RBT Data Collection and Graphing Questions and Answers

B is correct. An observable, measurable definition describes a movement a second observer could see and count — pushing the worksheet off the desk qualifies, and two RBTs watching the same session would agree on whether it happened.

A is wrong because "frustrated" is a private event you infer rather than see; you cannot count frustration. C is wrong for the same reason — "dislikes" is an interpretation of a mental state, not a behavior. D is wrong because "bad attitude" is a subjective label that different observers would score differently, which is exactly what an operational definition exists to prevent.

A is correct. Rate is a count divided by the time of the observation period: 12 ÷ 40 minutes = 0.3 responses per minute. Reporting rate rather than raw count lets sessions of different lengths be compared to each other.

B is wrong because 12 is the frequency (the raw count), not the rate — it has no time unit attached. C is wrong because it describes duration, which measures how long behavior lasts, not how often it occurs. D is wrong because it inverts the calculation into an average interval between responses; that is closer to inter-response time and is not what rate means.

D is correct. Latency is the elapsed time between the onset of a stimulus — here the instruction "Time to go outside" — and the beginning of the response. It answers "how long before he starts?"

A is wrong because duration measures how long the behavior lasts once it has begun, not the delay before it starts. C is wrong because inter-response time measures the gap between the end of one response and the start of the next response, which requires two responses and no instruction. B is wrong because partial-interval recording is a discontinuous estimate of whether behavior occurred within blocks of time; it cannot produce a time-to-onset figure.

B is correct. In partial-interval recording an interval is scored if the behavior occurs at any point within it, even for one second. A behavior occurring briefly in every interval scores 100%, which overstates how much of the session it actually filled.

A is wrong — underestimation is the known bias of whole-interval recording, which requires the behavior to persist for the entire interval. C is wrong because discontinuous procedures are commonly chosen precisely for behaviors with unclear boundaries; it is continuous duration recording that needs clean onset and offset. D is wrong because interval data yields a percentage of intervals, which graphs perfectly well.

A is correct. Whole-interval recording scores an interval only when the behavior occurs throughout its entire length, so a two-second glance at the worksheet does not count. That conservative bias is exactly why it is typically paired with behaviors you are trying to increase.

B is wrong because partial-interval scores the interval for any occurrence, which would credit the brief moments the supervisor wants excluded. C is wrong because on-task behavior is continuous rather than discrete — counting "how many times" someone is on task is not meaningful. D is wrong because permanent product measures a physical outcome left behind, which does not capture whether the client was engaged throughout.

D is correct. Momentary time sampling requires the observer to look only at the designated moment — typically the end of each interval — and score whether the behavior is occurring right then. Between those moments the observer is free to do other things, which is why it suits RBTs running a group.

A is wrong because that describes partial-interval recording. B is wrong because that describes whole-interval recording. C is wrong because counting every occurrence is frequency recording, a continuous procedure that defeats the labour-saving purpose of time sampling.

A is correct. Permanent-product recording measures the lasting physical outcome the behavior produced, after the fact. The completed worksheet still exists when the session is over, so the RBT never had to watch the responses happen.

C is wrong because that is frequency recording performed in real time as the behavior occurs. B is wrong because that is duration recording, which also requires live observation. D is wrong because that is partial-interval recording, a discontinuous live-observation procedure. None of the three leaves behind a product that can be measured later.

C is correct. Total count IOA divides the smaller count by the larger count and multiplies by 100: 15 ÷ 18 = 0.833, or 83%.

B is wrong — 88% does not follow from these numbers under any standard IOA formula. A is wrong for the same reason; 94% would require the counts to be much closer together, such as 17 and 18. D is wrong because it divides the larger by the smaller (18 ÷ 15 = 120%). IOA is a measure of agreement and can never exceed 100%, so any answer above 100% signals the calculation was inverted.

