Your supervisor can only make good decisions with good data, and as an RBT you are the person collecting most of it. This lesson covers the measures you will use every day when you record every instance of a behavior.
Why it matters
Every change to a learner's program (keep going, add a prompt, fade a reinforcer, try a new approach) depends on the data you bring back from sessions. Continuous measurement gives the most complete picture: you record every occurrence of the target behavior for the entire observation period. Nothing is estimated and nothing is sampled.
Practice questions on this task (RBT task A.1, in Data Collection and Graphing) are mostly short scenarios: a behavior is described and you choose the measure that fits, or an RBT does something with a stopwatch and you name what was measured. If you can match a behavior to the dimension that matters, these questions become reliable points.
Key ideas
- Continuous means complete. You capture each instance from the start of the observation to the end. Methods that sample behavior, like interval recording and momentary time sampling, are discontinuous; see discontinuous measurement.
- Pick the dimension the team cares about. Ask what question the data has to answer:
- How often? Count it (frequency), and convert to rate when sessions differ in length.
- How long? Time it from start to finish (duration).
- How quickly does it start after a cue? Time the gap from the instruction to the start of the response (latency).
- How much time passes between responses? Time the gap from the end of one response to the start of the next (interresponse time, or IRT).
- Clear boundaries make counting possible. Frequency works best for behaviors with a clear beginning and end that last about the same amount of time each time, such as hitting, requesting, or throwing.
- Rate and IRT are linked. When responses come closer together (shorter IRTs), more of them fit into the same amount of time, so rate goes up.
- Follow the operational definition. Data only means something if you record exactly what the definition says counts, including when an episode starts and when it ends.
- Record in the moment. Tally, start, and stop timers as the behavior happens. Reconstructing data from memory after the session is not measurement.
Definitions
- Frequency (count): the number of times a behavior occurs during the observation. Some textbooks use "frequency" to mean rate, so read each question carefully.
- Rate: count divided by observation time. Twelve requests in 30 minutes is 0.4 per minute, or 24 per hour. Rate lets you compare sessions of different lengths.
- Duration: how long a behavior lasts, from onset to offset. You can report the total duration per session or the average (mean) duration per episode.
- Latency: the time from the onset of a stimulus, usually an instruction, to the start of the response.
- Interresponse time (IRT): the time between the end of one response and the start of the next response of the same behavior.
Examples from a session
Five learners, five measures
- Frequency (home): Ava, age 4, hands you a picture card to ask for an item 9 times during a 2-hour session. You make a tally mark each time.
- Rate (home): Tuesday's session with Ava lasts only 1 hour and she makes 6 requests. Her count dropped from 9 to 6, but her rate rose from 4.5 to 6 per hour.
- Duration (school): During a 30-minute math block, Marcus, age 8, is out of his seat three times. You time each episode from the moment he stands until he is seated again, as his definition says: 6 minutes in total, or a mean of 2 minutes per episode.
- Latency (clinic): You tell Jordan, a 15-year-old practicing job skills, "Please start folding the towels." He picks up the first towel 25 seconds later.
- IRT (day program): At lunch, Dana, an adult client, starts each new bite about 2 seconds after finishing the last one. Her plan aims to lengthen that gap so she eats more safely.
- Non-example (not latency): You time the 40 seconds between two of Jordan's questions, "Are we done yet?", with no instruction in between. Nothing cued the second question, so this is IRT, not latency.
Notice that each measure answers a different question. The behavior plan, written by your supervisor, tells you which one to use for each target. If a plan seems to call for a measure that does not fit what you are seeing, raise it with your supervisor instead of switching on your own.
Common mistakes / exam traps
- Comparing raw counts across sessions of different lengths. Twelve hits in 60 minutes (0.2 per minute) is less than 9 hits in 30 minutes (0.3 per minute). Convert to rate first.
- Counting long or uneven behaviors. If one tantrum lasts 10 seconds and another lasts 15 minutes, a count treats them as equal. Duration shows the difference.
- Starting the clock at the wrong moment. Duration runs from the onset to the offset of the behavior itself, not from the start of the session or from your instruction.
- Filling in data later. Estimating after the session makes the data unreliable, even when your guess feels close. Record what you observed and tell your supervisor about anything you missed.
Exam trap
Latency vs IRT. Both time a gap before a response; what starts the clock differs. Latency starts at a stimulus, such as an instruction or cue. IRT starts at the end of the previous response of the same behavior. If nothing cued the response, it is not latency.
Exam trap
Latency vs duration. When a question asks how quickly a learner responds after an instruction, the answer is latency, not duration. Latency ends when the response starts; duration is how long the response itself lasts.
Quick check
A teacher wants to know how much of a 20-minute work period Mateo, a 10-year-old student, spends actively working on his worksheet. Which measure fits best?