7/26/2019 10361444

1/14

TECSE 23:2 7789 (2003) 77

Using Relationship-Focused Intervention to

Enhance the SocialEmotional Functioning of

Young Children with Autism Spectrum Disorders

This study investigates the effectiveness of relationship-focused intervention onthe social and emotional well-being of children with autism spectrum disorders.Relationship-focused intervention is a general approach to developmental in-

tervention that encourages and supports parents to enhance their use of responsive in-teractive strategies during routine interactions with their children. The sample for thisstudy consisted of 20 young children diagnosed with autism or pervasive develop-mental disorder and their parents. Parents and children received weekly interventionsessions for 8 to 14 months. These sessions focused on encouraging parents to use a

Responsive Teachingcurriculum to promote childrens socioemotional development.Comparisons of pre- and postassessments indicated that the intervention was successfulat encouraging mothers to engage in more responsive interactions with their children.Increases in mothers responsiveness were associated with significant improvementsin childrens social interaction, as well as in standardized measures of their socialemotional functioning. These results indicate that relationship-focused interventionholds much promise for enhancing the socialemotional functioning of children withautism spectrum disorders.

Gerald MahoneyandFrida Perales

Case Western ReserveUniversity

Address: Gerald Mahoney, Mandel School of Applied Social Sciences, Case Western Reserve University, Cleveland, OH 44106.

Relationship-focused (RF) intervention is a general ap-proach to developmental intervention that encourages

parents to use responsive interactive strategies (e.g., takeone turn and wait; follow the childs lead) during routineinteractions with their children. This approach to interven-tion is derived from child development research reportedover the past 30 years that has consistently indicated amoderate relation between the degree in which mothersengage in responsive interactions with their children andchildrens level of cognitive functioning (Beckwith &Cohen, 1989), language (Bornstein, Tamis-LeMonda, &Haynes, 1999; Hoff-Ginsburg & Shatz, 1982), and socio-emotional behavior (Kochanska, Forman, & Coy, 1999).These findings have been reported for diverse populations

of parents, including middle-socioeconomic status (SES)parents and lower-SES parents (Beckwith & Cohen, 1989),teenage mothers (Fewell, Casal, Glick, Wheeden, & Spi-ker, 1996), and Caucasian (Tamis-LeMonda, Bornstein,Baumwell, & Melstein Damast, 1996) and Black (Brad-ley, 1989) mothers from North America and Europe(Vereijken, Ricksen-Walraven, & Kondo-Ikemura, 1997),and Japan (Bornstein, 1989). The children in these stud-

ies have included typically developing children, childrenat risk due to prematurity or poor environmental condi-

tions, children with mild and moderate developmentaldelays (Mahoney, Finger, & Powell, 1985), and childrenwith autism spectrum disorders (ASD; Siller & Sigman,2002).

Consistent with these research findings, several inter-ventions that use similar responsive interactive strategieshave been developed to address most aspects of childrensdevelopmental functioning, including cognition (Mahoney& Powell, 1988), communication (MacDonald, 1989;Manolson, Ward, & Dodington, 1995), and socioemo-tional functioning (Greenspan & Weider, 1998). At least13 studies have reported that RF intervention can en-

hance childrens cognitive and language functioning (e.g.,McCollum & Hemmeter, 1997). However, only one studyhas been published reporting the positive effects of RFintervention on childrens socioemotional functioning(Greenspan & Weider, 1997). This study was a chart re-view of 200 children diagnosed with ASD who receivedRF intervention over a 2- to 8-year period. However, sev-eral methodological problems obscured the contribution

7/26/2019 10361444

2/14

78 Topics in Early Childhood Special Education 23:2

of RF intervention to these outcomes: (a) the interven-tion was implemented on an irregular basis, (b) there wasno documentation of how RF intervention was carriedout by parents, and (c) nonstandardized measures andclinical judgments were used to assess child outcomes.

