randomized trial of two parent-training programs for ...€¦ · been reported by webster-stratton...

Journal of Consulting and Clinical Psychology1984, Vol. 52, No. 4, 666-678

Copyright 1984 by theAmerican Psychological Association, Inc.

Randomized Trial of Two Parent-Training Programs for FamiliesWith Conduct-Disordered Children

Carolyn Webster-StrattonDepartment of Parent and Child Nursing

University of Washington

Clinic mothers of 35 conduct-disordered children were randomly assigned to awaiting list control group, 9 weeks of individual therapy, or 9 weeks of therapist-led group therapy based on a standardized videotape modeling program. Mothersand their children were assessed at baseline, immediately after treatment, and 1year later by home visits, twice-per-week telephone reports, and questionnaires.One month after treatment, both groups of treated mothers showed significantattitudinal and behavioral improvements compared with untreated controls. Ad-ditionally, the children in the two treatment groups showed reductions in childnoncompliance compared with control children. At the 1-year follow-up, not onlywere most of the changes in mothers' behaviors maintained, but both treatmentgroups of children continued to show significant reductions in noncompliant anddeviant behaviors. There were no significant differences on any of the attitudinalor behavioral measures between individual and videotape modeling group discussiontherapies at the immediate or 1-year follow-up. Total therapist time was approx-imately 251 hr for the entire individual group and 48 hr for the entire videotapediscussion group. Although both treatments seem to offer equivalent and sustainedimprovements for parents and conduct-disordered children, the therapeutic efficiencyof the videotape modeling group format is more cost-effective.

In response to the large numbers of childrenwith conduct disorders and the shortage ofprofessional personnel, there has been an in-creasing emphasis on training parents as ther-apists for their own children. Several excellentreviews (Berkowitz & Graziano, 1972; Gra-ziano, 1977; Johnson &Katz, 1973;Moreland,Schwebel, Beck, & Wells, 1982; Pawlicki,

This research was supported in part by University ofWashington School of Nursing Biomedical Research Ser-vices Grant and Graduate School Research funds.

The author appreciates the skillful assistance of BarbaraHummel in the preparation of the manuscript. The authoris also grateful to a number of people who assisted inextensive work related to data collection and data man-agement: Jeanne Bourget, Janet Cady, Melanie Calder-wood, Judy Cantor, Jayne Eriks, Maxine Fookson, DonGoldstein, Terri Hollinsworth, Margaret Jarvis, LizLeCuyer, Sharon McNamara, Judi Withers, and BerniceYates. Appreciation also goes to Matthew Speltz, who par-ticipated in the planning of the study as well as the conductof the therapy. Finally, special thanks goes to Bob Abbott,John Stratton, and Mary Hammond for statistical anddesign consultation.

Requests for reprints should be sent to Carolyn Webster-Stratton, Department of Parent and Child Nursing, SC-74, University of Washington, Seattle, Washington 98195.

1970) have concluded that behavioral parenttraining is an effective intervention for con-duct-disordered children. The majority ofthese parent training programs have beenbased on individual therapy and have incor-porated direct feedback techniques such as livemodeling, role playing, and behavioral re-hearsals. However, these individualized pro-grams are costly, time consuming, and inef-ficient and therefore are incapable of meetingthe increasing demands. In a recent reviewMclntyre et al. (1983) noted that none of the43 studies of therapy for conduct-disorderedchildren had considered an analysis of cost-effectiveness. There is a clear need to enhancethe cost efficiency of parent training programsif they are to be more widely available.

Potentially efficient and cost-effective train-ing methods can be based on group therapywith the use of standardized videotape mod-eling programs. Videotape approaches havethe advantage of mass dissemination and lowindividual training costs. Videotape modelingprograms designed to teach parents the use ofthe time-out technique have been evaluatedin four studies. Nay (1976) found videotape

666

PARENT TRAINING FOR CONDUCT-DISORDERED CHILDREN 667

modeling alone to be as good as videotapemodeling plus role playing and better thanwritten presentation or a lecture in teachingparents time-out skills. Flanagan, Adams, andForehand (1979) found videotape modeling tobe superior to written presentation, lecture,and role playing. O'Dell, Mahoney, Horton,and Turner (1979) and O'Dell, Krug, Patter-son, and Faustman (1980) found videotapemodeling plus individual checkout with atrainer to be superior to live modeling com-bined with role-played rehearsal. O'Dell,O'Quin, Alford, O'Briant, Bradlyn, and Gie-benhain (1982) also found that videotapemodeling was equal to live modeling with re-hearsal in teaching parents reinforcementskills. However, one limitation of all thesestudies was the emphasis on a single target-parent behavior rather than on evaluating theeifects in terms of child behavior changes orparent-child interactions.

Webster-Stratton(1981a, 1981b, 1982) de-veloped and evaluated a comprehensive, stan-dardized, therapist-led, videotape modelingdiscussion program that trained groups ofparents in general ways of interacting andcommunicating with children and in a varietyof operant techniques for handling behaviorproblems. The program resulted in significantattitudinal and behavioral improvements inboth the mothers and children when comparedto controls. Moreover, most of the mother andchild interactional changes were maintained1 year later.

However, despite these promising resultswith the use of videotape modeling methodsfor groups of well-educated, nonclinical par-ents, there have been no studies with less well-educated clinical parents who have childrenwith clinically significant behavioral problems.Moreover, it is unknown whether the videotapemodeling therapist-led group discussion ap-proach is as effective as the more widely uti-lized individualized parent training based ondirect feedback techniques.

The purposes of this randomized study were(a) to evaluate the short- and long-term effec-tiveness of a standardized, therapist-led, vid-eotape modeling, group discussion programin altering parent attitudes and parent-childinteractions in a clinical population and (b)to compare the effectiveness and cost efficiencyof such a videotape-based program with the

more widely utilized individualized therapythat employs feedback, rehearsal, and livemodeling.

