Once data are graphed, the next step is saying what the data path is doing. Spotting a trend early helps your supervisor keep what works and fix what does not.
Why it matters
ABA teams usually make decisions by looking at graphed data rather than by running statistics. This is visual analysis. Your supervisor makes the treatment decisions, but you see the learner every day, and you are often the first person to notice that a skill has stalled or that a problem behavior is creeping up.
Reading trends belongs to Domain A, Data Collection and Graphing (13 of 75 scored questions). Practice items give you a series of data points and ask for the direction, or whether the learner is moving toward the goal. When the trend is worrying, the RBT keeps following the plan and tells the supervisor.
Key ideas
- Trend is direction over time. Look at the data path within one phase, from left to right, and ask: is it going up, going down, or staying flat?
- Three directions:
- Increasing (ascending): values rise across sessions.
- Decreasing (descending): values fall across sessions.
- Zero trend (no trend): values stay at about the same height, even if they bounce a little.
- Trend also has steepness (sometimes called its degree). A steep slope means fast change; a gentle slope means slow change. "Increasing, but only slightly" is a more useful description than "increasing."
- Use several points, not one. A single high or low session is not a trend. A common rule of thumb is at least three points moving in the same general direction.
- Good or bad depends on the goal. Rising data are good news for a skill you are teaching (independent requests) and bad news for a behavior you are trying to reduce (aggression).
- Read within phases. A phase change line marks a change in conditions, such as baseline to intervention. Describe the trend inside each phase, then compare phases. Data points are not connected across a phase change line.
- Your job is to notice and report. Changing a program because of a trend is a clinical decision that belongs to your supervisor.
Definitions
- Trend: the overall direction of a data path across time, along with how steep that change is.
- Level: where the data sit on the vertical (y) axis, often summarized by the mean or median value within a phase. "Tantrums are at a high level" describes level, not trend.
- Variability: how much the data points bounce up and down around the trend. High variability hides trends, so more data are usually needed.
- Stable data (steady state): data that show little variability and no clear upward or downward trend over several sessions. They make it easier to see whether a change in conditions made a difference.
Three graphs, three trends
- Increasing: Mateo, age 6, works on independent hand washing at a clinic. His percentage of steps completed independently over five sessions is 20, 30, 45, 50, 65. That is a clear increasing trend, which is what the team wants.
- Decreasing: Ms. Lee, an adult in a day program, has a plan to reduce loud verbal outbursts. Her daily count over five days is 14, 11, 10, 7, 5. The decreasing trend is moving toward the goal.
- Zero trend: Priya, age 12, is learning to request a break at school. Her daily count of independent requests is 2, 3, 2, 2, 3. The values bounce slightly, but the path is flat: no progress yet.
Examples from a session
Jordan, an RBT, updates the graph for Kayla, a 4-year-old learning to label emotions at home. Over eight sessions her percentage correct went 40, 50, 55, 60, 60, 60, 60, 60. Early on there was an increasing trend, but the last five points are flat. Jordan describes it to his supervisor: "Kayla rose over the first four sessions, and since then she has stayed flat at 60 percent." He names the direction and the change without guessing at the cause.
Later, Kayla's mom mentions a new medication. Jordan adds this to his report because it may help explain the data, and he does not change prompting or reinforcers on his own.
Another learner, Dev, has a plan to reduce hitting. Right after a new extinction procedure starts, the graph shows hitting rising for two sessions. This could be an extinction burst, a temporary increase that is common early in extinction. Jordan keeps following the plan as written and reports the rise to his supervisor that day.
Non-example: Jordan writes, "Kayla's trend is 60 percent." That number tells where the data sit (level), not which way they are moving (trend), so it is not a trend description.
Common mistakes / exam traps
- Calling one point a trend. One unusual session (a sick day, a substitute teacher) is not a trend. Look at the overall path.
- Mixing up trend and level. Data can sit at a high level with no trend, or at a low level with an increasing trend. Level is where the data are; trend is which way they are going.
- Describing one trend across a phase change line. Describe each phase separately, then compare them.
- Changing the program yourself. Even when a trend clearly looks bad, an RBT reports it to the supervisor rather than adjusting procedures independently.
Exam trap
The trap: "increasing" is not the same as "improving." Direction tells you which way the data move; the goal tells you whether that is good. For a skill being taught, up is progress. For a behavior targeted for reduction, such as elopement, up means things are getting worse.
Exam trap
The trap: trend, variability and level all describe the same graph. If a question asks about the direction of the data over time, the answer is trend. If it asks about how much the points bounce around, the answer is variability. If it asks about how high or low the data sit, the answer is level.
Quick check
A learner's daily count of elopement attempts over six days is 3, 4, 6, 7, 9, 10. How should an RBT describe this data path?