Behavioural Interventions and Young Children with Autism

The issue of the benefits and costs of applied behavioural treatment of children with autism is revisited in these summaries of recent findings.

The first section concerns the effects of low-intensity intervention compared to those of "eclectic" provision … albeit with questions about the significance levels of the differences observed and about their value for predicting outcomes in further cases. Reference is then made to the experience of the behavioural interventionists themselves, and the nature of the stresses associated with their role.

An alternative to ABA-style intervention is described which involves video-modelling of positive behaviours.

The outcome of the high court case in Dublin is noted (concerning the contention between a family and the Education Department over payment for individual ABA treatment as opposed to placement of the child within existing provisions).

The final section concerns the outcome of early behavioural intervention for a sample of young children with autism, monitored by a research team from Southampton University, which demonstrated effectiveness in enhancing certain areas of functioning but with variations in outcomes according to child characteristics and with no changes brought about in the core diagnostic symptoms and problem behaviours … highlighting the ongoing question of how to identify factors that best predict effectiveness.

M.J.Connor May 2007


Effects of Low-Intensity Behavioural Treatment

The report of the research completed by Eldevik et al (2006) begins with a reference to a range of findings … including those of Lovaas (1987) … which have shown that behavioural intervention can enhance the language and adaptive performance of children with autism.

One issue that has been highlighted by a number of authors concerns the intensity of the intervention (in terms of the number of hours per week during which the behavioural treatment is implemented), and it has been argued (eg by Green 1996) that positive outcomes are correlated with high intensity … perhaps as many as 40 hours per week over an extended period.

In those studies where the intensity has been pitched somewhere between 20 and 30 hours per week, there have been reported gains (in IQ measure, for example) but of a more modest degree.

One practical problem for comparing the different studies and drawing implications is the variation in the factors that may be related to outcomes, such as age at intake, initial measures of IQ, duration of treatment, and the training or supervision offered to the staff implementing the treatment.

Meanwhile, it is acknowledged that there may be practical issues which underlie arrangements where there is less intensity of intervention, such as costs or the availability of staff. It may also be the case that parents or professionals are concerned about the stress that might impact upon the child (or other members of the family) as a result of high intensity which will also limit the child’s opportunity for play and interaction with other children. In such cases, it would be a deliberate decision to combine a low intensity of behavioural treatment with time to interact with peers in a group setting.

The authors report that, as far as they have been able to see, there has been no outcome study concerned with those children with autism whose intervention programme has combined a relatively small number of hours of individual behavioural treatment (fewer than 20 hours per week) with placement in a group setting.

Accordingly, this present (retrospective) study was initiated in order to observe the effects of this kind of joint arrangement. Two groups of children were identified … a group of children who received low intensity individual behavioural treatment for two years, and a comparison group who received individual special educational treatment of an eclectic kind but of a similar intensity.

All the children had been formally diagnosed with autism, were below 6 years of age at the start of the intervention, had no medical condition that would confound the effects of the intervention, whose one to one treatment had been implemented for between 10 and 20 hours per week, and whose records included pre- and post- intervention measure of intellectual functioning, adaptive behaviour, and language.

The children were then divided into two sub-groups based upon the nature of the interventions that had been provided … an ABA group and a group who had received a combination of two or more approaches.

All the children attended mainstream kindergarten or infant school classes for at least 20 hours per week, and individual sessions were conducted in a separate room.

The behavioural group (10 boys and 3 girls) received an intervention based upon the ABA principles set out by Lovaas with initial work devoted to establishing basic tasks, like responding to simple requests from an adult, imitation of gross motor actions, matching of objects, puzzles, and shape sorting. More complex skills followed, such as imitation of fine motor or oral behaviours, and recognising objects and actions upon request; and functional use of words was taught once vocal imitation and basic receptive language had been acquired. Play and social skills were also targeted with a view to establishing eventual symbolic play and cooperative play. The basic approach was that of operant conditioning including differential reinforcement, shaping, chaining, and prompts (and prompt fading).

The eclectic group (14 boys and 1 girl) were exposed to various approaches, and all children were treated with at least two interventions drawn from alternative communication, total communication, sensori-motor therapies, TEACCH procedures, and some elements of ABA such as imitation and matching.

