Observer drift
Observer drift is an unintended change, over time, in the way an observer applies the operational definition when recording behavior.
After weeks of recording the same behavior, an observer may slowly start scoring it differently from the written definition. They may begin counting events the definition excludes, ignoring borderline events it includes, or loosening the onset and offset rules. The change is usually gradual and unintentional, which is why it is hard to notice from the inside.
Drift is a serious threat to data. A graph may show the behavior getting better or worse when the only thing that changed is how it was measured. The supervisor may then continue an ineffective procedure, or change an effective one, based on a false picture. When two observers drift together in the same direction, interobserver agreement can stay high even though both are inaccurate.
Common safeguards include clear operational definitions with examples and nonexamples, periodic retraining (booster training) on the definitions, regular IOA checks with a supervisor or trained colleague, and practice scoring video samples against a standard. RBTs can help by rereading definitions often, using the definition rather than memory, and asking the supervisor when a new kind of response does not clearly fit.
Example and non-example
| Example | Over several months an RBT gradually starts counting loud talking as "screaming", which the definition limits to high-pitched vocalizations above conversational volume, so the screaming data begin to rise without any change in the client. |
|---|---|
| Non-example | A client starts behaving differently because they know a new observer is watching. That is reactivity, a change in the client, not in the observer. |
Don't confuse it with
| Reactivity | Observer drift is a change in how the observer records; reactivity is a change in the client's behavior caused by being observed. |
|---|---|
| Procedural fidelity | Observer drift affects data collection; drift in how an intervention is carried out is a procedural fidelity problem. |
Test yourself
-
Question 1A plan defines crying as tears plus vocalizing for at least 3 seconds. Over several months, an RBT gradually begins counting brief whines without tears as crying. What is this called?
Show answer
A: Observer drift. Observer drift is an unintended, gradual change in how an observer applies an operational definition over time. It makes later data incomparable with earlier data. Periodic retraining on the definition and IOA checks help catch it.
- B: Treatment drift is a gradual change in how the intervention is carried out, not in how behavior is recorded.
- C: Prompt dependence is a client relying on prompts to respond; it is not a recording error.
- D: Reactivity is a change in the behavior of the person being observed because they are being observed.
Task A.8 · 40-hour curriculum section C
Answer every question to see your score.
On the RBT exam
This term belongs to task A.8 Unreliable data and poor procedural fidelity in content area A, Data Collection and Graphing, which is 13 of the 75 scored questions.