Key Points
- A phase is defined by what is being held constant; the moment a procedure is deliberately changed, one phase ends and the next begins.
- Comparing phases only means something if the measure stayed the same, so switching recording method partway through breaks the comparison the phases exist for.
- Phases are named, dated, and given a reason in the record, which is what both the lines on the graph and the account in a progress report are built from.
Treatment Phase Explained
Programs do not run under one set of conditions from beginning to end. A treatment phase is a period during which the conditions are held constant on purpose - no change of procedure, schedule, prompt level, or setting - so the data gathered inside it describe a single arrangement. When something is changed deliberately, that phase closes and a new one opens.
The familiar sequence begins with baseline, where measurement runs with no teaching procedure in place, and moves into intervention once the baseline data have settled. After that, phases record the program's actual history: a prompt level faded, a reinforcement schedule thinned, a new instructor or setting introduced, and eventually maintenance, where the target is probed periodically rather than taught.
Keeping phases clean matters because all of a program's evidence is a comparison between them. Changing two things at once leaves nobody able to say which one worked, and changing how the behavior is measured is worse still: a baseline taken by frequency and a treatment phase taken by interval recording produce two numbers that were never comparable in the first place.
How long a phase runs is a data decision rather than a calendar one. It continues until the pattern in it is readable - a stable level, or a direction clear enough to act on - and it ends when it has answered its question or plainly stopped working. Ending a phase on one unrepresentative session is how a team ends up chasing noise; leaving a failing phase running for weeks is the more common and more expensive version of the same mistake.
Each phase is named, dated, and recorded with the reason for the change, and everything downstream is built on that record: the phase change lines on the graph, the narrative in a progress report, and a supervisor's ability to see what has already been tried. Where the program lives in a system that versions it, the phases document themselves; where it lives in a binder, they get reconstructed later from memory.
Phases are also the shape of the clinical argument. A case showing a baseline, an intervention, a documented change when the data stalled, and a maintenance phase after criterion demonstrates a program reading its own data - which is what a funder reviewing continued care is looking for, and what an undifferentiated year of sessions cannot show.