Home / RBT Study Guide / Data collection and graphing (Domain A)

Domain A · 13 questions · 17% of the exam

Data collection and graphing (Domain A)

Domain A covers how you measure behavior, summarize the numbers and keep a graph current. It has eight tasks (A.1 to A.8) and accounts for 13 of the 75 scored exam questions, or 17%.

Last reviewed November 10, 2026 against the BACB RBT Test Content Outline (3rd ed.)

What Domain A covers

Domain A is about the numbers behind every decision in applied behavior analysis. The RBT Test Content Outline (3rd ed.) lists eight tasks here, A.1 through A.8, and gives the domain 13 of the 75 scored questions, or 17% of the exam.

The division of labor is simple. Your supervisor (a BCBA or BCaBA) decides what to measure and how. You collect the data exactly as the plan says, enter it, keep the graph current and speak up when something looks off. Your supervisor reads the graph and decides whether the plan changes. Because those decisions rest on what you record, this domain rewards precision more than memorized vocabulary.

Measuring behavior: tasks A.1 to A.3

A.1 Implement continuous measurement

Continuous measurement means you record every instance of the behavior for the whole observation. The outline names frequency, duration, latency and interresponse time. Frequency is a count. Duration is how long a response lasts. Latency is the time from an instruction or cue to the start of the response. Interresponse time (IRT) is the time from the end of one response to the start of the next.

In session

Priya tells Leo, "Shoes on." He starts reaching for his shoes 12 seconds later and finishes in 90 seconds. The 12 seconds is latency. The 90 seconds is duration. If he puts his shoes on three times today, 3 is the frequency.

The full breakdown, including when each measure fits, is on the continuous measurement page.

A.2 Implement discontinuous measurement

Discontinuous measurement samples behavior instead of capturing all of it. You divide the observation into intervals and mark each one. With partial interval recording you mark the interval if the behavior happened at any point in it. With whole interval recording you mark it only if the behavior lasted the entire interval. With momentary time sampling you look at the moment the interval ends and mark whether the behavior is happening right then.

These methods give an estimate, not a true count. Partial interval tends to overestimate how much of the time a behavior occurred, and whole interval tends to underestimate it. See discontinuous measurement for a worked example of both errors.

A.3 Implement permanent product recording

Permanent product recording measures a behavior by the lasting result it leaves behind, so you do not have to watch it happen. Completed math problems, folded towels and a made bed are all products you can count afterward. It only works when the behavior reliably leaves a product and you can be confident who produced it. This task is covered alongside data quality on the data reliability and procedural fidelity page.

Measurement methods at a glance

MethodWhat you recordTypical use
Frequency (count)Each occurrenceBehaviors with a clear start and end, such as hitting or requests
DurationHow long each occurrence lastsTantrums, time on task
LatencyTime from the cue to the start of the responseSlow starts after an instruction
Interresponse timeTime from the end of one response to the start of the nextPacing, such as time between bites
Partial intervalWhether the behavior occurred at any point in the intervalBehaviors targeted for decrease, often brief and frequent
Whole intervalWhether the behavior lasted the entire intervalBehaviors targeted for increase, such as staying on task
Momentary time samplingWhether the behavior is occurring at the instant the interval endsWhen you cannot watch one client continuously
Permanent productThe lasting result of the behaviorWorksheets, chores, completed assembly tasks

Describing and graphing: tasks A.4 and A.5

A.4 Enter data and update graphs

Data that stays on a clipboard helps no one. Task A.4 expects you to transfer what you recorded into the data system and keep the graph up to date, following your workplace's timeline. On a standard line graph, time (sessions or dates) runs along the horizontal x-axis and the measure of behavior runs up the vertical y-axis. Each session gets one data point. Points are connected within a phase but not across a phase change line.

A missed session gets no data point. It does not get a zero, because zero means the behavior was measured and did not happen. The reading ABA graphs page walks through each part of the graph.

A.5 Describe behavior and environment in observable and measurable terms

Write what a camera would capture. "Sam was frustrated" is an inference. "Sam pushed the worksheet off the desk and put his head down for two minutes" is observable and measurable. The same rule applies to the environment: "the room was chaotic" becomes "two peers were shouting and the fire drill bell rang." Clear descriptions let two people record the same event the same way. More examples are on the operational definitions page.

Summarizing data and keeping it trustworthy: tasks A.6 to A.8

A.6 Calculate and summarize data

The outline names three calculations.

Rate matters whenever sessions differ in length. Ten responses in 20 minutes (0.5 per minute) is a higher rate than twelve responses in 40 minutes (0.3 per minute), even though twelve is the bigger count.