Despite frequent recommendations for RF interven-tion for children with ASD (e.g., Dawson, 1991; Green-span, 1992; Prizant & Wetherby, 1989; Rogers & Dilalla,1991), the approach is generally considered to be a prom-ising intervention rather than best practice (Dawson& Osterling, 1997). Hesitancy to fully endorse this ap-proach not only stems from its limited empirical supportbut also is related to the lack of theoretical clarity regard-ing the mechanisms by which RF intervention promoteschildrens development and socialemotional functioning.

Child development experts typically explain the de-velopmental influence of responsive interaction in termsof its reported effect on childrens attachment relation-

ship with their mothers (De Wolff & van Ijzendoorn,1997). This focus has lead many to assume that the ef-fects of RF intervention are mediated through its signifi-cance on the attachment relationship between childrenand their parents. However, this rationale could be inter-preted as implying that RF intervention is primarily amethod for addressing deficiencies in parents interactiverelationships with their children. Yet, because empiricalevidence of deficient attachment relationships betweenmothers and their children with ASD is lacking, there isno legitimate basis for postulating that the developmen-tal or socialemotional disturbances that children withASD manifest in early childhood are related to the qual-

ity of their parents relationship with them.One alternative explanation for RF intervention is

that enhanced maternal responsiveness encourages chil-dren to learn and use the behaviors they need to attainhigher levels of socialemotional and developmental func-tioning. Regardless of the quality of their attachmentrelationship with their children, the more responsive par-ents become, the more opportunities children have tolearn the developmental behaviors they need to achievehigher levels of functioning. This explanation is compat-

ible with findings from a recent study by Siller and Sig-mund (2002). These investigators found no differences inthe quality of interactions between mothers and childrenwith autism compared with mothers and children withdevelopmental disabilities and with mothers and typicalchildren who were matched for developmental age. None-theless, the more mothers of children with autism en-gaged in responsive interaction with their children, thehigher the levels of communication functioning their chil-dren attained at 1, 10, and 12 years of age (Siller & Sig-mund, 2002).

In a similar manner, RF intervention may promotethe foundational behaviors that underlie the developmen-tal and socialemotional functioning of children whohave ASD by enhancing parents responsiveness, regard-less of whether there are inadequacies or deficiencies inthe attachment relationship between them and their chil-dren. In this study, we examine the effect of RF interven-

tion on the socioemotional behavior of children with ASDusing a newly developed early intervention curriculum,Responsive Teaching (RT; Mahoney & MacDonald, inpress). Similar to most RF intervention curricula, RT fo-cuses on teaching parents to use responsive interactionstrategies to address their childrens individualized de-velopmental needs. Several features distinguish RT fromother RF intervention models. First, RT is a comprehen-sive intervention that has 19 predefined interventionobjectives designed to address four developmental do-mains: cognition, communication, motivation, and socialemotional functioning. As indicated in Table 1, these ob-jectives, referred to aspivotal intervention objectives, are

global developmental behaviors with two characteristics:They have been reported in the research literature to beinfluenced by maternal responsiveness (e.g., Ainsworth &Bell, 1974; Carpenter, Nagell, & Tomasello, 1998; Feld-man & Greenbaum, 1997; Kochanska, 1997, 1998;Kochanska et al., 1999; Landry & Chapieski, 1989; Lead-beater, Bishop, & Raver, 1996; Mahoney, 1988a, 1988b;Mahoney et al., 1985; Mahoney, Fors, & Wood, 1990;Mahoney & Neville-Smith, 1996; Mangelsdorf, McHale,Diener, Heim Goldstein, & Lehn, 2000; Nelson, 1973;

TABLE 1. Responsive Teaching Pivotal Intervention Objectives

Developmental domain

Cognition Communication Socialemotional functioning Motivation

Social play Joint activity Trust/attachment Interest

Initiation Joint attention Empathy/intersubjectivity Persistence (success)

Exploration/manipulation Vocalization Cooperation Enjoyment

Problem solving Intentional communication Self-regulation Feelings of competence

Practice Conversation Feelings of control

7/26/2019 10361444

3/14

Relationship-Focused Intervention 79

Thomas, Chess, & Birch, 1968; van den Boom, 1994,1995), and they have been identified in contemporarychild development theory and research as being criticalprocesses for each of their respective domains of func-tioning (e.g., Ainsworth, Blehar, Waters, & Wall, 1978;Bates, 1979; Bowlby, 1969; Bruner, 1983; Piaget, 1963).