Method

Subjects

The study was conducted in a psychiatric and behavioralclinic in a pediatric hospital. To encourage referrals, theclinic announced that it had a specialized program forthe treatment and evaluation of children with conductproblems. Subject children were screened by an intakecall followed by an office appointment. Criteria for studyentry were the following: (a) The child was between 3 and8 years old. (b) The child had no debilitating physicalimpairment, intellectual deficit, or history of psychosis,(c) The primary referral problem was the child's oppo-sitional behaviors (e.g., refusal to follow requests, tantrums,aggression), (d) Parents agreed to home visits, weekly tele-phone calls, and random assignment to specific treatmentgroups. Because the program was offered as part of aregular clinic service, the families had to pay the regulartherapy fee, from $5 to $50 per session, depending uponfamily income.

Forty families referred by pediatricians, psychiatrists,school or mental health personnel, nurses, or parentsthemselves were admitted to the study. Once subjects wereaccepted for entry, a phone call was made to the projectsecretary, who opened a sealed envelope designating theassigned group (waiting list control, individual therapy, orvideotape modeling group therapy). Families were con-tinuously assigned at random to one of the three groups.Three subjects dropped out during baseline observationsprior to starting treatment, and 2 subjects dropped outafter the first two treatment sessions. Thus data will bepresented on the 35 subjects who completed immediateposttreatment assessments.

Study children included 25 boys and 10 girls, with amean age of 4 years, 8 months. Nineteen of the 35 children(54%) were from father-absent families. The mother's meanage was 30 years and the father's was 32 years. The averagenumber of children per family was 1.9. The mean socio-economic status score was 51.8 (Social Class 4) indicatingthat the average family was lower middle to lower class,as determined by Hollingshead and Redlich's (1958) Two-Factor Index of Social Position. Fifteen of the familiesindicated that they had had recent contact with ChildProtective Services because of child abuse reports. Table1 presents the characteristics of parents and children forthe entire sample and for each of the three treatmentgroups. There were no significant differences betweengroups on any of the demographic variables.

ProcedureBaseline assessment, which consisted of two home ob-

servations, twice-per-week telephone reports of child be-haviors, and paper-and-pencil questionnaires, was com-pleted by 35 families. After baseline data collections, thetwo treatment groups attended a series of nine weeklytherapy sessions while the waiting list control group hadno parent training contact except for the twice-per-weektelephone reports of their child's behaviors. Three months

668 CAROLYN WEBSTER-STRATTON

after baseline all three groups were again retested on thesame measures. Subsequently, each parent in the waitinglist control group was randomly assigned to individualtherapy or the videotape modeling group therapy and thenassessed immediately after treatment on the same measures.To determine long-term treatment effects, 31 of 35 originalsubjects were retested on most of the same measures at1 year after treatment.

TreatmentTraining curriculum. The training curriculum was kept

as similar as possible across the two parent-training pro-grams. For the first 4 weeks, both programs included amodification of the interactional model (Hanf & KJing,1973; Kogan & Gordon, 1975), and for the last 5 weeksboth focused on teaching parents a specific set of operanttechniques. Parents in the videotape modeling, therapist-led group discussion program attended a mean of 8.5(±1.3) sessions for a total of 16.8 (±2.6) hr, and parentsin the individual treatment attended 9.1 (±.60) sessionsfor a total of 15.8 (±3.5) hr.

Training methods. Although the content and sequenc-ing of training were comparable for both programs, theprocess of training differed markedly. In the videotape

modeling, therapist-led group discussion program, parentsobserved (in groups of 8-10 parents) videotapes of modeledparenting skills. The program consisted of 180 videotapevignettes showing parents and children engaged in bothdesirable and problematic interactions at the dinner table,in the living room or bathroom, during telephone calls,and so forth. After the presentation of one or more 2-minparent-child vignettes, the therapist led a focused discus-sion of the important interactions and skills and elicitedparents' reactions, ideas, and questions about the material.Because the children did not attend the sessions, the parentsdid not receive any direct feedback on their interactionswith their children, nor did they rehearse the modeledskills. A more complete description of this program hasbeen reported by Webster-Stratton (1981b).

The individual treatment consisted of one-to-one sessionsbetween the therapist, parent, and target child. In thesesessions the therapist modeled "live" many of the parent-training skills. Parents role-played and rehearsed the mod-eled skills with their child while the therapist watchedthrough a one-way mirror and gave direct feedback to theparent via a "bug-in-the-ear." In addition to providinggeneral parent training concepts, the individual sessionsalso focused on training directly related to the target child'sspecific behavior problems.

Table 1Demographic Variables for the Randomly Assigned Treatment and Control Groups

Group

Demographic variable

Child's mean age (months)Mean number of children

in familyMother's mean age (years)Socioeconomic status"'1' (ns)

Social Class 2Social Class 3Social Class 4Social Class 5

Mean total scoreChild's sexb (ns)

MaleFemale

Marital status1" (ns)Single/divorcedMarried

Income11 (ns)Welfare$9,000-$20,000$20,000+

Child protective referral' (n)

Total(« = 35)

58.2 (17.6)

1.9(0.8)30.6 (5.7)

110131151.8

2510

1916

11111315

Waitinglist control( » = 11)

59.1 (15.4)

2.2(1.1)31.5(7.2)

0245

57.6

83

65

3534

Individualtreatment( « = 11)

62.5 (23.8)

1.5 (0.5)31.4(4.6)

0542

48.1

65

65

3446

Videotapemodelingdiscussion(n = 13)

53.7(13.1)

2.0 (0.7)29.1 (5.3)

1354

49.7

112

76

5265

Note. Numbers in parentheses are standard deviations. There were no significant differences between the three groups."Based on Hollingshead and Redlich's (1958) Two Factor Index of Social Position (education and occupation).b Reflects actual numbers of families in each category.c Number of parents who indicated at baseline that they had previously been reported to Child Protective Servicesfor child abuse.


