2.7 Trend Identification, Data Quality, and Fidelity Risks

Trend Identification, Data Quality, and Fidelity Risks

The final measurement skill in this chapter connects graph reading to responsible practice. The 2026 RBT Test Content Outline includes identifying trends in graphed data and describing risks associated with unreliable data collection and poor procedural fidelity. Trend is the general direction of data over time. Data may increase, decrease, remain stable, or show no clear pattern because of variability. RBTs should be able to describe what they see, but they should not independently decide that a treatment should be changed unless the supervisor gives that direction.

Trend is easiest to inspect when data are graphed across repeated sessions. If aggression frequencies are 12, 10, 8, 5, and 3 across consecutive comparable sessions, the trend is decreasing. If independent requests are 2, 3, 5, 8, and 11, the trend is increasing. If engagement percentages are 68, 70, 69, 71, and 70, the data look stable. If points jump from 20 to 90 to 35 to 80, variability is high and the direction may be unclear. RBTs can report these patterns using objective language: Data for independent requests increased across the last five sessions.

That is different from saying the intervention is working because of me or the client has mastered requesting. Those conclusions require supervisor review and mastery criteria.

Data quality affects trend interpretation. A beautiful graph can still be misleading if the data were collected inconsistently. If one RBT counts every vocal protest and another counts only protests lasting more than 10 seconds, the graph may show change caused by observer differences. If duration data are estimated at the end of the session, a decrease might reflect memory error. If interval timers are missed during busy parts of instruction, data may undercount behavior during the hardest moments. The RBT should recognize these risks and report them rather than letting the graph speak with false confidence.

Procedural fidelity means implementing the plan as written and trained. Poor fidelity can create data that look like treatment failure or treatment success for the wrong reason. Suppose the plan says to provide reinforcement immediately after independent requests, but the RBT often delays reinforcement while completing paperwork. Requests may decrease, but the issue may be implementation rather than an ineffective plan. Conversely, if the RBT gives extra prompts not specified in the plan, skill data may improve temporarily while independence does not. Fidelity is not about blaming staff.

It is about knowing whether the procedure was actually implemented before interpreting outcomes.

Data Quality and Fidelity Risk Register

Trend should also be considered with level and variability. Level is the overall value or height of the data. A decreasing trend in aggression from 60 to 45 per hour may still require urgent supervisor attention because the level remains high. A stable trend in independent toileting at 95% may suggest maintenance, while a stable trend at 10% may signal lack of progress. Variability can make trend harder to read. The RBT's role is to keep collecting accurate data and report patterns and conditions, not to force a trend line that the data do not support.

A scenario shows why data quality and fidelity belong together. An RBT sees that a client's latency to start work increased over two weeks. Looking back, the RBT realizes that three sessions used a different instruction because a substitute teacher said Please get going instead of the written instruction Start worksheet. In two sessions, the RBT forgot to start the timer until after repeating the direction. The graph may still show longer latencies, but the supervisor needs to know that antecedent wording and timing fidelity changed.

Without that report, the team might adjust reinforcement, task difficulty, or prompting based on contaminated data.

RBTs should develop a habit of separating observation, calculation, and interpretation. Observation: The client left the seat 8 times in 30 minutes. Calculation: Rate was 0.27 episodes per minute. Trend description: Rates increased across the last three sessions. Interpretation needing supervision: The function changed or the intervention should be revised. This separation keeps the RBT within scope and gives supervisors clean information for clinical decision making.

When data are unreliable, the risks are practical and serious. The team may continue a procedure that is not helping, stop a procedure that is helping, set goals that do not match the client's needs, communicate inaccurate progress to stakeholders, or miss safety concerns. Poor data can also affect trust among team members. The RBT protects the process by using definitions consistently, collecting data in real time as much as possible, entering data accurately, asking for feedback, and reporting problems quickly.

Trend identification is not about being a statistician. It is about reading graphs carefully enough to notice direction, level, variability, missing points, and context. A prepared RBT can say, The graph shows a decreasing trend in duration, but two data points are missing because the timer failed, and the last session included a schedule change. That kind of statement is practical, objective, and useful. It supports high-quality supervision and better client service.