Continuous vs. discontinuous measurement

Every data sheet answers a hidden question: did the observer catch every instance of the behavior, or only a sample of it? That single distinction shapes what your numbers mean and how much you can trust them.

Why it matters

As a BCBA, you choose measurement systems that staff, teachers, and caregivers can actually use. Recording every response gives the most complete picture, but it is not always realistic. A teacher with 25 students cannot run a stopwatch on one child all morning. Sampling methods make data collection possible in busy settings, but they trade accuracy for convenience, and each one tends to err in a predictable direction.

If you don't know those directions, you can misread a graph. A behavior can look worse, better, or more stable than it really is because of how it was sampled. Task C.5 asks you to tell the two families apart and predict the error of each sampling method.

Key ideas

• Continuous measurement captures every instance. The observer records each occurrence of the target behavior during the whole observation period. Event recording (count and rate) and timing measures (duration, latency, and interresponse time) are continuous when they are used this way.
• Discontinuous measurement samples behavior. The observer scores intervals or checks at set moments instead of recording each instance, so some instances, or parts of instances, go undetected. The result is an estimate, not a true count or duration.
• The three main discontinuous methods are whole-interval recording, partial-interval recording, and momentary time sampling. Each is usually reported as a percentage of intervals scored.
• Each method has a typical bias. Whole-interval recording tends to underestimate how much of the period the behavior filled. Partial-interval recording tends to overestimate it, and it undercounts high-rate behavior because several responses in one interval score as a single interval. Momentary time sampling can err in either direction and easily misses brief behaviors.
• Shorter intervals reduce error. The smaller the intervals, the closer a discontinuous estimate tends to come to the continuous record, at the cost of more observer effort.
• Prefer continuous when it is feasible. Use discontinuous methods when continuous recording is impractical, for example when behavior is very frequent, lacks a clear start and end, or the observer has other duties.

Definitions

• Continuous measurement: any procedure that detects all occurrences of the target behavior during the observation period.
• Discontinuous measurement: any procedure in which some occurrences may go undetected, because behavior is scored by interval or at set moments instead of instance by instance.
• Whole-interval recording: an interval is scored only if the behavior lasts through the entire interval.
• Partial-interval recording: an interval is scored if the behavior occurs at any point during it, even once or briefly.
• Momentary time sampling (MTS): an interval is scored only if the behavior is happening at the exact moment the interval ends.
• Planned activity check (PLACHECK): a group version of momentary time sampling, where the observer counts how many people in a group are engaged in the behavior at each check.

Examples from a session

• Continuous in a clinic. Priya, an RBT, works one-to-one with Mateo, age 7. She taps a counter app every time he asks a peer to play and times each tantrum from onset to offset. Every instance is captured.
• Partial-interval at home. A caregiver, Mr. Osei, marks each 15-second interval in which his daughter Ama, age 4, puts objects in her mouth. Ama mouths a toy for one second near the start of an interval, and the whole interval is scored. The graph will suggest she mouthed objects for more of the session than she did.
• Whole-interval in a high school. Ms. Reyes scores an interval only if Jordan, age 16, stays engaged with his reading for all 30 seconds. Jordan reads for 25 seconds, checks his phone, and returns. That interval is not scored, so his engagement is underestimated.
• Momentary time sampling in a classroom. While teaching, Mr. Lee glances at Hana, age 10, each time a vibrating timer goes off every 2 minutes and notes whether she is in her seat. A quick trip out of her seat between checks is never recorded.

Common mistakes / exam traps

• Mixing up which method errs which way. Partial-interval overestimates; whole-interval underestimates; MTS can do either.
• Thinking any timer makes data continuous. Timing every instance is continuous. Checking only when a timer beeps is momentary time sampling.
• Choosing sampling when continuous is easy. If a behavior is countable, not too frequent, and someone is free to watch, continuous recording is the better choice.

Quick check

Related topics

• Measuring occurrence: count, rate, and percentage (C.3)
• Measuring time-based dimensions: duration, latency, and IRT (C.4)
• Designing interval recording and time-sampling procedures (C.6)
• Judging the validity and reliability of measurement (C.8)
• Choosing a measure that captures the critical dimension within real-world limits (C.9)