At intake and following intervention, the children had been tested on cognitive measures such as the Bayley Scales of Infant Development or WPPSI/WISC or Stanford-Binet Scale; the Reynell Scales had been used to assess language performance; adaptive behaviour had been assessed by the Vineland Adaptive Behavioural Scales; and non-verbal ability by subtests from the Merrill-Palmer Scales. Information about behavioural needs and symptoms were gathered from records, parental interviews, or direct observations of the children, and focused upon existence of words, level of affection shown, presence or absence of toy and peer play, stereotypical behaviours, tantrums, and progress in toilet training.

Examination of the data led to the conclusion that, after two years of treatment, the behavioural group had made statistically-significant larger gains on measures of intellectual functioning, language and communication, and behavioural pathology.

However, the general gains were small and of questionable clinical significance. As such, these results fell short of the outcomes reported elsewhere … for example, Lovaas (op.cit) had referred to the achievement of intellectual functioning falling into the average range in nearly half of his experimental group receiving intensive behavioural treatment.

The authors speculated that the difference could reflect the fact that the children monitored in this present exercise had an average pre-treatment IQ of 41 while children in most other reported studies had a higher initial mean IQ (between 50 and 65); or that the treatment provided to the current sample of children was of a much lower intensity.

On the other hand, the behavioural improvements appeared to be greater than those cited in the Lovaas control group who received less than 10 hours of behavioural treatment, and this was speculated to reflect the difference in hours of input in favour of this current group. Alternative explanations concerned, respectively, the improvements and refinements in treatment techniques that will have been developed over the intervening 20 years or so, or methodological differences/limitations which would limit the validity of direct comparisons between different study outcomes.

(For example, children in different studies may not be directly comparable on variables such as age at intake, duration of treatment, and the amount of supervision and training available to the staff. [The current writer – MJC – would cite also the possibility of "therapy drift" as a further potential confounder.])

In any event, Eldevik et al note that the eclectic group whose performance they examined did not make positive progress in IQ, language, and adaptive skills (in fact a decline in age-related scores was noted) despite access to elements of well established treatment protocols chosen to match the particular profile of needs, and of a level of intensity similar to that of the behavioural treatment provided to the other group.

However, the authors acknowledge that, given the retrospective nature of this study, they were not able to monitor the quality of implementation so that it is possible that the poorer outcomes here were not due to limitations in the eclectic interventions per se but to the way in which they were delivered.

(Nor were they able to gather information about precisely how much time was spent on the various elements of the eclectic treatment.)

These confounds and the associated questions will require further research with a view to determining what precisely underlies the apparently greater benefits of behavioural intervention … such as whether or not the significant differentiator is the use of systematic reinforcement and systematic prompt and prompt-fading procedures within the behavioural treatment.

In their concluding summary, the research team recognised limitations in the study, notably the retrospective nature of the methodology and the lack of opportunity to ensure random assignment to groups. Nevertheless, they held that the results suggest low intensive behavioural treatment to be more effective than low intensive eclectic treatment albeit not achieving the degree of improvement cited in those studies involving greater intensity; and that replacing some of the one to one behavioural treatment with time in a mainstream (kindergarten) class was not successful in improving adaptive and social skills.

The Experience of Behavioural Interventionists

The paper by Elfert and Mirenda (2006) reports on their survey of the experiences of, and stressors upon, staff providing one to one home-based behavioural treatment to young children with autism.

These authors cite the recognition that has grown in recent years of the significance of providing early intervention in the case of children with autism; and they note the wide use of applied behavioural analysis (notably the discrete trial teaching method by which to encourage new skills involving intensive one to one working).

Typically the behavioural interventionists (BI) are students within the fields of education and/or psychology who are trained to use the behavioural methods by which to enhance the children’s cognitive, communicative, and social skills.

The very nature of this role … ie the intensiveness of the work, the probability of behavioural problems in the children, the commonly slow rate of progress, and inevitable awareness of within-family anxieties and sensitivities … is thought to give rise to challenges and pressures.

There has been no previous study of the experiences of BIs, but the authors suggest that they may be comparable to at least some degree with those of other types of home-visiting professionals such as health visitors or early years advisors/educators who (as set out by Wasik and Bryant 2001) may often experience unclear role definitions, limited training, problems in maintaining professional boundaries with the family, or conflicts with other professionals involved.

