RE.P O R T RESUMES03 016 997 CG out 184

GROUP PLAY THERAPY AND TANGIBLE REINCORCERS USED TO MODIFYTHE BEHAVIOR OF EIGHT- YEAR -OLD BOYS.

BY- CLEMENT, PAUL W. MILNE, D. C.CALIFORNIA UNIV.. LOS ANGELESPEPPERDINE COLL., LOS ANGELES, CALIF.EDRS PRICE HF*.$0.25 NC -$O.92 21P.

DESCRIPTORS-. *BEHAVIORAL SCIENCE RESEARCH, *GRADE 3, *PLAYTHERAPY, *GROUP THERAPY, *REINFORCEMENT, PERSONAL ADJUSTMENT,PERSONALITY PROBLEMS, EMOTIONALLY DISTURBED, REWARDS,THERAPISTS, SPONTANEOUS BEHAVIOR, BEHAVIOR CHANGE, ANXIETY,STATISTICAL ANALYSIS, FACTOR ANALYSIS, BENDER- GESTALT,CALIFORNIA TEST OF MENTAL MATURITY, RORSCHACH TEST, BEHAVIORPROBLEM CHECKLIST, CHILDREN'S MANIFEST ANXIETY SCALE

THE PRESENT STUDY WAS DESIGNED TO DETERMINE WITHIN GROUPPLAY THERAPY - -(l) THE COMBINATION OF EXPERIENCES WHICHPRODUCE THE GREATEST CHANGE IN BEHAVIORS AND THE EMOTIONALSTATE, AND (2) WHETHER TREATMENT INCLUDING TANGIBLE REWARDSPRODUCE MORE CHANGE THAN TREATMENT EXCLUDING THEM. THE 11THIRD -GRADE BOYS REFERRED BY TEACHERS BECAUSE OF THEIR SHY,WITHDRAWN BEHAVIOR WERE RANDOMLY ASSIGNED TO THREE PLAYGROUPS. GROUP XA, COMPOSED OF FOUR BOYS, WAS TREATED BY THETHERAPIST (T), RECEIVING TANGIBLE REINFORCEMENTS IN THE FORMOF TOKENS FOR ASSERTIVE AND SOCIAL APPROACH BEHAVIORS. GROUPXB, COMPOSED OF FOUR BOYS, WAS SIMILARLY TREATED,, EXCEPT THATNO REINFORCEMENTS WERE GIVEN. GROUP CA, WITH THREE BOYS,SERVED AS THE CONTROL. RESULTS INDICATED THAT ALL GROUPSFAILED TO EXHIBIT CHANGE IN PRODUCTIVITY AS MEASURED BYTEACHERS'S GRADES, ANXIETY, GENERAL PSYCHOLOGICAL ADJUSTMENT,AS REPORTED BY MOTHERS. IN THE TWO REMAINING DEPENDENTVARIABLES, XA IMPROVED MORE THAN XB, AND XB IMPROVED MORETHAN CA. XA EXHIBITED INCREASED SOCIAL ADJUSTMENT AS MEASUREDBY RESEARCH ASSISTANTS DURING THE THERAPY SESSIONS, ANDDECREASEV PROBLEM BEHAVIOR AS REPORTED BY THE MOTHERS.LIMITATIONS TO THE STUDY ARE -- (l) THERE WAS ONLY ONETHERAPIST WHO ALSO SERVED AS THE SENIOR INVESTIGATOR, (2) THEVERY SMALL SAMPLE SIZE, (3) THE DIFFICULTY OF GENERALIZINGFINDINGS, AND (4) THE QUESTIONABLE VALIDITY OF SUBJECTIVEMATERNAL. REPORTS. (AUTHOR)

4 1,'1;.t

11°:!... 4",' "1:N14;"'-7-irr.... n .l 4......, , ,,,.... , , , - . +" - - -

Is%4IDUCLA Medical School, Harbor General Hospital

Department of PsychiatryPaul W. Clement

and

Group Play Therapy and Tangible Reinforcers Usgd toModify the Behavior of Bight-year-old Boys

D. COUktiley ma.umC.7.) Pepperdine CollegeC:1

Group and individual play therapy techniques ale routinelyused in most child guidance and psychiatric clinics throughoutthe country. Although the therapists who use these techniquesbelieve in their efficacy in removing unwanted and in establish-ing more acceptable behavior patterns, few systematic studieshave been reported on play therapy. Masling (1966) went so faras to say, "Research in child therapy is absolutely non-existent."While the situation is not quite this bad, there has been ageneral failure to demonstrate that play therapy techniquesproduce predictable changes in behavior and personality (Ginott,19611 pp. 135-158; Levitt, 1957; 1963).