Therapists. In order to control for therapist effectsacross the two treatment programs, there were only twotherapists (one female, one male) who both led the vid-eotape modeling discussion program and then equally di-vided the mothers in the individual treatment. Both weredoctorally trained psychologists who had specialized ed-ucation and previous experience in counseling and parenttraining. Total therapist time was approximately 251 hrfor the entire individual treatment and 48 hr for the entirevideotape group.

Parent Report MeasuresAchenbach Child Behavior Checklist (CBCL). The

CBCL, consisting of 118 behavior-problem items, has beenshown to discriminate clinic-referred from nonreferredchildren. Intraclass correlations were 0.98 for interparentagreement, 0.84 for 1-week test-retest reliability, and 0.95for interinterviewer reliability (Achenbach & Edelbrock,1981).

Eyberg Child Behavior Inventory (ECBI). The ECBI,a 36-item inventory, is applicable for children 2-16-years-old. The inventory has been shown to correlate well withindependent observations of the children's behaviors andto differentiate between clinic-referred and nonclinicalpopulations. Reliability coefficients for the ECBI scalesrange from 0.86 (test-retest) to 0.98 (internal consistency)(Eyberg & Ross, 1978; Robinson, Eyberg, & Ross, 1980).

Parent Daily Telephone Reports (PDR). The PDR de-veloped by Chamberlaine (1980) consists of a list of 19negative and 19 prosocial behaviors commonly expressedby children. During intake, parents were asked to selectthose negative and aggressive behaviors they felt were majorproblems and those positive behaviors that would be par-ticularly pleasing to them if performed by their child.These shorter, individually tailored checklists were usedas the basis for the phone calls conducted biweekly fromthe time of intake until the posttreatment assessment.During phone calls, the checklist was read to the mothers,who were then asked to report on the occurrence or non-occurrence of the specific behaviors for the previous 24hr. After asking about the positive and negative behaviorson the PDR, the interviewer then asked about the occur-rence of spanking. Families in the control group were alsocalled throughout the period of the study. All telephonecalls were made by the same interviewer throughout thecourse of the study. Previous studies (Chamberlaine, 1980;Patterson, 1974) have reported test-retest reliability of thePDR from 0.60 to 0.82.

Home ObservationsAll families were observed according to the Dyadic Par-

ent-Child Interaction Coding System (DPICS; Robinson& Eyberg, 1981). The DPICS consists of 29 separate be-havior categories covering parent and child behaviors thatare coded as present or absent for each 5-min segment.Because many of these families had only one parent livingat home and only one child, only mother-problem childdyadic interactions were analyzed. From the 29 behaviorcategories, five separate variables were formed for the be-havior of mothers: total praise (labeled and unlabeledpraise), total critical statements, total commands (directand indirect commands), total no opportunities (vague,interrupted, or chain commands given by parents in sucha way that there is no opportunity for the child to comply),

and direct command ratio (direct commands/total com-mands, including no-opportunity commands). For the tar-get child there were two variables: total child deviancy(whine + cry + hit + swear + yell + destroy), and non-compliance ratio [noncompliance/(compliance + non-compliance)]. Noncompliance was defined as failure torespond to a command within 5 s after a command wasissued.

These behavior observations were obtained by makingtwo home visits at each assessment period. All observationstook place for 30 min between 4:30 p.m.-7:30 p.m., withall family members present. Mothers were instructed tomaintain their daily routine as much as possible exceptthat they were requested to remain in two rooms, to ignorethe observer, and to avoid having visitors, making or re-ceiving telephone calls, or watching television. Home ob-servations were made by extensively trained observers whowere blind to the hypotheses and group membership ofthe subjects. To maintain accuracy, observers had weeklytraining sessions and practiced on videotaped interactions.To assess reliability, two observers were used on approx-imately 50% of all observations. Reliability was calculatedin two ways: agreements/(agreements + disagreements)and Pearson product-moment correlations between ratersfor each individual behavior dimension. Mean overall in-terrater agreement was 78.6%, and the product-momentcorrelations calculated between observers for each behaviorare shown in Table 2.

Teacher Report MeasuresBehar Preschool Questionnaire (PBQ). The PBQ was

completed only at the 1 -year follow-up assessment becauseso few of the children were in preschool the previous year.It is a 36-item measure designed to be filled out by pre-school teachers of children who are 3-7-years-old. Test-retest reliabilities have ranged from .60 to .99 (Behar, 1977).

Social Validity MeasuresConsumer satisfaction questionnaire. This measure,

adapted from the work of Forehand and McMahon (1981),consisted of 40 items with a 7-point Likert scale responseformat. This measure was given 3-4 weeks after treatmentand at 1-year follow-up and was not administered by thetherapist. Parents were told their responses would be anon-ymous. Statements were made to which the parent couldrespond from strongly agreelo strongly disagree. Responseswere transformed into scores 1 to 7, with 7 being the mostpositive. There were six subscales that measured parents'perceptions or attitudes about the following: child behaviorimprovement after treatment (11 items); format of treat-ment difficulty (5 items); treatment usefulness (5 items—live modeling, role playing, rehearsals, use of videotapes,group discussion); difficulty (7 items), and usefulness (7items) of specific parenting content and skills taught (ig-nore, time-out, play skills, commands); and the therapists(5 items). Internal consistency of the six individual subscalesranged from .71 to .90.

