Journal of Abnormal Psyo&ology1972, Vol. 79, No. 3, ZdP-.-'S

COMPARISON OF RECIPROCAL INHIBITION, PSYCHOTHERAPY,AND WAITING LIST CONTROL FOR PHOBIC CHILDREN '

LOVICK C. MILLEK ,2 CURTIS L. BARRETT, EDWARD HAMPE, AND HELEN NOBLE

Child Psychia ry Research Center, University of Louisville School of Medicine

Sixty-seven phobic children, ages 6-15, were randomly assigned to a 2 X 3 factorial,repeated-measures, cowriate design which included two male therapists and threetime-limited treatments: reciprocal inhibition, psychotherapy, and waiting listcontrol. Following 24 sessions or 3-mo. wait and at 6-wk. follow-up, 5s werereassessed by an independent evaluator and by parents. Results indicated a signifi-cant effect due to time and child's age. Clinical evaluation, using initial scores asthe covariate, showed no effects of treatment or therapist. Parents reported treat-ment effects for both target fear and general fear behavior. Therapies were equallyefficient, and all treatment effects were achieved with phobic children aged 6-10.

There is a clear need for more rigorousstudies of the nature and treatment of child-hood phobias. Despite the continued interestshown in the study of phobic disorders (An-drews, 1966; Lang, 1968; Marks, 1969; Rach-man, 1968; Weitzman, 1967), there has beenlittle systematic work with phobic children.Berecz (1968) found most of the evidence ontreatment of childhood phobia to come fromsingle case studies, studies without controlgroups, or what he labeled "one shot attemptsto 'prove' the effectiveness of certain tech-niques [p. 707]."

The purpose of this study was to comparethe effectiveness of reciprocal inhibition ther-apy, psychotherapy, and waiting list controlas methods for treating phobias in childrenaged 6-15. In addition, we were interested inthe comparative effectiveness of the procedureswhen conducted by therapists who differed inexperience.

METHODThe 2 X 3 factorial, repeated-measures, covariate

design of this study included two therapists, threetime-limited treatments, and three evaluations: pre-treatment, posttreatment, and a follow-up after 6-wk.treatment. The preevaluation was used as the covariatefor all measures.

Subjects

Efforts were made to obtain 5s aged 6-15 who rep-resented the domain of clinical level childhood mono-

'This research was supported in part by NationalInstitute of Mental Health Grant 13219.

2 Requests for reprints should be sent to Lovick C.Miller, Child Psychiatry Research Center, Universityof Louisville School of Medicine, 608 South JacksonStreet, Louisville, Kentucky 40201.

phobias and multiphobias from all socioeconomicclasses of the community. Referrals were solicited byletters from physicians, social agencies, schools, andthrough newspaper publicity. No fees were chargedand transportation was furnished for indigent children.A total of 148 children passed initial screening and wereevaluated for this project. Of these, 86 were acceptedand 67 participated fully. Table 1 shows the dispositionof the 148 children.

Inspection of the demographic data in Table 2reveals several differences between our sample and thegeneral population: We had fewer blacks, Catholics,and lower to middle socioeconomic status representa-tives and more upper-middle-class children. Thisoccurred despite our efforts to achieve a representativedemographic sample. However, intelligence datamatched the general population for all phobias includingschool phobia.3

Table 2 also shows the number of fears per child andthe distribution of target phobias among 5s. The

3 E. Hampe, L. C. Miller, C. L. Barrett, & H. Noble.Intelligence and school phobia. Unpublished manu-script, Child Psychiatry Research Center, Louisville,Kentucky, 1971.

TABLE 1

DISPOSITION OF PROJECT APPLICANTS

Category

RejectedJudged not phobicSymptoms cleared in diagnosisFailed project criteria0

AcceptedFull participantsDecided against treatmentPilot casesDropped out of treatment

Total evaluated

2V

291518

675

104

148

« Project criteria, other than phobia, were: (a) Wechsler In-telligence Scale for Children Full Scale IQ > 75; (6) not psy-chotic or brain-damaged; (c) one parent cooperative to theextent of bringing the child three times per week and contract-ing to permit videotape and sound recordings.

269

270 MILLER, BARRETT, HAMPE, AND NOBLE

TABLE 2

SAMPLE CHARACTERISTICS, DEMOGRAPHIC CHARACTERISTICS, AND FEARS rr PROJECT CHILDREN

Demographic characteristics

SexMaleFemale

RaceWhiteBlack

ReligionProtestantCatholicJewishOther

Socioeconomic statusLowerLower-middleMiddleUpper-middleUpper

IQ75-7980-8990-99

100-109110-119120-129130+

No. ofproj ectchildren

3730

643

441607

510172312

21018191332

%

5545

964

66240

10

715263418

31527281953

Generalpopula-tion %

4951

8119

613414

1324371313

31322251895

Sample characteristics

Age at intakeYounger

6789

10Older

111213

268

1111

611S

39

131616

9167

Sample characteristics — (Continued)

Age at intakeOlder

1415

Grade123456789

10

No. ofprojectchildren

43

41089

116

10432

%

64

6151313169

15643

Generalpopula-tion %

Fears

TargetSchoolSleep/aloneDarkSeparationDogsStormsHeightsGermsPhysical injuryElevatorsNakednessDeep water

No. of fears per child1234567

4664221111111

269

158711

6996332221111

391422121111

target phobia was the fear that was the most disablingfor the child and the one toward which treatment wasaimed. Fear of school composed 69% of the targetphobias, an unusually high percentage since it occursin less than 1% of the general population (Miller,Barrett, Hampe, & Noble, 1971). Conversely, fearswhich occur in the general population with an incidenceup to 21%, such as fears of storms, the dark, anddomestic animals, comprised 6% or less of the targetphobias. Fear of separation, which is often reportedto be the core problem in school phobias, was listed asthe target fear when separation was primary andfear was manifested in many situations. Fourteen ofour 46 school phobics showed such generalized separa-tion fear, but fear of school was deemed the mostdisabling. In summary, we believe that 5s in our study

represent the population of fearful children for whomparents are willing to seek help.