The present study was designed to shed some light on thefollowing two questions which are concerned with child psycho-therapy research: (1) Within the context of group play therapy,what combination of experiences produces the largest number ofchanges in behavior and in emotional state and/or the greatestdegree of change? (2) will a treatment approach which includesthe systematic reinforcement with, tangible rewards of preselectedbehaviors lead to more change than will a treatment approachwhich excludes tangible rewards?

For the present study "group play therapy" is defined asa situation in which a specially-trained person, the therapist,establishes a relationship with a group of children who exhibitbehavior patterns which are personally and/or socially unaccept-able. The therapist attempts to apply a behavioral model tomodify the behavior of the children so that there will be an

ti

7.7

llhis study was supported by grant P-389 from the AttendingStaff Association, Harbor General Hospital, Torrance, California.The authors wish to thank. the following for their assistanceon this project: Robert S. Dollarhide and Louis Kaplan of theTorrance Unified School District as well as the many principalsand teachers who cooperated; Katherine Clement, Anita Congelliere,Paula Heindel, and Lee Ann Robertson who were the observers; the

4 staff of the Department of Psychiatry, Harbor General Hospital;and the staff and students of the Department of Psychology,

0Pepperdine College.

0

Saskatchewan, Canada.2Now at Saskatchewan Penitentiary, Prince Albert,

..---10ARiMENT OF HEALTH, EDUCATION & WELFARE

OFFICE OF EDUCATION

THIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THE

PERSON OR ORGANIZATION ORIGINATING IT. POINTS OF VIEW OR OPINIONS

STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATION

POSITION OR POLICY.

increase in (1) productivity in school and other settings,(2) the positive aspects of the self concept, (3) socialattractiveness and the ability to get along with other people,and (4) the ability to cope with the problems of life, and sothat there will be a decrease in discrete problem behaviors.

We assume that "group play therapy" is not a single treat-ment approach uniformly practiced by many chin. therapists,bir rAthar 4t iS A 'MAGA nrinnApt Imarl to matognrim R verylarge number of poorly defined techniques used in a multitudeof combinations by many practitioners. Also we assume thatthere are significant differences between therapists both withina given "school" of play therapy as well as across schools. Theproblem, therefore, is that for the present study and data"group play therapy" can only refer to the specific techniquesemployed. Generalization to other approaches should be doneonly with great care.

Regardless of the approach, behavioral terminology may beused to describe what is going on in a given type of psycho-therapy. At the onset of treatment each child has a habit-family hierarchy of responses. Some of the more dominant behaviorsin the hierarchy are unacceptable to the child's parents, teachers,etc. Lower on the hierarchy are more acceptable behaviors. Byselectively reinforcing these more acceptable behaviors andfailing to reinforce the unacceptable behaviors, the frequencyof occurrence of the desired responses should increase and thatof the undesired responses should decrease.

A few attempts have been made to apply conditioningtechniques to groups of adults (e.g., Lazarus, 1961), butapparently no study has been published on conditioning techniquesapplied to groups of children in a therapeutic setting. Therehave been many studies, however, using operant techniques onindividual children, and most of them have reported positiveresults (e.g., Allen, Hart,_ Buell, Harris, and Wolf, 1964; Kerr,Meyerson, and Michael, 1965).

The primary hypothesis of the present study was thatchildren who are systematically reinforced with tangiblereinforcers will improve more than those who receive treatmentwithout tangible reinforcers, and those who receive treatmentwithout tangible reinforcers will improve more than those in acontrol situation.

