Measurement, Data Display, and InterpretationDifficulty 3/5
Two observers record a low-rate behavior using 10-second intervals. Most intervals contain no occurrence and interval-by-interval agreement is 96%. What is the concern?
- AAgreement is too high to be credible, so both observers should be retrained
- BInterval-by-interval IOA inflates when most intervals are empty; unscored-interval IOA is the appropriate check
- CThe observers should switch to partial-interval recording to raise sensitivity
- DAgreement cannot be computed for low-rate behavior without a longer observation window
Show answer and rationale
Correct: B. With mostly empty intervals, agreements on non-occurrence dominate and inflate interval-by-interval IOA. Scored-interval (for high-rate) and unscored-interval (for low-rate) IOA correct for this by only counting the intervals that matter.
- A — High agreement is not itself a reason to discard data; the metric is the problem.
- C — The recording method is not the issue; the agreement calculation is.
- D — Agreement can be computed; it simply needs a method suited to the rate.