- Applied Behavior Analysis Measurement and Data Collection Questions and Answers Flashcards

A behavior analyst is measuring stereotypic hand-flapping using a 10-second partial interval recording system across a 30-minute session. The behavior occurs for exactly 5 continuous seconds during the first interval and does not occur again. What does the data system record, and what is the primary measurement artifact introduced?

Answer: 1 interval scored; partial interval recording overestimates duration because any occurrence—regardless of length—scores the full interval

Partial interval recording scores an interval as positive if the behavior occurs at any point during it, regardless of duration. A 5-second occurrence in a 10-second interval yields a scored interval. The well-documented artifact is overestimation of duration: because the entire interval is marked, the data suggest the behavior lasted longer than it actually did. This is why PIR is most appropriate for low-frequency, long-duration behaviors where overestimation is less clinically misleading.

During interobserver agreement (IOA) assessment using the exact agreement method for interval data, Observer A scores intervals as: + + - + - - + + - + and Observer B scores: + - - + - + + + - +. What is the exact agreement IOA percentage?

Answer: 80%

Exact agreement IOA counts intervals where both observers recorded the same score (both + or both -), divided by total intervals × 100. Comparing interval by interval: +/+ (agree), +/- (disagree), -/- (agree), +/+ (agree), -/- (agree), -/+ (disagree), +/+ (agree), +/+ (agree), -/- (agree), +/+ (agree). That gives 8 agreements out of 10 intervals = 80%. Note that exact agreement does not weight agreements differently based on occurrence vs. non-occurrence, unlike scored-interval or unscored-interval methods.

A practitioner needs to measure latency for a child's compliance response to instructions. During a session, the instructor delivers 8 instructions. The child does not respond to 2 of them within the observation window. Which approach best preserves data integrity while allowing latency to be calculated for the session?

Answer: Assign the maximum observation window duration to the 2 non-responses and include them in the latency mean

Assigning the full observation window duration to non-responses is the most conservative and data-complete approach; it captures the fact that the child never initiated a response within the available time. Recording zero would misrepresent non-responses as instantaneous responses. Excluding non-responses biases the mean by discarding the worst latency observations, which is especially problematic when non-compliance is the clinical target. Switching to frequency data loses important temporal information. Ceiling-coding non-responses maintains the latency dimension and prevents artificial underestimation.

A researcher collects permanent product data on math worksheet completion across three students. Student A completes 18/20 problems correctly, Student B completes 20/20 but took twice as long, and Student C completes 15/20. The researcher wants to add a fluency dimension without switching measurement systems. Which augmentation is most technically appropriate?

Answer: Divide each student's correct responses by session duration to yield a correct-responses-per-minute (rate) measure alongside accuracy

Fluency in ABA is operationalized as accuracy combined with rate (responses per unit time). Computing correct responses per minute from permanent product data is technically straightforward, preserves the original count data, and directly measures the fluency dimension without introducing new observation systems. Weighting factors and z-scores are statistical transformations that obscure raw behavioral data and are not standard ABA measurement practice. Adding MTS introduces a second, independent measurement system rather than augmenting the existing permanent product measure.

An analyst uses a multi-element design to compare three intervention conditions. Data are collected using 5-minute whole interval recording for on-task behavior across all conditions. A consultant notes that the choice of whole interval recording may confound the comparison. What is the most precise explanation of this confound?

Answer: Whole interval recording systematically underestimates the true level of on-task behavior; if conditions vary in how they distribute brief off-task pauses within intervals, differences in observed scores may reflect measurement artifact rather than true condition differences

Whole interval recording requires the behavior to occur throughout the entire interval to be scored as an occurrence; even a momentary interruption results in a non-occurrence score. This systematically underestimates true duration/level. Critically in a multi-element design, if one intervention condition produces longer but infrequent off-task pauses while another produces many very short ones, the scoring outcomes can diverge dramatically even if total on-task time is identical—introducing a measurement artifact that mimics a condition effect. This is the core technical risk the consultant is identifying.

A behavior analyst is establishing inter-rater reliability for a new operational definition of 'aggression' that includes hitting, biting, and scratching as topographically distinct responses but counts them as a single response class. Observer A records 12 aggressive episodes; Observer B records 15. Within the episodes both observers recorded, they agreed on episode boundaries in 10 cases. Which IOA formula is most appropriate here, and what does it yield?

Answer: Trial-by-trial IOA is inappropriate; the analyst should use the occurrence IOA formula: (smaller count ÷ larger count) × 100 = 80%

When IOA is calculated for frequency data without interval structure, the standard approach is occurrence IOA (also called total count IOA or frequency ratio): divide the smaller total count by the larger total count and multiply by 100. Here: 12 ÷ 15 × 100 = 80%. This formula is convention for free-operant frequency measures. The 'exact count' formula shown in option B is not a standard ABA IOA method. Mean count-per-interval requires interval data that don't exist here. Scored-interval IOA applies to interval recording, not episode-boundary data, and misuses the '10 agreements' figure.