They go on to describe the Stress, Strain, and Coping Model (Osipow and Davis 1988) which suggest a complex interaction between stressors associated with a given role, and the resulting strain upon an individual and his/her coping resources which can undermine that person’s work performance … and which will become a further course of stress in its own right.

Occupational stressors may include role overload, role insufficiency, role ambiguity, role boundary, responsibilities, and aspects of the physical environment.

The authors’ own study set out to determine what specific aspects of the role BIs find most stressful, the relationships between specific forms of stress and the characteristics of the children and their families, and the nature of the interaction between stressors and strain.

The participants (final number = 65) were recruited from those providing one to one intervention to one of more children in the children’s own home and using ABA principles, who had worked for at least 6 months, and who had gained permission to be involved from the director of their employing organisation.

There were 63 women and 2 men, covering an age range of 20 to 46 years, with a mean of 27 years. Their hours ranged from 4 to 44 hours per week, with a mean of 29 hours. Their training varied widely in type and quantity, and the commonest formats were workshops/lectures (mean of 32 hours) and reading training manuals (mean of 17 hours).

Information about occupational stress, psychological strain, coping resources, the behavioural symptomatology of the children, and the family environments (including cohesiveness, or achievement orientation, etc) was gathered by the use of inventories and rating scales.

The outcomes indicated that the issues of role overload and role boundary were the significantly most stressful. Role overload infers a feeling of being less than well trained or competent for the task and of needing more help, plus an awareness of working against tight deadlines.

Almost all the BIs expressed a need for more support from senior staff and more in-service training in the areas of managing difficult behaviour in the children and in coping with family issues/tensions.

The role boundary stressors are typically concerned with uncertainty about authority lines and about having a number of people issuing instructions. BIs were also consistent in describing the stress associated with trying to meet the expectations of various others and of seeking to incorporate differing goals into their work.

Lines of communication between BIs and supervisors were described as unclear.

As far as family circumstances were concerned, the pattern of responses suggested that either no specific family factors were uniformly associated with BI stress or that the rating scale did not adequately cover the characteristics of challenging families that led to BI stress.

The latter view was thought to be the more likely given that the informal contents from many BIs referred to an issue not covered in the Family Environment Scales, viz, the stress associated with their being exposed to private and sensitive information about the families (and with conflict between BIs and family members, or perceived inconsistencies and lack of supportiveness).

Meanwhile, problems commonly arose in respect of the frustration and lack of rewards linked to the efforts made because of the typical non-responsiveness of the children … no eye contact, or no developing "relatedness".

The authors acknowledge some limitations of this study, including the variation in precisely how the BIs operated, including their different hours of input, and some question whether the findings can be generalised to all families of children with autism. However, they conclude that the findings have important implications for the training and support of BIs who provide intensive behavioural treatment. The pressure upon them, as well as those upon the children and the families, are potentially considerable (and, one might also infer, the background, training, support for, and resilience of, the interventionists are further significant factors to take into account when undertaking the already complex task of evaluating the effectiveness of given programmes and of identifying factors predictive of positive outcomes).

Video Modelling Interventions

Children with autism present unique challenges to carers and teachers, hence the critical need to continue the quest for effective interventions to enhance educational progress and behaviours.

So starts the paper by Delano (2007) who describes video-modelling as a versatile intervention, with roots in social learning theory, which involves observational learning and which appears well suited to the particular styles and needs of children with autism.

As set out by Haring et al (1987), video modelling involves the watching of videotaped examples of positive behaviour of the kind that one is seeking to teach the child. The model in the video recording may be another child, or an adult; or it could be the target child him-or herself filmed when performing in the desired way (even if the edited recording could have involved lengthy and repeated sessions of filming in order to capture the desired sequence of actions).

The approach is flexible and can be used in a range of setting and to target any of a range of types of behaviour … social, functional, etc).

Delano’s paper represents the outcome of a review of relevant studies and evaluations of this approach for children with autism, and the summary data were based upon initial questions concerning the characteristics of the children and settings, the type of modelling (self or other as model), the skill areas addressed, the methods, and their effectiveness.

The articles covered were those appearing in peer-reviewed journals between 1985 and March 2005, and which described experimental studies involving an independent variable (primarily the video modelling) and valid quantitative measures of the dependent variable (the changes in the target behaviour or skill).

Not included were those studies where the intervention was multi-modal and included video-modelling as one part of the package.

