10.6 Data Irregularity and Supervisor Escalation Lab
Key Takeaways
- Data irregularities include missing data, impossible values, sudden unexplained changes, mismatched graphs, unclear definitions, and fidelity barriers.
- The RBT should not alter data to make them look cleaner or more consistent.
- When data are unreliable, the RBT documents what is known, flags the concern, and seeks supervisor direction promptly.
- Poor procedural fidelity and unreliable data can lead supervisors to make incorrect decisions about client progress or risk.
- Escalation should be objective, timely, and routed through the correct chain of command.
Data irregularity scenario lab
The RBT opens the data system before a school session and notices that yesterday's aggression frequency is listed as 0, but the session note from another staff member mentions two episodes of hitting. The graph also shows a sudden drop in task refusal from 12 events per day to 0 for three days, even though the teacher told the RBT that refusal occurred several times. During today's session, the RBT's timer stops working for duration data, and the operational definition for refusal seems different from what the teacher is calling refusal. This is a data integrity lab, not a blame exercise.
The RBT should not edit another person's data, accuse a coworker, or quietly ignore the discrepancy. The RBT also should not make up missing values. The correct action is to preserve what is known, collect today's data as accurately as possible, document barriers, and escalate through the supervisor or chain of command. Data are used to make decisions about treatment effectiveness, safety, mastery, fading, and supervision. Unreliable data can lead to premature changes, missed risk, or unnecessary intervention changes.
| Irregularity | Example | RBT response |
|---|---|---|
| Missing data | No duration recorded for a crying episode | Mark missing according to system rules and notify supervisor |
| Impossible value | Session was 60 minutes but behavior duration entered as 90 minutes | Flag the entry; do not change someone else's data unless authorized |
| Note-data mismatch | Note describes hitting but frequency is 0 | Report discrepancy objectively |
| Sudden trend shift | Target behavior drops to 0 after data collector changed | Ask supervisor to review definitions and fidelity |
| Equipment failure | Timer stops during duration recording | Record what is known and note timing limitation |
| Definition confusion | Teacher uses refusal for any slow response | Use operational definition and seek clarification |
During today's session, the plan defines refusal as pushing materials away more than 12 inches, saying no after a demand, or leaving the work area within 10 seconds of instruction. The teacher says the client refused because he looked away for five seconds. The RBT should score according to the operational definition, not the teacher's informal label. The RBT can document that the teacher reported concern about looking away and tell the supervisor. If looking away becomes clinically relevant, the supervisor may define and add it later. The RBT does not expand the definition during the session.
The timer failure requires immediate practical handling. If duration data are required and the timer fails mid-episode, the RBT should use an approved backup if available, such as a clock or second timer. If no accurate timing source is available for part of the episode, the RBT records the known start or end information and marks the data limitation according to workplace procedures. The RBT should not estimate a precise duration from memory and present it as measured. A note such as duration timing interrupted when timer battery failed; episode was already in progress at 2:14 p.m. and ended at 2:16 p.m.
by classroom clock is more honest than pretending exact stopwatch data exist.
Graph irregularities need supervisor attention because they can change clinical interpretation. A sudden drop to zero may reflect improvement, but it may also reflect a new data collector, missed sessions, changed definitions, environmental changes, or data entry errors. The RBT can identify the trend and the possible data quality concern without interpreting the treatment effect independently. The supervisor reviews the full context and decides whether to retrain staff, correct entries, annotate the graph, modify the plan, or continue collecting.
Escalation should be timely and neutral. Poor wording would be: Yesterday's RBT lied on the data. Better wording is: I noticed yesterday's note states two hitting episodes, while the frequency field shows 0. I did not change the entry. Today I used the written definition and recorded one hitting episode. Please advise whether you want a data review. This protects professionalism and keeps the focus on client care.
Data integrity checklist:
- Read the operational definition before scoring.
- Record data during the session or at the approved time, not from vague memory.
- Use the assigned measurement system and backup tools.
- Mark missing, partial, or invalid data according to workplace rules.
- Do not change another staff member's data unless specifically authorized.
- Do not inflate progress by scoring prompted responses as independent or nonoccurrence as zero when observation did not occur.
- Report mismatches, equipment problems, definition confusion, sudden trends, and fidelity barriers to the supervisor.
- Keep escalation objective and tied to client care.
The RBT may also notice their own fidelity issue. Suppose the plan requires partial interval recording during a 20-minute group, but the RBT was pulled into a hallway conversation and missed three intervals. The ethical action is to mark those intervals as missed or follow the agency's invalid data rule, then tell the supervisor. The RBT should not fill in no occurrence because nothing was noticed. Not observing is different from observing that behavior did not occur.
This lab integrates measurement, graph interpretation, documentation, supervision, and ethics. Reliable data are part of benefiting the client. When data are messy, the RBT does not hide the mess. The RBT makes the limits visible so the supervisor can make better decisions.
An RBT notices that a session note says hitting occurred twice, but the frequency field for that day says 0. What should the RBT do?
A timer fails during duration recording. Which response best protects data integrity?
A teacher calls a slow response refusal, but the plan's refusal definition does not include slow responding. What should the RBT score?