Skip to main content
Measurement Foundation

Baseline Data Collection: Measure Before You Intervene

Baseline data captures how a behavior is occurring before any intervention begins. It is the comparison standard that lets you prove an intervention actually worked. Learn how much to collect, how to judge stability and trend, and how to avoid the mistakes that invalidate the comparison.

What is Baseline Data?

Baseline data is behavior data collected before an intervention begins, in order to establish the current, pre-intervention level of the behavior. It answers a simple but essential question: how often, how long, or how intensely is this behavior already happening under existing conditions? Because it describes the behavior before you change anything, baseline becomes the yardstick you measure against later.

Without a baseline, you cannot demonstrate that a behavior changed, let alone that your intervention was responsible for the change. Baseline turns "it seems better" into a defensible, data-based conclusion. It is foundational to the broader data collection process, to writing and monitoring behavior intervention plans, and to the assessment work described in the functional behavior assessment guide.

In single-case design terms: baseline is the "A" phase. The intervention is the "B" phase. You compare the B phase back to the A phase to judge whether — and how much — the behavior changed.

Why Baseline Data Matters

It proves the intervention worked

Change can only be demonstrated against a starting point. Baseline gives you that starting point, so a drop (or rise) after intervention is measurable rather than a matter of opinion.

It enables data-based decision making

Teams decide to continue, modify, or stop an intervention by comparing current data to baseline. Without the reference line, those decisions are guesses.

It anchors IEP goals and progress monitoring

A measurable IEP goal states a present level and a target. Baseline is the present level, and progress monitoring charts movement from it toward the target.

It is the "A" phase of single-case design

In ABA and applied research, baseline is the control condition. Comparing the intervention phase to a stable baseline is what allows a functional, causal claim.

How to Collect Baseline Data: Step-by-Step

1

Choose the Same Measure You Will Use During Intervention

Pick the dimension and method you will keep using once the intervention starts — frequency or rate, duration, latency, or interval percentage. Whatever you choose for baseline must stay the same during intervention, because changing the measure between phases invalidates the comparison.

Pro Tip:

Match the measure to the behavior: use frequency/rate for discrete countable behaviors, duration for how long behaviors last, and interval recording for continuous or high-rate behaviors.

2

Decide Conditions, Times, and Settings

Identify when and where the behavior is relevant and plan to sample across those conditions. Collecting only at the "worst" time or in a single setting produces a baseline that does not represent typical behavior.

Questions to answer:

  • •Which times of day is the behavior most and least likely?
  • •Which settings, activities, or people are involved?
  • •How will you sample across them rather than cherry-picking?
  • •Are conditions being kept the same throughout baseline?
3

Collect Until the Data Are Stable and Representative

Gather at least 3–5 data points across different days, and ideally continue until the pattern is stable (low variability) or shows a clear trend. Low-rate behaviors may need a longer window to be captured fairly. Critically, do NOT implement any part of the planned intervention during this phase.

Key considerations:

  • •At least 3–5 points across different days as a minimum
  • •Continue until stable or a clear trend emerges
  • •Extend the window for low-rate behaviors
  • •Keep all conditions unchanged — no intervention yet
4

Check Stability, Trend, and Variability

Graph the baseline and evaluate three things: stability (how consistent the points are), trend (the direction — increasing, decreasing, or flat), and variability (how much the points bounce). Interpret what you see before deciding whether baseline is ready.

How to interpret what you see:

  • •Stable, flat, low-variability baseline: clearest starting line
  • •Baseline already trending toward the goal: harder to attribute later change to the intervention
  • •Worsening baseline: strengthens the case that an intervention caused improvement
  • •High variability: usually collect more data before intervening
5

Then Begin the Intervention

Once baseline is stable and representative, introduce the intervention and keep measuring the exact same way. Continue charting so that every intervention data point can be compared directly back to the baseline reference line.

Pro Tip:

Mark a clear phase-change line on your graph where baseline ends and intervention begins so the comparison is obvious to anyone reviewing the data.

