Abstract:
Discrete-trial teaching (DTT) is often used to teach academic and other skills to learners with intellectual disabilities. Traditionally, DTT data have been analysed using whole-sessions methods in which data collected across the duration of the study are aggregated into single-session bins. The aim of this thesis was to analyse DTT using sequential analysis, especially Markov transition matrices (MTM) and probability chains, to uncover any unsuspected sequential dependencies that may be affecting the quality of the teaching. Therapy sessions consisting of DTT were video-recorded in their typical setting (home or school) for 8 therapist-learner dyads (consisting of 8 therapists and 5 children with intellectual disabilities). From the video recordings all components of discrete trials were coded and time-stamped. The aim of Study 1 was to identify, and then correct, therapists’ within-trial treatment integrity errors. Between 110 and 1531 discrete trials per dyad were included. MTM identified treatment integrity errors for all dyads. Errors that were consistent across all dyads included learner self-corrections, response prompt errors, and incorrect application of error correction procedures. With 4 dyads, programme consultants were advised of the errors so that therapist re-training could be conducted. At follow-up, increases in treatment integrity were observed for 3 of the 4 dyads. Study 2 had three aims. First, to extend the findings from Study 1 and investigate how within-trial error-correction treatment integrity errors affected learner responding on acquisition trials; second, to introduce a method of evaluating between-trial treatment integrity for the error-correction procedures that had a between-trials component; third, to compare the relative effectiveness of concurrent and interspersed task sequencing procedures using extended Markov chain analyses. The results showed that learners were less likely to make a correct acquisition trial response following within-trial error-correction errors than if the procedure was administered as prescribed; that the between-trials error-correction treatment integrity was less than 60% for all dyads; and that learners may not be receiving the intended benefits of interspersed training. This thesis provided a demonstration of how clinically relevant information that can be obtained when DTT data are analysed on a within-sessions basis.