Treatments

Waiting list control. This procedure was designed tobe comparable to waiting list periods commonly foundin child guidance clinics. Professional contact was keptto the minimum necessary for evaluation. Operation-ally, this meant that after evaluation, parents weretold that: (a) their child's phobia was serious enoughto need treatment; (6) they had been placed on awaiting list; (c) while on the waiting list, they wouldbe asked to come for evaluation "at times;" and(d) the research team would not be available to advisethem during the waiting period "because we cannot

TREATMENT OF CHILDREN'S PHOBIAS 271

be with you to help you carry out the advice." Parentsand referral sources seemed to be comfortable with thisrationale, and thus there was little professional contactthat could be regarded as treatment for childrenassigned to waiting list control.

Reciprocal inhibition therapy. Systematic desensi-tization, the main form of Wolpe's (1958) reciprocalinhibition therapy, was employed with every child inthis treatment group. However, in order to adapt thetechnique for use with children, the therapist employeda broader repertoire, primarily based on learningprinciples.

A child and one parent were usually seen togetherfor about 15 min. to review progress and to deal withtherapy-related problems. Then, the therapist workedwith the child alone. No constraint was placed on thetherapist in coordinating the treatment with significantpersons in the child's life such as school personnel or,for a child who feared water, a lifeguard. Occasionally,the therapist spent the entire session with the parentpresent in order to deal with resistance problems,parental guilt, or other clinical phenomena that thetherapist judged to be likely to threaten continuationof the child in treatment. Where parent-child interac-tion patterns appeared to reinforce fear behavior,behavior therapy principles were employed to restruc-ture contingency schedules, for example, eliminatingtelevision during school hours for a school phobic whostayed home. Assertive training was used for inhibitedchildren. In one case, special breathing control tech-niques were utilized to control chronic crying thatinterfered with systematic desensitization operations.For cases in which parents' problems were not specifi-cally related to the phobia and threatened to divertthe therapist from the child's pathology, our socialworker was available to treat the parents.

In the first session, the therapist established rapport,explained the rationale of the treatment, taught theparent to relax as a model for the child, started relaxa-tion training with the child, and assigned a homeworktask of relaxation practice 10 min. a day. By thefourth session, relaxation training was usually completeas was construction of initial fear hierarchies. Fromthis point, systematic desensitization proceeded aswith adults (Barrett, 1969a; Paul, 1966; Wolpe, 1958).A relaxed child was asked to imagine a scene that hehad listed as arousing little anxiety. If he could imaginethe scene comfortably, he was asked to switch to apleasant scene. Then he was asked to imagine the nextscene in the hierarchy. If imagining a scene caused thechild to signal anxiety, the therapist asked that thescene be "switched off" and that the pleasant scene beimagined again. At this point, the therapist eitherrepeated the scene or revised the hierarchy to providean item that aroused less anxiety. Periodically, thechild was asked to describe scenes as a check on thevividness of imagery. When all items of a fear hierarchycould be imagined comfortably, an "in vivo" test wasarranged. If unsuccessful, systematic desensitizationcontinued.4

4 For a more detailed description of the use of sys-tematic desensitization with children, a 16-mm. blackand white film, "I'm Afraid . . . Systematic Desensi-

In summary, the purpose and method of reciprocalinhibition therapy was to help both parents and childto develop alternative responses which would inhibitfear and would allow the child gradually to experiencethe fear-inducing situation without anxiety.

Psychotherapy. Play therapy in a well-equippedplayroom was used for children aged 6-10 assigned tothe psychotherapy treatment group. Older childrenwere seen in the playroom or in interview therapy,whichever seemed more appropriate for the child.Overall, a strategy much like that outlined by Lippman(1956) was followed. Work with parents and those out-side the family was essentially the same as for reciprocalinhibition therapy.

Psychotherapy, whether in the playroom or inter-view room, concentrated on the child's "inner experi-ence," his hopes and his fears, particularly his aggressiveand sexual fears and dependent needs. The therapistfocused on other problems of living as well as on theprimary fear, and children were encouraged to examineand formulate both behavioral strategies for copingwith stress and the affect accompanying these efforts.Regressive fantasies and behavior were accepted andinterpretations at preconscious levels were frequentlymade, but depth interpretations were rare. For example,a child's worry while at school about his mother'shealth was more likely to be interpreted as guilt overactual forbidden behavior or as sibling jealousy ratherthan as an indication of oedipal rivalry. When secondarygains became apparent, the therapist encouraged theparents to remove such gratifications as extra sleepor television watching on school days, or the securityof mother's bed at night. In therapy sessions, childrenwere encouraged to vent their anger and disappointmentassumed to arise from these deprivations.

In brief, the child in psychotherapy was encouragedto act and "talk out" his feelings and conflicts. Empha-sis was placed on affective expression and cognitiveawareness as preconditions for developing alternativecoping mechanisms more adequate than fear.

