Phases of Single Subject Design: Baseline, Intervention, and Reversal

If you want to study how a specific intervention changes one person’s behavior – not a group, but a single individual – you need a research design built for exactly that. Single Subject Design (SSD) is that tool. Widely used in clinical psychology, applied behavior analysis, and educational research, SSD tracks how an individual responds to a treatment over time. But what makes it rigorous – and not just anecdotal – is its structured use of distinct phases. These phases, the baseline, the intervention, and the reversal, each serve a specific scientific purpose, and together they allow researchers to move from simple description all the way to establishing causation.

What is single subject design?
The three core phases of single subject design
Phase A: the baseline
Phase B: the intervention
Phase A again: the reversal
Descriptive, correlational, and causal knowledge: a progression
Why data stability is non-negotiable
Limitations of the reversal phase
A structured path to understanding individual behavior

What is single subject design?

Single-subject experimental design (SSED) focuses on the repeated measurement of a single participant’s behavior across time and conditions. Unlike group studies that average results across many people, SSD zooms in on one individual, tracking changes in a dependent variable as conditions shift. This makes it especially valuable when large-scale experiments are impractical, when population sizes are small (such as in rare disorders), or when the goal is to evaluate a specific intervention’s impact on a specific person.

Research published in the American Journal of Speech-Language Pathology describes SSEDs as an important tool for developing and implementing evidence-based practice, precisely because the focus on the individual allows researchers to ask questions that traditional group designs cannot feasibly address. The study is divided into distinct phases, and the participant is observed under one condition per phase – making the comparison between phases the engine of scientific inference.

The three core phases of single subject design

SSD operates through a structured sequence of phases. Each phase must meet certain standards – particularly around data stability – before the researcher moves to the next. This careful progression is what allows the design to support different levels of scientific knowledge: descriptive, correlational, and causal.

Phase A: the baseline

The baseline phase is where everything begins. During Phase A, a baseline is established for the dependent variable – this is the level of responding before any treatment is introduced. Think of it as the control condition. The researcher observes and records the target behavior in its natural state, without any manipulation.

The purpose is straightforward: to establish a clear, stable pattern that can serve as a point of comparison. Typically, the baseline should include at least three data points, and there should be an identifiable pattern. If the baseline is erratic or unstable, it becomes impossible to determine later whether any change was actually caused by the intervention or was simply part of natural fluctuation.

Researchers examine three dimensions of baseline data: level (the overall amount of the behavior), trend (whether it is increasing, decreasing, or stable), and variability (how much it fluctuates from measurement to measurement). The purpose of the baseline phase is to establish the existing levels and patterns of the behavior of interest, thus allowing for future performance predictions under the continued absence of intervention.

This is also where the first level of scientific knowledge is generated: descriptive knowledge. The baseline tells us what the behavior currently looks like, how often it occurs, and in what pattern – nothing more, nothing less.

Phase B: the intervention

Once the baseline data has stabilized, Phase B begins. The researcher introduces the independent variable – the intervention – and continues to measure the dependent variable (the target behavior) across multiple sessions. There may be a period of adjustment during which the behavior of interest becomes more variable and begins to increase or decrease, and the researcher waits until it reaches a new steady state before drawing conclusions.

The intervention phase generates correlational knowledge. By comparing the behavior during Phase B to the behavior during Phase A, the researcher can detect a relationship between the intervention and changes in the target behavior. If anxiety levels drop significantly after a therapeutic technique is introduced, that’s a meaningful correlation. But correlation alone doesn’t prove the intervention caused the change – something else happening at the same time could account for it. That’s exactly why the reversal phase exists.

Just as with the baseline, data stability matters here. The researcher should not move on prematurely. According to research on single-subject experimental designs, the steady-state strategy – allowing behavior to become fairly consistent before shifting conditions – makes any effect of the treatment much easier to detect, because it minimizes noise in the data.

Phase A again: the reversal

The reversal phase – sometimes called the withdrawal phase – is what separates a rigorous single-subject experiment from a simple observation. Here, the intervention is removed, and the researcher returns to the conditions of the original baseline. The purpose of the reversal phase is to determine whether the behavior would have remained unchanged if the intervention had not been introduced. If behavior returns to near-baseline levels after the intervention is withdrawn, this is powerful evidence that it was the intervention – not some outside variable – driving the change.

This is where causal knowledge is established. In an ABA design, there is a baseline condition (A), followed by a treatment condition (B), followed by a return to baseline (A). The logic is that if the dependent variable changes when the treatment is introduced and then reverts when it is removed, the treatment is almost certainly responsible.

