By Matt Harrington, BCBA · Behaviorist Book Club · April 2026 · 12 min read

Reducing Biases in Clinical Judgment with Single-Subject Treatment Design

Overview & Clinical Significance

Clinical judgment is a necessary component of behavior-analytic practice, but it is also susceptible to a range of cognitive biases that can compromise treatment effectiveness and client welfare. This course examines how single-subject treatment designs serve as a systematic safeguard against these biases, drawing upon work by Moran and Tai (2001) that identifies specific biases and describes how experimental methodology can mitigate their influence on clinical decision-making.

Behavior analysts make high-stakes decisions daily: whether to continue or modify an intervention, whether a functional relationship has been established, whether a client is making meaningful progress, and whether treatment goals have been met. Each of these decisions is potentially vulnerable to bias. When biases go unchecked, they can lead to ineffective treatments being continued, effective treatments being prematurely abandoned, or conclusions being drawn that are not supported by the data.

The psychological literature has identified numerous biases that affect professional judgment. Pathology bias leads clinicians to overinterpret behavior as symptomatic. Confirmatory bias leads practitioners to seek and attend to information that supports their existing hypotheses while ignoring contradictory evidence. Hindsight bias creates the illusion that outcomes were predictable after the fact, reducing the motivation to use systematic assessment methods. Misestimation of covariance leads to perceived relationships between variables that do not actually covary.

For behavior analysts, the commitment to single-subject experimental design is both a methodological and an ethical stance. Single-subject designs, including reversal (ABAB), multiple baseline, alternating treatments, and changing criterion designs, provide the structure needed to demonstrate functional relationships between interventions and behavior change. This is not merely an academic exercise but a practical tool for protecting clients from the consequences of biased judgment.

The relevance of this topic extends beyond individual treatment decisions to the broader credibility of behavior analysis as a discipline. If practitioners rely on uncontrolled clinical judgment rather than systematic experimental methods, the field loses its empirical foundation. At a time when behavior analysis faces scrutiny from insurers, regulators, and allied professionals, maintaining methodological rigor is essential for the discipline's continued growth and acceptance.

This course is particularly timely given the rapid growth of the profession and the increasing number of practitioners who may have limited experience with single-subject research methodology. Ensuring that all practitioners understand how single-subject designs protect against bias is critical for maintaining treatment quality across the field.

Background & Context

The study of cognitive biases in professional judgment has a rich history in psychology and medicine. Decision science research has demonstrated repeatedly that even trained professionals are susceptible to systematic errors in judgment when they rely on intuition rather than structured decision-making processes. These findings apply to behavior analysts just as they apply to physicians, psychologists, and other professionals who make complex decisions under conditions of uncertainty.

Pathology bias is the tendency to interpret ambiguous information in pathological terms. In behavior-analytic practice, this might manifest as interpreting normal developmental variation as symptomatic, over-attributing behavior to pathological functions, or setting treatment goals that reflect the clinician's expectations rather than the client's actual needs. This bias is reinforced by the clinical context itself, where practitioners primarily encounter individuals with identified problems, creating a base rate error in which pathology seems more prevalent than it is.

Confirmatory bias is perhaps the most insidious threat to clinical judgment. Once a practitioner forms a hypothesis about the function of a behavior or the effectiveness of an intervention, they tend to seek confirming evidence and discount disconfirming evidence. In the absence of controlled experimental design, confirmatory bias can lead practitioners to maintain ineffective interventions because they selectively attend to instances that appear to support the treatment while ignoring instances that suggest it is not working.

Hindsight bias distorts the practitioner's perception of their own judgment accuracy. After an outcome is known, the practitioner tends to believe they would have predicted it, reducing the perceived need for systematic assessment. This bias undermines the motivation to collect baseline data, implement experimental controls, and rely on visual analysis rather than subjective impression.

Misestimation of covariance leads practitioners to perceive relationships between variables that are not supported by the data. For example, a practitioner might conclude that a particular antecedent reliably evokes problem behavior based on a few salient instances while overlooking the many instances in which the antecedent was present without problem behavior occurring. Without systematic data collection, these illusory correlations can drive treatment decisions.

Moran and Tai (2001) argued that single-subject treatment designs provide a structured methodology for overcoming these biases. By requiring repeated measurement, establishing baseline stability before intervention, and demonstrating experimental control through systematic manipulation of the independent variable, single-subject designs force practitioners to confront the data rather than relying on subjective judgment. The design itself serves as a debiasing tool.

This work fits within a broader tradition in behavior analysis that values methodological rigor not as an end in itself but as a means of protecting client welfare. The founders of applied behavior analysis emphasized the importance of demonstrating experimental control in clinical settings, and this emphasis distinguishes behavior analysis from many other clinical disciplines.

