform a nonfunctional task. It is inter-esting to note that in some cases. a portion of the sample did

25 31

Table Z-3

Relotieseitip itehrson Th. Snosessalog sodlautPlivlialielive Nature silk Sot

Minim Smiles &Miming Studies Addreseingal Studies lbsceptive Skiffs initiative SIM,

folaseedSeriel t $

Storee1t4"bocianun1 4

Distrdnded-Seriel 1

Distributed-Concurrent B

11480%)

3475%)

01094)

1 (17-5%)

3120%)

1(25%)

1(100%)

(87.5%)

Table 2-4

Inrelidono id Gonereitiodoe Remalte Accordingto die tollintivollecoptive Mauro vitae NM 4

Studios addreesing Studies addressinginitiative skillsreceptive skills

Number of ieudiee

Percent funclionei

Studies reporting elteal scene priseraiisetion

Studies nniortiog 100%poem motion inunediatiris

Studies reporting 10094_on eito.harden training

Studies menthol less then100% genessiindion evenafter hutber trussing

lb

15 814%)

5133%)

3410%)

7444%)

12

121100%)

12(100%)

0107%)

a

32

respond as trained. However, the giility of an individual togeneralize a nonfunctional response such as "head-tapping to astranger in the woods" '.41 of questionable value in that person'slife.

Another aspect of the Itincovie. and Koegal study (197S) is thatthe children who failed to generalize were furtha trained, andAU attempt was made to discern the variables to which the chil-dren attended. The subjects who failed to generalize respondedto the therapist raising his hand. the therapist letting go of thesubject's hand. the initial prompt. and the furniture placement.The child.en were overselective, attending to physical move-ments or objects rather than to the verbal commands or thepeople themselves. Thus. when the children were prompted in astrange environment. they called upon past experiencetheir response: These, resuits,suggestAhet-neterateTiv romnenttraining would be advantageous since many, relevant cueli'andstimuli would be present in both training and nontraining situa-tions.

1 tee of Behaviors within the Subject's Repertoire

/sing forms of behavior already within the subject's repertoirehis implications for generalization. in a study by Charlop (1983).three echolatic children generalized while three mute childrendid not. Significant to these results is that part of the task was toecho the experimenter prior to handing the experimenter anobject. Perhaps the use of a behavior already in the child's reper-toire facilitated the generalization process. When the child wasnot allowed to echo. performance dropped considerably.Ctiarlop suggests that the incorporation of verbal response mayhave facilitated acquisition and generalization for the echohilichildren in that they could more actively participate in the task.

The ionic children had no such opportunity.

The Purchase Power of the Skill

.1fter functionality. the next issue which surfaces is the purchaseLewin of the behaviors. especially when failure to generalize isreported. Satiation and diminished motivation are factorsrelated to trial sequencing which seem to impact successfulsyneralization. A massed sequence requires many highly repeti-tious trials which might prompt boredom with the task and rein-forcer situation. whereas a distributed sequence involves less

27

33

repetition and the opportunity to pmform other behaviorsbetween trials.

Satiation

Massed trial sequences require highly repetitious responding bythe children. Reinforcement (food or verbal praise) is most oftenoffered on a continuous schedule (Buxom & Koegel, 1975) oron a ratio schedule. usually a variable ratio schedule of 3:1(Chariot,. 1983: Simic & Bucher. 1980). Two to 20 correct trialswere required prior to the generalization probes. Char lop (1983)as well as Rincover and Koegel (1975) reported less than 100%generalization. Furthermore, Simic and Bucluir (1980) reportedno generalization even though reinforcement was offered during

---ssinsokfixation probe sessions. Perhaps the children grew tired ofthe offiiiitt-reinfoLcoment.

However, massed trial sequences do not always lead to satiationand problems with generalization. Panyim and Hall (1978)required two retarded girls to reach criterion or complete 200trials. Waldo. Guess. and Flanagan (1982) required six correctresponses for each of three receptive labels before moving toanother set. However, movement to another portion of the setwas allowed when criterion was achieved on one portion.Indeed, the potential for overlearning and insufficient rewardwere -high risk. In both of these studies, generalization did occurafter massed-concurrent trials, but was less successful aftermassed-serial trials. Thus, the massed nature of the trials. withcontinuous reinkucenwnt did not always lead to satiation.Rather. another feature within the trial sequencing must havebeen a significant variable in the generalization results.

While overlearning may facilitate fluency (White & Haring,1980). insufficient payoff risks extinction of the skill. Ability toperform the task may not have been represented by performanceif the individual is satiated on the offered consequences. Thelaw of diminishing returns is likely to be applicable with respectto the motivational power of a piece of candy to awoke a behaviorwhen the child has already received many pieces of candywithin a short period of time.

Motivation

An issue related to satiation is the motivation of the individualto maintain the skill. Some rehiforcers may be less susceptible to

2834

satiation than others and thus maintain the individual's motive-thin to parftnin the task.

Bateman (1975) applied Premack's-principle that, "for any pairof responses, the independently more probable willreinforce the less 'probable one" (p. 604) to show that to ccupa-tional behavior of two semely retarded people could be modi-fied. Bateman showed that the subjects were willing to pert wm aless preferred skill to earn time to perform a desired skill. Thechance to perform a desired skill was a natural consequence ofthe subject's behavior and provided sufficient motivation tocomplete the less preferred skill.

Ahhough the less preferred skills wine quite repetitive (windingballs of wool and plug assembly), as were the preferred activities(sewing a batlunat and sealing plastic bags), succesdul general-ization resulted. While food often seemed to risk satiation. theopportunity to work cen,a preferred task maintained motivationin this case. So, the motivating power of the offered reinforce-ment was perhaps mime determinant of successful generaliza-tion than the sequencing employed.

Another issue related to motivation is the presence ar absenc:e ofreinforcement during generalization probes. If generalizationtrials are not artificially reinforced and there is no reinforcementintrinsic to the generalization setting. generalization trials tendto promote extinction of the behavior. Meisel and Rinouver(1977) addressed this issue. They trained six autistic children toimitate a behavior. modeled by a therapist. when provided withthe verbal cue, "Do this." Each correct response was come:*'lated with food. All six children generalized the receptiveskill. However. without reinforcement the skill did not transferto the generalization setting. A noncontingent reinforcementschedule (NCR) was implemented in further maintenanceprobes. The first NCR recovered correct responses by the chil-dren. Further NCR probes failed to recover correct responding orresulted in low responding.

Therefore, while failure to generalize was reported in severalstudies. the reported data indicate that generalization may haveinitially occurred but was extinguished for lack of reinforee-meet. indeed, if there is no positive consequence for a behaviorin the natural environment. the behavior may not generalize ormaintain.

