1992) 259, NUMBER 2 (summer 1992)

TEACHING SOCIAL SKILLS TO STUDENTS WITH AUTISM TO INCREASE PEER INTERACTIONS INANINTEGRATED

FIRST-GRADE CLASSROOM

DEBRA M. KAmps, BETrY R. LEONARD, SUE VERNON, ERINP. DUGAN, AND JOSEPH C. DELQUADRI

UNIVERSITYOFKANSAS

AND

BETH GERSHON, LINDA WADE, AND LOUISE FoLK

SHAWNEEMISSION PUBUC SCHOOLS

Weinvestigatedthe useofsocialskillsgroups tofacilitate increased socialinteractionsforstudents with autism and their nonhandicapped peers in an integrated first-grade classroom. Socialskills groupsconsistedoftrainingstudents andpeers in initiating,responding, and keepinginteractions going; greeting others and conversing onavariety of topics; giving and accepting compliments; taking turns andsharing; asking for help and helping others; and including others in activities. Training occurred duringthefirst 10minof20-minplaygroups, fourtimesper week. Using a multiple baselineacross subjects design, resultsdemonstratedincreases in the frequency of, time engagedin,and durationofsocialinteractions,aswellastheresponsivity ofstudentsandpeers to each other. Results were maintained when studentsweremonitored andgivenfeedback onsocial performanceinplaygroupsandduringfollow-up.

DESCRIPTORS: autism,socialskills, integration,peersocial interaction

Studentswith autismexhibitamultitudeof

aso-cial and antisocial characteristics (Autism Society ofAmerica, 1990). By

definition,

appropriatesocial

behavior implies positive or atleast functional

in-teractionwith others. Consequently,therehasbeen

an increase inresearch investigations thatuse peer

strategies (e.g., modeling, prompting, tutoring) as a vehide for increasing learning and improving

social relationships of students with autism and

otherdevelopmental disabilities (e.g.,Carr& Dar-cy, 1990; Charlop,

Schreibman,

& Tryon, 1983;

Kamps, Locke, Delquadri, & Hall, 1989; Odom & Strain,

1986;

Sasso, Hughes, Swanson, &

No-vak, 1987; Shafer, Egel,&

Neef,

1984).

Inaddition totheeffortstoincreasesocialskills

and interactions amongchildren with autism,

fur-ther consideration must be given to the effects of

This research wassupportedbyGrantH023C00024from theU.S.DepartmentofEducation.

Address correspondence and reprint requests to Debra Kamps,Juniper Gardens Children's Project, 1614 Wash-ingtonBlvd.,KansasCity, Kansas66102.

trainingin socialgroups andintegratedschool en-vironments (Brown, Ragland, & Fox, 1988;

McEvoyetal., 1988).These issuesareofparticular

concernforhigh-functioningstudents with autism.

Thissubgroupisoftenofnormal intelligence;thus,

expectationsarethatsocialskillsareatnear-normal levelsaswell.Yet manyhigh-functioningstudents with autism are typically quite ineffective socially. Like low-functioning children with autism, they

oftendesiretobealone, adhere rigidlytostructure

and schedules, are disinterested in others

(partic-ularlypeers), and perseverate on irrelevant objects

ortopics.

This study investigated the use of social skills

groups for high-fanctioningstudents with autism toimprove social interactionskillswithpeers. This

study differs from previous work in several ways.

Studentswere in anelementary school, totally

in-tegratedinto aregularfirst-gradedassroom. Treat-mentwas neither target- norpeer-oriented but was

conductedin agenericfashionforall group

mem-bers. Thus, treatment wasconsidered tobe part of

the first-gradecurriculum. Social skillsinstruction 281

was implemented using portions of published cur-ricula (McGinnis & Goldstein, 1984; Walker, Hops, &Greenwood, 1988).Theseprograms have

had limited testing among students with autism (Sasso, Melloy, & Kavale, 1990), with primary

investigationsconducted for students with behavior disorders and other mild disabilities (McConnell, 1987; Schloss, Schloss, Wood, & Kiehl, 1986).

