5.6 Procedural Fidelity, Error Correction, and Supervisor Feedback

Fidelity keeps the teaching procedure interpretable

Procedural fidelity means the procedure is implemented as written and trained. In prompting, chaining, shaping, and generalization, small deviations can change what the client learns. A prompt delivered too early may prevent independence. A prompt delivered too late in an errorless procedure may allow repeated errors. A skipped task-analysis step may make chain data inaccurate. A shaping criterion raised too quickly may reduce responding. A generalization probe with extra hints may no longer measure generalization. RBTs are responsible for implementing with fidelity within their role and for asking for direction when they are not sure how to proceed.

Fidelity starts before the first trial. The RBT should review the target, materials, discriminative stimulus, prompt hierarchy, reinforcement schedule, error-correction procedure, mastery or fading criteria, and data codes. If the data sheet has codes such as I, G, M, PP, FP, NR, and E, the RBT should know what each code means in that program. If the program says to use a three-second time delay, the RBT should know whether to count silently, use a timer, or follow a trained rhythm.

If the procedure includes physical prompting, the RBT should know the authorized prompt type, any assent or dignity safeguards, and what to do if the client resists.

Error correction is not one universal routine. In one DTT program, an error may lead to a model prompt, a brief practice response, and a transfer trial. In a chaining program, an error may be corrected at that step and the routine continues. In a shaping program, a non-criterion response may simply not contact the target reinforcer, while another appropriate response may receive different support. In a generalization probe, the plan may require recording the error without correction until the probe ends. The RBT follows the specific protocol because error correction changes learning conditions.

Supervisor feedback is part of competent RBT practice. BACB expectations for RBTs include practicing under ongoing supervision and receiving effective training that can include instructions, modeling, rehearsal, feedback, and observation of service delivery. An RBT should not treat feedback as criticism to avoid. Feedback is how the team keeps implementation aligned with the plan. A supervisor may observe that the RBT is leaning toward the correct card, repeating instructions, fading prompts too quickly, or reinforcing after the wrong response.

The RBT should listen, ask clarifying questions, rehearse the corrected step if needed, and apply the feedback in later trials.

Data integrity is closely tied to fidelity. If the RBT accidentally gives a gestural prompt and then records the response as independent, the supervisor may think the client has stronger stimulus control than they do. If the RBT marks a chain complete without noting full physical prompts, the team may fade too quickly. If the RBT does not record that a generalization trial occurred with a substitute teacher instead of the regular teacher, the data may hide an important people variable. Honest data is not about making the session look good. It is about making treatment decisions possible.

When an implementation error occurs, the RBT should handle it professionally. The RBT can continue safely within the protocol if the next step is clear, document the event according to workplace expectations, and tell the supervisor. For example, "On trial 4, I gave a model prompt immediately instead of waiting the 3-second delay; I recorded the trial as prompted and resumed the delay procedure on trial 5." That is more useful than hiding the error or changing the data. If the error affects safety, dignity, confidentiality, or client rights, the RBT follows workplace escalation procedures promptly.

The same approach applies when the procedure seems ineffective. The RBT should avoid statements such as "this plan does not work" without data. Better information includes the number of trials, prompt levels, errors, reinforcers used, motivating operations observed, setting variables, and client behavior. For example, "Across 15 trials, the client responded independently twice, needed full physical prompts on 9 trials, and pulled hands away during 4 partial physical prompts. The edible reinforcer was available but consumed only once." This gives the supervisor actionable information.

Procedural fidelity protects the client, the data, and the treatment plan. The RBT does not need to be perfect to be professional, but must be accurate, receptive to feedback, and quick to seek supervision when competence or clarity is limited. That is especially important for Chapter 5 skills because prompting, fading, chaining, shaping, and generalization can look simple while depending on precise timing, criteria, and data.