Second, RT includes a series of intervention topicsthat interventionists can use to explain the rationale of thisintervention to parents. Intervention topics describe howeach of the pivotal intervention objectives contributes tochild development. They are designed to help parents un-derstand how RT strategies are thought to promote thedevelopmental or socioemotional outcomes they desirefor their children.

Third, RT is a holistic intervention. Whereas theintervention topics discussed during each session are tai-lored to childrens individualized developmental concerns,all areas of childrens functioning are addressed at the

same time, regardless of the focus of a particular session.Because the same qualities of parental responsiveness arereportedly associated with childrens cognitive, commu-nication, and socioemotional development, RT uses thesame strategies to address each of these developmentaldomains. Thus, if intervention focuses on one domain ofdevelopment, such as communication, the RT strategiesthat parents are asked to use with their children are thesame strategies that are used to address the other threeareas of functioning.

The children who participated in this evaluation eachreceived comprehensive intervention services that weredesigned to address their full scope of developmental

needs. In this article, we focus only on the effect this inter-vention had on childrens socialemotional functioning.Other findings reported from this project indicated thatchildren with ASD made significant increases in theirrate of cognitive and language development; these im-provements were related to the degree to which parentsbecame more responsive with their children; and theseintervention effects were similar for developmentallymatched groups of children with ASD and children withdevelopmental delays who did not have ASD (Mahoney& MacDonald, in press).

This study included children and parents who re-

ceived Responsive Teachingbut did not include a com-parison group who did not receive this treatment. Tooffset the threats to validity associated with this type ofresearch design (e.g., observed treatment effects couldbe attributed to factors such as maturation or history),following the recommendations by Shadish, Cook, andCampbell (2002), intervention outcomes are analyzedin relation to the logic model of Responsive Teaching,which is depicted in Figure 1. Thus, our analysis does notonly focus on whether children made improvements onsocialemotional behaviors from pre- to postinterventionbecause these could be attributed to factors such as mat-

uration or history. Rather, the analysis focuses on whetherimprovements in socialemotional functioning that oc-curred during early intervention were associated with thekinds of changes in parental responsiveness that werepromoted through the RT curriculum.

In summary, this study is designed to examine twoquestions regarding the effect of RF intervention on thesocialemotional functioning of children with ASD dur-ing 12 months of intervention. First, do children who re-ceive the RF intervention make significant improvementsin their socialemotional functioning? Second, are chil-drens intervention improvements in socialemotionalfunctioning related to the apparent effect of the RF in-tervention on their parents responsiveness?

METHOD

Participants

Participants were 20 children who had been diagnosedby their physicians as having autism or pervasive devel-opmental disorder (PDD) and their mothers. The par-ticipants were recruited over a 15-month period fromOctober 2000 through December 2001. The sample wasreferred to the center where this project was conductedfrom three sources: (a) county service coordinators for thePart C early intervention program, (b) family physicianswho thought that the childs problems warranted earlyintervention, or (c) referrals by other parents who had at-tended the center. Eighty percent of the children were un-der 3 years of age when they enrolled, and the other 20%

were 4 and 5 years of age. Services were provided at nocost to parents.

A total of 26 children and their families were ini-tially enrolled in this study. Six of these children andfamilies were not included in the final sample; two dis-continued services prior to 8 months, and 4 had incom-plete data required for analyses.

Table 2 presents the demographic characteristics ofthe parents and families. The average age of the motherswas 34.1 years, and the majority were Caucasian (95%)and married (100%). The mothers had an average of15.5 years of education, and half worked part- or full-

time. Families had an average of 2.0 children, and themajority (60%) had incomes that were in the middle toupper-middle range.