METHOD

Subjects and Setting

Three malestudentswith autismparticipatedin thestudy.Allstudentswereconsideredhigh

func-tioning, asindicatedbyintellectualcapabilities, lan-guageskills, and academicperformance, but were

lacking in social skills (e.g., few interactions with

nonhandicapped peers, frequent periods of

isola-tion, limited spontaneous speech, and few initia-tions ofconversations). The first-grade dassroom

consisted of 11 nonhandicapped students, 3

stu-dents with autism, 2 students with physical dis-abilities, thefirst-grade teacher, and a

paraprofes-sional.

Mike was 7 years 5 months old, with an intel-ligence scoreof101 andfirst-grade academic skills.

Languageskillswereappropriate; however, he also used rote or"canned" phrasestocommunicateand

perseverated on inappropriate topics. Mike often

whinedandcomplainedin a loud,disruptivevoice

whenchanges in routineoccurred, andpreferredto

be alone,

frequently

ignoring initiations

by

peers and adults.

Adamwas 7years 9 monthsof age, witha full scaleintelligence score of71. Language character-isticsindudedfrequentinitiationsofirrelevant top-ics with perseveration on the weather, dates,

ge-ography,andsoforth. He

performed

at

first-grade

level, but had some difficulties in

comprehension

and taskcompletion.Adam

responded

to

questions

andinitiations,

although

hespentmoretimealone than withpeers duringgroups and at recess.

Billwas 7 years 7 monthsold,with afull scale intelligence scoreof70. Deviant characteristics

in-duded perseveration on irrelevant

topics,

imma-turity,severemoodswings,andaggressionathome.

Billtypicallychose to be alone; however, he would remain in physical proximity to peers.

Dependent Measures

Socialinteraction. Observations were conduct-ed during play group sessions to determine the frequency of, time engaged in, and duration of social interactionsamong targetstudentsand their

nonhandicapped peers. Data were collected using the SocialInteraction Code developed by Niemeyer and McEvoy(1989) and NEC 8300 laptop com-puters using 5-min samples taken in random order. This code is a computerized system that records

social interactions with initiations, responses, and duration of interactions as variables. Initiations weredefined as motor or vocal behavior (e.g., shar-ing, assistshar-ing, touching), dearlydirected to a peer, that attempted to prompt a social response.

Re-sponses were defined as motor or vocal behavior

that acknowledged a replyto an initiation within 3 s. Interactions were thus defined as reciprocal

social behaviors that occurred as a result of an initiation-response sequence.

Social skills behavior. A 21-item social

skills

ratingscale was used to measure theoccurrenceof particular social behaviors by targets during the play group sessions. Items were scored using a 3-pointLikertscale(1 =never orseldomoccurring, 2 = sometimes, 3 = very often). Fourteen ofthe itemsmeasuredsocialskills, indudingsharing ma-terials;seekingorofferinghelp; initiating, listening,

andresponding; interacting with one or more peers

(for

extendedperiods); giving andaccepting com-pliments; accepting peer ideas; and inviting others to join in activities. Seven of the items were

de-scriptive ofappropriate behavior and responsiveness during the group, induding being nondisruptive,

having apleasant demeanor, not displaying autistic

behavior, displaying appropriate behavior (no whining), stayingingroup, and following

instruc-tions.

Interobserver agreement. Interobserver agree-ment was assessedby conducting twoobservations concurrentlyduringthe5-minsamples for 17% of

theplaygroup sessions. Agreementwas

computed

by dividingthenumber ofagreements

by

the

num-ber ofagreements plus

disagreements

and multi-plyingby 100. Agreementpercentagesfor the

fre-quency of social interactions ranged from 33% to 100% across conditions, with an overall mean of

87.5%. Agreement percentages for the time en-gaged in social interactions ranged from 25% to

100%, with an overall mean of 80.6%.