Method

Sub tsta

Eleven third-grade 'joys from the Torrance UnifiedDistrict constituted the SA. The mean age was 8 yearswith a range of 8 years 2 months to 9 years 3 months.IQ from the California Test of Mental Maturity (CTMM) ,

..2»

School10 monthsThe moan1963

S-Form, was 100 with a range of 80 to 123. All of the boyswere Caucasian; had appropriate speech; had never received anyplay therapy or psychotherapy before; did not exhibit psychoticor sociopathic behaviors; did not present symptoms of aperceptual-motor handicap; and met Fish and Shapiro's criteriafor a Type III child with a "mild" global severity rating (Fishand Shapiro 1965). Most Type III children have a formaldiagnosis within the psychoneuroses. All Ss lived with bothnatural parents except for two boys. One g lived with hisnatural mother and his step father; the otTer S lived with hismother only, his father having died one year before the studybegan. The mean number of years of formal education for thefathers was 13.4 years with a range of 11 to 16. The mothers'mean was 12.4 years with a range of 11 to 14.

Selection Procedure

Sampling was done from 2,761 third-grade children from105 classrooms aod 33 grade schools. Each principal was askedto give a letteri to each of his third grade teachers requestingthat the teacher send a form letter from the investigators tothe parents of the two boys in her class who most closely fitthe following description: "(1) socially withdrawn, very quiet,introverted, and friendless; (2) lacking in spontaneity;(3) maladjusted; and (4) not so withdrawn that he is unable toattend school regularly," The letter to the parents brieflydescribed the study and asked the parents to contact the seniorinvestigator if they would like to participate.

A group appointment was made for the mothers who applied.At this first meeting the mothers completed a social anddevelopmental history for their boys. They also filled out aBehavior Problem Check List (BPCL), a Q-Sort (Block, 1961,pp. 154-156), and the Children's Manifest Anxiety Scale(Castaneda, McCandless, and Palermo, 1956). The mothers wereasked to take their boys to their family physicians or pedia-tricians for a complete medical examination.

About one week after the mothers were seen each boyreceived a psychological evaluation covering intellectual and

personality functioning. Group administrations were used forfigure drawings, the Bender-Gestalt, and the CTZIM. CardsIII, VI, VIII and X of the Rorschach and a sentence completiontest were administered individually.

Based on the evaluation procedure, 12 Ss were selected Whomet the criteria previously listed. One S rad to undergo major

411 letters, instructions, questionnaires, etc. used inthe present study which have not previously been published maybe obtained in mimeographed form from the senior author.

. 3 ."

surgery during the second week of the research program; there-fore, he had to drop out, leaving a final N of 11.

Experimental Design

Inde pendent variables. There were two independent variables:(1) the hype ortHaWriaMinistered to the Ss and (2) timei.e., the number oi group play therapy sessions. The experi-mental model was for a two-factor experiment with repeatedmeasures on the second factor with unequal groups, i.e., n1 = 4,B2 = 4, and n3 = 3 (Winer, 1962, pp. 374-378).

Bp in the first experimental group (XA) went through theevaluation procedure, came to the hospital once a week withtheir mothers spent a inimam of 50 minutes in each group play

isession, met in a group of four boys in the playroom, weretreated with the procedures suggested by Ginott (1960) for doingGroup, Ps chotherap with Children and received tangible rein-forcements w en asseETTWIEri3Eial.approach behaviors occurred.The tangible reinforcements consisted of brass tokens Whichcould be accumulated to purchase candy, trinkets and small toysat the end of each therapy hour (cf Ayllon & Azrin, 1965). Thetokens could also be saved across therapy sessions to obtainthe more expensive items. XA received a total of 14 therapysessions.

The second experimental group (XS) received the sametreatments administered to a except the tangible reinforcerswere excluded.

The as in the control group (CA) went through the evaluationprocedure, came to the hospital once a week with their mothers,and met in a group of three boys in the playroom. This groupdid not have a therapist, but they were observed through a ono-way mirror. They were allowed to do as they wished except toendanger anyone's physical safety and to damage the playroomand furnishings. One observer was always available to intervenetemporarily in an emergency; however, active intervention wasonly required once.

The original 12 Ss were randomly assigned to the 3 groups.Each group was seen on a weekday at 9:00 a.m. for 14 consecutiveweeks during the spring of 1966.

Mothers' guidance groups were held concurrently with thetherapy and control sessions. The mothers were seen on thesame day and at the same time as their sons were seen. Themothers' groups were run according to the suggestions of Ginott(1960, pp. 169-189), i.e., they were guidance rather thancounseling or psychotherapy groups.