Length of Treatment

A treatment period of 24 1-hr, sessions (three timesper week for 8 wk.) and a 6-wk. follow-up were selectedfor two reasons. First, a waiting period of about 3-mo.for the waiting list control group was tolerable formost parents and was shorter than that existing atother local treatment centers. Thus, the 14-wk. periodwas ethically defensible. Second, Wolpe (1958) reportedthat the median number of sessions for adult phobia

was 24. This number was greater than the mediannumber of sessions reported for child guidance clinics(Heinecke, 1960; Hood-Williams, I960; Imber, Frank,Nash, Stone, & Gliedman, 1957; Levitt, 1957a, 1957b,1959, 1963; Miller, 1967c; Ross & Lacey, 1961). Thus,24 sessions seemed to provide a sufficient test of boththerapies.

tization of a Phobic Child," has been directed andedited by Marion Harcourt. Distribution of the filmis currently being arranged. The authors are alsoindebted to Mrs. Harcourt for her valuable editorialcomments on this paper.

272 MILLER, BARRETT, HAMPE, AND NOBLE

Measures of Outcome

Severity scores computed from the parents' ratingsand from the primary evaluator's ratings were usedas outcome measures. The severity score was deter-mined by rating the intensity of the target phobia andits extensity (that is, the degree to which the fearaffected the child's life sphere) on a 7-point scale(1-7) and was computed as follows:5

Severity Score = V Intensity X Extensity

In addition, two measuring instruments were completedonly by the parents: The Louisville Behavior CheckList (LBCL; Miller, 1967a, b; Miller, Barrett, Hampe,& Noble, 1971a; Miller, Hampe, Barrett, & Noble,1971) and the Louisville Fear Survey for Children(LFSC; Miller, Barrett, Hampe, & Noble, 1972). TheFear Scale of the Louisville Behavior Check List andthe total score of the Louisville Fear Survey for Childrenwere also used as outcome measures. Both instrumentsmeasure the child's general tendency to focalize anxietyaround specific objects.

Parental Judgments

The ratings to give the severity score were madeby one parent, usually the mother, in the presence ofthe social worker (HN) who insured that the parentunderstood the ratings. The LBCL and the LFSC weredone at pretreatment, posttreatment, and follow-up.

Primary Evaluator

The severity score for the target phobia was judgedby a clinician (EH) called the primary evaluator. Thisrating was selected as the outcome criterion for thestudy. In making his ratings, the primary evaluatorhad available information from the intake interviewwith the social worker, his own initial interview withthe parent and child and child alone, and the resultsof the following assessment procedures: (a) LBCL;(ft) School Behavior Check List (Miller, 1972; Ross,Lacey, & Parton, 1965); (c) Child's Fear Thermometer(Walk, 1956); (d) LFSC completed by the parent andchild independently; and (e) a behavioral fear testand 5-day check. The behavioral fear test was con-structed and conducted at the first interview by theprimary evaluator, if appropriate. For example, achild who feared elevators would be asked to ride onein the presence of the evaluator. For fears such as fearof storms, the primary evaluator was forced to rely on

6 The authors are indebted to Goldine C. Gleser,statistical consultant, for suggesting the severity score,covariant analysis, and other important methodologicalissues. The multiplication of Intensity X Extensity toobtain the severity score most closely approximatesclinical reality. A phobia with high intensity and zeroextensity has no clinical interest, while phobias involv-ing both parameters increase the clinical importancegeometrically. Taking the square root of the productnormalizes the distribution assuring independence ofmean and variance.

the report of those present when the fear occurred.The 5-day check consisted of calling the family for5 consecutive week days to check on the child's reactionwhen confronted with the feared situation or object.6

During the initial phase of the project, the primaryevaluator was not informed as to the therapist and typeof treatment when making his judgments. However,as a result of budget limitations, only one qualifiedclinician was available for the primary evaluator role,and he was housed in the same small facility as the restof the project staff. Consequently, it became impossibleto keep him entirely blind. Even if all chance remarks(in the waiting room, for instance) could have beencontrolled, the primary evaluator and parents and childat the posttreatment and follow-up interviews wouldhave been under pressure to expose or conceal thisinformation. Consequently, to insure that cases wouldbe judged on an equal basis at posttreatment andfollow-up, we provided the primary evaluator with thetherapist's name and type of treatment for all children.Initial evaluations were made prior to random assign-ment. We believe the gains made in acquiring uniforminformation and in freeing the primary evaluator fromconstraints in the inquiring process offset the lossattributable to an unknown bias.

Replication of Primary Evaluator Ratings1

A sample of 36 of the primary evaluator's pretreat-ment and 16 follow-up interviews were videotaped.These videotapes and all other case material, exceptratings made by the primary evaluator at case con-ference, were made available to a clinician called theindependent rater who had no other role in the project.The independent rater was not informed of the nameof the therapist or type of treatment and rated theintensity and extensity of the target phobia. A severityscore was computed from his ratings in the same manneras for the primary evaluator.

Therapists

The senior therapist (LCM) is a Diplomate (ABPP)clinical child psychologist with 20 yr. of experience inpsychodiagnostics and psychotherapy with children.His prior use of systematic desensitization with chil-dren constituted the pilot work for this project. Thejunior therapist (CLB) received a PhD degree inclinical psychology shortly before the project beganand had minimal experience with children. He had com-pleted a study using systematic desensitization therapywith adult phobics (Barrett, 1969a).

6 Copies of all rating procedures and instrumentsused in this study are available from the authors.Please send requests to Lovick C. Miller, Child Psy-chiatry Research Center, University of Louisville Schoolof Medicine, 608 South Jackson Street, Louisville,Kentucky 40201.

7 Roger Gardner served as the independent rater.At the time, he was within 3 mo. of receiving hisPhD in clinical psychology from the University ofLouisville.