" 35

The Environmental Relevance of the Skill

It was hypothesized that the extent to which the training stimuliresembled stimuli in the natural environment was indicative ofgeneralization success or failure. Thus, a skill taught in a naturalenvironment would generalize more successfully in a naturalenvironment domestic, vocational. community) than a skilltrained in a contrived environment (e.g., therapy..school). Stokesand Baer (1977) suggested that, when training for generalization,the trainer "use stimuli that are likely to be found in the general-ization setting in the training setting as well" (p. 304). In conskl-eration of this issue, the studies were reviewed in terms of theimpact of the types of environments (natural and contrived) inwhich training and generalization probes occurred.

The above hypothesis was not fully supported by the studiesreviewed (see Table 2-5). The similarity between the trainingenvironment and the generalization environment. in and ofitself. was not found to be a major factor in generalizationresults. Hnsvever, the studies did not thoroughly test thishypothesis for the following reasons:

(1 )The sample size woe small (e.g., only one study examinedtraining in a natural environment only and probed in acontrived).

(2)Erivirownental variables within the studies were notnecessarily con:rolled.

Since the results of the studies reviewed neither proved ordiscisimed Stokes and Baer's position, two specific factorswithin the area of environmental relevance, discrimination ofstimuli and independence, were examined.

Discrimination Stbituli

To meet the demands of change, a child needs the ability toselect relevant stimuli from his environment. Here, sequencingduring training gained Islevance. It was hypothesized thatincreased opportunities to make appropriate discriminationswould facilitate generalization. Schroeder and Baer (1972) hypo-thesized that there wet a greater opportunity to compare inconcurrent training than serial training. In serial training, thesublect made one response repetitiously to a fairly constant stim-ulus or set of stimuli. On the other hand. in concurrent training.several items are trained simultaneously. thus the individualmust attend to the discriminating features of each task.

30

36

4C

r.

prOttifi

1,401PN V 16.13intar,

un1t griraltie9 %cot pstuadatt

alwAlad

slueposs pi

miquins

(domes tiossay sosuasapussogfigwomoV 0111Puldi uti

INIVITIPusule no wimosiimis eislassistwaLP Pedmq

ii 014'01

Three of the reviewed studies specifically compared serial andconcurrent training and related these sequences to generaliza-tion results (Palaver' & Hall. 1078: Schram ler & Beer. 1872:Waldo. Guess. a Flanagan. 1982). In an cases, concurrenttraining proved to be superior to serial training in facilitatinggeneralization.

However. concurrent versus serial training may not have beenthe only variable influencing the dhcrimination of stimuli. Forexample, Lovas', Kongo!. Simmons. and Long (1973) exandnedgeneralization across stimuli with 20 autistic children inbehavior therapy. Imitative language, meaningful speech. andsocial and self-help skills were trained concurrently. Thesebehaviors were trained as alternatives to inappropriate behav-iors (self-stimulation and echolalia). "To assess generalization.the children were observed in a room separate from, and notassociated with. the training situation and in the company of anunfamiliar adult. The room was equipped with toys (wagon.paper and crayons. a hobo doll. etc.)" (p. 139). The child'sbehavior in the new setting with the new stimuli was observed.In addition. the child's behavior at home was observed. Whenthe children stayed hoses with their parents who had learned thetechniques. they generalized the bthavior to that setting.Hower. the children who returned to a state hospital did notgeneralize to that setting

Though generalization occuned in scene cases, the reason for theoccurrence was not dear. Levees, at al. (1973) attributed thegeneralization success to the varied stimuli: however, anotherinterpretation is that parent training facilitated generalization.

Independency

It was hypothesized that those skills which allowed a steptoward independence in a natural environment wouldgeneralize more successfully than skills which left the subjectdependent on others in the environment. Performance of a skillwas considered independent if a subject could perform or main-tain the skill after instructional support was withdrawn. Sell-direction and self-initiation were two qualities considered repro-sentative of independent behavior.

A distributed sequence was found to be used with behaviorsleading to independent functioning more often than a massedone. Distributed sequences tenthid to involve a set of responses

32

eaii naturally cued by the previous response. For maniple.Mahone et al. f 39791 trained four retarded subjects to ride a busto a last-food restaurant. order food items, and pay for the items.The subjects sucxesslully acquired these skills oind generalizedthem to a new setting. The subjects were responsible for theirown behavior and independent of trainers. These skills allowedthe subjects to operate independently In their natural environ-ment.

hi contrast. many skills taught In a messed seifuenve left thesubject dependent upon another person. Clark and Sherman19751 trained three children to respond to questions about

pictures. The child did nut successfully generalize these skills tonew items. Performance of this skill was dependent uponpresentation of cards and questioning by an experimenter. It wasnot until after additional discrimination training that the chil-dren generalized the skill. Even so. the skill could WA beperformed independently.

Conclusion

The nature of the task. the environment. and the task's impactare factors which influence: skill generalization as much. if notMOD!, than the schedulina of the trials. Puthermore. this reviewof the literature found that the term "skill generalization" isused nebulously, is difficult to define, and is difficult to trainand

4.1.teralization is affected by deviation from the natural cues andmisequences present in the environment. It is important for a

%kill to be necessary is a natural environment. So. generalizationseems to he affected by time. setting, people, and the nature ofthe behavior. Stokes and Ham. f 1977) state: "Perhaps the mostdependable of all generalization programming mechanisms is...the transfer of behavioral control from.the teacher-experimenterto stable. natural contingencies that tan be trusted to operate inthe environment to which the subject will return, or already

f f .15:1).

Certain trial sequencing formats tend to contain factors whichinfluence the ability to generalize. However, across.the-boardimilusions about any specific format could noVbe drawn

without ismaidetation of trial sequence characteriatics.,Some ofthe best predictors of success are whether the skills taught arefunctional. whether they allow for self-initiation and indepen.

33

39

dente. and whether they are reinforced in a natural connect:From the studies, it was noted that distributed sequences tend tohave features important in promoting geninalization. Indeed, itmay be maned that trial sequencing itself does not ..Eject general-ization. but rather that the type skill trained should determineboth the type of trial sequencing used and the generalizatineresults. In the current literature. the impact of trial sequencingupon generalization is not an issue sully separable fromperhaps more significant, extra-sequence influences.

Itataremate

Bateman. S (1978). Application of Prensack's generalization on rein-forceinant to modify occupational behavior in two severelyretard individuals. American Punta, of Mental Deficiency. 79.904-510.

tIarlop. M. H. 11983) The effects of echo/she on acquisition andgeneralization of reca.:.iive labeling in autistic diddles. 104111101 ofApplied Behavior Analysis, 111-128.

Clark, H. N. & . , 1. A. (1974 Teachim generative use of sentenceanswers to three forms of questions. journal of Applied BehaviorAnalysis. 0.321.330.