A.7 Identify trends in graphed data

Trend is the direction the data path is heading. An increasing trend climbs from left to right, a decreasing trend falls, and a zero trend stays flat. Two other features describe a graph: level (how high or low the points sit on the y-axis) and variability (how much they bounce from point to point).

Whether a trend is good news depends on the target. Rising correct responses is the goal. Rising aggression is a reason to alert your supervisor. The graphs page linked under A.4 describes each pattern in words, with sample numbers.

A.8 Describe the risks of unreliable data and poor procedural fidelity

Unreliable data means the numbers do not match what actually happened. Procedural fidelity means running the plan the way it was written. When either one slips, the graph stops telling the truth. A supervisor might end a program that was working, keep one that was not, or move a client forward before a skill is solid.

The usual check is interobserver agreement, where two people record the same session independently and compare. Your part is to follow the written definition, record during the session and not from memory, and tell your supervisor about any session where data was missed or the plan was not followed. A worked agreement calculation is on the data reliability page.

How Domain A questions tend to read

A common question format describes a short session scenario and asks you to name the measure, pick the right calculation or choose what the RBT does next. A few habits help.

Common mix-up

Frequency and rate are not the same thing. Frequency is the raw count. Rate is the count divided by time. If an answer choice gives a number "per minute" or "per hour," it is a rate.

On the exam

Read the last sentence of the question first. "Which measurement procedure" and "what should the RBT do" are different questions about the same scenario, and the distractors are usually correct answers to the other one.

Practice questions

Original questions in the style of the exam. Pick an answer before you open the explanation.

1. Priya tells Leo to put on his shoes. He begins reaching for them 12 seconds after the instruction. Which measure does the 12 seconds represent?

  1. Response duration
  2. Interresponse time
  3. Latency to respond
  4. Rate of responding
Show answer

C. Latency to respond

Latency is the time from an instruction or cue to the start of the response. Duration would be how long putting on the shoes takes once he starts, and interresponse time is the gap between two responses, not between a cue and a response.

2. An RBT divides a session into 15-second intervals and marks an interval whenever hand flapping happens at any point during it. Which procedure is this?

  1. Partial interval recording
  2. Whole interval recording
  3. Momentary time sampling
  4. Permanent product recording
Show answer

A. Partial interval recording

Marking an interval when the behavior occurs at any point in it is partial interval recording. Whole interval recording requires the behavior to last the entire interval, and momentary time sampling checks only the instant the interval ends.

3. Sam calls out 18 times during a 45-minute class. What is his rate of calling out?

  1. 0.25 per minute
  2. 2.5 per minute
  3. 4 per minute
  4. 0.4 per minute
Show answer

D. 0.4 per minute

Rate is count divided by time: 18 / 45 = 0.4 per minute. The 2.5 answer comes from dividing the wrong way around (45 / 18), which gives minutes per call-out and not a rate of behavior.

4. A graph shows instances of aggression across six sessions: 12, 10, 9, 7, 5 and 4. How would you describe these data?

  1. An increasing trend in the data
  2. A decreasing trend in the data
  3. High variability with no trend
  4. A stable level with zero trend
Show answer

B. A decreasing trend in the data

Each point is lower than the one before, so the data path falls from left to right: a decreasing trend. Variability describes how much points bounce around, and these points change steadily in one direction.

5. Jayden has a busy session and fills in his frequency tallies from memory after the client leaves. What is the main risk?

  1. The client will show an extinction burst in the next session
  2. The behavior will be measured with a discontinuous method
  3. The supervisor may change the plan based on inaccurate data
  4. The graph will need a phase change line added for that day
Show answer

C. The supervisor may change the plan based on inaccurate data

Data recorded from memory may not match what happened, and the supervisor uses those numbers to make treatment decisions. Recording late does not change the measurement method or create a new phase. It makes the data unreliable.

Common questions

How many RBT exam questions are about data collection and graphing?

The RBT Test Content Outline (3rd ed.) assigns Domain A, Data Collection and Graphing, 13 of the 75 scored questions, which is 17% of the exam. The domain contains eight tasks, A.1 through A.8.

What is the difference between continuous and discontinuous measurement?

Continuous measurement records every instance of the behavior during the observation, using measures such as frequency, duration, latency and interresponse time. Discontinuous measurement samples behavior in intervals (partial interval, whole interval or momentary time sampling) and gives an estimate of how much the behavior occurred.

Do RBTs make graphs and interpret them?

The outline lists entering data and updating graphs (A.4) and identifying trends in graphed data (A.7) as RBT tasks. The RBT keeps the graph current and can describe what the data path is doing. The supervisor interprets the data and decides whether the plan changes.