Two Programs, One Task Analysis
Marcus, 6, is learning to wash his hands independently at school. The BCBA wrote a 10-step task analysis. Your written instructions specify forward chaining: teach step 1 to criterion, then steps 1-2, and so on. You have run 22 morning sessions. Marcus performs steps 1-3 independently at 100% and then stops and waits; your data sheet shows 0% independent on steps 4-10. His mother reports that at home he washes his hands "all by himself" once she turns the water on. You also learn that the classroom aide runs the same program in the afternoon using total-task training, prompting every step, and that she has never been observed or given a fidelity checklist.
1. Is Marcus's 0% on steps 4-10 evidence of a skill deficit? Explain what the data do and do not support.
Read the worked answer
No. In forward chaining, steps 4-10 have never been taught in your session, so 0% independent is the expected product of the procedure, not a finding about Marcus's repertoire. Reporting it as a deficit confuses a programming artifact with client performance.
The 100% on steps 1-3 plus the stop-and-wait pattern is more informative: the chain has likely come under the control of the prompt sequence, with the therapist completing the rest after step 3.
The mother's report undercuts a deficit interpretation further - the behavior occurs in another context with a different SD condition (water already running). What the data do not tell you is what Marcus can do on the full chain given one opportunity and no prompts. That requires a probe.
2. What is wrong with the current arrangement, and what do you do as the technician before your next session?
Read the worked answer
Problems:
- Two incompatible teaching procedures are running for one target. Total-task prompting in the afternoon delivers prompts for every step, while forward chaining deliberately withholds them; the client is getting contradictory contingencies and the two data sets are not comparable.
- The aide has had no fidelity check, so what she calls total-task training is unknown.
- The measurement systems differ, so afternoon and morning percentages cannot be pooled.
Before the next session, as the technician:
- Do not unilaterally switch procedures. Report the conflict to the BCBA/supervisor and request a single specified procedure and a decision about the aide's program.
- Run a brief probe of the whole chain - one opportunity, no prompts, score each step - so the BCBA has data on Marcus's entry repertoire.
- Ask whether the aide should be trained and checked for fidelity, and document what you observed and what you were told.
3. The BCBA switches to total-task training with least-to-most prompting. How would you measure progress so the change is detectable, and how would you handle the mother's report?
Read the worked answer
Measure two things separately:
- Step accuracy: for each of the 10 steps, code independent / verbal / model / physical / incorrect. This lets you see whether prompting is being faded even when top-line accuracy is flat.
- Independence: percent of steps scored independent, and a separate count of whole-chain completions with no prompt at any step. Do not count prompted steps as correct on the independence measure - that is how prompt-bound progress gets misread as skill acquisition.
Add a periodic no-prompt probe so the trend is not confounded by the prompt schedule.
For the mother's report: treat it as a lead, not as data. Ask her to record what she sees (simple frequency or yes/no per handwash), or arrange a probe during a home session, and report the discrepancy to the BCBA as a generalization question - the SD conditions differ (water already running), which is a programming issue rather than a disagreement to be settled.