19 studies met the criteria for inclusion.

In 12 studies, an adult or peer provided the model and the child with autism would, for example, watch some play activity and then be provided with the same play materials and his or her subsequent behaviour would be observed.

In 7 studies, self modelling was involved … typically requiring the child’s behaviour to be recorded over a period of time with subsequent editing to leave only the appropriate target behaviours on the final tape.

In 2 studies, the researchers sought to compare video modelling with live modelling with some indication of the greater effectiveness of video modelling.

(It was suggested by Charlop-Christy et al [2000] that the video condition could produce the better outcomes because the child is helped to focus upon the relevant cues and the procedure is rewarding in itself; and no social demands are imposed upon the child.)

The majority of studies were concerned with social-communicative behaviours, such as social imitation or language production, while others focused upon functional living skills, perspective taking (theory of mind), and the elimination of problem behaviour.

Overall, the results cited in the studies suggested that video modelling procedures can be effective in producing positive gains in all these areas, although 5 of the studies produced mixed results.

The author speculated that these latter results (from studies which emphasised modelling by another person to increase social communicative skills) could be explained in terms of there being insufficient intensity of instruction in the video procedures alone, and initiation of contact could only be evoked by the use of this approach alongside another approach. It was also considered possible that self modelling may be more effective in improving social initiations. A final hypothesis suggested that the critical matter could be the particular characteristics (weaknesses) among participant children such as poor language skill or high levels of challenging behaviour.

Meanwhile, where maintenance of behaviours and their generalisation were assessed, the results were positive.

Delano summarises by referring to the overall positive nature of the outcomes of the existing video modelling initiatives, but recognises that additional research is required in order to increase the numbers of children with autism who will have participated, and to have a wider spread of ages represented (since most existing evidence relates to children below 12 years); and also to be able to cite studies where treatment fidelity has been ensured.

The implications drawn by Delano include the probability that children with autism who can attend to a video for several minutes without resorting to difficult behaviour will respond well to video modelling; that this approach may prove valuable in dealing with some of the core deficits in autism; and that it has the advantages of ease of use in any setting and generalisability of the skills acquired.

His conclusion has it that video modelling is a valuable approach to add to the repertoire of practitioners. It may be of particular use with children who have autism and who have deficits in language and attention given the few demands made upon language and attention. Nor is there a need for direct social interaction with a teacher, while the visual format is likely to suit the autistic "style".

High Court Case

This case with, on one side, a family seeking the funding for ABA intervention for a child with autism, and, on the other, representatives of the Ministers for Education and for Health in Ireland, was described in a previous paper in this series (Applied Behaviour Analysis: Lovaas Intervention Revisited. October 2006).

With regard to that part of the case dealing with ongoing educational provision … (the family sought ABA funding: the Education and Health Departments offered provision within existing services) … the judgement delivered at the end of March 2007 involved the collation and analysis of evidence and argument which had occupied 68 days!

The judge, in his brief summation in advance of the still-awaited full exposition, stated his conclusion that … "…The evidence has not been sufficient to determine that the model of educational provision made by the Department of Education and Science for the plaintiff is not an appropriate one. I have concluded that it is an appropriate specific educational provision, and it follows that I will not be making the declarations sought in that regard or granting any mandatory injunctions as sought."

(The present writer – MJC – would infer from this that decisions about provision do typically include two elements.

Firstly, there is the very difficult matter of seeking to predict/prove that a given intervention, such as ABA, will be effective – significantly effective – for a given child. Much of the court hearing picked over details of research findings, their generalisability, and methodological issues. It seems still to be the case that effectiveness, however defined, will involve an idiosyncratic interaction of factors concerned with the child, with the staff delivering the programme, with the family, and the programme content. When one considers, further, such matters as "therapy drift", staff supervision, or stress linked with input intensity, one can appreciate the risks associated with any claim that a reported outcome observed in a sample of children with autism in place X and at time Y will be replicated in the case of a different child, in a different setting, involved with different staff and at a different time. All this still being so, the conclusion of the paper providing a literature review [Connor 1998] appears to remain appropriate.

Secondly, decisions about ongoing educational provision involve, frequently, a comparison of benefits/costs applicable to alternative approaches. The implication is for an analysis of both approaches so that, as in this present case, the Department or LEA must not only offer evidence by which to justify the doubts raised about the approach requested, but also be ready to provide chapter and verse by which to justify its own preferred approach.