Cam M. F... Vogelsberg, P. 1.. & Williams. W. (1981). Mc ta of classroompublic transportation instruction on generalization to the naturalenvironntent:gouned of the Association for Severely ffustdi-capped. N. 49,-53.

Cuvu. A I., Lent It. B.. it Bunko's. I.. S. (19711). Teaching lanitorial skillsto the mentally retarded: Acquisition. generalisation. and mainte-nance. journal of Applied Rahman. Analysis, I t. 345-355.

Genie. E. 11974). The training and generalization of a conversationalspeech form in nonverbal retardate*. journal of Applied BehaviorAnalysis, 7. 137-149.

CuriiN D a Baer. D. M. (1973). An analysis of Individual differences ingeneralization between receptive and productive language inretarded children. Journal of Applied Behavior Analysis. 8. 31)-329.

Hill. I . Wehztan. P.. & Horst. G. (1982) Toward generalization ofapt.tropriete Jaime and social Wavier in severely handicappedyouth-pinball machine use. Journal of the Association forSeverely Handicapped. 5.38-44.

34

40

Hung, 11 W. (1977). Generalization a-curiosity" questkming behaviorin statistic chilblain. foam& of Behavior Therapy and Experi-mental Psychiatry, 237-245.

Koeged. K. L. a Rincover. A. (1977). Research an the differences betweengeneralisation and maintenance in extra-therapy responding.Journal of Applied Behavior Analysis, 10. 1-12.

Lowers. a L. merit L. shamans, I. Q.. 8 Lon& J. S. (1973). Somegeneralization and follow-up inessures on autistic children inbehavior therapy. Journal of Applied Behavior Analysis. ti, 131 -165.

Lowe. M. L & Cum. A. I. 11979). Teaching coin summation to thementally retarded. Journal of Applied Behavior Analysis, it. 483.409.

Marholin. U.. Moak. K. M.. Totwlette. P. L. Berger. P. L. &Doyle. D. A. (1979). "111 have a big mac, french fries. targetcake. and apple pie."...or teaching adaptive communitykills. Behavior Therapy. 10, 238 -248.

Ate Le en. I.. F. & McLean. I. E. (1974). A language training program fornonverbal autistic children. Journal of Speech and Hearing Diem-dent. 39.105-193.

Mulligan. M.. Guess. 11, Hohmet. I.. & Brown. F. ( 1981). The individual-ited Lurriculum sequencing mode! (11: Implications from researchten massed. dyttribute d. or spaced trial training. Journal of theAssam. whim lathe Severely Handicapped, 5.325 -335. ,

Mulligan. M.. Lecy.L. & Guess, D. (1902). Meets of massed. distributed.end mated trial sequencing an severely handicapped students'performance. Journal of the Association for tilt Severely Handl-I lipped. 7.48 -61.

Hannan. M C. & Hall. V. (1978). Mitts of Redid versus anicumuit taskwiwinem,ing an acquisition. maintenance. and gesunalizetian.Journal of Applied Behavior Analysis. 11.67-74.

KilitintPt. A. & Karma K. L (19751. Setting generality and stimuluse.antral in outtalk children. journal of Applied Behavior Anal-ysts. 5,235-246.

Si bonnier. C L & Baer, D. M. (1972). Effects of concurrent and serialtraining on generalized vocal imitation in retarded children.Developmental Psychology. ft. 293-301.

35

41

Shalt. & Bucher. 11,(111110). Devekmment of spontaneous minding{ inlongues, deficient children. Journal of Applied Behavior Anal-

3. 523-521a

Stevem-Long. J. Schwarz. acquantirm andgenerelization compound ocatencechild. Behavior Therapy. 7. 307.404.

Stokes. T. F.. & Bser. 0.114. (1977). An implidt techookey at esneraliza.tom. *Imola/ Applied Behavior Analysis. l0.349 -387.

Guam D.. & Flanasen. 8. 119112). Effects of concurrent andeiriiittrabeisemp_seceptive labeling by seine* minded individ-uals. 'Gonad cif tlEittociotien-foriltaatrynorjekvidicaPPed.

White. aft, & Haring. Briceptimud Thachk% CaUfatheigOhio: BO sod Howell Company.

Zifferialatt. S. ht. Burton. S. M. Horner. R. & White, T. (1977). Eina:r-Whine generelkaition effects amuse autistic tinkling. journal ofAutism and Childhood Schisoithienia. 7.337447.

42

SAL_EttIONITORING AND SKILLGENERALIZATION: A REVIEW OFCURRENT RESEARCH

Kathleen A. Liberty

When settings, managers. contingencies, am sequences, cues. orprompts change. students often fell to display acquired behav-iors. This failure to perform under CirCUMANICati which differfrom training * been called a failure to aeneralize, and aunt;tutes a serioid threat to the functional impact of instructicealprowents. Consequently, current research efforts are directed atidentifying practical sohationa to the problem. Much of thecurrent newerch taint in genendbadion isdirected at modifyingthe training environment andlor instructiond procedures tofacilitate generalization. Another approach is to teach the bandi-capped person to control the variables that seam to affectgeneralisation. In most training situations, antecedents (cuesand prompts) and consequences (e.g., rd) are controlledby teachers. therapists. and other awegivms. If the handicappedperm can control those variables herself, the differencesbetween settings may be minimized. and ientenslization facili-tated. Self-control techniques. in which the individual controlsthese variables. offer hope far facilitating generalization.

The three most common self-control techniques include self-instruction (control of antecedsmts), self-reinforcement, and self-monitoring. Self-monitoring may be defined as the act ofcounting or otherwise registering ov.irtly the occurrence of thetarget behavior. The act of self - monitoring implies that the indi-vidual perform ors observable behavior which functionallyidentifies the completion for initiation) of a particular behavior.Such signaling is an integral component of antecedent andconsequent contra; as well. sham self-instruction implies thatthe individual determine when one behavior has beencompleted and another is to be initiated, and since self-rein-forcemeat implies Abet the individual determine when abehavior has been completed (Kezdin. WOO). Therefore. theunderstanding and examination of self-monitoring is an essen-tial step in the study of self-control.

43

The Act of Self4fenitiming

Self- monitoring involves (at a minimum) two overt behaviors:the target behavior (i.e., the behavior that is self - monitored) Auldthe act of self-monitoring i self. While there is obvious binnedin the effects of self - monitoring on the target behavior, the act ofself-monitoring itself has not been systematically investi-gated(Brigham, 1978).

One important sat of questions revolves around the instrunninta-lion of self-monitoring. Self-monitoring behaviors that havebeen utilized by subjects in studies in clede: marking tallies on apiece of paper with a pencil (Jackson & Martin. in press; Braden.Hall. & Mitts. 1971: Gottman & McFall, 1972: Nelson. Upinski. &Black. 1978: Zegiob, Klass, & Junginger, 1978); coloring squareson a piece of paper (Nelson, Hay. Devany. & Koslow-Green.1980): pushing a Minna on a computer (Brodsky. Lepage.