Results

Immediate Posttreatment Effects

Analysis of covariance (ANCOVA) was usedto statistically compare posttreatment scores


Table 2Interrater Reliability Coefficients for EachBehavior Dimension on Home Visits

Behavior category& subject

Reliability coefficient"

Range Mdn

MotherTotal commandsTotal no opportunitiesDirect commandsTotal critical statementsTotal praise

ChildTotal deviancyNoncompliance

.92-.9SJ3-.92.S3-.94.S7-.99.92-.99

.91-.98

.72-.94

.97

.83

.90

.94

.95

.91

.89

Note. Behavior categories are from the Dyadic Parent-Child Interaction Coding System (Robinson & Eyberg,1981)." Reliability coefficients computed as the median corre-lation between scores from two observers during homevisits.

adjusted for baseline values. The decision toadjust for initial baseline scores was based onthe existence of differences at baseline on somevariables that could have masked differentialtreatment effects. A one-way three-group AN-COVA was first performed on each of the de-pendent variables followed by preplannedcomparisons. Both treatment groups com-bined were compared with the control group,and then each treatment group was separatelycompared to the control group as well as toeach other. For each dependent variable theDunn-Bonferonni tables were used to deter-mine the critical values in order to correct forthe number of individual comparisons. Theerror term from the three-group ANCOVA wasemployed using Pinny's correction for ad-justing for mean differences in the covariate(Winer, 1962). Because of the small samplesizes, multivariate analysis of covariance wasnot used. The number, magnitude, and con-sistency of significant differences with the smallsizes suggest that the contribution of multi-variate tests to controlling alpha level wouldnot change the interpretation of the results.

Parent attitudinal measures. There weresignificant differences on all six of the parentreport variables by the three-group ANCOVA.Compared to controls, the combined treat-ment group of mothers reported significantly

lower scores on the Achenbach Total ChildBehavior Problem Scale, t(3l) = 2.99, p< .05;and significantly fewer total number and in-tensity of behavior problems on the EybergInventory, ;(31) = 3.11, p < .05, and £(31) =4.39, p < .001. The telephone reports of targetbehaviors also indicated that treated mothersobserved significantly fewer negative behaviorsin their children, ;(31) = 5.05, p < .001, andsignificantly more prosocial behaviors, t(3l) =2.75, p < .05, than the control mothers. Inaddition, treated mothers reported signifi-cantly less use of spanking, /(31) = 3.36,p < .01.

Further analyses compared each treatmentgroup separately to the control group. For theindividualized treatment group, as comparedto the control group, four out of six variablessignificantly improved in the predicted direc-tions. For the videotape modeling discussiongroup, as compared to the control group, fiveout of six variables improved significantly.Those variables that did not achieve signifi-cance using the conservative Bonferonni didshow change in the predicted directions. Whenthe two treatment groups were compared, nei-ther was significantly different from the otheron any of the six attitudinal variables. Table3 presents mean scores and standard deviationsfor each of the attitudinal measures.

Behavior summary variables. All five be-haviors by mothers showed significant differ-ences by three-group ANCOVA. Behavioral data(DPICS) for the combined treatment groupsrevealed that four of five behavior summaryvariables significantly changed in the predicteddirection. The combined treatment groups ex-hibited significantly fewer total commands,?(31) = 4.15, p < .001, fewer ineffective com-mands, ?(31) = 3.07, p < .05, fewer criticalstatements, t(3l) = 3.41, p < .01, and signif-icantly more praise, t(31) = 3.11, p < .05, thanthe untreated control group of mothers. Onlythe direct command ratio did not change inthe predicted direction.

Additionally, child behavior variableschanged in the predicted direction toward lesschild deviancy and noncompliance for thecombined treatment groups. Total child non-compliance was significantly lower for thetreated group than for the control group,/(31) = 3.04, p < .05. However, the noncom-pliance ratio, which took into consideration


the number of commands, was of borderlinesignificance, /(31) = 2.39, p < .08. Similarly,total child deviancy was lower, but not at astatistically significant level, /(31) = 2.19,p< .10.

Each treatment group was next separatelycompared to the control group. For the in-dividual therapy group, two out of five motherbehaviors changed significantly in the pre-dicted direction. For the videotape modelingdiscussion group, four of five changed in thepredicted direction. Each group of treatedchildren showed a significant reduction in totalchild noncompliance. However, the decreasein the noncompliance ratio in each treatmentgroup did not reach significant levels. Althoughboth treatment groups had a definite trend

toward fewer child deviancy behaviors, it wasnot statistically significant in either group.

Despite the fact that the videotape modelingtreatment group seemed to be slightly moreeffective on the behavioral measures for moth-ers, when the two treatment groups were com-pared, there were no significant differences onany of the mother or child behavioral measuresexcept for the mother direct command ratio,which was significantly lower for the videotapemodeling treatment group. Table 4 presentsthe means and standard deviations for eachof the behavioral measures.

One month after treatment, each treatmentgroup evaluated its program. All treatedmothers reported positive evaluations in termsof usability, acceptability, usefulness, and child

Table 3Attitudinat Measures Before Treatment (Pre) and Immediately After Treatment (Post)

Dunn-Bonferroni multiple comparison (t)

Parent attitude scoreand group

Achenbach ChildBehaviorInventory

CONVTGIT

Eyberg Child BehaviorInventoryProblem score

CONVTGIT

Intensity scoreCONVTGIT

Daily telephone reportsNegative behaviors

CONVTGIT

SpankingCONVTGIT

Prosocial behaviorsCONVTGIT

Means Pre

73.0 (28.8)65.0(19.1)64.9(31.6)

17.9(8.2)21.8(6.9)22.3 (4.9)

147.8 (29.9)151.6(27.2)168.7(21.3)

6.6 (4.4)7.1 (2.5)6.9(1.4)

3.2 (3.9)3.1(4.1)4.5 (5.3)

12.3(16.7)6.2(4.1)6.8 (2.5)

Unadjustedmeans post

60.741.538.9

15.19.79.0

131.5104.5112.6

6.13.23.1

2.4.18.81

13.78.38.8

(33.6)(16.2)(28.5)

(8.9)(5.9)(7.9)

(29.7)(20.6)(18.6)

(4.2)(1.3)(1.4)

(1.8)(4.0)(1.8)

(22.6)(6.1)(3.4)

Adjusted BTG vs. VTG vs. IT vs.means post CON CON CON

55.9043.70 2.99* 2.40 2.79*41.16

16.589.12 3.11* 2.61* 2.81*8.12

138.91106.51 4.39*** 3.89** 3.76**106.31

6.213.09 5.05*** 4.47*** 4.34***3.06

2.37.20 3.36** 3.46** 2.42.79

8.5511.44 2.75* 2.71* 2.1010.88

VTG vs.IT

0.50

0.05

0.02

0.06

0.94

0.53

Note. BTG = both treatment groups; CON = waiting list control (n = 11); VTG = videotape group discussion (n13); IT = individual therapy (n = 11). Numbers in parentheses are standard deviations.*p < .05. ** p < .01. *** p < .001.


improvement (Table 5). There were no sig-nificant differences between the two treatmentgroups.

