Frequency · Rate · Duration · Latency · IRT · IOA · Graphing · Data Interpretation
Accurate measurement is the foundation of ABA practice. RBTs collect data on behavior to allow their supervising BCBA to make informed programming decisions. Understanding what and how to measure is essential for every session.
Data drives decisions in ABA. Without accurate measurement, the team cannot know if a behavior is improving, worsening, or staying the same. The RBT's role is to collect data precisely and consistently as directed by the supervisor.
Behavior can be measured across several dimensions — count/frequency, rate, duration, latency, IRT, and magnitude. The BCBA selects which dimension is most meaningful for each target behavior.
RBTs implement data collection procedures exactly as trained; they do not design measurement systems. They graph data as directed and report concerns to the supervising BCBA promptly.
Data must be accurate and honest. Never fabricate, alter, or omit data. Inaccurate data leads to incorrect clinical decisions and harms clients.
| Dimension | Definition | Example | Best Used When |
|---|---|---|---|
| Frequency / Count | Number of times behavior occurs | Child hit table 4 times | Discrete behaviors with clear start/end |
| Rate | Frequency per unit of time | 4 hits per 10 min = 0.4/min | Comparing across sessions of different lengths |
| Duration | Total time behavior lasts | Tantrum lasted 8 minutes | Behaviors where how long matters |
| Latency | Time from SD to behavior onset | Responded 3 sec after instruction | How quickly client initiates behavior |
| Inter-Response Time (IRT) | Time between two consecutive responses | 45 sec between each step attempt | Behaviors affected by pacing or timing |
| Magnitude | Force or intensity of behavior | Loud vs quiet vocalization | When severity of behavior is the concern |
| Celeration | Rate of change in behavior over time | Graphed on Standard Celeration Chart | Precision teaching contexts |
Frequency = raw count (5 times). Rate = frequency ÷ time (5 per 20 min = 0.25/min). Rate is preferred when session lengths vary — it allows fair comparison across sessions.
Start a timer when behavior begins, stop when it ends. Can measure total duration per session OR average duration per occurrence. Used for behaviors like on-task time, tantrum length, self-stimulatory behavior.
Measure from the moment the SD (instruction/prompt) is delivered until the client begins the response. Important for measuring compliance speed and prompt dependency reduction.
Time between the END of one response and the START of the next. Used when responses are chained or paced — e.g., spacing between self-injurious responses to target deceleration.
| Method | Type | Description | Best For |
|---|---|---|---|
| Event Recording | Continuous | Tally each occurrence | Discrete, low-frequency behaviors |
| Duration Recording | Continuous | Time behavior from start to stop | Duration of behavior |
| Latency Recording | Continuous | Time from SD to response start | Response speed |
| IRT Recording | Continuous | Time between consecutive responses | Response pacing |
| Whole Interval | Discontinuous | Behavior must occur ENTIRE interval to score + | Sustained behaviors (on-task) |
| Partial Interval | Discontinuous | Behavior occurs ANY PART of interval = + | Detecting behavior presence; overestimates |
| Momentary Time Sampling (MTS) | Discontinuous | Check at END of interval; record if occurring | Ongoing/high-frequency behaviors |
| Permanent Product | Continuous | Measure outcome/artifact of behavior | Written work, assembled items |
Two independent observers collect data simultaneously on the same behavior. IOA measures how consistently the behavior is being measured. Acceptable IOA is typically ≥80%. Low IOA = measurement problem — not necessarily a behavior change.
Agreements ÷ (Agreements + Disagreements) × 100. Most stringent method. Used for event and interval data. Example: Observer 1 records [+,+,−,+,−] and Observer 2 records [+,−,−,+,−] → 4 agreements, 1 disagreement → 4/5 = 80%.
Smaller count ÷ Larger count × 100. Used for frequency data. Less precise — two observers could agree on total but disagree on WHEN behavior occurred. Example: Observer 1 = 10, Observer 2 = 8 → 8/10 = 80%.
Used with interval recording. Compare each interval agreement. Similar to point-by-point but applied to interval data.
Recording the lasting outcome (product) of behavior rather than the behavior itself. Examples: number of math problems completed correctly, number of puzzle pieces assembled, words written. Advantage: can be measured after the fact; no need to observe in real-time.
| Element | Description |
|---|---|
| X-axis | Sessions or days (time) |
| Y-axis | Behavior measurement (frequency, rate, duration, %) |
| Data Points | Individual session data plotted as dots |
| Phase Lines | Vertical solid lines marking condition changes (e.g., baseline to intervention) |
| Phase Labels | Labels above each phase (e.g., "Baseline," "DTT Intervention") |
| Condition Change | When a new procedure begins; marked by phase line |
| Trend Line | Direction of data path: increasing, decreasing, or flat/variable |
| Level | The overall value/height of data in a phase |
Direction of data over time. Accelerating (increasing) trend for behavior targeted for increase = good. Decelerating (decreasing) trend for behavior targeted for reduction = good. Variable trend may indicate inconsistency in implementation.
The average or overall magnitude of the data within a phase. Changes in level between phases (especially abrupt) indicate the intervention had an immediate effect.
How much data points fluctuate. High variability makes it difficult to detect trends. The BCBA analyzes variability to determine if behavior is stable enough to change conditions.
Semi-logarithmic chart used in precision teaching. Y-axis is logarithmic (equal ratios). Used to track celeration — rate of change in rate. The RBT may be asked to plot on an SCC; the BCBA interprets it.
The simplest graph structure. A = baseline (no intervention), B = intervention. Phase line separates them. RBT graphs data; BCBA determines when to change phases.
Select the best answer for each question, then click "Check Answer." Submit all when finished to see your score.
High-retention mnemonics to lock in the concepts that matter most on the RBT exam.
Frequency = raw count. Rate = count ÷ time. Use RATE when sessions vary in length so you're comparing fairly.
Latency: time until behavior STARTS (after SD). Duration: how LONG behavior lasts. Latency ends when behavior begins.
Partial interval records + if behavior occurs ANY PART of the interval — tends to OVERestimate. Whole interval UNDERestimates.
Agreements ÷ (Agreements + Disagreements) × 100. Most stringent. ≥80% acceptable.
X-axis = time (sessions/days). Y-axis = behavior measure. Phase change line = vertical solid line at condition change.
Count/measure the lasting product of behavior (worksheets, assembled items) — no need to observe in real time.
Click a card to flip it
Start with the six measurement dimensions: Frequency, Rate, Duration, Latency, IRT, Magnitude. Know the difference between frequency (raw count) and rate (per unit time). Then learn the graph parts: X-axis=time, Y-axis=behavior, phase line=condition change.
Study the four recording methods (event, duration, latency, IRT) and three interval methods (whole, partial, MTS). Know which over/underestimates. Practice IOA calculations: point-by-point vs total count.
Focus on when to use each measurement dimension clinically, IOA calculation methods and acceptable thresholds, and reading graphs for trend/level/variability. Know permanent product and when it applies.
High-yield: Rate=frequency÷time; Partial interval=overestimates; Whole interval=underestimates; Point-by-point IOA=most stringent; Latency=SD to response start; Phase change line=vertical solid line; IOA ≥80% acceptable.
Frequency=count; Rate=per time; Latency=SD to start; Duration=how long; IRT=between responses; Partial interval overestimates; Whole interval underestimates; MTS=momentary check; IOA point-by-point=agreements÷total; Phase line=vertical line at condition change.