Not every behavior can be counted with a tally. Some happen too fast, too frequently, or without a clear start and end point. That's where Partial Interval Recording comes in.
Partial Interval Recording (PIR) is a discontinuous measurement method that divides an observation period into equal time blocks. If the target behavior occurs at any point during an interval, that interval is scored as an occurrence. If it doesn't occur at all, it's scored as a non-occurrence. The result is reported as a percentage of intervals with occurrence.
It's not a perfect measure of frequency or duration. It's a practical, efficient way to track behaviors that would otherwise be difficult to capture accurately, and when used correctly, it produces data that reliably guides clinical decisions.
Theralytics makes Partial Interval Recording much easier by providing automated interval timers, real-time mobile prompts, digital scoring, and instant percentage calculations, allowing clinicians to focus on observation rather than manual timing and paperwork.
What Is Partial Interval Recording in ABA?
Partial Interval Recording is a type of interval recording where an observer divides a session into equal time blocks and marks each interval based on whether the target behavior occurred at any point within it.
Scoring rule: If the behavior occurs even once during the interval, the whole interval is marked "+". If it doesn't occur at all, the interval is marked "–".
Reported as: Percentage of intervals with occurrence. Formula: (Number of "+" intervals ÷ Total intervals) × 100
Quick examples:
- "Talking out of turn" during 30-second intervals. Any call-out within an interval = "+"
- "Hand flapping" during 10-second intervals. Any flap within an interval = "+"
- "Shirt picking" during 15-second intervals. Any instance within an interval = "+"
Key characteristic: PIR overestimates relative to true duration. A behavior that occurs briefly at the start of a 30-second interval fills the entire interval in the data. This is expected, understood, and factored into how the data is used.
When to Use Partial Interval Recording in ABA
PIR is the right choice in specific situations. Knowing when it fits and when it doesn't is what separates good data collection decisions from poor ones.
Use PIR when:
- The behavior occurs at a high rate or is brief in duration, making continuous recording impractical
- The behavior lacks clear start and end points, making frequency counting unreliable
- You want to decrease a behavior, PIR's tendency to overestimate provides a conservative measure of reduction
- Continuous observation isn't feasible due to setting demands or multiple learners
- You need a practical method that still produces meaningful, actionable data
Do not use PIR when:
- You need an exact count of how many times the behavior occurred → use frequency or rate
- You need to know how long the behavior lasted → use duration recording
- Constant observation across many learners is impractical → consider momentary time sampling
- The behavior is so low-rate that most intervals will be "–" and PIR loses sensitivity → use frequency recording
- PIR percentages are consistently near 0–10% or 90–100%, meaning the method has lost its ability to detect meaningful change
Partial Interval Recording vs. Whole Interval vs. Momentary Time Sampling
Understanding PIR means understanding where it sits relative to the other interval recording methods.
One-sentence decision rule:
- Decreasing a behavior → PIR
- Increasing a sustained behavior → WIR
- Monitoring group behavior or when continuous observation isn't possible → MTS
Classroom example:
- PIR for "calls out" every 30 seconds, mark "+" if it occurred at all
- WIR for "on-task behavior" across entire 1-minute intervals
- MTS to check if each student is "seated" at the 2-minute mark
Home example:
- PIR for "shirt picking" in 15-second intervals
- WIR for "brushing teeth for the full 30 seconds"
- MTS glance at 1-minute marks for "playing independently"
Pros and Cons of Partial Interval Recording
Pros:
- Feasible during long sessions or when managing multiple learners
- Sensitive to decreases in high-rate behavior, making it well-suited for reduction targets
- Requires less continuous attention than frequency or duration recording
- Captures behaviors that are too brief or frequent to count individually
- Provides both a frequency estimate and duration estimate from one data sheet
Cons:
- Overestimates relative to true duration and frequency. A brief instance fills the whole interval
- Percent of intervals is not the same as true rate or duration. Interpret accordingly
- Interval length significantly affects accuracy. Too long and overestimation increases. Too short and observer burden increases
- Not suitable for low-rate behaviors where most intervals will be scored "–"
- Does not tell you how many times the behavior occurred within an interval, only whether it did
How to Run a Partial Interval Recording Session: Step by Step
Step 1: Define the Target Behavior
Write a clear, observable, measurable behavioral definition. Specify exactly what counts as an instance and what doesn't. Every observer measuring the behavior must agree on the definition before data collection begins.
Example: "Talking out of turn" = any vocalization directed at the teacher or peers without first raising a hand and being called on. Does not include whispering to self or humming.
Step 2: Set the Observation Window
Choose a consistent, realistic observation period. 15–30 minutes of typical instruction works well for most classroom or session-based targets. Keep the window the same across sessions so data is comparable.
Step 3: Choose Your Interval Length
Interval length is one of the most consequential decisions in PIR. Get this wrong and your data will be less useful.