D is correct. Interobserver agreement checks the measurement system, not the client. High agreement gives confidence that the numbers on the graph reflect the behavior rather than one observer's idiosyncratic scoring.

A is wrong because IOA says nothing about the client's consistency — both observers are watching the same behavior at the same moment. C is wrong because IOA increases staffing cost during those sessions rather than reducing it. B is wrong because there is no general requirement for two staff in every session; IOA is conducted periodically on a sample of sessions.

A is correct. By convention the horizontal x-axis carries the passage of time — sessions or dates — and the vertical y-axis carries the dependent variable, the behavior measure. This is what lets a reader see change over time at a glance.

B is wrong because it reverses the convention, producing a graph that no supervisor could read quickly. C is wrong because two variables cannot share one axis. D is wrong because graphing conventions in applied behavior analysis exist precisely so that graphs are interpretable without study; labelling does not make a reversed graph acceptable.

B is correct. A vertical phase-change line marks where conditions changed — baseline ending and intervention beginning, for example. Data paths are not connected across it, so the reader can compare the two conditions cleanly.

A is wrong because missing sessions are shown by a break in the data path, not a phase line. C is wrong because mastery is usually shown with a horizontal criterion line or a note, not a vertical one. D is wrong because outliers are never simply disregarded — every data point stays on the graph, and unusual sessions are explained in the notes.

B is correct. The four baseline points sit flat near 1 request per session with little bounce, and after the vertical phase-change line the points climb steadily from 4 to 16. That is a stable baseline followed by a clear increasing trend.

A is wrong because only the baseline is flat; the intervention phase rises sharply, so the description does not hold across all ten sessions. C is wrong because it reverses the direction — the values go up, not down. D is wrong because variability describes how much data bounce around; these points move in a consistent direction with very little scatter, which makes them easy rather than hard to interpret.

B is correct. Total duration is 4 + 7 + 2 = 13 minutes. As a proportion of the 60-minute session that is 13 ÷ 60 = 0.2167, or about 22%.

A is wrong because 13 is the total number of minutes, not a percentage — the conversion step was skipped. C is wrong because it appears to divide the mean episode length (about 4.3 minutes) by something, or reports mean duration as a percentage; either way it does not answer the question asked. D is wrong because it triples the correct figure, as if each of the three episodes lasted 13 minutes rather than 13 minutes in total.

B is correct. Preparing for data collection includes solving problems when materials are missing. Recording accurately in the moment on any available paper, then transferring it to the correct form the same day, preserves the integrity of the data.

A is wrong because a lost session of data creates a hole in the graph that a supervisor needs for treatment decisions; the problem was solvable. C is wrong because reconstructing data from memory produces numbers that are not real data, which is a documentation and ethics violation, not just sloppiness. D is wrong because writing one client's data on another client's form is a confidentiality and record-keeping problem even with the name crossed out.

C is correct. Inter-response time measures the interval between two consecutive responses — specifically from the end of one to the onset of the next. It describes how spread out responding is, which is why it is used alongside DRL procedures.

A is wrong because that is latency, timed from a stimulus rather than from a previous response. B is wrong because that is duration, the length of a single response. D is wrong because that is total duration across the session, a summed measure that says nothing about the gaps between responses.

D is correct. IOA only tells you something if the two observers score independently. Once one observer copies the other, high agreement is guaranteed and reveals nothing about whether the measurement system is being applied consistently.

A is wrong because speed is irrelevant to the validity problem, and copying would if anything be faster. C is wrong because both observers are authorised members of the client's team, so no confidentiality breach has occurred. B is wrong because billing depends on services delivered and documented, not on how the IOA check was conducted.

D is correct. It names specific topographies — hitting, kicking, biting — and explicitly resolves the common edge case of near-misses. Two observers reading it would score the same events the same way.

A is wrong because intent cannot be observed; observers would have to guess at what the client meant. C is wrong because "violent" is a summary label, not a description of movement, and each observer would draw the line differently. B is wrong because "losing control" is both an inference and vague about what behaviors count, which is the opposite of what an operational definition needs to do.

