A.7 Identifying trends in graphed data

What the outline says

Task A.7 in content area A, Data Collection and Graphing (13 of the 75 scored questions): "Identify trends in graphed data."

Three things to look for

• Trend: the overall direction of the data path over time. It can be increasing, decreasing, or flat (no trend). Steepness shows how fast the data are changing.
• Level: where the data sit on the vertical axis, such as high, low, or around a certain value.
• Variability: how much the data points differ from one another. Points that jump up and down are highly variable; points that stay close together have low variability.

Data with little variability and no clear trend are called stable. Stable data make it easier to tell whether a later change came with a change in conditions.

These features are separate. Data can trend upward while also bouncing a lot, and two clients can have the same average level with very different variability.

Reading a graph step by step

• Read the vertical-axis label. Know what is being measured and in what unit.
• Decide which direction is the goal. For a skill being taught, the goal is usually up. For behavior targeted for reduction, the goal is down. For latency, a decrease means the client is starting sooner.
• Look at each phase separately. Describe the trend, level and variability within it.
• Compare phases at the phase change line. Did the level shift? Did the trend change direction? Did variability change? Was the change immediate or did it start after several sessions?
• Look at the most recent sessions. Is progress continuing, flat, or going the wrong way?

Base a trend on several sessions, not one. A single unusual point should be checked against the data sheet and reported, but it does not make a trend.

Common patterns

• Steady rise in a skill: progress in the desired direction.
• Flat data after early gains: a plateau. The skill has stopped improving.
• Decreasing trend in a skill: performance moving away from the goal.
• Lower level in intervention than baseline, both flat: a change in level when conditions changed.
• No change between phases: the intervention data look like baseline.
• A brief rise when extinction starts, then a steady fall: a pattern consistent with an extinction burst.
• Baseline already moving in the desired direction: important for the supervisor to know before intervention begins.
• A return to baseline levels when baseline conditions are put back: the behavior changed along with the conditions.

If conditions stay the same, a clear trend is the best guide to where the next data point will fall.

What you report, and how

Report what the graph shows, in plain observable terms: "Percentage correct has risen over the last seven sessions," or "Hitting has increased sharply over the last three sessions." Pass along anything else you know that might matter, such as a caregiver's report of a medication change, illness or schedule change.

Do not add conclusions the graph cannot show. Whether a skill is mastered, why behavior changed, and whether to change prompting, criteria or procedures are the supervisor's decisions. Report promptly, especially when problem behavior is increasing or progress has stalled, rather than waiting for the next scheduled meeting.

How this connects to other tasks

You can only read trends from data that were collected and entered accurately (A.1 to A.4, A.8) and summarized correctly (A.6). Knowing common phases, such as baseline, intervention, generalization and maintenance, helps you understand what each part of the graph represents. Reporting what you see is part of seeking clinical direction and documenting variables that affect progress (E.2, E.3).

Common mistakes

• Confusing trend with level. Trend is direction; level is where the data sit.
• Calling a single high or low point a trend.
• Forgetting which direction is the goal. A decrease is good for problem behavior and for latency to follow an instruction, but bad for a skill being taught.
• Reading data across phases as one data path instead of comparing phases.
• Reporting interpretations, such as "the client is more motivated" or "the new reinforcer worked," instead of what the graph shows.
• Changing the plan in response to the data instead of reporting to the supervisor.

How it's tested

Expect graph items that ask which description fits the data: increasing, decreasing, stable, variable, a change in level, or a delayed change. Read the axis label and phase labels first. Scenario items often ask what the RBT should do after noticing a pattern; the answer is to report it to the supervisor, not to change the procedure.