& Zeller. 1970: Epstein. Widener, & Maim. 1975):crossing off numbers on a form (Hayes & Caviar, 1980); inesahnja telegraph key (Hayes & Nelson. 1983); marking a " + " or a " "in squares on a form (Homer & Brigham, 1979); marking offsmarm on a form (Lipinski & Nelson, 1974) actuating a wristcounter (011endick. 1981): and actuatine a hand-held ftequencycounter (Willis & Nelson. 1982). Unfortunately, investigations ofthe relationship between the form of self - monitoring andchanges in the target behavior, the fam of self-monitoring andthe speed of training, the form of self-monitor's* and the relia-bility of self -mnitoring. and the Wm of self-monitoring and themaintenance of self-monitoring have not been reported.

Mother aspect of the act of self-monitoring that is of importanceis the timing of self-monitoring (e.g., after each behavior. beforeeach behavior, after several behavims, during an interval ofresponding). Although timing may influence the effect of self-monitoring (Nelson & Hayes, 1981: Nelson at al., MO: Albion.1983). it is often neglected in experimental reports.

Another important variable. relating both to the act of self-moni-toring and its effects on the target behavior is the reliability ofself-monitoring. Unfortunately. many investigators fail to reportreliability. When it is included, reliability is reported as aPearson product moment correlation, which indicates reliabilitycf roues in terms of level changes. but does not estimate actualwxuracy. For example. self-monitoring of 1. 2. and 3 wouldcorrelate perfectly from 1.0) with an observer's scores of 10. 20.

38 44

and 30. while accuracy calculated by traditkeel reliability esti-mate formulas (e.g., issmements + agreements + disagreementsx 100) would be 10%. In addition, reliabilities we pigmented asmeans across sessions =the across terbjectic such data do notdemonstrate how reliabilities may fluctueo, and make directcomparisons of the day to day changes in both self-mountainsaccuracy and the effects of self - monitoring impossible to deter-mine.

The reliability of subjects has ranged from .06 and .17 (Hayes& Cavior, 1980). to .52 and .80 (Lipinski & Nelson. 1976). to .90and .95 (Willis & Nelson. 1982). In a series of studies, Nelson.Hey. Devany. and Koslow-Green (1980) relented rellabliltles (esPearson product moment correlations) ranging from .07 to .79 forregular elementary school children in first, third, and fifthgrades. Accuracy increased across age for self-monitoring posi-tive verbalizations. and decreased across age for counting inap-propriate verbalizaticers. Reliability ,h3r handicapped subjectshas been reported as 80% and 87% (Horner & Brigham, 1979)and 96 to 99% (Jackson & Martin. in press). In tithae studies.however, investigators have reported that subjects who had beentrained to self-monitor were not observed to de so after trainingceased (Braden, Hall. 8 Mitts. 1971; Seymour & Stokes. 1976;Gottman & McFall. 1972; Zegiob. Klukas, & funginer. 1978:fames. 1981).

Accuracy of self-monitoring may be Unproved by both notifyingsubjects that reliability will be checked (Hayes & Horn. 1982)and by reinfon:ing accuracy Turlonvitz. O'Leary. &icommith. 1975: Fixen. Phillips. & Wolfe, 1972: Risley & Hart.1968). 1 Infortunately. specific positive reinforcement for self-monitoring may confound examination of changes in targetbehavior (e.g.. Epstein. Webster. & Miller, 1975: Epstein. Miller.

Webster. 1976). reduce the probability of the generalization ofself -m unitoring itself (Stokes & Beer. 1077). and/or establish self-monitoring as a behavior which...competes with the targetbehavior (Epstein. Websier>ti-M111 er, 1975). In addition to ques-tions on the relationship isetween effects on target behaviors andsell-monitoriNtserllibility. the issue is of concern because thereliability.-orself-rnonitoring is one measurable aspect of thesun cars of self-monitoring training. If. as proposed. self-moni-toring is to be used to promote cross-setting skill generalization.it is to be expected that self-monitoring will be trained. System-atic evaluation of the success of training methods will involvemeasures of the accuracy of self - monitoring itself.

45

Another focus of concern in studying self-monitorim is bowsublects were trained to self-monitor. Most often. researchtwshave used a combination of verbal directions. demonstrations,practice opportunities, reinforcement. sad error correctionprocedures teach self-monitoring, and training is usuallyconducted in a single session (Gardener, Class, St Cole: 1983:Haves & Nelson. 1983; Lipinski & Nelson, 1974: Mattel &Barnhart. 1981). Often experimenters report that additionlprompting. feedback, and reinforcement ant required to estab-lish or maintain reliable self-monitoring of the target behaviorfollowing training or in other settings (e.g.. 011andick. 1981).Unfortunately, experimental (kits on the acquisithm, mastery.maintenance. and generalization of the act of self-monitoring arenot presented in the experiinenhs1 literature. Therefore, ques-tions as to how to train the act of self-monitoring to produce reli-able self-monitoring, which itself maintains, and which gtmeral-izes across settings (and/or across target behaviors). remainunexplored.

The Impact of Self-Monftering on Tenet Behaviors

Self-monitoring was originally introduced into the researchliterature as a data collection mechanism, and was used inexperiments where the target behavior was of a private nature(e.g.. marital fights). or where external observers were obtrusive.or when long periods of data collection were required. However.researchers quickly noted that self-monitoring itself producedchanges in the target behavior. and self-monitoring became atherapeutic intervention strategy (Kazdin, 1974; Kazdin. 1980).The mall:wily of studies in the area of self-monitoring have beendirected at: (1) the effects of self-monitoring (as an interventionstrategy) on a target behavior (e.g.. Zahn & Bornstein. 1980:Homer & Brigham. 1979): (2) the impact of different variables onthe effects produced by self-monitoring (e.g., Kazdin. 1974:Nelson. Hay. Dummy. & Koslow-Green. 1980): and (3) a compar-ison of the effects of self-monitoring with other interventionsle R.. Jackson & Martin, in press: Hanel & Martin. 198e; Rosen-baum & 'Nehmen, 1979).

In general, when self-monitoring is used as an interventionstrategy. it has resulted in increases when the target behaviorwas defined as "positive" and an increase was a therapeuticgoal. and in decreases when the behavior was defined as "nega-tive." However. such effects have not been shown consistently.

4.46

Many variables which might affect thedegree of effect seff-mopi-toting has on the target behavior have been studied &

Horn. 1982. present a review of the literature in the area). Impor-tant issues include:

11) The direction of change desired kw the target behavior.In some studies. behaviors with a postive valence(increase desired) show more effects than behaviorswith a negative valence Millis & Nelson. 1982).However. behaviors with a motive valence showhigher levels of self-monitoring reliability. In general.behaviors with higher response strength and -axialsensitivity" show greater effects (Hayes & Horn. 1982).

i21 Overt or covert observations of sublects engaged in self-monitoring. Overt observations produce more accurateself-monitoring and may produce increasedeffects overcovert observation (Lipinski & Nelson. 1674).