De endent variables. Table 1 lists (1) the dependentvaria es, the indices to them, (3) the source person

I

Tab

le 1

Sum

mar

y of

the

Dep

ende

nt V

aria

bles

Dep

ende

nt v

aria

ble

(DV

)In

dex

to D

V41

1111

6.

Person providing data

Whe

n m

easu

red

dom

mlli

nli1

1111

0111

1111

1111

1111

111i

l

Productivity

Gra

deson

report card

Teachers

Quarterly during

academic, year

Anxiety

4111

.041

Social adjustment

CMAS

Ss Mot

hers

10Pre-therapy,

7th session, and

14th session

Play room

observations

General psychological

Q-sort

adjustment

Prob

lem

beh

avio

rsBev

Research assistants

Mot

hers

Mothers

Thr

ough

out e

ach

of the 14 play

sess

ions

1114

101.

0101

1111

1111

1111

1111

11m

ainI

ON

IOR

Pre-- therapy,

7th session, and

14th session

now

leco

mm

eclim

oom

mol

imm

e

Pre-

ther

apy,

7th session, and

14th session

providing the data, and (4) when each index was measured. Thedependent variables were taken from the definition of playtherapy given in the Introduction. Valid indices to thesedependent variables were sought which would provide highlyobjective and reliable data.

Much of the present data was obtained by directly observingthrough.. a one-way mirror the behavior of the children duringthe group play sessions. A time sampling technique was used torecord the observations. The observer (0) recorded the behaviorof each kby observing him every session for 12 one-minuteperiods taken at four-minute intervals. As is indicated inTable 1, the time sampling provided the "social adjustment "

data.

Ten classes of behavior were recorded: (1) statements toSs, (2) statements to the therapist (T), (3) questions to Ss,74) questions to T, (5) non-verbal vocalizations, (6) verbaland symbolic aggression, (7) physical aggression (8) solitaryplay, (9) social play, and (10) proximity. Specific criteriawere established for each class of behavior. The proximitymeasure was made as follows: Strips of tape were laid on theplayroom floor so that a grid of rectangles approximately oneyard square was formed. If the S was in the same square asanother child during any part of the 60 seconds the S was beingobserved, the S was scored for proximity.

Only one check mark per S per class of behavior per 60seconds was allowed, making tEe maximum possible score 12 perA per behavior per play session.

Because of marked session-to-session variability, the datawere grouped into blocks of sessions, i.e., sessions 1-4, 5-9,and 10-14. This grouping made the data more manageable andeffected easier statistical evaluation.

Attendance at all sessions was not 100%; therefore, thefollowing procedure was used for estimating raw scores formissing Ss. The mean of the S's raw score from the sessionpreceding and following the missed session was computed.Second, the mean raw score obtained by the Sts group on theday he missed was computed. The mean of those two meansserved as the Ws estimated raw score for the day he missed.Host of the boys attended all of the sessions.

During the course of the study only one 0 recorded thebehavior of a given group. Before the study Vegan, therefore,a check was made on inter-observer reliability. Three observerswho were using the time sampling system for the first timeobserved a pilot group for one session. Pearson product-momentcorrelation coefficients yielded its of .86, .88, and .93 forthe three possible pairings of the three Os. A secondreliability check. was run on the final therapy session at the

T,v.t...v.*.vc.?r,-r...vz+n!Par~tiNeovr.-smftr,reVrYe'y "TF.',ntort'e,SAWVn

end of the present study. The r's at this time were .97 9)and .99, indicating extremely high inter-observer reliability.

Statistics. As has already been mentioned, the primaryanalynriarEibsed on a 3 X 3 factorial design with repeatedmeasures on the second factor (Winer, 1962, pp. 374-378)eSupplementary evaluations included (1) single-factor analysesof variance for XA, ami CA talmn individually (WinDr,19622 pp. 105-132), (2) trend tests (Winer, 1962, pp. 132-135),and (3) tests on all pairs of means using the Tukey (a)procedure (Winer, 1962, pp. 87-89).

Personnel and Rama.

The therapist for XA and XB was the senior investigator.The leader of the three mothers' groups was the juniorinvestigator. The three observers were volunteer researchassistants. They were all young homemakers who had volunteeredto work specifically on the present project and were notacquainted with any of the Ss.