One-Year Treatment Effects

One year later, long-term effects were as-sessed by retesting 31 of the 35 subjects on

most of the same measures plus one additionalteacher report measure. Of the four familieslost to follow-up, three were from the controlgroup and never showed up for treatment, andthe fourth did not return for the follow-upassessments. Thus, 97% (31 of 32) of the fam-ilies who completed the treatment programswere reevaluated at 1 year.

Table 4Home Visit Behavioral Measures Before Treatment (Pre) and Immediately After Treatment (Post)

Dunn-Bonferroni multiple comparison (t)

Behavior summaryscore and group

Mother totalcommands

CONVTGIT

No. of noopportunities torespond

CONVTGIT

Direct commandratio"

CONVTGIT

Critical statementsCONVTGIT

Total praise1'CONVTGIT

Child Totalnoncompliance

CONVTGIT

Noncomplianceratio"

CONVTGIT

Total devianceCONVTGIT

Pre

31.1 (22.5)28.5 (15.1)29.2 (29.9)

9.6 (6.3)9.3 (6.1)

11.2 (11.2)

.57(1.2)

.48(1.7)

.48(0.11)

11.5 (9.7)11.7 (6.5)10.3 (9.6)

1.7 (1.6)3.6 (3.7)2.1 (2.1)

7.77 (6.7)6.42(4.1)5.41 (4.57)

.36 (.2)

.33 (.2)

.28 (.2)

14.7 (15.7)12.5 (9.0)18.4 (24.1)

Unadjustedmeans post

44.0 (34.0)19.3 (12.2)23.5 (14.6)

20.2 (23.5)6.0 (5.4)8.6 (7.7)

.55 (.10)

.43 (.14)

.50 (.12)

17.4 (19.7)3.8 (1.7)7.8 (7.5)

2.5 (1.9)11.9 (7.1)10.7 (9.5)

7.00 (5.33)2.92 (2.94)2.50 (2.40)

.28(0.1)

.18(0.1)

.15(0.1)

11.2 (12.0)4.3 (5.1)5.4 (7.3)

Adjustedmeans BTG vs. VTG vs.post CON CON

43.0319.67 4.15*** 4.07***24.09

20.596.71 3.07* 2.81*7.46

.54

.43 0.87 2.12

.57

17.043.38 3.41** 3.76**8.68

3.5510.58 3.11* 2.66*11.21

6.512.96 3.04* 2.69*2.93

.28

.19 2.39 1.74

.16

10.704.30 2.19 2.215.86

IT vs. VTG vs.CON IT

3.19** 0.75

2.55 0.15

0.56 2.70*

2.21 1.46

2.77* 0.24

2.62* 0.02

2.22 0.58

1.60 0.54

Note. BTG = both treatment groups. CON = waiting list control (n = 11); VTG = videotape group discussion (« =13); IT = individual therapy (« = 11). Numbers in parentheses are standard deviations.a Direct command ratio is calculated by number of direct commands/total number of commands (including commandswith no opportunities to respond). b Total Praise includes labeled plus unlabeled praise. c Child noncompliance ratiois calculated by noncompliance/compliance plus noncompliance.*p<.Q5. * * / > < • 01.*** p<. 001.


Table 5Consumer Satisfaction Measure ResultsImmediately After Treatment

Group

Individualtreatment

Consumersatisfaction

subscale

Behaviorimprovementposttreatment"

Format oftreatmentdifficulty11

Format oftreatmentusefulness11

Difficulty of skills'Usefulness of skills0

Attitude towardtherapists'1

(n =

M

68.3

26.5

32.535.643.3

31.9

11)

SD

6.2

4.8

3.411.06.1

4.5

Videotapemodelingdiscussion(n =

M

69.2

28.2

32.336.646.8

33.3

13)

SD

6.6

3.9

3.59.13.4

2.5

Note. All comparisons were nonsignificant.0 Range of scores possible = 11-77. b Range of scores pos-sible = 5-35.c Range of scores possible = 7-49.

Because no significant differences existedbetween the control group after its delayedtreatment and the two treatment groups aftertheir treatments, the 1-year follow-up data andanalyses combined the delayed treatment sub-jects with their respective treatment group. Forthe 1-year follow-up data, analysis of variancewas performed to determine if there were anysignificant differences between the individu-alized and videotape modeling group discus-sion treatments. Next, repeated-measuresanalyses of covariance were conducted to de-termine if there was any interaction of timeand type of treatment program. Finally, thefollowing planned comparisons were per-formed for each treatment group: (a) pretreat-ment versus 1-year posttreatment, and (b) im-mediate posttreatment versus 1-year post-treatment. Paired / tests were performed todescribe changes over these time periods. Foreach dependent variable the Dunn-Bonfer-onni tables were again used to determine thecritical values in order to correct for the num-ber of individual comparisons.