13) Reliability of self-monitoring has not been associatedwith effects obtained: The relationship'between accu-rate self-monitoring and changes in the target behavior.ere ant established (Rosenbaum & Drabinan. 1979:Meuse .es).

13..i effect of self-monitoring on target behaviors has been.spared with other interventions. but it is usually applied

in a oniliinlion with either self-reinforcement or self-ic.struc-'ion (Kazilin. 1980). In general. self-control techniques havebeen as effective as externally-controlled interventions inchat ,ring behavior (Rosenbaum & Drabman. 1979. present areview of the literature in this area).

An examination of the methodology of many studies indi-i ates that the training provided in self-monitoring mayconfound examination of effects on the target behavior1 Albion. 1983: Nelson & Haves. 1981: Kazdin. 1980). Furexample. in some studies the subject is not told whichbehavior the target until baseline data have been collected.and then training in self-monitoring includes statements as tothe desirability of changing the behavior in a certain direction

Reline & Martin, in press: Hayes & Nelson. 1983:Gardner. Ones. & Cole. 19113). Subsequent changes in thetatted behavior may then be the result either of the self-moni-toring or of the goal statements during training (Albion.1983).

Another confounding variable is present when the subject isreinforced by the experimenter for self-monitoring le.g..

41 4 7

Fixen. Phillips. a Wolf. 1972: Coyos, Michael. a Martin.1979: Matson & Reinhart. 1981: Shapiro & Klein, 1980): theniinforcement is than an additional consequence forperforming the behavior (since sell-mcmitoring follows thebehavior). Any increase in behavior. therefore, may be aresult of the "extra" experimenter-delivered minfmcementand not lust self-monitoring (Epstein. Webster. at Miller.1975: Zegiob, Makes, it fungingra. 1978: Guyon. et al. 1979:Albion. 1983). Other possible confounding factors incite%very short experimental phases (e.g., 20 minutes). failure toreport reliability, and vague procedural descriptions (Albion.19b3). Despite these problems, however. self-monitming doesseem to produce chancel in the lathavior which is self-moni-tored.

Three major theoretical explanations for effects of self -moni-toring on target behaviors have been developed:

(1) Self-monitoring provides a feedback loop; the indi-vidual observes and records changes in her ownbehavior and then adjusts the target behavior(Kanter, 1990).

:2) Self-monitoring acts to clarify the relationshipbetween the behavior and its consequences;behavior chew is the result of the increasing an re-spondence gm the believer) of the relationship(Each lin. 1979).

(3) The events surrounding the Mine' and continuinguse of self-monittxring (e.g., training procedures.materials used to self-monitor) function as stimuliwhich cue environmental consequences, and thoseconsequences result in changes in the targetbehavior (Nalsce b Hayes, 1981).

However, until additional carefully designed studies accummu-late. these explanations must amain theoretical.

If self-monitoring is to be used to facilitate cross-setting general-ization of the target skill. the impact of self-monitoring onchanges in the target behavim me of import. Far more critical.however. is the value of self- monitorim in maintaining andextending behavior chances established in more traditionalmethods. For example. a (target) behavior may be acquired andmastered through traditional externally-amtraled instructionaltechniques: self-umbel skills may then be taught to the studentto maintain treatment gains or to skill gemealization. Insuch cases, the crucial issue is value of self-monitoring for

42

maintaining treatment gains during post-treatment periods innontraining settings.

&lanais,. investigators have expanded the study of self-moni-toring to include the use of self-monitoring to maintain targetbehaviors and to facilitate cross-setting generalization. Drahmen.Spitaliuk. & O'Leary (1973) taught a combination of self-controlskills (including self-monitoring) to a group of disruptive chil-dren in an elementary school. The self - control skills producedmaintenance of gains in the target behavior and generalized toother periods of the school day. In a replitatiun. Turkewitz,O'Leary, & Irottsmith 11975) produced similar results. but failedto show cross-setting generalization. Holman & Baer (1979)taught six chifdren to count pages of academic work in writingand math. Effects of self- monitoring were measured 64th during

the individual experimental sessions and in the children'sregular classrixim (cross-setting generalization). Following base-line, the children were taught to count each page completed by

mcw lug a bead on a wrist bracelet. A goal was established for thenumber of pages to he worked, and a white bead inserted on theneuter by the experimenter. The child was directed to try to get

to the white bead. During the self-monitoring phase*. the childvas praised for self-monitoring, both in the 1:1 situation and in

the classroom. and fm reaching the goal (i.e.. the white head).Similar goal setting and reinforcement were not present in base-

line. MI measurement of the effects of self - monitoring cm thetarget behaviors Om.. on task, off task, and disruptive behaviors)was I unfounded. although the authors reported changes in thedesired directions in both the experimental and the classroomsettings The act of self-monitoring transferred to the classroom

ii.e.. genericlizeill. with high levels of accuracy (reliability of sell-

tori ng during the study was 90% or better).

Stokes and Baer (1977) have reported that one method of facili-tating generalization is to establish 4 response that generates astimulus that will occur in both training and nuntraining situa-tions ("mediate generalizatitm''). Holman & Baer (1979) suggest

that the cross-setting transfer of effects on the target behaviorswas mediated by "the discriminative properties of the bracelet

lie .1 minter). which served to remind subjects (1) to work. (2) tocomplete a certain amount of work. (3) to measure ongoing prog-

ress. and 14) finally to seek reinforcement from the teacher for

having done so" (p. 4142). Holman and Baer (1977) conductedtwo follow-up measures. for months 2-5 and months 7-10following the conclusion of the study, anti self-monitoring wasdurable throughout, although some cif the gains made in target

43

49

behaviors were not maintained. Increasing attention to self.moriitoring itself may he a function of the promise of self-moni-toring and other self-control skills for promoting crosseettingskill generalization (ICazdin. 1980: Holman & Beer. 1979: Baer,Holman. Stokes, Fowler, & Rowbury. 1981; 1"urkewitz, O'Leary.& Ininsmith. 1975; Kurtz & Neisworth. 1976). Further studies ofthe use of self-monitoring to maintain and generalize skilledbehaviors may help to identify not only the effects of such self-control training. but also the durability and means of any gains,Such questions are of Tmmedlate importance and value in anyevaluation of the potential of self-control techniques forpromoting skill generalization.

