Scatterplot assessment

A grid makes time patterns visible

A typical scatterplot places dates or observation periods on one axis and time blocks or routines on the other. The observer marks whether the target behavior occurred, how often it occurred, or which intensity category applied in each eligible cell.

The visual can reveal clustering during late afternoons, particular transitions, one setting, or selected days. It can also show a diffuse pattern that argues against chasing one time window.

Define the cell before collecting data

A useful plan states:

• the observable target response
• the cell’s start and end events
• whether frequency, occurrence, duration, or intensity is recorded
• the exposure needed for an eligible cell
• rules for multiple events in one cell
• missing, interrupted, and not-applicable codes
• who records and how observers are trained
• relevant schedule, health, staffing, communication, and setting information

Without an exposure rule, a blank cell could mean zero behavior, no observation, cancelled activity, absence, or missing documentation.

Use time blocks that match the question

Very large blocks hide meaningful variation. Very small blocks increase workload and can create unstable patterns from a few events. Choose intervals that fit the behavior’s frequency, routine, decision, and available observation.

A lunchtime question may use routine-based cells. A full-day pattern may use 30-minute blocks. Keep the definition stable within the comparison period and version the form when it changes.

Occurrence and intensity are different

A yes-or-no mark reports whether at least one event occurred. It says nothing about whether there was one brief event or several long events. Frequency and duration can add detail when observers can record them reliably.

If intensity matters, define observable categories. Avoid vague labels such as mild, moderate, and severe unless each has measurable criteria and safety meaning. Report injury and emergency actions separately.

A fictional 30-block example

Omar’s team observes six predeclared routine blocks on each of five school days, creating 30 eligible cells. The defined response occurs in 9 of 30 cells, or 30%. Five of the nine occurrence cells follow schedule transitions.

The pattern suggests closer observation around transitions. It cannot show that transitions caused the response or that any particular consequence maintained it. Four occurrence cells fall elsewhere, and the 21 zero-occurrence cells still matter.

The next week, the team uses ABC recording around transitions and collects schedule, communication-access, sleep, pain, and staffing information. It keeps the original 30-cell denominator unchanged.

Temporal association differs from function

Function concerns the relation through which consequences influence future behavior. A scatterplot arranges no experimental comparison. Many events can vary with time, including activities, people, medication timing, fatigue, hunger, noise, transport, communication access, and observation quality.

The Hanley, Iwata, and McCord review defined functional-analysis research through direct measurement under at least two conditions involving manipulation of an environmental variable. That boundary helps explain why a scatterplot supports hypothesis development rather than a functional conclusion.

Follow patterns with better-matched evidence

Possible next steps include behavioral interview, direct ABC observation, ecological assessment, record review, medical or interdisciplinary referral, or a qualified clinician’s decision about experimental analysis. Select the next step based on the question, risk, burden, and confidence needed.

If a cluster disappears after correcting missing data or exposure, document the correction. Preserve the original version and reason for change.

Measure data quality

Useful metrics include eligible cells recorded divided by cells due, observer-agreement samples completed divided by samples due, and records entered within the target time divided by eligible records. Report missing cells by reason and age.

High completion cannot repair an ambiguous definition. Calibrate observers using the same examples and sample agreement across relevant times and settings.

Review the display with raw counts nearby. Dense shading can make one occurrence look the same as ten unless the legend preserves magnitude. When the team changes interval length, target definition, or intensity code, start a clearly labeled version and avoid joining unlike cells into one trend.

Keep interpretation person-centered

Ask the person what occurs during high-density periods and which conditions feel difficult or supportive. Preserve AAC, breaks, medical care, and ways to decline participation. A pattern should lead to better questions and environmental support rather than blame.

The BACB Test Content Outline covers measurement, visual analysis, assessment, environmental variables, and functional analysis as examination content. The current BACB Ethics Code addresses competence, client involvement, assessment, medical needs, risk, data, documentation, and evaluation for covered professionals.