Baseline Data Example

Suppose the target behavior is call-outs during independent work, measured as a frequency count during a 30-minute work period each day. Before any intervention, the team records five days of baseline:

DaySettingCall-outs (30-min period)
MondayIndependent math work11
TuesdayIndependent math work9
WednesdayIndependent reading12
ThursdayIndependent math work10
FridayIndependent reading11

Interpretation:

  • Stability:The counts stay in a tight 9–12 band — the baseline is stable.
  • Trend:There is no upward or downward drift; the data are essentially flat.
  • Variability:Low — only three call-outs separate the highest and lowest days.
  • Decision:With five stable points averaging about 10–11 call-outs, this baseline is representative and ready. If a self-monitoring intervention then drops call-outs to 2–3 per period, the change is clear and defensible because it is measured against this reference line.

Common Baseline Mistakes to Avoid

✗

Collecting too few data points

"Declaring a baseline after one or two observations"

✓Gather at least 3–5 points across different days before drawing conclusions
✗

Starting intervention over an unstable baseline

"Baseline bounces from 2 to 18 and you intervene anyway"

✓Keep collecting until the data are stable or show a clear trend
✗

Changing the measure between phases

"Baseline used frequency, but intervention data uses duration"

✓Use the exact same dimension and method in baseline and intervention
✗

Intervening during baseline

"Adding prompts or supports while "still collecting baseline""

✓Keep conditions unchanged — the intervention starts only after baseline
✗

Baseline too short for a low-rate behavior

"Three days of baseline for a behavior that occurs about once a week"

✓Extend the window so low-rate behavior is represented fairly
✗

Prolonging baseline for dangerous behavior

"Extending a no-treatment phase while a student is unsafe"

✓Balance rigor with safety — do not unnecessarily delay needed intervention

Frequently Asked Questions

What is baseline data?▼

Baseline data is behavior data collected before an intervention begins, in order to establish the current, pre-intervention level of the behavior. It describes how often, how long, or how intensely the behavior is already occurring under existing conditions. Baseline is the comparison standard against which you later measure whether an intervention actually changed the behavior.

How much baseline data do I need before starting an intervention?▼

Collect enough data points to see a stable, representative pattern. As a rule of thumb this is at least 3–5 data points gathered across different days, and ideally you continue until the data are stable (low variability) or show a clear trend. Low-rate behaviors may need a longer window to be represented accurately. The goal is a trustworthy picture of the behavior, not a fixed number of days.

What is the difference between stability, trend, and variability in baseline?▼

Stability refers to how consistent the data are from point to point. Trend is the overall direction of the data — increasing, decreasing, or flat. Variability (or "bounce") is how much individual points scatter around that trend. A stable baseline with low variability and no strong trend gives the clearest starting line for judging an intervention.

Should I ever intervene during the baseline phase?▼

No. The purpose of baseline is to capture the behavior under existing, unchanged conditions. Introducing supports, prompts, or the planned intervention during baseline contaminates the comparison and makes it impossible to attribute later change to the intervention. Keep conditions the same throughout the baseline phase.

What if the behavior is dangerous — can I shorten baseline?▼

Yes. Ethics and safety take priority over experimental rigor. For dangerous or severe behavior you should not unnecessarily prolong baseline; collect the minimum representative data needed and move to intervention, or use a design that does not require an extended no-treatment phase. Document the clinical rationale for a shortened baseline.

Why does baseline matter for IEP goals and single-case design?▼

IEP goals must state a present level and a measurable target, and progress monitoring compares later data back to that starting point — that starting point is the baseline. In single-case design, baseline is the "A" phase that the intervention "B" phase is compared against. Without baseline you cannot demonstrate that change occurred or that your intervention was responsible for it.

Chart Baseline and Intervention in One Place

Classroom Pulse graphs baseline and intervention phases together, with clear phase-change lines and trend analysis, so you can see whether an intervention is working at a glance.