TREATMENT OF CHILDREN'S PHOBIAS 273

TABLE 3

AGREEMENT OP SEVERITY SCORE RATINGS FOR TARGET PHOBIA BY1 PRIMARY EVALUATOR AND INDEPENDENT RATER

Rating period

PretreatmentFollow-upAll ratings

Agreement within indicated rating scale range

0

n

37

10

%

84319

±.l-.5

n

112

13

%

301225

±.6-1.0

n

114

15

%

302529

±1.1-1.5

n

91

10

%

256

19

1.5

«

224

%

6128

N

361652

The social worker (HN) began by screening thechild's parents by telephone. If the child's problemsseemed to indicate phobia, an intake interview wasscheduled. Parents were asked to complete the LBCL,the LFSC, and to give demographic information.Videotaping was discussed and written permission wassecured. When the intake interview further indicatedphobia, the primary evaluator interviewed the parentsand child together and then the child alone. Beforecompleting the interview with the child, the primaryevaluator conducted the behavioral fear test, if prac-tical. Following the primary evaluator's interview,the 5-day check was made. Also a School BehaviorCheck List was requested from the child's teacher,and psychological testing (Wechsler Intelligence Scalefor Children, Bender-Gestalt, Wide Range Achieve-ment Test) was done (see Footnote 6). At case confer-ence, the primary evaluator and the social workerpresented their findings to the research team. Theteam members independently rated the fears, and aseverity score was computed for each. If three of thefour conferees agreed that the child was phobic, andif the mean severity of the phobia was 3.0 or higher,the child was accepted.

Next, the child was assigned at random to therapistand treatment. Random assignment was conducted bythe social worker according to a prearranged schedule,keeping an approximate balance between younger(6-10) and older (11-15) males and females in allgroups. Following assignment, parents were seen againby the social worker who explained the treatmentprocedures and obtained the parent's written agree-ment to bring the child three times weekly for 8 wk.(or for 24 sessions) and for later evaluations, After8 wk. for the waiting list controls, or after 24 treat-ment sessions for the reciprocal inhibition and psycho-therapy groups, the posttreatment interview with thearimary evaluator was scheduled. The procedure inthe postinterview was essentially similar to the pre-interview. At this time, a School Behavior Check Listwas requested from the child's teacher. The socialworker again obtained the parent's severity scores for;he child's phobias, the LBCL, and the LFSC. Sixveeks after postinterview, the follow-up interview was:onducted using the above procedures.

At follow-up, if any children in the control groupwere found to still need treatment, it was given by ateam member. The social worker helped parents findadditional treatment from agencies or private practi-tioners for children who had been treated unsuccess-fully. Regardless of outcome of treatment, children andparents were asked to agree to our contacting them fromtime to time for follow-up.

RESULTS 8

Table 3 shows agreement of primary evalua-tor and independent rater severity scores forthe target phobia. For 48 of the 52 cases(92%), agreement was within ±1.5 points ona 7-point scale. In 38 cases (72%), agreementwas within ±1.0. Table 3 also shows thatthere was greater absolute agreement at follow-up. This was due to the shift from a normaldistribution of ratings at pretreatment to abimodal restricted range distribution at fol-low-up. Thus, the probability of absoluteagreement was higher at follow-up than atpretreatment.

A median test (Siegel, 1956, pp. 111-116)on the two raters' severity score ratings wasnot significant (X2 < 1.00, df=\,p> .05).Thus, we assume that primary evaluator'sratings represent clinical judgments of phobicseverity in children.

Figure 1 shows the mean primary evaluatorseverity scores for the target phobia at pre-treatment, posttreatment, and follow-up, andTable 4 gives means and standard deviationsfor all groups.

8 The assistance of Kay Scott, University of LouisvilleSchool of Medicine Computer Center, in performingmost of the analyses for this paper is gratefullyacknowledged.

274 MILLER, BARRETT, HAMPE, AND NOBLE

70

5.0

8 4.0

3.0

2.0

1.0

o--- R I T N = 21A— PT N = 23•— WLC N = 23

Pre- Post- Follow-up( 8 t h week) (14 th week)

EVALUATION

FIG. 1. Mean primary evaluator severity score forreciprocal inhibition, psychotherapy, and waiting listcontrol 5s at pretreatment, posttreatment, and follow-up.

Figure 1 and Table 4 indicate that therewas a considerable reduction in primary eval-uator severity score over time. A one-wayanalysis of variance on these scores at pre-treatment, posttreatment, and follow-upshowed that this reduction was significant(F = 69.29, df = 2/60, p < .05). However, a2 (Therapists) X 3 (Treatments) repeated-measures analysis of covariance, using thepretreatment score as the covariate, revealedno effect of treatment (F = 1.20, df = 2/60,p > .05) or of therapist (F = 2.26, df = 1/60,p > .05). Thus, the significant reduction inprimary evaluator's severity score ratingsacross time was not due to either a treatment(reciprocal inhibition therapy, psychotherapy,waiting list control) or a therapist effect.

Table 4 also shows that the parent's severityscore ratings on the target phobia reflect aview of outcome that differed from that of theprimary evaluator. A 2 X 3 repeated-mea-sures analysis of covariance, using the parentseverity score at pretreatment as the covariate,showed a significant effect of treatment(F = 4.47, df = 2/58, p < .05) but no effectof therapist (F = 1.41, df = 1/58, p > .05).There was no significant interaction of treat-ment and therapist (F < 1.0). Orthogonal com-

parisons of mean parent severity scores showedthat reciprocal inhibition therapy and psy-chotherapy did not differ in their effect(F < 1.00, df = 1/56, p > .05, at posttreat-ment and follow-up). These scores did show,however, that both therapies were more effec-tive than waiting list control procedures (post-treatment F = T.93,df= 1/56, p < .05; follow-up F = 13.22, df = 1/56, p < .05).