Self - Monitoring bySeemly Handicapped Individuals

Severely handicapped subjects have rarely participated in inves-tigatiomi of self-monitming, although self-monitoring has beenused as an intervention agent in studies with moderately andmildly retarded subjects (e.g.. Gardner, Clues, & Cole. 1983:Nelson. Lipinski. a Black. 1978; Zahn a Bornstein. 1980; Zegiohet al.. 1983; Shapiro & Klein, 1980). The effects of self.moni-toting as an intervention agent with mildly and moderatelyretarded workers have replicated results with nonhandicappedsubjects. Unfortunately. confounding effects of externally-controlled reinforcement and training procedures. as well asprocedural questions of training, instrumentation, and relia-bility of self-monitoring, limit the applicability of such studiesto questions of major importance in the study of self-monitoringas a method of facilitating skill goner; ligation.

In addition. studies have failed to document the reliability ofself-monitoring. the efficiency of the training methods used, andwhether the act of self-monitoring itself has maintained orgeneralized. Although the ad of self-monitoring itself may betreated like other behaviors trained. specific problems in themethod of training used may not only confound effects of self-monitoring on target behaviors te.g, if self-monitoring is rein-forced). but inhibit generalization and maintenance of self-moni-toring with other behaviors. In addition. the act of self-moni-toring can not be trained in isolation. The subject must betrained to <,..ount something. If the target skill is also beinginstructed or intervened on. instruction in self-monitoring mayinterrupt that process. Using traditional instructional

" 50

prucedures for both behaviors may then result in confusion, orin problems with both the target behavior and the instructedresponse of self-monitoring. Finally. specific studies relating tothey use of self-monitoring to facilitate cross- setting generaliza-tion of target skills have not been reported. The promise of self**intro! strategies for facilitating generalization makes it neces-sary to investigate with greater precision how to most effectivelytrain self-monitoring. the maintenance of self-monitoring. andthe transfer of self-monitoring across settings and acniss behav-iors. as well as to clintinue to study the effect of self - monitoringon behavior and the transfer of those effects to other settings.

A series of studies has been conducted to probe specific ques-tions relating to the use of self - monitoring to facilitate skillgeneralization with severely handicapped individuals (seeLiberty & Fatah. lima: Liberty & Meth. 1983b: Liberty. 19tr3).

The purpose of the first study was to examine the acquisition ofself-monitoring by a lieverely handicapped student through anacinilani I. training procedure. which eliminated theonfounding effects of other training procedures while permit -

trnat integration into ongoing instruction. Measures of theindependent use of self-monitoring and the reliability of self-monitoring were used as indices of at quisition. In addition. the*atm ts of spit-monitoring on the target behavior were examined.

The subject was Pi years and five months old. He attended aset unitary program for severely handicapped youth. located in at

high school. The experimental design consisted of a base-line phase. two training phases. and a contingency changephase the hawhilt pro's ided a measure of the rate of production(it the sublect During the training phases. the subject was taughtto at tliate a t limiter placed .m the table next to his work. Thetraining prig pitore consisted of interrupting work (whichifeia rerillon .ement available for work) fur nonactuation ofthe * minter. which was then prompted by gestures (no verbaldare teals gic cut. Delay of reinforcement was avoided by self-monitoring. No spec ific reinforcement for self-monitoring wasphilt,141Pit Iii the WI and training phase. the subject wore the«meter tni his wrist In the last phase. a general contingency forbehavior was introduced.

The .ntielance training procedure pnaluced rapid accelerationittlialatipttql find very reliable self-monitoring. Self-moni-

toring itself was positively reinforced. presumably by the,sudible i.lic.k of the counter and the display mil-over. Self-moni-

5 145

luring produced changes in the target behavior (rate of produc-tion) bidet:mina that self-monitoring acted as a (self-controlled)positive reinforcer sufficient to maintain performance of thetarget behavior. A similar effect was shown by Nelson at(ism. who found that first and fourth graders recorded higherfrequencies of the target behavicor when the behavior of self-monitoring was coloring squares with a crayon. The authorsspeculated that the subjects engaged in higher frequencies of thetarget behavior in cyder to color squares (self-monitoring), andcolored "extra" squares as well. Holman and Baer (1979) alsofound that subjects prompted teachers to praise them for self-monitoring when they reached the white bead.

Investigators /sporting studies in whichself-monitoring aloneproduced i in target behaviors have speculated thatsuch increases y be attributed to (1) the nature of the training.(2) additional. exter nally-controlled. reinforcement for self-monitoring, (3) motivational . properties c if self-monitoringpmduced by covert self-evaluaticm, or (4) that all of the eventssurrounding the act of self-monitoring and training in self-moni-toring function as stimuli that cue envionnumtal consequencesthat. in turn. effect behavior change (Nelson ik Hayes. 1981). Inthe current study. the hillning procedures for self-mon:tiringexcluded any statements about desired changes in the targetbehavior and avoided any additional externally-controlled rein-forcement. It is also unlikely that the subject, who was unable tocount or recognize numerals, in covert comparisons ofthe numeral counter displays from session to session (self-evalu-ation). As an alternative. it may be hypothesized that thebehavior of pushing the counter, accompanied by an audible"dick" and a changing display. acted to reinforce not only thebehavior of self-monitoring. but the target behavior as well.

The purposes of the second study were to:(1) Examine the maintenance of self-monitoring skills.12) Determine if self- monitoring generalized within stim-

ulus classes and across responses. and if so, the extentand nature of such generalization.

13) Examine the Affect of self-monitoring on the targetbehaviors.

The same subject participated. The first study had concluded343 days prior to the start of this study; during the interveningperiod. the std.*t had not worked on production or self-moni-tored

de

52

This study incorporated a multiple baseline design across twobehaviors. one performed in the classroom (production) and one

in a cafeteria work settong (assembly of sack lunches). Thephases for production were Baseline I. Self-monitoring I. Base-line II. and Self-monitoring II. The phases for bagging sacklunches were Baseline 1. Opportunity to self-monitor (intro -

clih simultaneously with Self-monitoring f of production),Thin to self-monitor. and Baseline IL

Production did not maintain at a commensurate level doting the343 intervening days. Self-monitoring did. however, maintain athigh Levels of reliability and independence. Production wasconsisluated by instructor-controlled reinforcement. while self-monitoring was consequated by subject-controlled reinforce-ment (counter click and display roll-over). Behaviors that areconsequated by self-contmlle.d reinforcement are more resistantto extinction than are behaviors cimsequated by externally-controlled reinforcement (Weiner & Dubanoaki. 1973), andperhaps that is why self-monitoring maintained at higher levelstikes did production.

The sohjei t did not actuate the counter. even when he had theopportamtv to do so. and therefore training in self-monitoringwas introduced. .triti .s.curate self-monitoring acquired in foursessions. The independent and reliable self.numitoring ofpouluttion did not transfer to self -nuniitoring of bagging. In this

study. the two behaviors were not of the same response class.and wen. performed under conditions (e.g.. supervision, stim-ulus materials. setting. time of day( totally unlike one another.Transfer rriry have (simnel' if the settings were more alike; forexample. it the subject had counted bags in his classroom. withthe same supervisor present.