The characteristics of the children who participatedare presented in Table 3. The children were an average of32 months of age at the start of intervention, and 60%were boys. The Transdisciplinary Play-Based Assessment(TBPA; Linder, 1993) was used to estimate the childrenscognitive and language development at the beginning ofintervention. Each of the play and social behaviors chil-dren produced during a 30- to 40-minute unstructured andsemistructured play observation were transcribed and

7/26/2019 10361444

4/14

80 Topics in Early Childhood Special Education 23:2

coded according to their developmental age level as re-ported in the Developmental Rainbow (Mahoney &Perales, 1996). Developmental ages for cognitive andlanguage development were estimated by independentraters based on the highest age level of developmental be-haviors the children consistently produced (i.e., morethan 10 times) during the course of the observation foreach of the four developmental domains. Interrater reli-ability for developmental age estimates was calculatedfor 20% of the observations, yielding a Pearson correla-

tion coefficient of 0.92. Results from the TBPA indicatedthe children had moderate delays in cognitive and lan-guage functioning at the onset of intervention.

Consistent with their diagnoses of autism or PDD,these children also exhibited severe socialemotional prob-lems as indicated by their scores on the Temperamentand Atypical Behavior Scale (TABS; Bagnato, Neisworth,Salvia, & Hunt, 1999). Their overall TABS scores wereat the 1st percentile, and their scores on three of the foursubscales were below the 10th percentile. Scores for all

FIGURE 1. Responsive teaching logic model.

7/26/2019 10361444

5/14

Relationship-Focused Intervention 81

four subscales were clinically significant, ranging from1 standard deviation (Hypersensitivity) to 3 standarddeviations (Detached).

Procedures

Parents and children received intervention during weekly1-hour individual sessions that were conducted either ina center-based setting or in parents homes. Interventionsessions were directed by one of four early interventionspecialists, each of whom had a masters degree in a dis-cipline related to working with young children with dis-abilities (two speech pathologists, a school psychologist,and an educator). Information regarding the interven-tion is presented in Table 4, which shows the interventionoccurred over a 1-year period (M = 11.4 months). Al-though parents usually were scheduled to participate inone session each week, on average they received 30.9 ses-

sions during the year. Intervention objectives addressedin these sessions were distributed across the four develop-mental domains addressed by the RT curriculum. How-ever, consistent with childrens developmental profiles,most of the sessions (76.7%) addressed cognitive and com-munication objectives.

Each intervention session focused on (a) helping par-ents learn one to two new RT strategies that they couldincorporate into their interactions with their childrenduring daily routines and (b) encouraging parents to con-tinue using strategies that they had previously learned.Each RT strategy is designed to help parents accentuateone of five different components of responsive interac-

tive behavior: reciprocity, contingency, shared control,affect, and match. Interventionists used the RT curricu-lum guide (see sample in the appendix) to individualizethe content to the childs individualized intervention ob-jectives. The curriculum guide recommends several top-ics and strategies for each pivotal intervention objectivethat can be used during a session. For each session, in-terventionists select the one to two topics and strategiesthat are most pertinent. When necessary, topics and stra-tegies presented in previous sessions are repeated. Thisprocess continues until the child makes substantial im-provement on the objective or else needs to move to a

complementary intervention objective.Most intervention sessions included a Family Ac-tion Plan that recommended intervention activities forparents to carry out with their children at home. Al-though parents reported that they were successful at fol-lowing through with these plans for more than 50% ofthe sessions, they did not follow through with recom-mendations for almost 15% of the sessions (see Table 4).

At the end of intervention, parents were asked toestimate the amount of time they devoted to carryingout early intervention activities with their children. Par-ents indicated they used RT strategies approximately

18.6 hours each week, which is an average of more than2.5 hours per day. These data suggest that parents per-ceived themselves to be carrying out RT throughoutmany, if not most, of the daily interactions they had withtheir children.

Data Collection

The data used for this study were collected at the begin-ning and end of the intervention. Preintervention data werecollected over a 2-week time frame. In the first week, there

was a 5- to 10-minute videotaped observation of moth-ers and children playing with each other using a standardset of developmentally appropriate toys. After this, moth-ers were asked to complete a questionnaire that includeda scale to assess childrens socialemotional functioning,the Infant Toddler Social Emotional Assessment(ITSEA;Carter & Briggs-Gowan, 2000). In the second week, theTABS was administered to mothers through a telephoneinterview.