The goal for all concerned – parents, researchers, and funding agencies – is to identify specific and valid factors by which to predict progress in any given type of programme, and to determine how to match a child’s demographic, cognitive and psychological profile to a particular style of intervention.)

Interim Research Report (Southampton)

A research team from the University of Southampton (Remington et al 2007) has assessed the effect of early intensive behavioural intervention for pre-school children (EIBI).

The intervention was of 2 year duration and involved the use of the behavioural approach with children in their own homes with an evaluation of outcomes.

Two groups of children with a diagnosis of autism were identified. One group received the EIBI for an average of 25 hours a week for the two year period; the second group received their LEA’s standard provision for the same period. Before and after testing was completed.

The children’s ages ranged from 30 to 42 months.

The EIBI training was delivered by a team of 3-5 tutors, with input from the parents, and covered language, cognition, play and adaptive behaviour.

The results indicated that EIBI can be effective even when the intensity is lower than that of some earlier studies and delivered in the children’s homes (with some loosening of the direct controllability of the input).

Positive changes were noted in intelligence scores, language and daily living skills. Some improvement (but not significant) was also noted in motor skills, and social/social communication skills.

Meanwhile, no evidence was reported of increased psychological tensions/stress in the parents involved in EIBI.

The findings indicated that the greater effects were observable among children with IQ or mental age scores at the higher end of the range, and with better social and communication skills.

However, apart from a small change in early social communication skills, EIBI led to few reductions in the diagnostic symptoms of autism or in problem behaviours (in contrast to parental reports of improved social behaviour).

The differences between groups were evident after one year and maintained during the subsequent years; but the size of the gains in the EIBI group did not increase by the end of that 2nd year.

The author recognised that there remain questions to be answered about the factors that predict effectiveness – which children will benefit most from which methodology – and about long-term maintenance of gains.

(In other words, there is confirmation of the likely benefits of the behavioural approach for young children with autism, albeit with the proviso that it is one of a number of approaches, that expectations might best not be pitched at too high a level in terms of change in the core characteristics, and that the matter of identifying predictive factors for given approaches remains salient.)


* * * * * * *

M.J.Connor May 2007



Charlop-Christy M., Le. L., & Freeman K. (2000). A Comparison of video modelling with in-vivo modelling for teaching children with autism. Journal of Autism and Development Disorders 30 537-552

Connor M. (1998). A review of behavioural early intervention programmes for children with autism. Educational Psychology in Practice 14(2) 109-117

Delano M. (2007) Video Modelling interventions for individuals with autism. Remedial and Special Education, 28(1), 33-42

Eldevik S., Eikeseth S., Jahr E. & Smith. T (2006). Effects of low-intensity behavioural treatment for children with autism and mental retardation. Journal of Autism and Developmental Disorders, 36(2) 211-224

Elfert M. and Mirenda P. (2006). The experiences of behaviour interventionists who work with children with autism in families’ homes. Autism, 10(6), 577-591

Green G. (1996). Early behavioural intervention for autism: what does research tell us? In G. Green (Ed) Behavioural Intervention for Young Children with Autism: A Manual for Parents and Professionals. (29-44) Austin, Texas : Pro-Ed Publishing

Haring T., Kennedy C., Adams M., and Pitts-Conway J. (1987). Teaching generalization of purchasing skills across community settings to autistic youth using videotape modelling. Journal of Applied Behaviour Analysis. 20, 89-96

Lovaas O. (1987). Behavioural Treatment and normal educational and intellectual functioning in young autistic children. Journal of Consulting and Clinical Psychology 55(1) 3-9

Osipow S. and Davis A. (1988) The relationship of coping resources to occupational stress and strain. Journal of Vocational Behaviour 32(1) 1-15

Remington B., Hastings R., Kovshoff K. et al (In press) A field effectiveness study of early intensive behavioural intervention : outcomes for children with autism and their parents after two years. American Journal on Mental Retardation

Wasik B. and Bryant D. (2001) Home Visiting : Procedures for Helping Families.

Thousand Oaks, CA : Sage Publishing

This article is reproduced by kind permission of the author.

© Mike Connor 2007.

Back to NAS Surrey Branch Welcome Page