- High-rate or brief behaviors: 10–30 seconds
- Lower-rate or longer behaviors: 1–5 minutes
- General rule: Shorter intervals produce more accurate data but require more observer effort. Match the interval to the behavior's natural cadence.
- Best practice: Gather baseline observational data first. Know roughly how often the behavior occurs before choosing your interval. If a behavior occurs roughly once per minute, a 30-second interval will capture it reliably. A 5-minute interval will miss most occurrences.
Step 4: Prepare Your Data Sheet and Timer
Set up a printed PIR grid or a digital tool with "+ / –" boxes for each interval. Program an interval timer or use ABA data collection software that provides automated interval prompts. Manual clock-watching during live observation is error-prone and splits your attention.
Step 5: Observe and Score
At the moment the behavior occurs within an interval, mark "+" and you're done with that interval. You do not need to track how many times it occurred or how long it lasted within the block. If the behavior doesn't occur at all during the interval, mark "–" when the interval ends.
Stay consistent with your rule. Do not upgrade a "–" to "+" based on a near miss.
Step 6: Calculate Percentage of Intervals
At the end of the session:
% of intervals = (Number of "+" intervals ÷ Total intervals) × 100
Example: 14 "+" intervals out of 40 total = 35% of intervals with occurrence.
Step 7: Graph and Interpret
Plot % of intervals on the y-axis across session dates on the x-axis. Add phase change lines when interventions are introduced or modified. Look for level, trend, and variability across phases.
Worked Example (With Complete Sheet)
- Scenario: “Talking out of turn” during a 20‑minute class, 40 × 30‑second intervals. Below is the full interval grid for one session (8 rows × 5 intervals).
- Key: “+” = occurred; “–” = did not occur.
- Tally: 14 “+” out of 40 intervals
Calculation: (14 ÷ 40) × 100 = 35% of intervals with talking out of turn. - Second session (summary): 24 “+” out of 40 = 60%.
Trend: 60% → 35% (improved). Important: This reflects fewer intervals with any occurrence, not fewer total call‑outs per se.
How to Calculate Interobserver Agreement (IOA) for Partial Interval Recording
IOA is what verifies that your data is trustworthy. Two trained observers using the same definition and the same intervals should get similar results. If they don't, something is wrong with the definition, the training, or the procedure.
BACB standards generally recommend IOA checks on at least 20–33% of sessions, with a target of 80% or higher.
1. Interval-by-Interval (I×I) IOA
What it measures: Agreement on both occurrence and non-occurrence across every interval.
Formula: (Number of intervals with agreement ÷ Total intervals) × 100
Example: Two observers, 10 intervals. They agreed on 8 intervals (both scored "+" or both scored "–"). IOA = (8 ÷ 10) × 100 = 80%
Best for: Most standard PIR applications. Gives a complete picture of agreement across all intervals.
Limitation: When behavior is very low-rate, many "–" agreements can inflate IOA artificially. A pair of observers can agree 90% of the time simply because the behavior rarely occurs.
2. Scored-Interval IOA
What it measures: Agreement only on intervals where at least one observer scored an occurrence ("+").
Formula: (Agreed "+" intervals ÷ Intervals where at least one observer scored "+") × 100
Example: 5 intervals had at least one "+" between the two observers. Both scored "+" in 3 of those 5. IOA = (3 ÷ 5) × 100 = 60%
Best for: Low-rate behaviors where "–" agreements would inflate I×I IOA. Scored-interval IOA removes the noise of non-occurrence agreement and focuses only on whether observers agree when behavior is present.
3. Unscored-Interval IOA
What it measures: Agreement only on intervals where at least one observer scored a non-occurrence ("–").
Formula: (Agreed "–" intervals ÷ Intervals where at least one observer scored "–") × 100
Example: 7 intervals had at least one "–". Both scored "–" in 5 of those 7. IOA = (5 ÷ 7) × 100 = 71.4%
Best for: High-rate behaviors where "+" agreements dominate and would inflate I×I IOA. Unscored-interval IOA removes that inflation and focuses agreement analysis on the non-occurrence side.
Which IOA Method to Use
How to Graph Partial Interval Recording Data
Good graphing turns raw data into decisions. A poorly labeled graph is just as problematic as a poorly collected data sheet.
Axes:
- X-axis: Session date or session number
- Y-axis: Percentage of intervals with occurrence (0–100%)
What to include on every graph:
- Target behavior name
- Interval length used (e.g., 30-second intervals)
- Observation window (e.g., 20-minute session)
- Setting (e.g., classroom, home, clinic)
- Who collected the data
- Phase change lines with labels (Baseline, Intervention A, Intervention B)
What to look for:
- Level: Is the overall percentage high or low?
- Trend: Is it moving in the right direction across sessions?
- Variability: Is the data stable or jumping around? High variability may indicate inconsistent implementation or measurement error.
How to Explain Partial Interval Recording Data to Caregivers and Teachers
PIR data is meaningful to clinicians. It's often confusing to everyone else. Translating it clearly is part of the job.
