Validity
Valid measurement directly measures the behavior that matters, using a dimension relevant to the question, under the conditions and times that matter.
A valid measurement answers the question the team is actually asking. In behavior analysis, validity depends on three things: measuring the socially significant target behavior directly, measuring a dimension of it that is relevant (for example, duration rather than count for long tantrums), and collecting data at the times and in the settings where the behavior matters.
Measurement can lose validity in several ways. Measuring a stand-in rather than the behavior (rating "frustration" instead of counting hitting), measuring the wrong dimension, or observing only during easy activities when the problem happens during transitions all produce data that may be precise but do not reflect the real problem. Discontinuous methods with very long intervals can also distort how much behavior occurs.
Validity is separate from reliability and accuracy. Data can be consistent and correct about what was measured and still be invalid because the wrong thing was measured. The supervisor designs the measurement system, and the RBT supports validity by recording the defined behavior, in the defined way, at the scheduled times, and by telling the supervisor if the behavior of concern seems to be happening mostly when data are not being taken.
Example and non-example
| Example | A plan targets a client's aggression during community outings, so the supervisor schedules data collection during outings rather than only during calm clinic sessions. |
|---|---|
| Non-example | Counting how many times a student says "sorry" as a measure of their aggression would not be valid, because it does not measure aggression directly. |
Don't confuse it with
| Reliability | Validity asks whether the right thing is measured; reliability asks whether measurement gives consistent values. |
|---|---|
| Dimensions of behavior | Dimensions are the properties that can be measured; validity requires choosing the dimension that fits the question. |
Test yourself
-
Question 1A plan measures a child's cleanup by counting the toys in the toy bin at the end of play. The RBT learns that an older sibling often helps put toys away. What is the main problem with these data?
Show answer
D: The product may not reflect the client's own behavior. Permanent product data are valid only if the product was produced by the client's target behavior. The RBT should report the problem so the supervisor can decide how to handle it.
- A: Toys in a bin are a lasting result that can be counted later; the issue is who produced it.
- B: If a sibling helps, the count is more likely to be too high than too low.
- C: Cleanup can be measured several ways, including by its products.
Task A.3 · 40-hour curriculum section C · RBT Ethics Code 2.06
-
Question 2A team considers measuring a teen's studying by counting pages of handwritten notes at the end of each study period. What is the main weakness of this product measure?
Show answer
A: Studying does not always produce notes, so pages miss some studying. A permanent product is a good measure when every occurrence of the behavior produces the same product. Reading or reviewing leaves no notes, so a page count would not capture much of the studying.
- B: Pages can be counted easily and checked by a second observer; reliability is not the main weakness.
- C: Counting pages is done afterward and does not require watching the teen.
- D: A page count is a count of products, not a duration.
Task A.3 · 40-hour curriculum section C
-
Question 3A client's worksheet comes back with every answer correct, but the RBT gave hand-over-hand guidance on most of the items. The plan measures accuracy with permanent product recording and also asks for prompt levels. What should the RBT do?
Show answer
C: Record the prompts given for each item, as the plan requires. A product shows the result, not how it was produced. Prompt data recorded alongside the product show how much of the work was independent.
- A: The worksheet looks perfect, but scoring it as independent ignores the guidance that produced it.
- B: Discarding the worksheet loses the record of the session.
- D: Altering the product changes the data and hides what actually happened.
Task A.3 · 40-hour curriculum section C · RBT Ethics Code 2.02
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.