RBT Data Collection and Graphing Study Guide

Observable and measurable definitions come first

Everything in this domain rests on one skill: describing behavior so that two people watching the same session would record the same thing. "Aggression" is not a definition. "Hitting, kicking or biting another person, including attempts that do not make contact" is, because it names topographies and resolves the obvious edge case.

Exam items in this domain frequently give you four candidate definitions and ask which is observable and measurable. The wrong options almost always contain a word describing an internal state — frustrated, upset, unmotivated, defiant, manipulative — or a summary label like "non-compliant" that different observers would apply differently.

Continuous measurement: every occurrence is captured

Continuous procedures record every instance of the behavior during the observation period. There are four you must be able to tell apart under time pressure, because the exam will describe a scenario and ask which one is being used.

Frequency (event recording)

A simple count of occurrences. Requires behavior with a clear beginning and end, and roughly consistent length. "He left his seat 12 times."

Rate

Frequency divided by the time of the observation. "12 occurrences in 40 minutes = 0.3 per minute." Use rate whenever sessions differ in length, because raw counts are not comparable across them.

Duration

How long the behavior lasts. Can be reported per occurrence or as a total across the session. Total duration is often converted to a percentage of session time.

Latency

Time from the onset of a stimulus (usually an instruction) to the onset of the response. It stops when the client starts, not when they finish — that is the single most common error on latency items.

Inter-response time (IRT)

Time from the end of one response to the beginning of the next. Pairs naturally with DRL procedures, where you are trying to spread responding out.

Discontinuous measurement: sampling, and its known biases

Discontinuous procedures divide the observation into intervals and sample rather than capture everything. They exist because an RBT frequently cannot watch a stopwatch continuously while also running a session. The trade-off is accuracy, and each procedure errs in a predictable direction — which is exactly what the exam tests.

Partial interval

Score the interval if the behavior occurred at any point during it. Overestimates. Typically used for behaviors you want to decrease.

Whole interval

Score the interval only if the behavior occurred throughout the entire interval. Underestimates. Typically used for behaviors you want to increase, such as on-task engagement.

Momentary time sampling

Score whether the behavior is occurring at the moment the interval ends. Frees the observer between moments, which makes it the practical choice for group settings and community outings.

Permanent product and IOA

Permanent-product recording measures the physical outcome the behavior left behind — completed problems, toys in the bin, dishes washed. Its advantage is that you do not have to observe live. Its limitation is that the product does not tell you who produced it or how much help they had.

Interobserver agreement checks the measurement system, not the client. Two observers record independently and their records are compared. Total count IOA divides the smaller count by the larger. Interval-by-interval IOA divides agreements by the total number of intervals. Agreement can never exceed 100% — if your answer does, you inverted the fraction.

Reading a line graph

Sessions go on the x-axis, the behavior measure on the y-axis. A vertical dashed line marks a phase change, and data paths are not connected across it so that conditions can be compared cleanly.

You will be asked to describe data using three features. Level is where the data sit on the y-axis — roughly the average value in a phase. Trend is the direction they are moving. Variability is how much they bounce from session to session. A phase can show a lower level and high variability at the same time, and a good answer names both.