The fear scale of the LBCL, which is com-pleted by parents, measures a generalizedtendency of children to respond with fear.Analysis of covariance as described aboverevealed an overall effect of treatment(F = 8.84, df = 2/58,^ < .05) but no effect oftherapist (F < 1.0, df = 1/58). Orthogonalcomparisons again showed reciprocal inhibi-tion and psychotherapy to differ from waitinglist control in mean fear scale score at post-treatment (F = 8.78, df = 1/57, p < .05) andat follow-up (F = 22.70, df = 1/57, p < .05),but not to differ from each other (F < 1.00,

TABLE 4

MEANS AND STANDARD DEVIATIONS or PRIMARYEVALUATOR AND PARENT SEVERITY SCORES TOR

TARGET PHOBIA AT PRETREATMENT, POST-TREATMENT, AND FOLLOW-UP

Assessment period

Group rated

Reciprocalinhibitiontherapy

Man

Psychodynamictherapy

MV

nWaiting list

controlMffn

All 5sMV

n

Pre-treatment

Pri-maryeval-uator

4.461.19

21

4.951.03

23

4.88.8223

4.801.03

67

Par-ent

5.491.15

20

5.651.49

20

5.551.07

23

5.561.22

63

Post-treatment

Pri-maryeval-uator

2.461.83

21

2.311.88

23

3.232.02

23

2.701.93

67

Par-ent

2.491.83

20

2.842.13

20

4.061.94

23

3.172.05

63

Follow-up

Pri-maryeval-uator

2.171.87

21

2.642.01

23

3.051.93

23

2.661.94

67

Par-ent

2.201.76

20

2.471.73

20

3.951.66

23

2.931.86

63

Note.—For parents' ratings, only 5s for whom data at allthree assessments was obtained are included.

TREATMENT OF CHILDREN'S PHOBIAS 275

df = 1/57, p > .05, at posttreatment andfollow-up). Analyses of the LFSC revealed asimilar effect of treatment (F = 3.39,df = 2/58, p < .05) but no therapist effect.

Two other results from these analyses bearmention: First, there were no variance differ-ences of the sort noted by Bergin (1963) andBarrett (1969b) for either parent or primaryevaluator ratings. Second, our choice of thecovariate design was based on clinical loreconcerning the relationship between initialand terminal severity. In this study, the corre-lation between pretreatment and follow-upseverity score was .28 for the primary evaluatorand .24 for parents. While these correlationsare significantly greater than zero, they arenot as high as clinical lore would lead us tobelieve and would not have affected our con-clusions if we had failed to covary initial values.

The experimental design did not provide forspecific assessment of an age effect, but ourexperience during the study suggested a rela-tionship between child's age and outcome oftreatment. Table 5 was constructed from theoriginal age stratification and shows the out-come by age and treatment.9 Thirty-one of38 young children, as opposed to 13 of 29older children, were successful (X2 = 7.86,df = I , p < .05). For younger children, 23 of24 (96%) who received treatment were suc-cessful, whereas only 8 of 14 (57%) of thewaiting list control children were successes(X2 = 6.42, df = 1, p < .05). No relationshipof treatment to outcome was found for theolder group. Together, reciprocal inhibitionand psychotherapy had 9 of 20 successes(45%), and waiting list control had 4 of 9(44%) successes (X* < 1.00, p > .05).

One-way analyses of variance to assessdemographic effects on success indicated thatsex, IQ, socioeconomic status, and chronicitywere unrelated to outcome. The same analysisindicated that the staff conference rating of;ne degree of initial parent motivation for

9 Analysis of variance is perhaps not appropriate forthese data since major differences in variance occurredin cells that were grouped by age and treatment.However, a one-way analysis of variance for age aloneising primary evaluator ratings showed a significantage effect with children aged 13 and above improving.ess than all other children except those aged 11, andchildren aged 11 improving less than children 10 andbelow (F = 5.35, df = 6/66, p < .05).

treatment was related to success. Childrenwhose parents were initially rated 1-2 (par-ents with high motivation) responded betterthan children whose parents were rated 5-7(low motivation).

DISCUSSION

We were able to draw no simple conclusionfrom our study of the effectiveness of reciprocalinhibition therapy, psychotherapy, and waitinglist control as methods for treating phobias inchildren aged 6-15. Our results, dependingupon which measure we accepted as a cri-terion, could support either side of the familiarcontroversy as to the effectiveness ofpsychotherapy.

Eysenck (1952, 1961) evaluating the resultsof psychotherapy with neurotic adults andLevitt (1957b, 1963) those with children con-cluded that neurotic adults and children, intime, improve, and there is little to indicate afacilitative effect of psychotherapy. Our resultsagree with this evidence for ratings made bythe primary evaluator. Second, there was amarked reduction of target phobia in all threegroups from preevaluation to follow-up.Eysenck, however, suggests that behaviortherapy may be effective, but our resultsshowed nothing to indicate an outcome differ-ence between psychotherapy and reciprocalinhibition therapy. Also, improvement occurredfor a relatively focalized disorder and with well-

TABLE 5RELATIONSHIP OP AGE TO OUTCOME or TREATMENT

TOR TARGET PHOBIA

Age andoutcome

Younger(6-10)

Success"Failure

Older(11-15)

SuccessFailure

N

Treatment

Recip-rocal

inhibi-tion

therapy

111

54

21

Psycho-dy-

namictherapy

120

47

23

Alltreated

5s

231

91144

Waitinglist

control

86

45

23

All5s

317

131667

• Success means that the child received a primary evaluatorseverity score rating of 2.9 or less on a 7-point scale (1-7) at6-wk. follow-up of treatment.