Attitudinal measures. There continued tobe no significant differences between the twotreatment programs at the 1 -year follow-up on

any of the parent attitudinal measures or onthe Behar teacher report measure. Repeated-measures analyses of covariance revealed nosignificant interaction of time and type oftreatment effect for the attitudinal measures.Furthermore, when 1-year follow-up data werecompared with pretreatment scores, reportsby mothers from both treatment groups in-dicated significant improvements (.001) on allthree attitudinal measures. There were no sig-nificant changes from immediate posttreat-ment results to 1-year follow-up results on anyvariable, indicating that all the significant at-titudinal changes noted for both treatmentgroups of mothers immediately after treatmentwere maintained 1 year later (see Table 6).

Behavioral data. There were no significantdifferences between the two treatment groupsat 1-year follow-up on any of the mother orchild behavior measures. Repeated-measuresanalyses of covariance revealed no significantinteractions of time and type of treatment forany of the measures. When 1-year follow-updata were compared with pretreatment, anal-yses of the mother behavioral data for the vid-eotape modeling discussion group showed asignificant decrease in mothers' critical state-ments, t(l4) = 3.28, p < .05, and a significantincrease in total praise, t(l4) = 4.72, p < .01.Child behaviors showed a significant reductionin total noncompliance, t(\4) = 5.00, p < .001,noncompliance ratio, t(l4) = 3.07, p < .05,and total deviance, t( 14) = 4.62, p < .01, frompretreatment to 1-year follow-up. Likewiseanalyses of the mother behaviors for the in-dividualized therapy showed a significant re-duction in critical statements, t(l5) = 3.17,p < .05, and a significant increase in totalpraise, t(l5) = 4.00, p < .01, when 1-yearfollow-up data were compared with pretreat-ment. Child behaviors also showed a significantreduction in total noncompliance, t(\5) =4.26, p < .01, noncompliance ratio, t(l5) =3.33, ,p < .05, and total deviance, *(15) = 3.33,p < .05. In both treatment groups the totalnumber of mother commands was reducedbut not at a statistically significant level. Thespecific type of command given was not sig-nificantly reduced from pretreatment levels foreither treatment program.

For both treatment groups there were nosignificant changes in mother behaviors fromimmediate posttreatment to 1-year follow-up,


indicating that the significant mother behav-ioral improvements reported immediatelyposttreatment were maintained 1 year later.However, child noncompliance and deviancybehaviors, which had shown only borderlinedecreases immediately posttreatment, contin-ued to show decreases 1 year later (seeTable 7).

At 1 year, there were no significant differ-ences between the two treatment groups interms of the mothers' evaluations of theirtreatment programs. Moreover, there was nochange in the mothers' evaluations from im-mediate posttreatment to 1 year later. Mothersstill perceived their children's behaviors as sig-

nificantly improved and the parenting skillsthey had learned as highly acceptable anduseful.

Out of the 31 treated families, 27 (87%)expressed no major concerns about their chil-dren 1 year later and did not wish furthertherapy. However, 4 families still perceivedproblems and felt the need for continued ther-apy. Three of these families were from theindividualized therapy group and 1 from thevideotape modeling therapy. Over the previousyear these 4 families and 2 others had askedfor and received further parent training assis-tance with the project; 5 were from the in-dividualized therapy group and received an

Table 6Short- and Long-Term Results of Attitudinal Measures for the Videotape ModelingTherapy Group and the Individualized Therapy Group

Pre-treatment

Attitudinal measureand group M SD

Parent reports AchenbachChild BehaviorInventory

VTG 61.67 18.2IT 71.00 33.9

Eyeberg Child BehaviorInventory

Problem scoreVTG 19.47 7.9IT 22.37 5.6

Intensity scoreVTG 144.0 30.3IT 166.62 22.9

Consumer Satisfaction ScaleBehavior improvement'

VTGIT

Difficulty of skills"VTGIT

Usefulness of skills'1

VTGIT

Teacher reportsBehar Preschool

QuestionnaireVTGIT

Immediateposttreatment

M

37.0042.69

8.679.13

102.87115.13

69.268.3

36.635.6

46.843.3

SD

14.229.7

6.28.6

22.219.3

6.66.2

9.111.0

3.46.1

1-yearfollow-up"-11

M

36.3334.44

8.476.56

104.27117.38

67.968.7

36.437.6

44.843.3

17.4315.73

SD

17.8*25.4*

5.8*8.0*

23.3*31.3*

6.88.0

7.06.2

5.26.4

8.58.1

Note. VTG = videotape group discussion (n - 15); IT = individualized therapy (n = 16).a Comparisons between 1-year follow-up and pretreatment; critical values from Dunn's multiple comparison tests.b All comparisons between VTG and IT at 1-year follow-up were nonsignificant.c Range of possible scores = 11-77.d Range of possible scores = 7-49.* p < .001.


average of 20 hr of additional parent training, and the deviant nature of the child. The meanand 1 family was from the videotape modeling number of child behavior problems at baselinediscussion therapy and received 1 hr of ad- was 67.6, which is well above Achenbach andditional therapy. Edelbrock's (1981) cutoff score for child de-

viancy (a score of 42 represents the 90th per-Discussion centile for this a&e Sroup).

Multiple assessment procedures were usedThe principal issue addressed by this study to compare the effectiveness of the two treat-

was the comparison of a low-cost, videotape ment groups. Two findings emerged consis-modeling, therapist-led, group-discussion par- tently. First, at the immediate posttreatmentent training program to a high-cost, individ- assessment, both treatment groups of mothersualized, one-to-one parent training program showed significant attitudinal and behavioralwith a clinical population. The group studied improvements when compared with untreatedwas "high risk" from a number of viewpoints, controls. In addition, the children in the treat-including the high number of single parents, ment groups showed reductions in deviant andthe low socioeconomic status, the low mean noncompliant behaviors compared with con-education, the high prevalence of child abuse, trol children, but these changes were of bor-

Table 7Short- and Long-Term Results of Home Visit Behavioral Measures for the Videotape ModelingTherapy Group and the Individualized Therapy Group