The playroom measured 19 by 8f feet. There was a one-way mirror at one end, and a sound system connected the playand observation rooms. The playroom contained a blackboard,a children's picnic table with attached benches, an easel, anda toy chest. The play materials consisted of crayons, fingerpaints, paper, chalk, building blocks, a doll house, two five-member doll families, four toy revolvers, a plastic war setwith soldiers, tanks, jeeps, etc., a Bobo punching bag, and asmall tape recorder.

Results

The results of the major statistical tests are summarizedin Table 2. Columns (1), (2), and (3) present the F ratiosfrom the two-factor analyses of variance. For the rime samplingdata the raw scores used in the analyses were the mean scoresfor each S from sessions 1-4, 5.9, and 10-14. The symbols incolumns (4), (5), and (6) indicate the outcomes of the single-factor tests performed for each play group uhen factor Al B2

or AB was significant on the two-factor analysis. The symbol"0" indicates that no change occurred, "+" that the groupimproved on the measure, and ?L" that the group got worse.

No Difference lietzsen proues4 No Changes over Sessions

Six measures of the dependent variables failed to produceany significant F's for groups, sessions, or groups-sessionsinteraction (see Table 2). The measures were achievement(i.e., the mean grade in reading, English, spelling, arithmetic,and social studies), boys' scores on the WAS, mothers' scoreson the CHAS, questions to 24 aggression (verbal and symbolic),and the Q-Sort.

- 7 -

Table 2

Summary of the

Results of the Two-factorand Single-factor Analyses

of Variance

Measure

(1)

Gro

ups

F Ratios

(2)

Sessions

(3)

Interaction

Results of Sinele-

factor Analyses of

Variance

(fi4

)(5

)XA

XB

CA

(6)

Criterion for

Improvement

Achievement

Citizenship

B4s? CMAS

Mothers' CMS

Statements to Ss

Questions to

ein

Non-verb. vocal,

Aggression (V &

(phys.)

Play (Solitary)

0436

9.310

.791

6.1r

-46173r6§4141

1102.79g

A5W

.32

2.58

2.33d

2061d

,4

38.94**%4

16099**.,

29.024"

Play (social)

Ilpximity,

Q-Sort

BPCL

.10d

.50e

113810

.15e

.466

3.44e

2051e

1075!

27°

1684e

364

6

.23e

37.234we

1.75

c.07°

O0

0Increase

O0

0Decrease

05f

2.67f

7.04

3056t

I.26A

3.284P-

.40rf

30781-

10574.11,

24452**L

15.75§4

9 09K-Ild

3000e

O0

0Decrease

o0

0D

ecre

ase

4-*

0 +4(4t.

0O

00

O0

0o

00

O0

O0

0

Increase

Decrease

Increase

Decrease

Increase

Increase

Decrease

O0

0Decrease

Increase

+it*

***

0Increase

2.35rm-0

00

Increase

.71f

4.*

00

Decrease

Improved

(+)

No change

(0)

Worse

(-

< .0

541

-7P

< .0

1B

ath

2 and

6df

bWith 2 and 12 df

eWith

4and 12 dr

dWith 2 and9 df

°with 2 and 18 df

(With 4 and 18 df

gWith 1 and 6 dt

42

012

13

16

01

0.1

1111

11b

'''ele.Trr:f0t1f11:1?-sVzrvevs,ter

Differences between Proups, No chasm orver Session1

In five cases the "groups" or "interaction" factor wassignificant, but the "sessions" factor did not reach signifi-cances In all of these instances single-factor analyses ofvariance revealed no reliable changes over sessions in groupXA, XB, or CA. Although Table 2 shows a reliable differencebetween groups for "citizenship," this difference existed onthe pre-therapy grades in citizenship (P < .05). For "questionsto Ss" the same kind of problem existed. The groups weredifferent on the very first play session (P < .05) n For thetime sampling data a significant F on the first play sessionvas taken to mean that due to sampling error the groups startedout different at the beginning of treatment. "Non-verbalvocalizations" and "aggression (physical)" produced significantinteractions only. The groups were different on "play (solitary)"(XA < XB, < .01; Lk CA, P < .05; and > CA, P 4: .01), butthe "sessions" factor was nor significant.

Changes, over Sessions

No indication was given by the preceding measures thatthe treatment procedures had any effect on the dependent vari-ables. The results are more positive for the remaining measures.