Clinical Implications

The clinical implications of understanding and mitigating judgment biases are extensive and touch every aspect of behavior-analytic practice. Implementing single-subject design principles in clinical settings is not just about conducting formal research but about adopting a systematic approach to treatment evaluation that protects against the biases that inevitably affect human judgment.

The most fundamental implication is the importance of baseline data collection. Without a stable baseline, the practitioner has no reference point against which to evaluate the effects of intervention. Confirmatory bias can easily lead a practitioner to attribute any improvement to the intervention when no baseline exists. By collecting baseline data and establishing stability before implementing treatment, the practitioner creates the conditions for an objective evaluation of treatment effects.

Repeated measurement is another critical safeguard. Biases are most influential when practitioners rely on a few data points or on their overall impression of the client's progress. Regular, systematic data collection across time generates a visual record that makes it difficult to sustain biased interpretations. When the data clearly show no change following intervention implementation, confirmatory bias has less room to operate.

Experimental control, whether through reversal, multiple baseline, or other design elements, provides the strongest protection against bias. By demonstrating that behavior changes when and only when the intervention is applied, the practitioner can rule out alternative explanations for behavior change, including maturation, regression to the mean, and coincidental environmental changes. In clinical settings, full experimental control may not always be feasible, but approximations of experimental control (such as staggering intervention implementation across behaviors or settings) still provide meaningful evidence.

For practitioners conducting functional behavior assessments, the biases described in this course are particularly relevant. The hypothesis-driven nature of FBA makes it vulnerable to confirmatory bias. A practitioner who hypothesizes that a behavior is maintained by escape may selectively attend to escape sequences and overlook instances of attention-maintained behavior. Using structured assessment procedures, including both indirect assessment and direct observation with standardized recording, helps mitigate this bias.

In treatment planning, pathology bias can lead to overly ambitious or inappropriate treatment goals. By grounding treatment planning in objective assessment data rather than clinical impression, practitioners can set goals that reflect the client's actual needs and abilities. Regular review of progress data guards against the tendency to continue pursuing goals that the data suggest are not being achieved.

For supervisors, this course highlights the importance of teaching supervisees to rely on data rather than impression. Many novice practitioners have not yet developed the skills to conduct visual analysis of single-subject data, and they may default to subjective judgment. Supervisors can model data-based decision-making and explicitly discuss the biases that threaten clinical judgment, creating awareness that is the first step toward mitigation.

The implications also extend to clinical documentation and reporting. When practitioners document their treatment decisions with reference to specific data patterns rather than general impressions, they create an accountability structure that resists bias. This documentation serves both the practitioner and the client by ensuring that treatment decisions can be reviewed and evaluated by others.

Ethical Considerations

The relationship between cognitive biases and ethical practice is direct and consequential. The BACB Ethics Code for Behavior Analysts (2022) contains multiple provisions that implicitly or explicitly require practitioners to guard against biased judgment, and understanding these connections strengthens the ethical foundation of practice.

Code 2.01 (Providing Effective Treatment) is perhaps the most directly relevant. Effective treatment requires accurate assessment, appropriate intervention selection, and ongoing evaluation of outcomes. Each of these processes is vulnerable to the biases described in this course. A practitioner who selects an intervention based on confirmatory bias rather than data, or who continues an ineffective treatment because of hindsight bias, is failing to provide effective treatment. Single-subject design principles provide the safeguards needed to fulfill this ethical obligation.

Code 2.18 (Continual Evaluation of the Behavior-Change Program) explicitly requires behavior analysts to use data to evaluate treatment effectiveness and to modify programs when data indicate that desired outcomes are not being achieved. This code is a direct mandate against the biases that lead practitioners to continue ineffective treatments. Without systematic data collection and analysis, practitioners cannot fulfill this requirement because their judgment will be distorted by the very biases this course addresses.

Code 3.01 (Behavior-Analytic Assessment) requires that assessments be conducted in a manner consistent with the best available scientific evidence. Assessments that are influenced by pathology bias or confirmatory bias are not consistent with this standard. The use of structured assessment procedures, including standardized functional assessment protocols and systematic data collection, helps ensure that assessment findings reflect the actual situation rather than the practitioner's preconceptions.

Code 2.13 (Selecting, Designing, and Implementing Assessments) requires behavior analysts to select assessments that are appropriate to the question being asked. When cognitive biases lead practitioners to rely on informal observation or unstructured clinical impression rather than appropriate assessment methods, they are not meeting this standard. Awareness of the biases that compromise informal judgment provides the motivation to use more structured approaches.