B is correct. Data errors are corrected through the documented correction process — which keeps an audit trail — and the supervisor is told because a wrong data point may already have influenced a treatment decision.

A is wrong because leaving known-false data in the record is a documentation violation, and the review that already happened is precisely why it must be flagged. C is wrong because silently erasing removes the audit trail that record-keeping requirements exist to preserve. D is wrong because a note beside an uncorrected point still leaves the graph showing a value the RBT knows is false.

A is correct. Frequency recording needs behaviors that are discrete — with a clear beginning and end — and of roughly uniform length, so each occurrence counts as one. Saying "help" fits both requirements.

B is wrong because continuous hand-flapping has no clean boundary between occurrences, so counting produces unreliable numbers; duration or interval recording suits it better. C is wrong because "how long" is a duration question. D is wrong because that describes momentary time sampling, a discontinuous procedure, not a count of occurrences.

D is correct. Interval-by-interval IOA divides the number of agreements by the total number of intervals (agreements plus disagreements) and multiplies by 100: 17 ÷ 20 = 0.85, or 85%.

B is wrong because 17 is the raw count of agreements, not a percentage. C is wrong because 88% would arise from a different pair of numbers, such as agreements on 15 of 17 intervals. A is wrong for the same reason — 94% would require agreement on roughly 19 of 20 intervals, which is not what the question states.

B is correct. Momentary time sampling requires the RBT to observe only at the end of each interval, freeing her attention in between — which matters when she is simultaneously responsible for safety in the community. It yields a usable estimate of how much of the outing involved conversation.

A is wrong because continuous duration recording demands unbroken attention on the stopwatch for two hours, which is incompatible with supervising safety. C is wrong because counting every utterance is even more attention-intensive and answers "how often", not "how much of the time". D is wrong because conversation leaves no physical product to measure afterward.

D is correct. Every intervention point sits below the baseline band of roughly 19–21, so the overall level has dropped — but the points swing between about 2 and 16 from session to session, which is high variability. Both features have to be reported together.

A is wrong because a decreasing trend would require the points to move consistently downward; here they zig-zag with no direction. B is wrong because the level plainly changed — no intervention point comes near the baseline band. C is wrong because it reverses the direction of the level change.

D is correct. It replaces every label with countable events: how many tasks were presented, how many were completed, and exactly what the client did on the remaining trials. Anyone reading it can picture the session.

A is wrong because "uncooperative" and "did not want to" are the same inferences as the original, just reworded. C is wrong because "rough day" and "struggled" are vaguer than what they replace. B is wrong because it adds a causal claim about sleep that the RBT did not observe — and stating an unverified cause in a clinical record is a bigger problem than the original vagueness.

D is correct. The first episode runs 4 minutes 30 seconds and the second runs 2 minutes. Total duration sums every episode: 4:30 + 2:00 = 6 minutes 30 seconds.

A is wrong because it reports only the first episode and ignores the second. C is wrong because it reports only the second episode. B is wrong because it measures from the start of the first tantrum to the end of the second, which includes the 15 minutes between them when no tantrum was occurring — that is elapsed time, not duration of behavior.

D is correct. You cannot record accurately unless you know exactly what counts as an occurrence and which procedure to use. Confirming both before the session is the core of preparing for data collection.

A is wrong because diagnosis and medication are not what determine how you score a behavior, and reviewing records beyond what your role requires raises confidentiality questions. C is wrong because scheduling is unrelated to today's data. B is wrong because family satisfaction, while important, has no bearing on how a behavior is measured.

B is correct. Interval data are reported as a percentage of intervals scored: 12 ÷ 30 = 0.40, or 40% of intervals.