Behavioral measureand group

Mother behaviorsTotal commands

VTGIT

Number of no opportunitiesto respond

VTGIT

Direct command ratioVTGIT

Critical statementsVTGIT

Total praiseVTGIT

Child behaviorsTotal noncompliance

VGTIT

Noncompliance ratioVTGIT

Total devianceVTGIT

Pre-treatment

M

26.9732.06

8.2711.19

.53

.50

10.1011.22

2.871.94

6.437.09

.38

.29

15.912.28

SD

14.123.9

5.710.1

.18

.10

5.810.5

2.91.9

3.86.6

.19

.19

12.213.3

Immediateposttreatment

M

19.0329.59

5.7311.56

.47

.52

4.038.06

11.6312.53

3.232.81

.23

.14

4.804.78

SD

12.021.2

5.110.7

.16

.49

2.36.8

6.39.8

2.72.6

.16

.09

4.61.4

1-yearfollow-up"'11

M

19.3024.78

8.3010.31

.42

.49

5.435.25

8.939.87

1.771.97

.18

.10

1.901.44

SD

13.224.6

8.011.4

.19

.16

4.6*5.8*

4.3**8.4**

1.4***2.9**

.13*

.08*

2.5**2.1*

Note. VTG = videotape group discussion (n = 15); IT = individualized therapy (« = 16)." Comparisons between 1-year follow-up and pretreatment; critical values from Dunn's multiple comparison tests.b All comparisons between VTG and IT at 1-year follow-up were nonsignificant.*p<.05. **p<.01.*** p<.001.


derline significance. One year later most of thechanges in the mothers and children weremaintained. Both groups of treated motherscontinued to be significantly more positive,less critical, and less negative in interactionswith their children. Both groups of motherscontinued to report their children as havingsignificantly fewer and less intense behaviorproblems. Most important, 1 year later bothtreatment groups of children showed signifi-cant reductions in noncompliant and deviantbehaviors. Thus the independent observationsof children's behaviors were congruent withthe mothers' reports of improved child be-haviors. The only deterioration from imme-diate to 1-year follow-up was a slight drop inmothers' praises, but these behaviors were stillsignificantly higher at follow-up than at base-line.

The second important finding that emergedconsistently was that there were no significantdifferences between the two treatments on anyof the attitudinal measures and on only oneof the behavioral measures immediately aftertreatment, and there were no differences onany measures at the 1-year follow-up. Threeprocedural variables distinguished the twotreatment programs: (a) context of treatment(group vs. individual), (b) modeling (live vs.videotaped), and (c) presence or absence ofrehearsal. Because the individualized programoffered direct feedback elements and greaterflexibility to focus on the parents' and child'sunique problems, it might have been expectedthat individualized therapy would have beensuperior to the videotape modeling, therapist-led discussion group, which offered less op-portunity for individual parent attention andno opportunity for direct feedback via bug-in-the-ear rehearsal. However, despite thesedifferences, the two treatments were strikinglycomparable in the changes induced in boththe attitudinal and behavioral measures. Eventhe behaviors identified on the tailored prob-lem checklist were reduced to an equal extentby the videotape modeling treatment and theindividualized therapy. Moreover, althoughother researchers (McMahon & Forehand,1983) have reported that parents are more sat-isfied in individual than in group treatmentprograms, in the current study there were nodifferences in parent evaluations between in-

dividual and videotape modeling treatments.Both groups were highly satisfied. In addition,attendance for both groups was excellent, andonly 2 treated subjects dropped out, one fromeach of the treatment groups.

Consistent with earlier videotape modelingstudies that were limited to teaching time-outand reinforcement skills (Nay, 1976; Flanaganet al., 1979; O'Dell et al., 1979; O'Dell et al.,1980; O'Dell et al., 1982), videotape modeling,therapist-led group discussion appears to behighly effective in training parents. The effec-tiveness was underscored in this study becauseparents with a wide variety of characteristicswere well trained. There are a number of pos-sible explanations for the potency of this typeof training. First, the group itself serves as apowerful source of support, reinforcement, andideas. Second, standardized videotape vi-gnettes, unlike live modeling, can portray awide variety of models in different settings andsituations. The variety of issues raised for dis-cussion by the videotapes may contribute tothe parents' ability to generalize skills to newsituations and problems, despite the lack ofdirect feedback.

More families assigned to individualizedtherapy sought out additional therapy over thesubsequent year than did parents assigned tothe videotape modeling group therapy. It ispossible that clients from individualized pro-grams are more attached to or dependent ontheir therapist to help them solve their specificparenting problems, whereas the videotapemodeling, therapist-led group discussion ap-proach may help parents solve problems moreindependently. It is also possible that parentstrained in groups may be better able to turnto other parents for advice and support whennew problems arise.

There are several limitations to the study.First, because of the ethical issue of with-holding treatment from families with conduct-disordered children, the control group wassubsequently treated after a 4-month waitingperiod. As noted earlier, both treatment groupswere significantly different from the controlsimmediately after treatment. However, at 1-year follow-up no comparisons were possiblewith a control group, because everyone hadbeen treated. Therefore it cannot be statedwith certainty that the continued mother and


child behavior improvements observed at 1year were due to the treatment programs aloneas opposed to maturational or other effects.However, the changes noted at 1 year werestrikingly similar to the immediate posttreat-ment results, strongly suggesting that the long-term changes were secondary to the program.In addition, there is considerable longitudinalresearch to suggest that conduct-disorderedchildren, if left untreated, do not outgrow theirproblems but continue to escalate in negativeand destructive interactions (Patterson, 1980).Most of the mothers and children in this studyseemed to have broken the "coercive cycle"(Patterson, 1980).

Both treatment groups consisted of severalcomponents. A second limitation of the studyis that it is unknown exactly what was thecritical ingredient in either program. In par-ticular, for the videotape program, it is un-known whether the videotapes, the group dis-cussion, or the combination was the activeagent. Likewise, it is difficult to interpret orcompare the effective components of the in-dividualized treatment.

