Measurement, generalisation and implementation
: exploring challenges facing autism intervention trials

Student thesis: Doctoral ThesisDoctor of Philosophy


The number and quality of randomised controlled trials testing social communication interventions for young autistic children has improved significantly over the past couple of decades. However, the autism intervention field faces a number of challenges, including limited evidence of impact on broader measures of generalised social communication skills. Two of the largest trials in the field are the Preschool Autism Communication Trial (PACT) and the Paediatric Autism Communication Trial – Generalised (PACT-G). Within the context of these trials and using a range of methodologies, this thesis outlines four studies aimed at informing efforts to better support children’s development of a range of social communication skills that generalise across contexts. The research considers three components of the underlying mechanism of parent-mediated interventions, namely parents’ implementation of therapeutic techniques, generalisation of child’s learning across contexts, and the measurement of downstream generalised social communication.

Change in parent interaction style has been shown to mediate child outcomes in a range of developmental intervention approaches. The first study (Chapter 3) reports key factors that parents experienced as barriers or facilitators in their use of PACT-G techniques with their children at home. The findings highlight the potential for implementation strategies, aimed at overcoming barriers experienced by parents, to increase parents’ use of strategies and, in turn, child outcomes. Indeed, one reason for using parents as mediators of intervention is to support generalisation of skills, which autistic individuals are widely considered to have difficulties with. Despite these widespread claims, to date, there has been limited rigorous research into autistic individuals’ generalisation of higher- order skills such as communication. The second study (Chapter 4) presents a systematic review of generalisation of social communication skills following early autism interventions. The findings show evidence of generalisation of social communication skills across people and settings, suggesting that the generalisation difficulties associated with autism may not be as pervasive or profound as often implied. Further evidence of generalisation is reported in Chapter 5, in which structural equation modelling demonstrated that autistic children generalise social communication skills more strongly from home to school than vice versa, and more strongly from home into an unfamiliar research setting than from school. Finally, in Chapter 6, the comparison of intervention effects and psychometric properties of two measures of downstream global social communication skills highlights key considerations in the ongoing efforts to find a ‘sensitive’ outcome measure in early autism social communication trials. In particular, it emphasises the importance of considering the mechanism by which an intervention is hypothesised to bring about changes across different behaviours, which will inform selection of appropriate outcome measures across different interventions. To support such efforts, we illustrate ‘signature of change’ plots that can be used in meta-analyses.

Overall, this thesis highlights several areas that could be instrumental in understanding how to improve current intervention models and the design of trials in order to better support children’s development of social communication. Key implications, relative strengths and limitations, and areas for future research are discussed.
Date of Award1 Jul 2022
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorTony Charman (Supervisor) & Andrew Pickles (Supervisor)

Cite this