Reliability for the social skills rating scale was

assessed by having observers

independently

com-pletethe scale for 32%ofthe sessions. Agreement was computed by scoring plus or minus for each

item, then calculating the number of agreements

divided by the total of agreements and disagree-ments. Reliability ranged from 52% to95%, with an overall mean of73%.

Experimental Design and Conditions

A multiple baseline design across

subjects

was used to evaluate theeffectivenessof the social skills

training.

Baseline.Baselineconsistedof20-minplay

ses-sions four times per week during which a target

student and3

nonhandicapped

peers wereprovided

with two activities(e.g.,

making

partyhats, creating collages, dressing up, playing house,

making

pup-pets, or playing games). The entire dass partici-pated in the play groups; thus, four play groups

participatedsimultaneously

during

the 20-min

ses-sion. Instructions were given to stay in thegroup, be polite to

friends,

and playwith the

designated

activities. Noprompts weregivenduring

baseline,

exceptto returnifastudent leftthegroup.Baseline

lasted 2 weeks for Mike (eight sessions), 4 weeks forAdam (16 sessions), and 6 weeks forBill (24

sessions).

Socialskillstraining. Trainingwasconducted for individual groups during the first 10 min of the 20-min sessions.

Specific

skills were selected frompublished curricula(e.g., McGinnis& Gold-stein, 1984; Walker etal., 1988)and

induded

(a)

initiating an interaction, responding to initiations,

andkeepinganinteraction going;(b)conversations, greetings, and topics; (c) giving and accepting com-pliments; (d) taking turnsandsharing;(e) helping

othersandaskingforhelp; and (f)

induding

others inactivities. Trainers werethe firstauthor,the

dass-room

teacher,

the

learning

disabilities

teacher,

and the dassroom

paraprofessional

for

Mike,

Adam,

and Bill

(with

their

peers)

and a fourth group,

respectively. Social skillstraining lasted for 2 to 3 weeksperskill.

Feedback

for

social skills. This conditionwas

implemented

following

the completion of social skills

training

and consisted of 20min of free

play

withteachermonitoring

(i.e.,

astarwas placed

by

thestudent'snameon amonitoring formifhewas

engaged

insocial interactionat 1-min checks) and

feedback

(i.e.,

students wereshown the formonce

during

the session and again at the

end).

Follow-up. Inthefinalmonthof the school year (1 month after the feedback condition)

free-play

groupswereheldtwotothree times perweek. The

classroom teacher restated the social skills at the

beginning

of

play.

Monitoringwasreducedtotwo

5-minchecks for social interaction per group, with

feedback

provided

to thewhole dass. RESULTS

Results indicated improved social

performance

for targetstudentsand peers. Positive changeswere

noted for social interactions among students and social skill behaviors.

SocialInteractions

Thefrequencyof socialinteractionsinvolvingthe

3 targetstudents (i.e., as initiator or responder) is presented in Figure 1. The average numbers of social interactions per 5 min during baseline

con-ditions were 0.6, 2.7, and 1.5 for Mike, Adam, andBill,respectively. During social skills training, the average frequencies were 9.5, 7.4, and 6.1.

During the last half of the play group (posttraining

probes), the average numbers of interactions were 3.5,4.8, and5.6,respectively. During thefeedback andfollow-upconditions, interactions occurred on the average of five to six times.

Theduration of socialinteractions involving the target students is depicted in Figure 2, with in-creases following implementation oftraining and feedback conditions. Thebaseline means of dura-tion per 5-min samples (300 s) were 9.8 s, 30.3

20-Bse- Social Feedback Follow-line Skills Up 15 . A

A~~~,Tralning'

4= Probes 10 5

-IMIkel:

20 Base- Social Feedback i

Follow-line . Skills . up 'ID z 0 F 15 I -10 UA cc ~~~~~~~~~Adam: 0.