Many researchers extend the design further into an ABAB design by reintroducing the intervention after the reversal. ABAB designs have the benefit of an additional demonstration of experimental control with the reimplementation of the intervention. Practically speaking, many clinicians prefer ending on a treatment phase (B) rather than leaving the participant without an effective intervention at the conclusion of the study.

Descriptive, correlational, and causal knowledge: a progression

One of the most important features of single subject design is that it doesn’t just aim for one type of knowledge – it builds through all three levels systematically.

Descriptive knowledge comes from the baseline phase. It tells us how the behavior currently appears without any interference. Correlational knowledge emerges from comparing the baseline to the intervention phase – we can now see a relationship between the treatment and behavior change. Causal knowledge is achieved through the reversal. According to the What Works Clearinghouse, a causal relationship is demonstrated when the data across all phases document at least three demonstrations of an effect at a minimum of three different points in time – something the ABAB design is specifically structured to achieve.

The final assertion that change in behavior was causally associated with the introduction of the intervention is what establishes the scientific credibility of single-subject designs. Without the reversal phase, you can describe and correlate – but you cannot confidently conclude causation.

Why data stability is non-negotiable

Throughout every phase of SSD, researchers must prioritize data stability before progressing. This principle – sometimes called the steady-state strategy – holds that moving to the next phase too early can compromise the entire study. If baseline data is still trending upward, for example, any improvement seen during the intervention phase may simply be a continuation of that pre-existing trend rather than a response to treatment.

Research on single-subject experimental designs identifies three types of evidence for treatment effects that researchers examine when comparing phases: changes in level (the overall magnitude of the behavior), changes in trend (the direction of change over time), and changes in variability (how consistent the data points are). A strong effect may show up in one or all three of these dimensions. For instance, an intervention may dramatically reduce overall levels of a problem behavior, or it may stabilize behavior that was previously highly variable – both are meaningful outcomes.

This reliance on visual inspection of graphed data is a hallmark of SSD. Single-subject research relies heavily on visual inspection – plotting individual participants’ data, carefully examining those data, and making judgments about whether and to what extent the independent variable had an effect on the dependent variable. This is quite different from group research, which relies on inferential statistics drawn from aggregate data.

Limitations of the reversal phase

The reversal design is powerful, but it has clear limits. The most significant constraint is that it only works when a behavior is reversible. If the intervention teaches a new skill – such as reading, communication, or a cognitive strategy – withdrawing the treatment won’t undo the learning. One major limitation of ABA designs is that they are not suitable for a target behavior that cannot be “unlearned.” In these cases, researchers turn to alternative designs such as the multiple baseline design, which introduces the intervention at staggered time points across different behaviors, settings, or individuals.

There are also ethical considerations. Before using a reversal design, researchers must determine that it is safe and ethical to withdraw the intervention, especially when the intervention is effective and necessary for the individual’s wellbeing. For example, if a behavioral intervention is successfully reducing self-injurious behavior, withdrawing it to complete a reversal phase may not be ethically justifiable.

A structured path to understanding individual behavior

The phases of single subject design – baseline, intervention, and reversal – form a deliberate and disciplined path. Each phase answers a different scientific question: What does the behavior look like naturally? Does the intervention correlate with change? And critically, did the intervention actually cause that change? The progression from descriptive to correlational to causal knowledge isn’t accidental – it’s the result of a carefully sequenced methodology that insists on stability at every stage before advancing. For researchers, clinicians, and educators working with individuals rather than populations, SSD offers a rigorous and flexible framework for generating real, actionable evidence about what works.

What do you think? If a reversal phase requires withdrawing a treatment that appears to be helping someone, where should the line be drawn between scientific rigor and ethical responsibility? And how might the requirement for data stability before advancing phases affect the pace and practicality of research in real clinical settings?

How useful was this post?

Click on a star to rate it!

Average rating 0 / 5. Vote count: 0

No votes so far! Be the first to rate this post.

References

PDF 📄

Table of Contents

What is single subject design? 🔗

The three core phases of single subject design 🔗

Phase A: the baseline 🔗

Phase B: the intervention 🔗

Phase A again: the reversal 🔗

Descriptive, correlational, and causal knowledge: a progression 🔗

Why data stability is non-negotiable 🔗

Limitations of the reversal phase 🔗

A structured path to understanding individual behavior 🔗