What is a prompt hierarchy in ABA?
A prompt hierarchy is an ordered set of levels of assistance - from full physical guidance through models and gestures down to independence - used consistently by a team so support can be faded as a learner acquires a skill.
A prompt is the help given to make a correct response possible: physical guidance, a model, a gesture, a verbal cue, a positional hint. Ranking them from most to least intrusive gives the team a shared vocabulary for how much help was actually given.
Teams generally work in one of two directions. Most-to-least prompting starts with enough support to guarantee success and fades it; least-to-most starts with the smallest prompt and escalates only as needed. Which one fits depends on the skill and the learner.
The goal is always independence, so the prompt level is recorded on every trial. Watching that level fade over time - not just the percentage correct - is how a BCBA knows a skill is genuinely being acquired rather than propped up.
Prompts are not all the same kind of help. Response prompts act on the learner - physical guidance, a model to imitate, a gesture, a verbal cue - while stimulus prompts change the materials themselves, for instance making the correct choice larger or closer before fading that difference away. Teams often combine them, but they fade on different schedules.
The risk the hierarchy is designed to manage is prompt dependence: a learner who waits for help rather than responding, because waiting has reliably produced assistance. It develops easily when prompts are delivered a moment too early or faded too slowly, and it shows up in data as high accuracy that never converts into independence.
Timing is a lever alongside level. Inserting a brief, consistent delay between the instruction and the prompt gives the learner room to respond independently while still guaranteeing success if they do not - and it produces useful information, since a correct response during the delay is evidence the skill is emerging.
This is where prompt-level data earns its keep. Percentage correct alone can look identical whether a learner is responding independently or being fully guided, so a program that records only accuracy can appear healthy while making no progress toward independence. Recording the level of support on each trial is what makes that difference visible on the graph, and it is the signal a BCBA uses to decide whether to fade further or hold.