What a task analysis is
A behavior chain is a sequence of responses in which completing each step sets up the next: turning on the tap is the cue to wet your hands, and wet hands are the cue for soap. The finished task - clean hands, a made sandwich - is the reinforcer at the end of the chain.
A task analysis is the written breakdown of that chain into its steps, in order, each described as an observable action. It serves as both the teaching plan and the data sheet: every step is taught, prompted, and scored against the same list.
How to write a task analysis
There are three common ways to build the list: watch someone competent perform the task and note each action, draw on published or expert task analyses for the skill, or perform the task yourself and write down every step as you do it. In practice clinicians often combine them, then field-test the draft with the learner.
Size the steps to the learner, not to the task. A step the learner cannot learn after reasonable teaching is too big and should be split; a learner who already does several steps fluently may not need them separated. A handwashing analysis might run: turn on the water, wet hands, get soap, rub palms, rub the backs of hands, rinse, turn off the water, dry hands. For one learner, "get soap" might need to become "put hand under the dispenser" and "press the pump".
Write each step as an action that starts with a verb, and make sure everyone who runs the program uses the same version. When two clinicians work from different step lists, their data cannot be combined or compared.
Measuring the baseline
Before teaching, find out which steps the learner can already do. In a single-opportunity probe, the learner attempts the chain and the probe stops at the first step that is missed or not attempted; every remaining step is scored as not independent. It is quick and conservative, but it says nothing about steps later in the chain.
In a multiple-opportunity probe, when the learner misses a step, the clinician completes that step or sets it up, and the learner gets the chance to perform each remaining step. It shows every step the learner already knows, at the cost of a longer probe and some risk that the learner learns from the help during the assessment.
Forward, backward, and total task chaining
Chaining procedures differ in which steps are taught first and who completes the rest. None is best for every learner or task; comparisons in the research literature are mixed, so the choice rests on the learner, the length of the chain, and where the hardest steps fall.
| Procedure | How it works | A good fit when |
|---|---|---|
| Forward chaining | The first step is taught until it is mastered while the clinician completes or prompts the rest; then the next step is added, working forward through the chain. | The early steps are the hardest, or the learner benefits from always starting at the natural beginning of the routine. |
| Backward chaining | The clinician completes every step except the last, and the learner is taught the final step first; then the second-to-last is added, working backward. | The learner benefits from finishing the task and contacting its natural reinforcer on every trial from the first session. |
| Total task chaining | The learner performs every step in every session, with prompts as needed on each, and prompts are faded step by step. Also called total task presentation. | The learner already does many of the steps, the chain is short, or the routine happens naturally and every opportunity should be used for teaching. |
Scoring task analysis data
Score every step on every opportunity with the prompt level the learner needed, using the same hierarchy as the rest of the program; see the ABA prompt hierarchy. The session value is usually the percent of steps completed independently, which is what gets graphed and compared against the mastery criterion.
Keep the step-level scores, not just the percentage. Two sessions at 60% independent can fail on entirely different steps, and the step that keeps breaking down is the most useful thing a task analysis can show. That is the main reason task analysis is a measurement system of its own rather than a target scored with a single percentage, as the eight ABA measurement systems guide explains.
Mastery criteria for chains are usually written against the whole chain - for example, all steps independent across three consecutive opportunities - while fading decisions are made step by step. In forward and backward chaining, each newly added step typically has its own criterion before the next one is introduced.
Common problems and how to avoid them
Steps that are too large produce data that never move, because the learner is failing a step that is really three. A prompt on one step can also carry into the next: a physical prompt to pick up the towel easily becomes help with drying, too, unless the clinician deliberately withdraws it between steps.
Generalization needs its own plan. A learner who washes their hands independently at one sink may stall at a different tap, soap dispenser, or towel. Probing the chain in the settings and with the materials the learner will actually use, and across the people who will see it, is what makes the skill independent in daily life rather than in the teaching room.
How Cadence records task analysis
Task analysis is one of the 8 measurement systems built into Cadence. Each step of a chain is stored in order with the prompt level the learner needed and whether it was completed independently, and the session's value is the percent of steps completed independently - the number that is graphed and evaluated against the target's mastery criterion.
Because the step-level scores are kept, the step where a routine breaks down stays visible in the record instead of disappearing into one percentage, and prompted steps use the same configurable prompt hierarchy as the clinic's other targets.