Statements to Ss. Table 2 indicates that only XA exhibiteda reirairriaeasein the amount of verbal communicationbetween the early part of treatment and the latter part. Atrend test on group XA. indicated a significant linear trend(P < .01). For the time factor, sessions 1-4 are indicated as

," sessions 5-91 as "b2," and sessions 10-14 as "b3." Testson all pairs of means for XA showed that 131( b3 (2,

but la = b2 and b2 = b3 (see Fig. 1).

Statements to T. A reliable decrease in scores on thismeasure was defined-as improvement. The assumption underlyingthe definition was that shy, withdrawn children are overly-dependent on adults at the beginning of therapy. As theybecome less dependent on T, they talk to him less and to theother children more. Trend analysis for XB indicated a signi-ficant linear trend (P < .01) . Tests on all pairs of meansproduced the following results: b3 > b2 (P < .05), 1)1> b3(P < .01), but b2 = b3 (see Fig. 21.

LIM Both XA and XB changed during the courseof trikaffehtf-hbViVer, the boys in XA increased in the amountof time they spent playing with each other, whereas the Ss inXB decreased (see Fig. 3). Trend tests on XA revealed asignificant linoar trend (P < 601), but both the linear andquadratic trends were significant at the .01 level for 1036Xests on all pairs of means for IA demonstrated that bi< b(E < .01), b2 < b3 (P < .01), and bl = b2, and for XB b3, > b2(P < .01), bi> b3 < .01), and b2 < b3 < 05).

9

=',',-1'..01#44P0.1"4311,0ftri,

01/2

1 5 - 9 10 4. 11.4

PLAY SESSIONS

J

Fig, li Frequency of "statements to Ss" averaged for the first foursmiddle five, and last five play sessions,

-10..

col

1 4- 4 5 4. 9 10 40 24

PLAY SESSIONS

Pig. 2. Frequency of "statements to T" averaged for the first

four, middle five, and last five play sessions.

" "

PLAY SESSIONS

Fig. 3. Frequency of "social play" averaged for the firstfour, middle five, and last five play sessions.

-12

Proximit XA and XB exhibited reliable increases duringtreatment`reatmen on is measure (see Fig. 4 and Table 2). Trendtests indicated significant linear and quadratic trends forboth XA and JCB. Tests on all pairs of means showed that forXA 1314( IA (1!. .ca), b2 < 1,1 (po 4:.01), bi = b2, and for XB

b2 .01), bi 11/3 (E: < .01), and 52 4( 153 (2, 4: .01).

Behavior Problem Check List, Although the groups didnot differ on TErVFME57614 measures, they became reliablydifferent during treatment (see Table 2), Tests on all pairsof means for the groups factor revealed that XA4( XB ( 4:.01),XA = CA 2 and XB> CA (L4L.01). All three groups had lowerraw scores on the post-therapy measures than on the pre-therapymeasures, but the over-all time effects were not significant.A trend analysis of Xk's data showed a significant lineartrend (e4.1. .05), and tests on all pairs of means indicatedthat bi>. b3 (E < 405)2 bl = b2, and b2 = b3.

Mothers' aljective Evaluations

At the end of treatment the mothers of the 11 Ss wereasked to write about any change which they had observed intheir boys during the time they were coming to the groupsessions. Ten of the mothers listed changes. One said thather boy's stealing had gotten worse. All of the other motherslisted positive changes only. Even the mother of the stealingboy indicated that her son's teacher felt he had improved atschool.

The types of changes which were listed were "more assertive,""happier," "more concerned about doing well in school," "moreself assured," "now has a good attitude toward himself," "is'branching .outs," "his tears are less frequent," "more stable,""less shy," "tells stories to his dad and relates the happen-ings of the day," "neighbors report him as more aggressive,""teacher says he speaks up for himself naw" "not so fearful,""not so hard to get along with," "is enjoying his PE classnow," "plays more with kids outside," "more competitive,""more energetic," etc.

There was no apparent difference in quality or quantityof tinse responses between the three groups.

Discussion

A common belief found in many child guidance centers isthat emotional conflicts can lead to underachievement in schoolwork but that psychotherapy usually leads to an increase inacademic productivity (e.g., Bills, 1950a; 1950b). No suchincrease was noted in the Ss of the present study; howeverthe boys had no', been referred as underachieverso Inspectionof their grades suggested a fairly normal mean and distribution,indicating that on the average the boys had been working up to

col

12

13.