Code 1.10 (Awareness of Personal Biases and Challenges) directly addresses the issue at the heart of this course. Behavior analysts are required to be aware of how their personal biases may affect their professional work. The biases described by Moran and Tai (2001) are not personal in the sense of reflecting individual prejudice but are universal features of human cognition. Awareness of these universal biases and the use of systematic methods to mitigate them is consistent with the spirit of this code.

Code 4.06 (Providing Feedback to Supervisees) is relevant because supervisors have a responsibility to help supervisees identify and address their own biases. This includes modeling data-based decision-making, pointing out instances where biased judgment may be influencing treatment decisions, and teaching the methodological skills needed to implement single-subject designs in clinical practice.

There is also an ethical dimension related to intellectual honesty. A practitioner who ignores data that contradicts their hypothesis or who claims treatment effectiveness without adequate evidence is engaging in a form of professional dishonesty that undermines trust in the profession and harms clients.

Assessment & Decision-Making

Integrating bias mitigation into clinical assessment and decision-making requires a structured approach that goes beyond simply being aware that biases exist. Practitioners need concrete strategies for identifying when bias may be influencing their judgment and systematic methods for correcting course.

The first strategy is to formalize the assessment process. Rather than relying on informal observation and clinical impression, practitioners should use standardized assessment protocols that specify what data to collect, how to collect it, and how to interpret it. Functional behavior assessment, for example, should follow a structured protocol that includes indirect assessment (interviews, rating scales), descriptive assessment (direct observation with standardized recording), and, when appropriate and ethical, functional analysis. Each step provides data that can be evaluated against the practitioner's initial hypotheses.

The second strategy is to actively seek disconfirming evidence. This is the most direct antidote to confirmatory bias. Before concluding that a particular function maintains a behavior, the practitioner should deliberately look for evidence that contradicts this hypothesis. Are there conditions under which the hypothesized reinforcer is present but the behavior does not occur? Are there conditions under which the behavior occurs in the absence of the hypothesized reinforcer? These questions force the practitioner to confront evidence that might otherwise be overlooked.

The third strategy is to use decision rules based on visual analysis criteria. Rather than making subjective judgments about whether data show a meaningful change, practitioners should apply established visual analysis criteria: level, trend, variability, immediacy of effect, overlap between phases, and consistency of data patterns across similar phases. These criteria provide an objective framework for evaluating treatment effects that is less susceptible to bias than gestalt impression.

The fourth strategy is to involve others in the decision-making process. Peer consultation, team-based review of data, and supervisor oversight all provide external perspectives that can identify biases the individual practitioner may not recognize. When multiple clinicians independently review the same data and reach different conclusions, this disagreement is a signal that bias may be at play and that additional data or analysis are needed.

The fifth strategy is to document decision-making rationale in real time rather than retrospectively. When practitioners record their hypotheses, predictions, and reasoning before the data are available, they create a record that can be compared against actual outcomes. This guards against hindsight bias by making it impossible to retroactively claim that the outcome was expected.

A practical decision-making framework for treatment evaluation might include the following steps: Collect baseline data until stability is achieved. Implement the intervention and continue data collection. At predetermined intervals, conduct visual analysis using established criteria. Compare the data pattern to the predicted pattern based on the hypothesis. If the data do not match the prediction, generate alternative hypotheses and test them. Document all decision points, including the data that supported each decision.

This framework does not eliminate the need for clinical judgment but constrains it within a structure that minimizes the influence of bias. The practitioner's judgment is still needed to interpret data patterns, select design elements, and make practical decisions about implementation. But that judgment is informed and disciplined by the systematic methodology of single-subject design.

What This Means for Your Practice

The most important takeaway from this course is that you are not immune to cognitive biases, regardless of your training, experience, or intelligence. These biases are features of human cognition, not deficits of individual practitioners. The difference between practitioners who make good decisions and those who do not is often not the quality of their judgment but the quality of the systems they use to constrain that judgment.

Practically, this means committing to single-subject design principles in your clinical work, even when it is not required for research purposes. Collecting baseline data before implementing interventions. Using repeated measurement to track progress over time. Implementing at least approximate experimental control when possible. Conducting visual analysis using established criteria rather than relying on your overall impression.

It also means building habits that guard against specific biases. Before concluding that an intervention is working, ask yourself what alternative explanations exist for the observed change. Before concluding that a behavior has a particular function, ask what evidence would disconfirm that hypothesis and then look for that evidence. Before attributing a client's difficulties to pathology, consider whether the behavior might be adaptive or normative in context.

For supervisors, this course highlights the importance of teaching supervisees about cognitive biases and the role of experimental methodology in mitigating them. This is not a one-time lesson but an ongoing conversation that should be woven into supervision throughout a supervisee's training.

The bottom line is that single-subject design is not just a research methodology. It is a clinical tool that protects your clients from the consequences of biased judgment. Using it is not optional for ethical, effective practice.