Assessing the Within‐Trial Treatment Integrity of Discrete‐Trial Teaching Programs Using Sequential Analysis

Brand et al. (2017) looked at how we check staff accuracy during discrete-trial teaching. They swapped the old way—counting percent correct steps—for Markov transition matrices. These matrices map every move the therapist makes inside one trial.

The team tested kids with developmental delay or intellectual disability. They recorded each second of the trial to see if the therapist gave prompts, praise, or moved on too soon.

The new matrix caught tiny errors that percent-correct scores missed. For example, a therapist who skipped praise after a right answer still looked perfect on the old score sheet. The matrix flagged the missing step.

With the clearer picture, supervisors knew exactly which staff needed retraining and on which step.

Weinsztok et al. (2022) extend this idea. They show that bigger or better reinforcers can protect DRA from staff slip-ups. Brand shows how to spot the slip; Weinsztok shows how to cushion the fallout.

Lambert et al. (2021) also fine-tune DTT. They add quick discrimination trials after FCT so kids only ask for items that are really there. Both papers push for razor-sharp trials, one by measuring, one by teaching.

O'mara (1991) used graph theory to map equivalence links. Brand uses Markov chains to map therapist moves. Both give us math tools to see what the eye alone can’t.

If you run DTT sessions, switch from percent-correct checklists to Markov matrices. You will see inside-trial errors while they happen, not after the child stalls. Pinpoint the exact step the therapist skips—prompt, praise, or pause—and retrain only that step. Cleaner data, faster fixes, better learner progress.