Postintervention data collection began approximately1 year after childrens first assessment or after 8 monthsof intervention for children who transitioned early from

TABLE 2. Demographic Characteristics of Parents andFamilies

Variable M SD

Mothers

Age (yrs.)a 34.4 4.4

Education (yrs.) 15.5 2.6

Marital status (% married) 100.0

Race (% Caucasian) 95.0

Employed (%)

Part-time 50.0

Full-time 25.0

Fathers

Age (yrs.)a 36.0 6.0

Education (yrs.) 15.9 2.6

Family income level (%)

Low 20.0Middle 20.0

Moderate to high 40.0

# Children in family 2.0 .6

Participant child birth order (%)

First 40.0

Second 40.0

Third or later 20.0

aAge at time study began.

7/26/2019 10361444

6/14

82 Topics in Early Childhood Special Education 23:2

this program. The same procedures and measures thatwere used for the initial data collection were used at thistime.

The instruments used to assess childrens socialemotional functioning were both newly developed norm-

referenced, parent-respondent instruments. The TABSwas developed for children between 1 and 6 years of age.It has 55 items that assess parents perception of chil-drens temperament, attention, attachment, social behavior,play, vocal and oral behavior, senses and movement, andstimulation and self-injury. These items are arranged intofour subscales: detached, hypersensitive/active, underre-active, and dysregulated. This test was normed on a na-tional sample of 833 children, including 212 childrenwith disabilities. The four subscales, which were estab-lished through confirmatory factor analyses, correspondto the factor structure used for the Zero to Three Diag-nostic Classification (Greenspan & Weider, 1994). Datafrom the normative sample indicate that the TABS hashigh levels of testretest stability (r = .94) and internalconsistency (rs = .79 to .95).

The ITSEA is a 169-item, close-ended scale that as-sesses parental perceptions of internalizing (e.g., depres-

sion, withdrawal), externalizing (e.g., aggression, activity),and regulatory problems (sleeping and eating difficulties),as well as the social competencies (e.g., empathy, com-pliance) of children who are between 12 and 48 monthsof age. It was standardized on an ethnically and socioeco-nomically representative sample of 1,279 parents of chil-dren between 12 and 48 months of age. It is reported tohave high levels of testretest reliability and internal con-sistency. The test manual reports that it has moderate tohigh correlations with the Child Behavior Checklist (Ach-enbach & Rescorla, 2000).

The videotaped observations of parentchild inter-action were used to assess mothers interactive style and

childrens social interactive behavior. Mothers style of in-teraction was assessed with the Maternal Behavior RatingScale (MBRS; Mahoney, Powell, & Finger, 1986; Maho-ney, 1999). This 12-item scale assesses four dimensionsof parentingresponsiveness, affect, achievement, anddirectivenessusing a 5-point Likert scale. Previous re-search has indicated that this scale assesses parentingcharacteristics that predict childrens developmental growthand is sensitive to the effects of parent-mediated inter-ventions (Mahoney & Powell, 1988; Mahoney, Boyce,Fewell, Spiker, & Wheeden, 1998). Childrens social inter-active behavior with their parents was assessed with the

Child Behavior Rating Scale (CBRS; Mahoney & Whee-den, 1998), which consists of global rating items using a5-point scale for seven behaviors: attention, persistence,interest, cooperation, initiation, joint attention, and af-fect. Previous research indicated that the behaviors mea-sured by this scale differentiate childrens interactionswith adults (teachers) during instruction and free-playand are sensitive to differences in teachers styles of in-teraction (Mahoney & Wheeden, 1998).