10

9

8

7

6

5

14

2

*OM

MOO

XA 0-0XB

CA Ile-44.11

- 4 5 - 9 10 - 14

PLAY SESSIONS

Fig. 4. Frequency of "proximity" averaged for the first four,middle five, and last five play sessions.

-

to- a44 `,. 4v- ci,r,,

,

capacity before treatment began. The use of school gradesas a dependent variable was probably inappropriate for boyswith the presenting symptoms of these Ss.

Another common belief held by child therapists is thatanxiety decreases as therapy progresses. The measure ofonnrigm+11 used 4m 4.4.^ present velaccolvsel'h showed Mn vsnlinhla changetheduring the course of treatment. The length of treatment mayhave been too short to produce any consistent changes; but

Ss,

the S raw scores indicated that the lack of change in CMASscores may have had the same cause as the lack. of change ingrades, i.e. the mean anxiety score was close to the mean forall 8.. and 9-year-old boys, In order for the Ss as a group tohave improved on this measure would have requffed them toexhibit less anxiety than typical boys their age.

After reviewing the literature on child psychotherapyresearch, Levitt (1957; 1963) said, "...the inescapableconclusion is that available evaluation studies do not furnisha reasonable basis for the hypothesis that psychotherapyfacilitates recovery from emotional illness in children (1963)."The present study has provided data which contradict theimpression that child therapists are unable to demonstrate thatformal treatment makes a difference. The primary hypothesisstated in the Introduction appears to have been substantiated.XA improved on four variables; XB improved on two and gotworse on one; and CA, the control group, did not show anyimprovement. The present study included a truly matched controlgroup; whereas, although Levitt claimed to use a well-matchedcontrol group for his comparisons, his control and experimentalSs were not obtained from the same population and at the samepoint in time.

Eysenck. (1954; 1964) has made much of the idea ofspontaneous recovery occurring in psychoneurotic disorderswithout formal treatment, but Kiesler (1966) orAticized theconclusions of Eysenck.and claimed that "spontaneous recovery"is a myth, The results of the present study contradictEysencLis conclusions and agree with Kiesler. No spontaneouschanges appeared in CA during the period these control Ss werestudied,

The problem of evaluating parents' subjective reports ofimprovement in their children during and following therapy ispointed out in the present results. All of the mothers exceptone reported improvement in their boys. Their comments impliedmore improvement during the course of treatment than wasobserved on the objective data. The assumption has been madein writing this Discussion that the mothers' reports wereunreliable; however their subjective reports provided thekind of data typically used in clinical settings to evaluateoutcome of child treatment. The discrepancy between objectiveand subjective data points out the need for systematic studyof the causes of the obtained differences.

m 15

Ayllon and Azrin (1965) demonstrated the effectivenessof treating adult psychotic patients with metal tokens whichcould be traded for desired objects. The present study showedthat the same type of operant conditioning technique iseffective in treating shy,. withdrawn 8-year-old boys in agroup play situation. Fig. I indicates most clearly the impactthe tokens had on XAss behavior. Although some child therapistsmay be repulsed by the idea of giving brass tokens to thechildren they are treating, the use of such a procedure willprobably increase the therapeutic impact of the therapist onhis clients.

There is so much session-to-session variability in thebehavior of 8-year-old boys in a group play situation thatweek -t' -week. predictions are very precarious. Accurateprediction becomes more possible when T looks for general trendsbased on data averaged on groups of sessions, i.e., meaningfulpredictions may be made from month to month but not from week.to week.

The major criticisms of this study are as follows:(1) There was only one T. (2) The senior investigator and Iwere the same person. Ideally T should not be are of themajor hypotheses of the researc' project. (3) The samplesize was very small. (4) The sample represented only one typeof presenting problem out of the many which confront thetypical child guidance clinic. (5) The data do not lendthemselves to reasonable generalization to other groups repro-seating other kinds of problems. And (6) although an attemptwas made to avoid the major criticisms of psychotherapyresearch (e.g., Xiesler, 1966), the present paper did notsolve the most important problem of providing an adequateresearch paradigm which will, lead to systematic theory con-struction and modification.

