Key Points
- What matters is the direction across a run of sessions, not where the most recent point landed - one session is noise, a consistent series is evidence.
- Direction is read together with level and variability, because the same upward path means different things in steady data and in data swinging wildly.
- Behavior already improving during baseline is the classic trap: start teaching at that moment and the intervention inherits credit for a change it did not cause.
Data Trend Explained
Ask what a graph says and the answer has three parts: how high the data sit, how much they bounce around, and which way they are heading. That last part is the trend, and it is what answers whether a program is working - a target is judged by its direction across sessions rather than by wherever the latest point happened to fall.
Trend is read across a run of sessions, and drawing it helps the eye. A line of best fit through the points of a phase, or the split-middle line drawn by hand in classic single-case work, turns an impression into something two clinicians can agree about. The exact slope is rarely the point; what matters is whether the direction is consistent and whether it is the direction the goal wants - independent responses rising on an acquisition target, rate falling on a reduction target.
Direction alone can mislead, which is why it is never read alone. A gently rising path in data that swing from nothing to everything is a trend in name only, and a flat path at a high level of independence is a mastered skill rather than a stalled one. Variability is often the clinically interesting part: behavior becoming more predictable is progress that a slope does not show.
The trap is a direction that was already there before anything was done about it. Improvement underway during baseline gets credited to whatever started next, which is the practical reason baseline continues until the data settle rather than for a fixed number of sessions. The reverse case is just as easy to misread: a procedure introduced while behavior was worsening can look ineffective for weeks when it is in fact arresting a slide.
A trend is what turns data into a decision. Movement in the intended direction argues for staying the course; a flat run long enough to be believable argues for changing the procedure rather than running it longer; and movement the wrong way on a reduction target is a reason to revisit the function of the behavior rather than the teaching. Those are the calls a supervising BCBA makes at every review, and the graph is where they get made.
All of which depends on graphs that are current. Data captured in session and graphed as it arrives means a stalled target is visible in the week it stalls; data reconstructed at reporting time tends to deliver the discovery that a goal has been flat for months - the same information, arriving after the authorization period it could have changed.