Instructional Strategies and Interventions
Domain 4: Instructional Strategies and Interventions
This domain tests your ability to select and apply explicit, evidence-based teaching procedures for students who need intensive, repeated instruction. The exam favors strategies that are systematic, data-based, and designed for generalization.
Core rules and concepts
- Task analysis breaks a complex routine into teachable steps; chaining teaches the sequence. Backward chaining teaches the last step first, ensuring the student contacts reinforcement.
- Prompt hierarchies range from least to most intrusive or most to least, depending on the learner. The critical rule is planned fading; otherwise students become prompt-dependent.
- Time delay, errorless learning, and discrete trial training build accurate responding; naturalistic and incidental teaching build spontaneous use in routines.
- Video modeling, peer-mediated instruction, self-monitoring, and visual supports can improve social and functional skills.
- Reinforcement must be contingent, individualized, and thinned over time.
- Generalization is taught, not assumed: vary materials, people, settings, and cues, and program common stimuli.
Worked example
A student is learning to wash hands. The team writes a 10-step task analysis. The student can complete all but the last two steps, so they use backward chaining: staff completes steps 1-8, the student turns off the water and dries hands with a model prompt, then receives praise. Over sessions, staff fades prompts and requires earlier steps. The reasoning: the student experiences the whole routine, success is reinforced, and the teaching plan has a clear data system for fading.
Traps that cost marks
- Using a prompt hierarchy without a fading plan.
- Reinforcing every response forever instead of thinning.
- Teaching a skill in one setting and assuming it transfers.
- Using punishment before trying positive, function-based supports.
- Confusing an accommodation, such as a calculator, with instruction in the underlying skill.
- Collecting data that cannot show whether the intervention works.