20 Base- Social Feed-

Follow-line SSkills bock up

15 10 5 -0 . 10 20 30 40 50 60 70 SESSIONS

300 ase-i Soolial Feedbaok , Follow-line Skills Up 250 Probes ME 200 150 Training 100 0A 50 Z

IMike]:

C) 0

U. 300 Base- I Sooiai Feedback I

Foiiow-Figure2.lTime engagedin s i iisd an p

Z 250-0 200-z

0&

1 50-z 0

co

10 Ix

j~~~~~~~~~~Adami:

Z 300 Base- ooiai Feed-

Foiiow-iine *Skilis *back up

250 200-1 50-1 00 50-10 20 30 40 50 60 70 SESSIONS

Table 1

Duration, Initiation,andResponseRatios for Baseline (BL),SocialSkills(SS), Feedback forSocialSkills(FB),

andFollow-Up(FU)

Partici-pant BL SSl SS2b FB FU

Durationof interaction(inseconds)

Mike 6 8 29 24 42 Adam 10 12 15 16 19 Bill 5 12 19 66 79 Peers 15 14 23 15 34 Initiationsbystudents Mike 0.5 2 2 5 4 Adam 2 3 5 4 6 Bill 0.9 4 3 3 6 Peers 4 5 4 8 5 Initiationstostudents Mike 3 6 5 5 3 Adam 4 5 4 4 5 Bill 2 3 5 3 3 Peers 3 2 3 4 2

Responsivityof studentsto initiationsfromothers

Mike .28 .30 .76 .80 1.0

Adam .50 .56 .80 .67 .67

Bill .49 .56 .76 .90 .88

Peers .59 .50 .86 1.0 .88

Responsivityofothers tostudents' initiations

Mike .00 .63 .69 .95 .67

Adam .33 .30 .60 .72 .70

Bill .61 .61 .87 .93 .90

Peers .56 .75 .43 .79 1.0

* Meanofthefirst halfofsocialskillstraining condition. bMeanof the second halfofthe socialskillstrainingcondition. s, and 14.3s.Social interaction time

averaged

96.6

to 140.6 s

during

social skills

training

and 65 to

93 s

during

posttraining probes.

PeerSocial Interactions

Socialinteractiondatawere

occasionally

collected for nonhandicapped peers (a total of 46 probes

across conditions) for comparison purposes. The

frequencyof peerinteractions

averaged

3.8

during

baseline,

5.8

during

training,4.2

during

posttrain-ing probes, 10.2 during the feedback

condition,

and 5.9

during

follow-up.

The duration of social interactionsfor peers

averaged

69.7

s in

baseline,

94.0 sduring training, 72.0 s

during posmaining

probes, 147.2 sduring

feedback,

and 185s

during

follow-up.

Additional Data

Analyses

Duration

of

social interaction. An additional analysis ofsocialinteraction consistedof computing the average duration of each interaction for the target students and their peers. The duration of interactions rangedfrom 5 to 15 s duringbaseline and initial trainingconditions, with increases to42

s, 19 s, and 79 s forthe target students and 34 s

for peers

during follow-up (Table

1).

Initiation andresponsepatterns.Baseline initi-ations

by

the targetstudentsrangedfroman average of 0.5 to 2 initiations, with increases to means of

4to6during follow-up, equaling initiationsmade

by

peers

(M

= 5). Fewchangeswerenotedin the

numberof peer-to-target initiations or peer-to-peer initiations across conditions.

Response ratios (seeTable 1) were computedby

dividing

the number of responses into thenumber

ofinitiations, with a ratioof 1.0 depicting a

one-to-onecorrespondence (i.e., allinitiations were

an-swered). The responsivityofthe target studentsto

peersincreased frombaselinelevelsof less than .50

to .67to 1.00 in follow-up, equalingthe response levelsof peers. The levelofresponsivityfrom peers

tothe targetstudentsalsoincreasedtoratiosof.67 to 1.0

during

follow-up.