A is wrong because interval procedures do not count occurrences — an interval scored does not equal one instance of behavior. C is wrong because 12 intervals of 10 seconds is 2 minutes, not 12, and interval data are an estimate rather than a true duration. D is wrong because rate requires a count of discrete responses, which whole-interval recording does not produce.

D is correct. Permanent product measures only the outcome. Twelve toys in the bin looks identical whether the client did it alone, did it with hand-over-hand prompting, or a sibling did it — the product does not record who produced it or how.

A is wrong because not needing to observe live is the procedure's main advantage, not a limitation. C is wrong because counts of products graph like any other count. B is wrong because there is no systematic direction of error; the count is accurate, it is the attribution that is uncertain.

A is correct. Data reconstructed from memory are estimates presented as measurements. A supervisor reading the graph cannot tell the difference, and may change or discontinue a program on the strength of numbers that were never recorded.

B is wrong because this is not a mere preference — accuracy and timeliness are documentation requirements. C is wrong because a colleague who was not present could not produce the data either. D is wrong because "approximately right" is not a standard that applies to clinical data; fabricated-but-plausible numbers are still fabricated.

A is correct. Latency is the time from the onset of the stimulus to the onset of the response. Timing all the way to completion combines latency with duration and yields a number that is neither.

B is wrong because the stop point is incorrect, so the measure is invalid. C is wrong because starting at response onset would eliminate the delay that latency exists to measure. D is wrong because latency is inherently a time-based measure; a count could not answer "how long before he started?"

A is correct. Low agreement is a signal that the operational definition is ambiguous or that observers were trained differently. That is the supervisor's problem to solve, and it must be solved before the data can be trusted.

B is wrong because averaging two disagreeing observers hides the problem instead of fixing it, and produces a number neither observer recorded. C is wrong because experience does not establish which observer scored correctly; the definition may be at fault. D is wrong because repeating the check without changing anything is just waiting for a chance result that clears the threshold.

C is correct. Discontinuous procedures — partial interval, whole interval, momentary time sampling — deliberately sample rather than capture everything, trading some accuracy for practicality when an RBT cannot watch continuously.

A is wrong because capturing every occurrence is the definition of continuous measurement. B is wrong because the trade-off runs the other way: continuous measurement is more accurate, and discontinuous procedures carry known directional biases. D is wrong because discontinuous procedures are used across all behavior types and are especially common with stereotypy and engagement.

D is correct. Level refers to where the data sit on the vertical axis — the typical or average value within a phase. It is one of the three standard features read from a graph.

A is wrong because direction is trend. B is wrong because session-to-session bounce is variability. C is wrong because the number of sessions is simply the phase length and is read from the x-axis, not a characteristic of the data path.

C is correct. The definition leaves the boundary undefined, so each observer supplies their own threshold. Adding a criterion — for example, "moves more than three feet beyond the doorway" — resolves it.

A is wrong because both observers saw the same event; they scored it differently, which is a definition problem rather than an attention problem. B is wrong because speed is not the issue when both observers detected the event. D is wrong because elopement is a discrete behavior that frequency recording handles well once the definition is tightened.

D is correct. The question contains two separate claims — how often (frequency) and how long each episode lasts (duration) — so both dimensions must be measured to answer it.

A is wrong because frequency and rate measure the same dimension; rate is just frequency divided by time, so the pair says nothing about episode length. C is wrong because latency measures delay to onset and IRT measures gaps between responses; neither captures how long a tantrum lasts. B is wrong because two interval procedures would both give estimates of the same dimension and neither yields a clean count.

C is correct. The honest response is to preserve the record, mark clearly which portion is affected, and let the supervisor decide whether the data are usable. That protects both the graph and the audit trail.

A is wrong because data collected under the wrong definition measure a different behavior, so mixing it in corrupts the graph. B is wrong because deleting records and concealing the error is a documentation violation regardless of intent. D is wrong because retroactively adjusting numbers to what they "would have been" is fabrication — the RBT has no way to know what the correct definition would have yielded.