Avoid saying: "The behavior occurred in 35% of intervals."
Instead say: "During today's observation, we divided the session into 40 equal blocks of 30 seconds each. The behavior showed up in about 14 of those blocks. That's down from 24 blocks last week, which tells us the intervention is reducing how often the behavior is showing up throughout the day."
Key points to communicate:
- PIR tells us how much of the session was affected by the behavior, not the exact number of times it happened
- A decrease in percentage of intervals means the behavior is occupying less of the learner's time
- Phase lines on the graph mark when we made changes so you can see what the data looked like before and after
Clinical Decision Rules and Troubleshooting
If PIR percentages are consistently near 0–10%: PIR is losing sensitivity. The behavior is occurring in too few intervals to detect meaningful change. Switch to frequency recording for a more precise measure.
If PIR percentages are consistently near 90–100%: PIR is saturated. Almost every interval is scored "+" so changes in the behavior won't show up in the data. Shorten the interval length or switch to frequency or duration recording.
If observers are consistently below 80% IOA: Check the behavioral definition first. If the definition is clear, check observer training. If training is solid, check whether the interval length is appropriate for the behavior's rate.
If the behavior is briefer than your interval: Shorten intervals. A behavior that lasts 2 seconds observed in 5-minute intervals will fill almost every interval, making every session look the same regardless of actual change.
If observers feel overwhelmed: Lengthen intervals slightly, but understand the accuracy trade-off. Alternatively, reduce the number of targets being measured simultaneously.
For clinical decision-making: Most single-case design standards recommend at least 3–5 data points per phase before drawing conclusions about level or trend. More is always better when feasible.
Final Thoughts
PIR is one of the most practical data collection tools in ABA. Used correctly, with a clear behavioral definition, appropriate interval length, and regular IOA checks, it produces data that reliably guides treatment decisions for high-rate and difficult-to-count behaviors.
The challenge is applying it consistently across sessions, settings, and team members, especially when observers are managing other demands at the same time. Automated interval prompts, real-time data collection, and automatic graphing remove the administrative burden so clinical teams can focus on observation rather than timekeeping.
Theralytics supports interval recording directly within the platform. Set your interval length and observation window, use the mobile app's automated prompts during sessions, and let the platform calculate percentage of intervals and generate graphs automatically. All of it is HIPAA-compliant, works offline on iOS and Android, and feeds directly into your reporting and analytics alongside the rest of your clinical documentation.
Book a free 15-minute demo to see how Theralytics supports interval recording and discontinuous measurement across your clinical team.
Frequently Asked Questions About Partial Interval Recording in ABA
What is Partial Interval Recording in ABA? Partial Interval Recording is a discontinuous measurement method where an observation period is divided into equal time blocks. If the target behavior occurs at any point during an interval, that interval is scored as an occurrence. Results are reported as a percentage of intervals with occurrence.
When should I use Partial Interval Recording? Use PIR for high-rate or brief behaviors where continuous recording isn't feasible, when the behavior lacks clear start and end points, or when you're targeting a behavior for decrease. PIR's overestimation bias makes it a conservative measure of behavioral reduction.
What is the difference between Partial Interval and Whole Interval Recording? PIR scores an interval as "+" if the behavior occurs at any point during it, which overestimates relative to true duration. WIR scores an interval as "+" only if the behavior occurs for the entire interval, which underestimates relative to true duration. Use PIR for behaviors you want to decrease and WIR for behaviors you want to increase.
Why does Partial Interval Recording overestimate behavior? Because any brief occurrence within an interval fills the entire interval in the data. A behavior that lasts 2 seconds in a 30-second interval counts the same as one that lasts 29 seconds. This is expected and accounted for in how PIR data is interpreted.
How do I calculate IOA for Partial Interval Recording? Three methods apply: Interval-by-Interval IOA (agreements on all intervals ÷ total intervals), Scored-Interval IOA (agreements on "+" intervals ÷ intervals where at least one observer scored "+"), and Unscored-Interval IOA (agreements on "–" intervals ÷ intervals where at least one observer scored "–"). Use scored-interval for low-rate behaviors and unscored-interval for high-rate behaviors.
What interval length should I use? Match the interval to the behavior's natural frequency. High-rate or brief behaviors: 10–30 seconds. Lower-rate or longer behaviors: 1–5 minutes. Shorter intervals produce more accurate data but require more observer effort. Gather baseline observational data first so you know roughly how often the behavior occurs before choosing your interval.
How do I graph Partial Interval Recording data? Plot percentage of intervals on the y-axis and session date on the x-axis. Include the target behavior name, interval length, observation window, setting, and observer on the graph. Add phase change lines when interventions are introduced or modified.
Can I use Partial Interval Recording for multiple behaviors at once? Yes, but observer burden increases with each additional target. Use a data sheet with separate columns for each behavior. Digital tools with automated interval prompts make multi-target PIR significantly more manageable than paper-based systems.
.avif)