Each of the videotaped observations of motherchild play were coded independently by raters who hadreceived at least 40 hours of training on each scale and

TABLE 4. Intervention Data

Variable M SD

Length of intervention (mos.) 11.4 2.4Mean # of sessions 30.9 10.6

Targeted pivotal intervention objectives(% of session)

Cognition 26.8

Language/communication 49.9

Socialemotional 10.6

Motivation 12.6

Parent follow-through with familyaction plans (%)

None to marginal 14.6

Fair 34.5Excellent 50.9

Hours parents devoted tointervention activities (per wk) 18.6 18.1

TABLE 3. Child Characteristics

Children with ASDa

Variable M SD

Age (mos.)b 32.1 7.1

Boys (%) 60.0

Cognitive age (objective abilities)c, d 16.5 6.0

Cognitive age (symbolic skills)c, d 15.3 5.3

Expressive language agec, d 13.8 6.7

Receptive language agec, d 12.0 7.2

Socialemotional characteristics

Detachede 20.0 21.1

Hypersensitivity/hyperactivitye 39.6 14.6

Underreactivee 35.7 12.8

Dysregulatede 33.6 15.9

Overall atypical behaviore 55.1 36.3

aN= 20. bAge at time study began. cDevelopmental age in months.dTransdisciplinary Play-Based Assessment(Linder, 1993). eTemperamentand Atypical Behavior Scale (Bagnato, Neisworth, Salvia, & Hunt, 1999).Subscale scores are tscores with a mean of 50 and a standard deviation of10. The overall Atypical Behavior Index is a standard score with a meanof 100 and a standard deviation of 15.

7/26/2019 10361444

7/14

Relationship-Focused Intervention 83

who had attained at least 80% agreement within 1 pointon a 5-point Likert scale. Pre- and postintervention ob-servations were coded at the same time for groups offour to six participants each to ensure that the same rat-ing criteria were used for pre- and postintervention ob-servations. Observations were randomly sorted so thatpre- and postobservations for each participant would notbe coded consecutively, and the order in which pre- andpostobservations were coded was counterbalanced. Thisprocedure minimized potential bias to rate postinterven-tion measures higher than preintervention measures.

A second rater coded a random selection of 30% ofall observations to ensure that adequate levels of reliabil-ity were maintained. For the MBRS, interrater reliability,as estimated using the Spearman correlation, was r = 0.73.Raters attained 60% exact agreement and 99% agree-ment within one scale point. For the CBRS, interrater re-liability, as estimated using the Spearman correlation, was

r = 0.73. Raters attained 56% exact agreement and 100%agreement within one scale point. The level of reliabilityattained for these two scales is consistent with the levelsof reliability reported for previous studies in which thesescales were used (e.g., Mahoney et al., 1998).

RESULTS

Pre- and postintervention measures for each of the depen-dent variables are presented in Table 5. Repeated measuresof multivariate analyses of variance (MANOVA) were usedto analyze pre- and postchanges on each of the assess-

ment instruments listed in this table. Univariate analysesof variance were used to identify scale items or subscalesthat contributed to significant multivariate effects.

Effects on Mothers Styleof Interacting with Their Children

Multivariate analyses indicated that intervention changesin mothers style of interaction were significant, F(4, 16)= 7.24,p < .0001, 2 = .64. Consistent with the empha-sis of the RT curriculum, univariate analyses indicted thatmothers made significant improvements in Responsive-

ness, F(1, 19) = 22.94, p < .0001, 2

= .55, and Affect,F(1, 19) = 21.84,p < .0001, 2 = .54, but did not makesignificant changes in Achievement Orientation, F(1, 19) =1.31,p > .05, 2 = .06, and Directiveness, F(1, 19) = 0.86,

p > .05, 2 =.04.Over the course of intervention, measures of Respon-

siveness and Affect increased by 35% and 27%, respec-tively, whereas measures of Achievement Orientation andDirectiveness declined approximately 13% and 4%, re-spectively. Eighty percent of the mothers increased theirlevel of responsiveness during intervention. At preinter-vention, 10 of the mothers had Responsiveness ratings

that were below the midpoint of the scale. At postinter-vention, only three mothers had ratings below the mid-point.

Effects on Childrens Social

Interactive BehaviorResults from the MANOVA indicated significant improve-ments in childrens CBRS scores from pre- to postinter-vention, F(7, 13) = 8.59,p < .001, 2 = .82. Comparedto preintervention, postintervention children had signifi-cantly higher ratings in Attention, F(1, 19) = 13.60,p