However. wearing the wrist counter (but not actuating it)produced i !Unseal in bagging rata that was of practical signifi-tante to the subtect ifrom 41)% to 67.5% of normal. with a highrate of 114%7 the ruinnal rate), and high rates continued as self-monitoring was trained. Ow* the opportunity to self-monitorwas renuwed from prink:LW/lc bagging rate dropped. When theopportunity to self -umnitor bagging was withdrawn. baggingbegan decelerating.

Similar effects have been reported in other studies (Braden. et

.11 . Coffman h Mi.FaH, 197Z; Zegiob, of 01.. 19781. Thereseveral hvir)theses for the effects of self-monitoring when

the da,t of self-srsonitoring is itself not observed. The wrist

47 53

counter could serve as an SD for faster work. or it may act as areinforcer by mediating differences between settings. Stokes andBaer (1977) have reported dun one method of facilitatinggeneralization is to establish a response (in this study, self-moni-toringj. that summates a stimulus (the counter) that will occur inbath training and nontraining settings. In the present study,externally-contrOled reinforcement was more frequent duringproduction (VI 1.5' vs. VI 5') sad any self-controlled reinforce-ment exerted by the act of self-monitoring was also morefrequent during production. since pages were finished morequickly than sacks. These data support the use of seffinoni-toting to mediate cross-setting generalization. The counter,which was worn during more frequent reinforcement (duringproduction), may have acquired generalized reinforcing proper-ties that mediated generalization of faster work rates acrosssettings which differed in reinforcement schedules.

The purpose of the third study was to extend and replicateresults with different behaviors and with a different subject.Specific issues examined included:

(1) The effects of wearing a wrist counter on two behaviors.one instructed and the other uninstructed, prior to anytraining in the use of a counter. and whether effectsproduced in the training setting would transfer to theprobe setting.

(21 The effects on the instructed and uninstructed behav-iors of training self-monitoring of the instructed)ehavior. in both the training setting and the probesetting.

(3) Whether self-monitoring would transfer from thebehavior on which it was trained (i.e., the instructedbehavior) to another behavior (i.e.. the un;:istructedbehavior).

(4) Whether self-monitoring would transfer from thesetting in which it was trained to the probe setting.

The subject of this study was an eleven-year old girl (i.Q., 30).attending a special school for handicapped children. Thesubject's regular classroom served as the training setting. Fromone to four times per week. the subject attended another class-room in the school for thirty minutes each time. This classroomserved as the probe setting. Two target behaviors were measuredin both settings: two-word responses to questions. and two-wordinitiations. Two-word responses were differentially reinforcedin the training setting on a fixed ratio of 1: %Oita in the probesetting, they were reinforced on a variable schedule of about 1 to

48

54

8. two-ward initedions were rainfonied at approximately thesome ratio Wit 3) in both seldom.

Following baseline observations. sessions in the train* settingwere divided into two approximately equal periods. The siftictwore the counter during one of the two periods each day. In thethird phase, baseline conditions were rOustituted. In the fourthphase, the subject was trained to seff-monitor two-wordanswers. using an avoichince training procedure. No training inselknonitaring was provided in the probe setting. Instructors inthat netting were asked to ignore the wrist counter. In the fifthOuse. baseline =dittoes Weft reinstated.

Wearing the wrist counter without any training in self-moni-toring did not affect performance. Once self-monitaring wastrained. however, performance was affected dransaticall7. Theact of self-monitoring generalised across behaviors and acrosssettings for both behavior Self-monitoring improved cross-setting generalization of two-word answers from 19% to 49%performance deteriorated quickly once the opportunity to self-monitor in the probe setting was withdrawn. Seff-monftoringdid not affect cross-setting perk:4=mm of two-word initiations.Prediminery analyses of the result siegpot that self

ully mediated the differences in - " " betweenprobe and training settimg when there were no thermals, self-made:wing did not affect performance. Further, the Wed isdirectly linked to the self training. during which thecounter acquired discriminative len which functioned asself-controlled reinforcement in , probe setting.

The results of these studies suggest first that self-monitoring mayeffect a champ in a target behavior in the training setting. underUMW circumstimces. with severely handicapped subjects.Effects are not dependent an the accuracy of self-monikaing.These data support the findings with mildly and nonhendi-capped subjects. These studies also indicate that the avoithincetraining procedure can produce rapid acquisition of accuratesell-monitoring, which is maintained. and which can generalizewithin response classes, but not across response (kisses. andwhich can pneralize across settings. Third. these studiessupport the use of self-monittaing to mediate skill peterzlizationwhen the reinforcimient contingencies differ across settinp.Although these studies are limited to two subjects, results aresimilar to those reported with other groups of subjects, butexpand especially on the information on the use of self-moni-toring to facilitate skill generalization. Future research with

4,55

addliknald sewn* headkaPPed subtle:le will be conducted totest the results and hypothesis.

ihdereswee

Maim. F. 9184 A mothadologiad analysis of aelf-control in appiledsettings. ilehaviand Diffoniers. OP& 87-102.

Baer. D., Nohow. J.. Stakes. T.. Fowler. a, T. (1981). Uses ofMI-control technianes in peogrunnaing issBaen. in S. Bijouand B. Rule (Rds), Behavior libehtketion: Contributions to Educa-tion. Hillsdale. NJ Lamm* Basun Associates. Publishers.

lirkilima. T. (1979). Selteautsol. In C Catatonia T. Brigham (eds.).liancthesok of applied behavior analysis. New York: irvhaManPublishere.15.474.

Brehm, Hall. LL. (1971). The affect of self-emording an thedoss-room barrier of two aighthipade soulams. journal ofApplied Beteaview Analyvhs. 4,191499.

Brodsky. G., LePage. T., Qidring. j.. beer. R. 111178). Salf-evahaativeresponses in adolescent retardate& American founts' of MenialDeficiencyinte

R.. Mossy. D. 11974). Teaching self-control todisruptive children. Purnell of Abnormal Psychology. 52(1). 10-16.

Epstein. I... Miller. P.. Webster, j. (1976). The effects of reinfreditsconcurrent behavior m self-madtoristg. Behavior Therapy. 7.49-05.

Epstein. L. Webster. J., & Mac P. 111711). Accuracy end embroilingeffects of self-moat eming as a function of concurrent respcmdirmand reinhecement.lithevior Thenpy. 5.854-588.

Men. D.. Philips, R.. Ma M. (19111. Achievement plasm The relia-bility at .and peer and their effects onbehavior. Journal of Applies! Behavior 5.19.30.