276 MILLER, BARRETT, HAMPE, AND NOBLE

differentiated treatment and control proce-dures. This is significant since it avoids thecriticism of studies which grouped resultsacross diagnostic categories and labeled somenonclinical interventions "psychotherapy"(Eisenberg & Gruenberg, 1961; Eisenberg,Gilbert, Cytryn, & Moiling, 1961). Our resultsadd new material to the controversy, for theyshowed that neither behavior therapy nor themore standard psychotherapy to which it wascompared was more effective than the controlprocedure.

A second side of the effectiveness contro-versy argues that psychotherapy does indeedwork, but instruments and evaluative modelsare not adequate to the task of measuring psy-chotherapeutic outcome (e.g., Bergin, 1963;Frank, 1969; Heinicke, 1960; Hood-Williams,1960; Meltzoff, 1969; Strupp, 1963; Strupp &Bergin, 1969). Some of our results supportthis side of the controversy as well. Parentsreported significantly lower severity scores forthe target phobia when their child had beenin either reciprocal inhibition therapy or psy-chotherapy, with either therapist, rather thanin the waiting list control group. Similarly,parents of children in therapy reported signi-ficantly lower fear scores on the LBCL andLouisville Fear Survey than did parents ofchildren in the waiting list control group.These ratings did not agree with those of ourprimary evaluator and could be explainedaway as due to the parents' knowledge thattherapy had been given by a professional orto cognitive dissonance following the effort ofparticipating with the child in therapy (Fest-inger, 1957) or to expectancy (Frank, 1963;Goldstein, 1966). However, in explaining awaythe parents' ratings, another question is raised:Who should be satisfied by psychotherapeutictreatment with children, the parent or theprofessional evaluator (Lennard & Bernstein,1971)? Further, in dismissing the parents'ratings, we raise the question of what is a validcriterion. It is now well known that subjective,objective, and physiological measures on theconstruct "fear" do not necessarily correlate(Lang, 1968, 1970), and each investigator isforced to choose what he believes best repre-sents phobic reality. In our study, the readermust choose between an evaluator labeled

"objective" and the nonobjective parent whohas much to gain and much to lose.

There is yet another facet of the controversyfor which our data is relevant since our resultsshowed the effectiveness of psychotherapy tobe a function of the age of the child. Ford andUrban (1967) suggested that the appropriatequestion is not whether psychotherapy worksor not, but: "Which set of procedures is effec-tive when applied to what kind of patientswith which sets of problems and practiced bywhich sort of therapists [p. 359]?" Our resultsshow that phobia must not be viewed as adisorder that responds similarly to treatmentacross age groups. From our data, we conclude:(a) that for young phobic children therapy ishighly effective (96%) and is superior towaiting list control; (6) that the therapist'sexperience or the type of therapy (reciprocalinhibition therapy or psychotherapy) do notaffect change, and (c) that a and b indicatethat the crucial variables affecting change areunknown. However, the failure of eithertherapy to diminish phobia in children aged11-15 leads us to conclude that the first taskis to discover an effective behavior changetechnique for this age group. An approachsuggested by Coolidge, Wilier, Tessman, andWaldfogel (1960) assumes that in older schoolphobics, the phobia may be a manifestationof a severe character disorder, rather than aneurotic adjustment as in younger children.However, there is another equally valid expla-nation, namely, that strategies available tothe therapist for influencing young childrenhave no effects on older ones. Factors such asthe child's size, increased social anxiety, in-creased information processing capacity cou-pled with the shift to formal logical operations(Piaget, 1960), the waning influence of juve-nile court, and adult intolerance for infantilecoping mechanisms remarkably alter thetherapist's strategy. It may also be that child-hood phobia, like adult phobia (Marks, 1969),will respond differently as a function of thephobic object. On the basis of our experiencein this study, we suspect that different treat-ments will need to be developed for differenttypes of phobias.

The ambiguous results obtained in thisstudy have led us to reevaluate its design. We

TREATMENT or CHILDREN'S PHOBIAS 277

intended for this study to incorporate as manyfeatures of an ideal psychotherapy design aspossible. We were able to include the following:the use of an appropriate control group;stratified random assignment of 5s to time-limited therapy; use of a representative clinicalsample with a range in age, severity of dis-order, degree of disturbance, intelligence,socioeconomic status, and variety of phobicobject; use of same-sex therapists who differedin age and experience; identical assessment ofchildren at pretreatment, posttreatment, andfollow-up of treatment; control of initial valuesby covariance; replication of primary evalua-tor's ratings by an independent rater; andtreating parents and children ethically withoutsacrificing research standards. Of these fea-tures, the most difficult for us to apply wasrandomly assigning children. This meant thateven severely phobic children who presentedsymptoms such as panic-induced flight fromhome or school, marked depression, suicidalideation, or anorexia could be assigned towaiting list control. As clinicians, we cannotoveremphasize the difficulty we had in facingthis. However, the results appear to vindicatethe procedure and should encourage futureinvestigators in randomly assigning children.

The reader may suggest a number of otherfactors that could account for our results.These may range from concern that .Ss werenot "true" phobics or that the two therapieswere overlapping in technique to concernabout technical problems such as the properapplication of systematic desensitization orthe proper timing and depth of interpretationsin psychotherapy. Though we grant the possi-ble importance of such methodological issues,these conclusions seemed clear: (a) The crucialoperations for reducing phobic behavior andthe criteria by which to measure phobia arenot yet known, (b) Older phobic children donot improve with treatment which is successfulwith younger phobics.

However, the most important conclusionwe drew from our findings was that the psy-chotherapy research model represented bythis study is itself inadequate because of itsrigidity. A consequence of applying the fac-torial covariant model in its traditional formis that the investigator remains locked into a

study long after it has become clear that somecrucial variables should have been controlledor some procedures should have been modified.Early in this study, the therapists discoveredthat they were obtaining good results quicklywith young children, but were having no suc-cess with older ones. Such an age effect wasnot anticipated, and there was no way to takeadvantage of it within the design.