Nonetheless, the principal purpose of thisresearch was to develop a cost-effective, widelyapplicable program of treatment for familiesof conduct-disordered children. In the past de-cade there has been little increase in the ef-ficiency of relatively costly traditional parent-training approaches. The videotape modeling,therapist-led group discussion program de-scribed in this study was more cost-effectiveand efficient than the individualized treatment.Total therapist time was approximately 251hr for the entire individual group and 48 hrfor the entire videotape group. With videotapemodeling group discussion, the same amountof therapist time (approximately 16 hr) wasused to train five or six families as was usedto train one family with individual treatment.Therefore, videotape modeling discussiongroups enable therapists to train more familiesin the same amount of time. Although someindividuals may benefit from the inclusion ofindividual training in their therapy, the datapresented here strongly suggest that videotapemodeling, therapist-led small group discussionis effective with a wide variety of parents ofconduct-disordered children.

References

Achenbach, T. M., & Edelbrock, C. S. (1981). Behavioralproblems and competencies reported by parents of nor-mal and disturbed children aged four through sixteen.Monographs of the Society for Research in Child De-velopment, 46, 1-82.

Behar, L. B. (1977). The preschool behavior questionnaire.Journal of Abnormal Child Psychology, 5, 265-275.

Berkowitz, B. P., & Graziano, A. M. (1972). Training par-ents as behavior therapists: A review. Behavior Researchand Therapy, 10, 297-317.

Chamberlaine, P. (1980). A parent daily report measure.Unpublished doctoral dissertation, University of Oregon.

Eyberg, S. M., & Ross, A. W. (1978). Assessment of childbehavior problems: The validation of a new inventory.Journal of Clinical Psychology, 16, 113-116.

Flanagan, S,, Adams, H. E., & Forehand, R. (1979). Acomparison of four instructional techniques for teachingparents to use time out. Behavior Therapy, 10, 94-102.

Forehand, R. L., & McMahon, R. J. (1981). Helping thenoncompliant child: A clinician's guide to parent training.New York: Guilford Press.

Graziano, A. M. (1977). Parents as behavior therapists.In M. Hansen, R. M. Eisler, & P. M. Miller (Eds.),Progress in behavior modification (Vol. 4, pp. 251-298).New York: Academic Press.

Hanf, C., & Kling, F. (1973). Facilitating parent-childinteraction: A two-stage training model. Unpublishedmanuscript, University of Oregon Medical School.

Hollingshead, A. B., & Redlich, F. C. (1958). Social classand mental illness (pp. 398-407). New York: Wiley.

Johnson, S. A., & Katz, R. C. (1973). Using parents aschange agents for their children: A review. Journal ofChild Psychology and Psychiatry, 4, 181-200.

Kogan, K. L., & Gordon, B. M. (1975). A mother-instruc-tion program: Documenting change in mother-childinteractions. Child Psychiatry and Human Development,5, 189-200.

McMahon, R. J., & Forehand, R. L. (1983). Consumersatisfaction in behavioral treatment of children: Types,issues, and recommendations. Behavior Therapy, 14,209-225.

Mclntyre, T. J., Bornstein, P. H., Isaacs, C. D., Woody,D. J., Bornstein, M. T., Clucas, T. J., & Long, G. (1983).Naturalistic observation of conduct-disordered children:An archival analysis. Behavioral Therapy, 14, 375-385.

Moreland, J. R., Schwebel, A. I., Beck, S., & Wells, R. T.(1982). Parents as therapists: A review of the behaviortherapy parent training literature 197 5 to 19&1. BehaviorModification, 6, 250-276.

Nay, R. W. (1976). A systematic comparison of instruc-tional techniques for parents. Behavior Therapy, 6, 14-21.

O'Dell, S. L., Krug, W. W., Patterson, J. N., & Faustman,W. O. (1980). An assessment of methods for trainingparents in the use of time out. Journal of BehaviorTherapy and Experimental Psychiatry, 11, 21-25.

O'Dell, S. L., Mahoney, N., Horton, W., & Turner, P.(1979). Media-assisted parent training: Alternativemodels. Behavior Therapy, W, 103-110.

O'Dell, S. L., O'Quin, J. A., Alford, B. A., O'Briant,


A. L., Bradlyn, A. S., & Giebenhain, J. E. (1982). Pre-dicting the acquisition of parenting skills via four trainingmethods. Behavior Therapy, 13, 194-208.

Patterson, G. R. (1974). Interventions for boys with conductproblems: Multiple setting, treatments and criteria.Journal of Consulting and Clinical Psychology, 42, 471-481.

Patterson, G. R. (1980). Coercive family processes. Eugene,OR: Castalia.

Pawlicki, R. (1970). Behavior-therapy research with chil-dren: A critical review. Canadian Journal of BehavioralScience, 2, 163-173.

Robinson, E. A., & Eyberg, S. M. (1981). The dyadicparent-child interaction coding system: Standardizationand validation. Journal of Consulting and Clinical Psy-chology, 49, 245-250.

Robinson, E. A., Eyberg, S. M., & Ross, A. W. (1980).The standardization of an inventory of child conduct

problem behaviors. Journal of Clinical Child Psychology,9, 22-28.

Webster-Stratton, C. (198 la). Modification of mothers'behaviors and attitudes through a videotape modelinggroup discussion program. Behavior Therapy, 12, 634-642.

Webster-Stratton, C. (1981b). Videotape modeling: Amethod of parent education. Journal of Clinical ChildPsychology, 10, 93-98.

Webster-Stratton, C. (1982). Long term effects of a vid-eotape modeling parent education program: Comparisonof immediate and one year followup results. BehaviorTherapy, 13, 702-714.

Winer, B. J. (1962). Statistical principles in experimentaldesign. New Ybrk: McGraw-Hill.

Received November 4, 1983Revision received February 9, 1984 •

randomized trial of two parent-training programs for ...€¦ · been reported by webster-stratton...

Documents