Key Points
- The line records when the program itself changed, which is what allows a shift in the data to be attributed to that change rather than to time passing.
- The data path stops at the line and restarts after it - joining the two points across it would imply a continuity the conditions do not have.
- An unlabeled line explains nothing, so each phase is captioned with the condition it ran under and every procedure change gets a line of its own.
Phase Change Line Explained
A behavior graph plots sessions along the bottom and the measure up the side, and what it means depends on knowing when the program changed. The phase change line is that annotation: a vertical line at the point one condition ended and another began - baseline to intervention, one reinforcement schedule to another, a prompt level faded, a new instructor introduced.
Convention is precise about the drawing, because the drawing carries the claim. The data path stops at the line and resumes on the other side rather than crossing it, since connecting the last baseline point to the first treatment point asserts one continuous condition. The line runs the height of the plot and falls between two sessions rather than through a data point, and each phase carries a label above it naming what was running.
The purpose is comparison. Visual analysis reads the level, trend, and variability of one phase against the next, and the line is what defines the next. Without it, forty sessions are a single undifferentiated series, and a drop that coincided exactly with a new procedure cannot be told apart from a decline that was already underway.
Because the line is a factual claim about when something happened, it is not adjustable after the fact. Sliding it a session or two to where the data look best manufactures an effect, and it is the kind of error that survives review precisely because the graph reads more cleanly for it. The matching discipline is drawing a line for every change, including the ones that did not help.
In practice the lines are only as good as the record of what changed and when. A note that a prompt level was faded on a given date is what a phase line is drawn from, so systems that version programs and stamp those edits produce graphs that annotate themselves, while a graph assembled from memory at report time reliably has fewer lines than the case actually had.
The lines also do work outside the clinical review. A progress report showing baseline, intervention, and a labeled change of procedure argues that the program responded to its own data - a different and more convincing claim than one line sloping in the right direction.