The factorial covariant design is appropriateonly when both the dependent and indepen-dent variables are relatively well isolated andthe goal is to establish a functional relation-ship. At present, only psychotherapy analoguestudies (Levis, 1970) can approximate theseconditions. In our study, it was difficult toisolate psychotherapy and reciprocal inhibitiontherapy as independent variables. Even therelatively structured reciprocal inhibition ther-apy with its exciting history of analogue andclinical applications had to be applied withinthe life space constraints of children. Threeexamples illustrate this: First, the child'sdependence upon his parents necessitates,among other things, working to extinguishparent reinforcement of the child's excessivedependent and fear behaviors. Second, childrenseldom seek therapy and may have no motiva-tion for resolving the problem. Actually, enter-ing therapy may serve to reinforce avoidance.As long as the child comes to therapy andcomplies with the rituals, neither he nor hisparents may press for confrontation with thefeared stimulus. For such cases, we arranged adeadline for confrontation in order to generateanxiety which we could extinguish. Third,during treatment, the child's phobia maydecrease in significance relative to total familypathology, so that such a highly focused tech-nique becomes inappropriate. A dramatic ex-ample of this point occurred when the thera-pist was attempting to desensitize a child of afear of stairs, while unbeknownst to the thera-pist, the child was preoccupied with his recentexperience of riding in the car while his fatherwas engaged in a running gun battle withanother motorist.

For the dependent variable, we accepted theprevailing clinical view that phobia is a rela-tively invariant response to an invariant,identifiable stimulus. However, in the course

278 MILLER, BARRETT, HAMPE, AND NOBLE

of the research, it became clear that childhoodphobia is a varying response to stimuli whoseproperties may also vary. For example, a boyreferred for fear of the dark was seen by afield observer to resist going upstairs at night,yet he remained uncomplaining in a strange,totally dark room in our center. Also, we foundthat of our 46 school phobics, 13 were attendingschool periodically and 7 attended regularly,though with great distress. In order to capturethis elusive phenomenon, we used a variety ofmeasures. However, these measures gave con-flicting answers and complicated the analysisof results.

The state of knowledge in child psycho-pathology and its remediation calls for aresearch strategy that can function when vari-ables are less clearly understood but may beclarified as the study progresses. At this time,we cannot suggest the number of cases aninvestigator should consider before altering orreplacing his procedures. This largely is afunction of the clarity of the independent,dependent, and controlled variables. For ex-ample, when the probability of treatmentfailure is high, as in the case of Marks' (1969)agoraphobics and our older school phobics,small increases in the success rate would beclinically significant.

As an alternative to the factorial covariatedesign, we propose a dynamic model thatpermits rapid incorporation of feedback as theresearch progresses while maintaining the rigorof the traditional design. This dynamic modelalso insures that base rates may be accumu-lated under rigorously controlled conditionsand that treatment failures will be availableas a test for new techniques. We need, ineffect, an extended series of pilot studies con-tinuing until an effective treatment for aspecific disorder is achieved. In our view,failures from a relatively distinct treatmentprocedure represent a more relevant S matchthan would occur by matching on the manyvariables known to affect outcome (Luborsky,Chandler, Auerbach, Cohen, & Bachrach,1971). The closest approximation to this pro-posed dynamic model that we know of isMarks' work with adult phobics (Marks,1969). Our results, we believe, show clearlythat a similar sustained line of investigationis needed for childhood phobia.

REFERENCES

ANDREWS, J. D. W. Psychotherapy of phobias. Psy-chological Bulletin, 1966, 6, 455-480.

BARRETT, C. L. Systematic desensitization therapy versusimplosive therapy. Systematic desensitization therapymanual, Appendix B. (Doctoral dissertation, Uni-versity of Louisville) Ann Arbor, Michigan: Uni-versity Microfilms, 1969. No. 69-00438. (a)

BARRETT, C. L. Systematic desensitization versusimplosive therapy. Journal of Abnormal Psychology,1969, 74, 587-592. (b)

BERECZ, J. M. Phobias of childhood: Etiology andtreatment. Psychological Bulletin, 1968, 70, 694-720.

BERGIN, A. E. The effects of psychotherapy: Negativeresults revisited. Journal of Counseling Psychology,1963,10, 244-250.

COOLIDGE, J. C., WlLLER, M. L., TESSMAN, & WALD-FOGEL, S. School phobia in adolescence: A mani-festation of severe character disturbance. AmericanJournal of Orthopsychiatry, 1960, 30, 599-607.

EISENBEK.G, L., GILBERT, A., CYTRYN, L., & MOLLING,P. A. The effectiveness of psychotherapy alone andin conjunction with perphenazine or placebo in thetreatment of neurotic and hyperkinetic children.American Journal of Psychiatry, 1961,117,1088-1093.

EISENBERG, L., & GRUENBERG, E. M. The currentstatus of secondary prevention in child psychiatry.American Journal of Orthopsychiatry. 1961, 31,355-367.

EYSENCK, H. J. The effects of psychotherapy. Journalof'ConsultingPsychology, 1952, 16, 319-327.

EYSENCK, H. J. The effects of psychotherapy. In H. J.Eysenck (Ed.), Handbook of abnormal psychology.New York: Basic Books, 1961.

FESTINGER, L. A theory of cognitive dissonance. Evans-ton, 111.: Row-Peterson, 1957.

FORD, D. H., & URBAN, H. B. Psychotherapy. In P. R.Farnsworth (Ed.), Volume 18, Annual review ofpsychology. Palo Alto, Calif.: Annual Reviews, 1967.

