> ## Documentation Index
> Fetch the complete documentation index at: https://docs.sightlinebehavior.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Interval Recording
> Score behaviors at fixed intervals using whole interval, partial interval, or momentary time sampling methods.
* Percentage data comparable across sessions of different lengths
* Eligibility evaluations needing baseline-to-intervention comparisons
* Observing multiple students in rotation or dividing attention across a group
* Behaviors hard to count or time (off-task, engagement)
1. Choose an interval method (whole, partial, or MTS) from the method picker
2. Set interval length (common: 10s, 15s, 30s) and total session duration (15–30 min)
3. Define target behaviors with operational definitions
4. Optional: enable activity context or peer comparison
Interval recording divides a session into fixed time blocks and scores whether a behavior occurs during or at each interval. The three methods below differ in scoring rules and measurement bias. Pick based on what you want the data to show.
## Keyboard shortcuts
| Key | Action |
| ------------- | ----------------------------------- |
| `1–9` | Score behavior for current interval |
| `N` | Add note |
| `Cmd+Z` | Undo last score |
| `Cmd+Shift+Z` | Redo last undo |
| `Space` | Pause / resume session timer |
| `Escape` | Return to current interval |
| `Cmd+E` | End session |
| `Cmd+/` | Show all shortcuts |
## The three interval methods
Sightline supports three distinct interval recording approaches, and the choice matters more than it looks: the three methods can read the same session very differently. In a classic textbook demonstration, a behavior truly present about 55% of a session scored roughly 30% under whole interval, roughly 70% under partial interval, and roughly 50% under momentary time sampling. None of the three is wrong; each answers a different question, with a known direction of error. Choose one based on your assessment question:
### Whole Interval Recording
**Rule:** Score the behavior only if it occurs throughout the *entire* interval without stopping.
**Interpretation:** Conservative estimate. Any brief interruption cancels the score.
**When to use:** When you want to demonstrate that a behavior is pervasive and continuous, best for sustained behaviors like on-task engagement or positive peer interaction you hope to increase.
**Key property (underestimates):** A student on-task for 90% of an interval but briefly looking away will score 0 for that interval. Whole interval data tends to underestimate the true rate of the behavior.
**Example:** Observing on-task behavior with 15-second intervals over 20 minutes yields 80 intervals total. If the student scores 28 intervals, the result is 35% of intervals on-task: a conservative lower-bound estimate. Pair this with the caveat that actual on-task time was likely somewhat higher.
### Partial Interval Recording
**Rule:** Score the behavior if it occurs at *any point* during the interval, regardless of how briefly.
**Interpretation:** Liberal estimate. One brief instance scores the entire interval.
**When to use:** When you want to capture low-rate or short-duration behaviors that might be missed by whole interval recording, useful for off-task behavior, stereotypy, or verbal outbursts.
**Key property (overestimates):** A single two-second outburst in a 30-second interval scores the same as 30 seconds of continuous outburst. Partial interval data tends to overestimate the true rate of the behavior.
**Example:** Observing off-task behavior with 15-second intervals. If the student scores 52 of 80 intervals, the result is 65% of intervals off-task. Read it as an upper-bound estimate; actual off-task time was likely somewhat lower.
### Momentary Time Sampling (MTS)
**Rule:** At the end of each interval, observe the student's state at that *exact moment* and score accordingly. The timer pauses briefly so you can score before the next interval begins.
**Interpretation:** Least-biased snapshot. You capture a point-in-time sample of behavior.
**When to use:** The most practical method for long sessions or when observing multiple students in rotation. Provides the least biased prevalence estimate among discontinuous recording methods.
**Key property (point-in-time):** MTS does not track what happened *during* the interval, only the behavior present at the exact boundary. It can miss brief behaviors that occur and end between observation points. However, for behaviors that are relatively stable over time (engagement, on-task state, activity), MTS estimates are generally more accurate than partial interval.
**Example:** Observing academic engagement with 15-second intervals. At the end of each 15-second block, the timer pauses momentarily. You observe whether the student is engaged *right now*. If engaged at 42 of 80 observation points, the result is 53% of moments engaged: a reasonable estimate of the true proportion of the session during which the student was engaged.
## Scoring intervals in Sightline
Once you press Start, a behavior grid appears on screen with one row for each behavior and columns for each interval, scrolling forward as the session progresses.
**To score an interval:**
* **Click the cell** at the interval you want to score
* **Or press the number key** (1, 2, 3...) corresponding to the behavior
* The cell fills with color, indicating the behavior occurred
You can scroll back through the grid to view past intervals or correct errors; press **Escape** to return to the current interval.
### MTS and the paused boundary
For MTS, Sightline pauses the session timer at each interval boundary and displays a prompt ("Score now"). You observe the student's state at that exact moment, score, and the timer resumes. Whole and partial interval can opt into the same paused, prompted cadence (enable "Prompt each interval" in Setup), but for MTS it's mandatory, since the rateable moment *is* the boundary.
**Why:** MTS explicitly requires observation of the behavior state *at a specific moment*. Pausing gives you that moment to score without rushing. Sightline won't advance to the next interval until every behavior has been scored: there's no time limit once the bell sounds, so take the time you need.
## Reading the results
An interval session produces:
* **Percentage of intervals**, per behavior: (intervals scored ÷ total intervals) × 100
* **Total intervals**: actual count of intervals completed
* **Interval-by-interval grid**: color-coded visualization showing when behavior occurred
* **Trend**: running percentage across the session, useful for detecting drift or phase effects
* **Activity comparison** (if activity context was enabled): a separate percentage for each activity type — for example, large group instruction 40% off-task, independent work 65%, transition 80% — revealing whether behavior is activity-specific and which settings need targeted intervention
* **Peer comparison** (if a peer was observed): target vs. peer percentage and the difference between them
Every interval percentage is an estimate of true occurrence, biased in a known direction, and reading it means holding that direction in mind:
* **Whole interval** underestimates: a 35% on-task figure means true on-task time was likely somewhat higher.
* **Partial interval** overestimates, sharply for brief low-rate behaviors: a 65% off-task figure means actual off-task time was likely somewhat lower.
* **MTS** comes closest to a true duration estimate of the three, though accuracy still depends on interval length.
Because the result is a percentage, it compares directly across sessions of different lengths: a 20-minute and a 30-minute session both report standardized data, so a change between them reflects the behavior, not the length of the observation.
## Common mistakes to avoid
Whole interval for behaviors you want to increase (a conservative estimate suits on-task or compliance), partial for low-rate behaviors you want to decrease (stereotypy, outbursts that whole interval would miss), MTS for stable states and long sessions — for 30+ minutes it also reduces cognitive load compared to continuous observation. Mixing methods within a study produces incomparable data.
If your first session uses 15-second intervals and your second uses 30-second intervals, the data are less comparable. Standardize on an interval length before you start baseline.
Operational definitions must specify what counts as "on" and "off." Example: "on-task: eyes and materials oriented; off-task: eyes diverted for >3 seconds." Without this clarity, scoring becomes subjective.
Interval data alone don't contextualize severity. A 50% off-task rate could be typical or concerning depending on classroom norms. Always observe a comparison peer if possible.
Whole interval underestimates, partial overestimates. Introduce the method's bias once at the top of your observation section so readers weigh the numbers correctly, and don't compare percentages produced by different methods or interval lengths as if they were the same measure.
Before starting interval recording, collect 1–2 frequency sessions to establish raw rate. This gives context to your interval percentages.
***
For turning these percentages into report-ready sentences, see [Writing Up Observation Results](/writing-up-observation-results). For choosing between methods, see [Observation Methods](/observation-methods).