SocialSkills Ratings

The percentageof socialskillsrated as occurring sometimes oroften (i.e., a scoreof 2 or 3)

during

baseline averaged 18% to 36% for the target

stu-dents. The percentage of socialskillsobserved

dur-ing traindur-ing ranged from 38% to 61%, with peer ratings averaging 57%. During feedback, ratings ofskills ranged from 54% to 100%, withincreases

to92% to97%duringfollow-upfortarget students and peers. Appropriate behaviors were generally

observed duringplay groups, with baseline mean

ratings ranging from 73% to 85%, and means of 85% to 100% during follow-up.

DISCUSSION

Thefindingsindicated that social skills training

for students with autism conducted concurrently withnonhandicappedpeers was a viableprocedure

for increasing the frequency of, time engaged in, and duration of social interaction for children in play groups. Perhaps as important were increases inlevelsofinitiationsand responsesbytargets and the responsivity of the peers to targets.

These results add to our current knowledge in several ways. First, the results demonstrated the benefits of socialskills training as a dasswide cur-riculum area (Neel, 1988), instead oflimiting in-struction to students with handicaps (e.g.,

Gaylord-Ross, Haring, Breen, & Pitts-Conway, 1984) or using peer-mediated procedures (e.g., Knapczyk,

1989; Strain &Odom, 1986).This supportsother studies illustrating the social effects of interactive group instruction, suchas the use of affection ac-tivities (McEvoy et al., 1988), peer tutoring(e.g.,

Greenwood, Delquadri, & Hall, 1989; Kamps et al., 1989), and cooperative learning (Johnson & Johnson, 1984).

A second finding is that rather than focus on structuredlearning of many social behaviors as dic-tated by published curricula (e.g., McGinnis &

Goldstein, 1984), the groups were moresuccessful

when focusingonfewer behaviorswith more prac-ticeopportunitiesforthosebasic skills.This reliance on more concrete behaviorswithrepeated practice was probably a moreefficient approach, and

con-tributed to the successoftheprogram forthe stu-dents with autism who might otherwise have had

difficulty in processing more abstract social skills

(e.g., negotiating,

problem

solving) into appropri-ate behaviors (Sasso et al., 1990).

Despitefavorable findings and recommendations for future use of social skills training for these

children, there remains aneed to examine the

qual-ity of interactions, the effects ofsocial skills pro-grams longitudinally, and generalization of skills.

For example, experimental effectswereimmediate

during training, but were slower (approximately seven to eight sessions) during posttrainingprobes offree play (see Figures 1 and 2).

To summarize,these findings support the use of socialskills instruction in small-group formats that

indude nonhandicapped peers and students with autism. Inaddition, more research needs to be done in instruction and measurement of the quality of

socialinteraction,isolationofthespecificsocialskills and/or training components from identified

cur-ricula responsible for increased interactions, longi-tudinal study of socialization in integrated

envi-ronments, and replication of social skills training for moreseverely impaired children with autism.

REFERENCES

Autism Society of America. (1990). Autismfact sheet (NIH PublicationNo. 83-1877). Bethesda, MD: U.S. DepartmentofHealthand Human Services.

Brown, W.H.,Ragland,E.U.,&Fox,J.J. (1988). Effects ofgroupsocializationproceduresonthesocialinteractions ofpreschool children.Researrh in Developmental Dis-abilities, 9, 359-376.

Carr, E.G.,&Darcy,M. (1990). Settinggenerality ofpeer modeling in children with autism.Journal ofAutism andDevelopmental Disorders, 20,45-60.

Charlop, M. H., Schreibman, L., & Tryon, A. (1983). Learning throughobservation: Theeffects ofpeer mod-eling on acquisition and generalizationin autistic chil-dren.Journal ofAbnormal ChildPsychology, 11, 355-365.