Gardner. W., Cleo& T.. Cele. C. flan Soff-olool of dhouPdoowithal ruminations by a mentally retarded edult. AppliedAmmoch in biessul Retardigion. 4(1). 41-58.

50

56

Cottesen, & McFall. R. (1972). Sat effects in a peagiam forpotential high school dropouts: A time series analysis. faunal ofConsuMag and Clinical Psychology. 399.273.291.

Gorse. A.. Michael. J.. & Martin, G. (1979). Seff-recording training toteach retarded adults to minion* west behaviors of retardedclients. Behabilitatism Psychokigy. 26(4), 215 -221.

Hanel. F.. & Martin. G. (1UO). Self-eatraitoring. oaf oftaken reinfixcernent. and goal-setting to im odes withretarded clients. International journal of flessanth,3.505-517.

Haves, S.. & Caviar. N. 11980i. Multiple tracking end the reerity ofself-monitorialp U. Positive behaviors. Behavioral Assessmead. 2.283-298

Hayes. S. A Marin. R. (1983). Similar reactivity produced by externalcues and self-monibning. Behavior Modification. 712).183 -199.

Heiman. j.. A Baia. D. (1979). Facilitiding geworelizrAion of on-taskthrough self-einnitoring of academic tasks. hairnet of Autism midDevelopmental Disorders, 014). 429-445.

unser, it.. i Brigham. T. (1979). The effects of self-manitorimprocedures on lisp seedy behavior al two retarded chits's.=Education and Training of the Mentally Retanbrd. 14,18-24.

lai.ksors. D. & Martin. G. (in press). Additive affects of compaaents of aself.control package fax improving the production of mentallyhandicapped workers, journal at Practical Approaches toDevelopmental Harsiiirap.

Mines. T (1981). Self-monitoring of stuttering: Reactivity and accuracy.Rehavnw fiesearch and Ilse:pay. 19. 291-296.

Kanter. F. (19701. Self-monitoring Methodological limit:dimsand din-ital applications. Immo! of Consulting and Clinical Psycholopy.35.148-152.

Kaolin. A. (1980). Reactive seffosanitorinsc The effects of responsedesirability goal setting and feedback. Journal of Consulting andClinical Psychology. 42. 704-715.

Karlin. A. (1980). Behavior atodflicatkIn in applied settings. Home-wood. U.: The Dorsey Press.

Kim. P.. & Neisworth. (1976). Self-control possibilities for excep-tional children. Biceptioned Children. 42.212 -217.

315 7

Liberty. K,. & Past. M. (1993). Sali-nroallating I: The acquisition el self-nionftorkm and its enact on the rale of a severelyhandicapped Waimea lastrinte adoration ef severely hand-icapped chihken: Washington Rmearch Annualreport FY 82-83. (U.& Department of Ede n, Contract No. 308in-asst.) seeds: Ussienefty ef weehiestem itiverheeetal MacHon Una

Liberty. K., & Perth. N. (1983). Self-manNoring 4 Maintenance andlenerallestion of daring and its affects on two =Vetbeiewiras. hunixes for school= 4 severely handicapped chil-dren: Watinglen Research Organkratien: Annual report FY 82-in. N.S. Department of Woodier'. Contract Na. 300.82-0634.)Sean* Univeraty at Washinglan, Experimantai Education Unit.

Liberty. K. (1963). Seireoniturbilicerai self-monitoring trainingon expressive communication: Mediation across settings. Instautefor education of severely leasdicappod children: WashinetenResearch Ormakadion: Annual repel FY 83-83. (U.S. Depart-

Contract Na. 3011412-0934.) Seetthn UnlynniffYof WasMngton. Experimental Education Unit.

Lipinski. D.. a Nelsen. R. (1974). The reactivfty and unreliability of seW-monierring. af Cremating and Clinical Psychology. 42.118-123.

Metros, j.. & Renthart. T. (1981). Previewing treatment effects to thenatural envirennient: A precedent for training Instintikmalleedretarded adults. Behavior MedVication. 5. 27-37.

Rehm. R.. Hay. L.. Delany. 1.. &Koslow-Crean. L (19901. The reactivityand accuracy al chlidrerfs self-moniterimp Three experimards.Child Behavior Therapy, 2(3). 1-24.

Nelson. R.. & Hayes, S. (1091). Theoretical explanations tor reactivity inself-manitaring. Behavior Modification. 3.3 -14.

Nelson, L. Lipinski. D.. & Black. ). (1978). The reactivity of adult retar-dates' sell-mossitorins A commis= among behaviors ofdifferent valences. and a comparison with token reinfercreemst.The Psychological &card, .

011endk.k. T. (11181). Self-monftoring and seltadministered overcome-non: The modificatkm of nervous tics in children. Behavior Modi-fication, 5(1). 73-84.

Racblin. H. (1974). Sen-control. Behavioriate, 2.94 -107.

Risley. T., & Hart, W (1968). Desolopirie correspondin betweennonverbal behavior of pr*achool children. fountal of AppliedBehaves Analysis. 1.267.281.

ig

58

Rosenbatun. M., & Drabinan. R. (1979). Self-control training in the class-room: A review and unique. Jounna of Applied Behavior Anal-pus. 12.467 -485.

Rosins, L, & Martin. G. Ph press I. Sett -man tierment training to decreaseundesirable behavior of mentally handicapped adults. Rehabilita-tion Psychology.

Seymour. F., & Stokes, T. (1976). Self-recording in training girls toincrease work and evoke staff praise in an institution for(Menden. Journal of Applied Behavior Analysis. 9.41 -54.

Shapiro. K. & Klein. R. (1980). Self-management of classroom behaviorwilt, retarded/disturbed children. Behavior Modification. 4(3). 83-94.

Stokes. T . & Beer. D. (1977). An implicit technology of 'generalization.Journal of Applied Hehavior Analysis. 10.349 -367.

Turkewitz, H.. O'Leary. K.. & Ironsmith, M. (1975). Generalization and1114intimance of appropriate behavior through se)f-control. Journalof consulting and Clincial Psychology. 43(4). 577-583.

Weiner. H . & Uuhenoiki. R. (1973). Resistance to extinction as a func-tion of self- or externally determined schedules of reinforcement.Journal of Personolstv and Social Psychology. 31.905 -910.

Wills. S & Nelson. R (1982). The effects of valence and nature of targetbehavior on the accurac:y and reectivity of self-monitoring. &tray-omit Assessment, 4. 401-412.

Zegiob. L Klukas. N . & lunginser. I. (1978). Reactivity of self -mom-luring procedures with retarded adolescents. American Journal ofMental lrefo:irnuy. 83. 156-163.

Zohn. . & Hornstein. P (19011) Self-monitoring of work performancewith mentally retarded adults: Effects upon work productivity.work quality. and on-task behavior. Mental Retardation. 18. 19-25

5953