FRANK, J. D. Persuasion and healing. New York:Schocken,1963.

FRANK, J. D. Common factors account for effectiveness.International Journal of Psychiatry, 1969, 7, 122-127.

GOLDSTEIN, A. P. Prognostic and role expectancies inpsychotherapy. American Journal of Psychotherapy,1966, 13, 38-45.

HEINICKE, C. M. Research on psychotherapy withchildren: A review and suggestions for further study.American Journal of Orthopsychiatry, 1960, 30,483-493.

HOOD-WILLIAMS, J. The results of psychotherapy withchildren; A reevaluation. Journal of ConsultingPsychology, 1960, 24, 84-88.

IMBER, S. D., FRANK, J. D., NASH, E. H., STONE, A. R.,& GLIEDMAN, L. H. Improvement and amount oftherapeutic contact: An alternative to the use ofno-treatment controls in psychotherapy. Journal ofConsulting Psychology, 1957, 21, 309-315.

LANG, P. J. Fear reduction and fear behavior: Problemsin treating a construct. In J. M. Shlien (Ed.),Research in psychotherapy. Vol. 3. Washington, D. C.:American Psychological Association, 1968.

TREATMENT OF CHILDREN'S PHOBIAS 279

LANG, P. J. Stimulus control, response control, and thedesensitization of fear. In D. J. Levis (Ed.), Learningapproaches to therapeutic behavior change, Chicago:Aldine, 1970.

LENNARD, H. L., & BERNSTEIN, A. Dilemma in mentalhealth program evaluation. American Psychologist,1971,26,307-311.

LEVIS, D. J. The case for performing research on non-patient populations with fears of small animals:A reply to Cooper, Furst, and Bridger. Journal ofAbnormal Psychology, 1970, 76, 36-38.

LEVITT, E. E. A comparison of "remainers" and"defectors" among child guidance clinic patients.Journal of Consulting Psychology, 1957, 21, 316. (a)

LEVITT, E. E. Results of psychotherapy with children:An evaluation. Journal of Consulting Psychology,1957, 21, 189-196. (b)

LEVITT, E. E. A follow-up evaluation of cases treatedat a community child guidance clinic. AmericanJournal of Orthopsychiatry, 1959, 29, 337-349.

LEVITT, E. E, Psychotherapy with children: A furtherevaluation. Behavior Research and Therapy, 1963,1, 45-51.

LIPPMAN, H. S. Treatment of the child in emotionalconflict. New York: McGraw-Hill, 1956.

LUBORSKY, L., CHANDLER, M., AUERBACH, A. H.,COHEN, J., & BACHRACH, H. M. Factors influencingthe outcome of psychotherapy: A review of quantita-tive research. Psychological Bulletin, 1971, 75,145-185.

MARKS, I. M. Fears and phobias. London: AcademicPress, 1969.

MELTZOPF, J. Effectiveness of psychotherapy is amplydemonstrated. International Journal of Psychiatry,1969, 7, 149-152.

MILLER, L. C. Louisville behavior check list for males,6-12 years of age. Psychological Reports, 1967, 21,885-896. (a)

MILLER, L. C. Dimensions of psychopathology inmiddle childhood. Psychological Reports, 1967, 21,897-903. (b)

MILLER, L. C. The problem of measurement in theevaluation of results of psychotherapy with children.Unpublished manuscript, Bulletin No. 1, Child

Psychiatry Research Center, Louisville, Kentucky,1967. (c)

MILLER, L. C. School Behavior Check List: An in-ventory of deviant behavior for elementary schoolchildren. Journal of Consulting and Clinical Psy-chology, 1972, 38, 134-144.

MILLER, L. C., BARRETT, C. L. HAMPE, E. & Noble, H.Factor structure of childhood fears. Journal of Con-sulting and Clinical Psychology, 1972, in press.

MILLER, L. C., BARRETT, C. L., HAMPE, E., & NOBLE,H. Revised anxiety scales for the Louisville behaviorcheck list. Psychological Reports, 1971, 29, 503-511.

MILLER, L. C., HAMPE, E., BARRETT, C. L., & NOBLE,H. Children's deviant behavior within the generalpopulation. Journal of Consulting and Clinical Psy-chology, 1971, 37, 16-22.

PAUL, G. L. Insight versus desensitization in psycho-therapy. Stanford, Calif.: Stanford Press, 1966.

PIAGET, J. The psychology of intelligence. Paterson, N. J. :Littlefield, Adams & Company, 1960.

RACHMAN, S. Phobias: Their nature and control. Spring-field, 111.: Charles C Thomas, 1968.

Ross, A. O., & LACEY, H. M. Characteristics of ter-minators and remainers in child guidance treatment.Journal of Consulting Psychology, 1961, 25, 420-424.

Ross, A. O., LACEY, H. M., & PARTON, D. A. Thedevelopment of a behavior check list for boys. ChildDevelopment, 1965, 36, 1013-1027.

SIEGEL, S. Nonparametric statistics. New York:McGraw-Hill, 1956.

STRUPP, H. H. The outcome problem in psychotherapy.Psychotherapy, 1963,1, 1-13.

STRUPP, H. H., & BERGIN, A. E. Some empirical andconceptual bases for coordinated research in psy-cotherapy. International Journal of Psychiatry, 1969,7(2, Whole No. 2).

WALK, R. D. Self ratings of fear in a fear invokingsituation. Journal of A bnormal and Social Psychology,1956, 52,171-178.

WEITZMAN, B. A. Behavior therapy and psychotherapy.Psychological Review, 1967, 74, 300.

WOLPE, J. Psychotherapy by reciprocal inhibition.Stanford, Calif.: Stanford University Press, 1958.

(Received May 3, 1971)