Gaylord-Ross, R.J., Haring, T. G., Breen, C., & Pitts-Conway, V. (1984). Thetraining andgeneralization ofsocial interactionskillswith autisticyouth.Journal of

Applied Behavior Analysis, 17, 229-247.

Greenwood, C. R., Delquadri, J., & Hall, R. V. (1989). Longitudinal effectsof classwidepeertutoring.Journal of Educational Psychology, 81, 371-383.

Johnson, D. W., &Johnson,R. T. (1984). Cooperation in the classroom. Edina, MN: Interaction Book Com-pany.

Kamps,D.,Locke,P.,Delquadri,J.,&Hall,R. V. (1989). Increasingacademicskillsof studentswithautismusing fifthgrade peersastutors.Education and Treatmentof Children, 12, 38-5 1.

Knapczyk, D. R. (1989). Peer-mediated training of co-operativeplay between special and regularclassstudents inintegrated play settings. Education and Training of the Mentally Retarded, 24, 255-264.

McConnell, S. R. (1987). Entrapment effects and the gen-eralization and maintenanceofsocial skillstrainingfor elementaryschoolstudentswithbehavioral disorders. Be-havioral Disorders, 12, 252-263.

McEvoy, M.A.,Nordquist,V.M.,Twardosz, S.,Heckaman, K.A., Wehby,J. H., &Denny, R. K. (1988). Pro-moting autistic children's peer interaction in an integrated earlychildhood setting usingaffectionactivities.Journal of AppliedBehaviorAnalysis, 21, 193-200. McGinnis, E., & Goldstein, A.P. (1984). Skill-streaming

in the elementary schoolchild: Aguide for teaching prosocial skills. Chicago:Research Press.

Neel,R. S. (1988). Implementing social skills instruction inschools.Behaviorin OurSchools,3, 13-18. Niemeyer,J. A.,&McEvoy, M.A. (1989). Observational

assessment of reciprocal socialinteraction: Social in-teraction code (SIC). Vanderbilt/Minnesota Social In-teraction Project,VanderbiltUniversityandthe Univer-sityofMinnesota.

Odom, S. L., & Strain, P. S. (1986). A comparison of peer-initiation and teacher-antecedentinterventions for promoting reciprocal social interaction ofautistic pre-schoolers.JournalofApplied Behavior Analysis, 19, 59-71.

Sasso, G. M., Hughes, G. G., Swanson, H. L., &Novak, C.G. (1987). A comparisonofpeerinitiation inter-ventions in promoting multiple peer initiators.Education and TraininginMentalRetardation, 22, 150-155. Sasso, G. M., Melloy, K. J., & Kavale, K. A. (1990).

Generalization,maintenance,and behavioralcovariation associatedwith socialskillstrainingthrough structured learning. BehavioralDisorders, 16,9-22.

Schloss, P.J.,Schloss, C. N.,Wood,C. E., &Kiehl,W. S. (1986). A critical reviewofsocialskillsresearch with

behaviorally disordered students. BehavioralDisorders, 12, 1-14.

Shafer, J., Egel, A., &Neef, N. (1984). Trainingmildly handicapped peers to facilitate changes in the social in-teraction skills of autistic children. Journal of Applied Behavior Analysis, 17,461-476.

Strain, P., & Odom, S. (1986). Peer social initiations: Effective intervention for social skill development of ex-ceptionalchildren. ExceptionalChildren, 52, 543-5 52. Walker, H.M., Hops, H., & Greenwood, C. R. (1988). Socialskills tutoring andgames: Aprogram to teach socialskills toprimarygrade students. Delray Beach, FL:Educational Achievement Systems.

ReceivedJuly 8, 1991

Initialeditorial decision October31, 1991 RevisionreceivedJanuary 20, 1992 Final acceptance February20, 1992 ActionEditor, Samuel Odom