Measurement and Data Collection Flashcards

A BCBA is measuring self-injurious behavior (SIB) using a 10-second partial interval recording system across a 30-minute session. The behavior occurs continuously for 45 seconds, then stops. How many intervals will be scored as 'behavior occurred'?

Answer: Cannot be determined without knowing when the 45 seconds began relative to interval onset

The number of intervals scored depends on exactly when within the interval the continuous behavior began. If it started at the very beginning of an interval, it would score 5 intervals (4 full + part of the 5th). But if it started 8 seconds into the first interval, it would score 4 intervals. Without knowing the onset timing relative to the interval structure, the exact count cannot be determined — a critical limitation of partial interval recording.

When calculating IOA for duration recording across a 20-minute session, Observer A records 4 minutes and 30 seconds and Observer B records 5 minutes and 15 seconds. Using the smaller-over-larger method, what is the IOA percentage?

Answer: 85.7%

Convert both durations to seconds: Observer A = 270 seconds, Observer B = 315 seconds. Smaller-over-larger IOA = 270 / 315 = 0.857, or 85.7%. This method is standard for duration data and avoids the inflation that can occur with other IOA formulas when both observers record zero or near-zero durations.

A researcher wants to measure the latency of a student's response to teacher instructions. The student frequently emits vocal stereotypy between the instruction and the response. Which measurement decision best isolates true instruction-following latency?

Answer: Measure only trials in which no stereotypy occurs and report as a separate data stream

When a confounding behavior (stereotypy) routinely occurs between the SD and the response, the cleanest methodological solution is to separate trials by whether the confound is present, then analyze latency on clean trials as a distinct data stream. This avoids artificially inflating or deflating latency, preserves ecological validity, and allows the researcher to also analyze stereotypy rate as a covariate — a more sophisticated approach than simply excluding stereotypy duration.

A behavior analyst compares whole-interval recording to continuous duration recording for the same on-task behavior. Which statement accurately describes the expected systematic bias of whole-interval recording relative to continuous duration recording?

Answer: Whole-interval recording underestimates the proportion of time the behavior occurs

Whole-interval recording systematically underestimates behavior because it only scores an interval as 'behavior occurred' if the behavior was present for the entire interval. Any interval in which the behavior started, stopped, or was briefly interrupted is scored as absent, regardless of how much of that interval the behavior actually filled. This is the opposite bias from partial-interval recording, which overestimates.

An analyst is using a multiple-baseline design and needs to select a measurement system for 'initiation of social interactions' — defined as any unprompted verbal or physical contact directed toward a peer. Two behaviors (verbal and physical) can co-occur within the same initiation event. What is the most defensible primary measurement dimension?

Answer: Event recording of discrete initiations (counting the initiating event, not its topographies)

Because the target behavior is defined at the level of an initiation event (not its component topographies), and because verbal and physical contact can co-occur within the same event, counting topographies separately would produce double-counting artifacts. Event recording that treats each initiation as a single discrete unit best matches the operational definition, prevents inflation, and yields a frequency measure that is interpretable across sessions — critical for stable baseline interpretation in multiple-baseline designs.

During a functional analysis, a clinician calculates inter-rater agreement on occurrence and non-occurrence separately using the point-by-point method. Occurrence IOA is 62% and non-occurrence IOA is 98%. What does this pattern most likely indicate?

Answer: The behavior occurs at a very low rate, inflating non-occurrence agreement independently of true reliability

When behavior occurs at very low rates, non-occurrence agreement is inflated because observers will agree that the behavior was absent in most intervals simply by chance. An occurrence IOA of 62% reveals poor agreement on the actual instances of behavior — which is the clinically critical measure in a functional analysis context. Averaging occurrence and non-occurrence IOA in this situation masks the true unreliability of occurrence measurement. This is why reporting both separately is standard practice.