Finally there has been relatively little systematicresearch oa psychotherapy with children. One of the reasonsfor tbj.s lack of research has been the apparent difficultyin obtaining an adequate sample. The present study demonstratesan effective approach for obtaining reasonably homogeneousgroups of children. Adult samples could also be obtained inmuch the same manner; however, institutions other than schoolswould have to be used as the source. For example, allmini3n;ers, priests, and rabbis in a community could be con-tactel as the school principals were in the present study andasked to make referrals having particular types of problems.

Summary

11 third-grade boys were randomly assigned to three playgroups. The boys had been referred by their teachers becausecf shy2 withdrawn behavior. Group XA met in a play group(4 Ss), had a therapist, and received tangible reinforcements

- 16..

-11.4,4"

e.

for social approach behaviors. Group XB (4 Ss) was treatedsimilarly to XA except that no tangible reinrorcements wereused. Group CA (3 Ss) met in a play group without a therapistpresent. 14 play sessions were held for each group. XAimproved more than XB; XB improved more than CA; CA. showed noimprovement on the objective measures used. XA exhibitedincreased social adjustment and a decrease in discrete problembehavior. XB increased slightly in social adjustment. Allgroups failed to exhibit change in "productivity," "anxiety,"and "general psychological adjustment."

References

Allen, K. Eileen; Hart, Betty M.; Buell, Joan S.;Harris, Florence R.; & Wolf, M. M. Effects of socialreinforcement on isolate behavior of a nursery schoolchild. Child Development, 1964, 35, 5116.518.

Anion, T. &Azrtn, N. H. The measurement and reinforcementof behavior of psychotics. Journal of the ExatapentalAnalysis of Behavior, 1965, 87357-703.--

Bills, R. E. Nondirective play therapy with retarded readers.Zournak of Consulting Psycholoia, 1950, 14, 140-'149. (a)

Bills, R. E. Play therapy with well-adjusted readers. Journalof Consulting, Psychology , 1950, 14, 246-249. (b)

Bloek, J. The Q-Sort. Method in Personalit Assessment andPs chinTlic EteseanE7Wing e 9 17""Winget: Thomas,

Castaneda A. McCandless, B. R. & Palermo, D. S. Thechildren's form of the Manifest Anxiety Scale, ChildDevelopment, 1956, 27, 315-3264.

Cox, F. N. Sociometric status and individual adjustment beforeand after play therapy. Journal, of Abnormal sal Socialaulaga, 1953, 48, 354-h"16.

Eysenck, H. J. The effects of psychotherapy: An evaluation.Journal of Consulting Psychology, 1952, 16, 319-324.

Eysenek, H. J. The outcome problem in psychotherapy: A reply.Psychotheram: Theory, Research and Practice, 1964, 1,Y7-71350.

Fish, barbara & Shapiro, T. A typology of children's psyfAlia-trio disorders. I. Its application to a controlledevaluation of treatment. Journal of the American &dem.of Child pchlatrz, 1965,7757.75r

Ginott, H. G. Grou Ps chothera with Children: Tha TmimPractiee-of eravy.

withMcGraw-H114--

ria.

Kerr, Ne9 Meyerson, L.9 & Michael, J. A procedure for shapingvocalizations in a mute child. In L. P. Ullmann &L. Krasner (Eds.), Case Studies in Behavior Modification.flew York: Holt, RiEZHErt, & WinnOn, 1965, -7157365737U.

Kiesler, D. J. Some myths of psychotherapy research and thesearch for a paradigm. Psychological Bulletin, 1966, 65,110-136,

.18.

Lazarus, A. A. Group therapy of phobic disorders by systematicdesensitization. Journal of Abnormal and So_ cial psychol7.ogy, 1961, 63, 504-0.0.

Levitt, E. E. The results of psychotherapy with children: Anevaluation. Journal of glasultiu, paghsamx, 195?, 21,10.4.96.

Levitt, E. E. Psychotherapy with Children: A furtherevaluation. Behaviour Research and Therapy, 1963, 1, 45-.51.

Masling, J. One man's viewpoint on training. Newsletter:Section on Clinical Child Ps cholo Division of tlinpalPsychoiogyasft-AmericariMe o og ca Association71966, 5,No. 2.

Winer, B. J. Statistical Princi les in Experimental Ewa.New Ybrk:--REUFFXR11 $