Graphing and Monitoring Behavior Change: Single-Subject Design

Introduction to Single-Subject Design

• Definition of Single-Subject Design (SSD):

  • Also referred to as within-subjects design or single-case design.
  • Defined as a wide variety of research designs that utilize experimental reasoning—specifically baseline logic—to demonstrate the effects of an independent variable on the behavior of individual subjects (Cooper, Heron, & Heward, 2019).
  • Entails the systematic change, control, and measurement of the environment and scientific methods.
  • A defining characteristic is that each individual subject acts as their own control.

• Practical Contexts and Applications:

  • Determining whether a reading intervention improved a second-grade class's reading performance.
  • Evaluating if a student engages in on-task behaviors more frequently as a direct result of a mystery motivator intervention.
  • Assessing if employees sell more products following the implementation of a new incentive package.

Key Variables in Single-Subject Research

• Independent Variable (IV):

  • The environmental event, intervention, or treatment that is manipulated by the researcher or practitioner.
  • Designed to directly influence or alter the dependent variable.
  • Common intervention examples include token economies, Repeated Reading, running schedules, time-out procedures, or medication regimens.

• Dependent Variable (DV):

  • The specific target behavior that is measured to assess change.
  • Changes in response to the systematic introduction, alteration, or removal of the independent variable.
  • Common target behavior examples include physical aggression, reading fluency, body weight loss, or classroom talk-outs.

• Temporal Tracking of Variables:

  • Systematic and repeated measurement of the dependent variable occurs across three distinct timing phases:
    • Before intervention: Baseline phase ($A$).
    • During intervention: Treatment phase ($B$).
    • After intervention: Follow-up phase.

Purpose and General Characteristics of SSD

• Primary Utility in Applied Research:

  • Determines treatment utility by evaluating whether an intervention truly produces meaningful behavior change.
  • Provides actionable, immediate feedback regarding individual intervention efficacy.
  • Maintains high sensitivity to subtle behavioral changes over time.

• Core Methodological Characteristics:

  • Repeated Observation: Measurement of the same behavior continuously over time under specific, controlled conditions to establish behavioral consistency.
  • Standardized Implementation: Execution of procedures by trained observers using consistent observation techniques to demonstrate experimental control and ensure reliable comparisons across time (attaining high Inter-Observer Agreement).
  • Detailed Description of Conditions: Comprehensive operational definitions of all environmental conditions to strengthen internal and external validity.
  • Baseline Condition ($A$): Recording of target behavior in the complete absence of the intervention, maintained long enough to establish data stability.
  • Treatment Condition ($B$): Recording of target behavior during the direct implementation of experimental manipulations, maintained long enough to establish data stability.
  • Cost Effectiveness & Flexibility: Highly adaptable to practice settings, inexpensive to conduct, and effective for testing clinical hypotheses, monitoring individual progress, and studying low-incidence populations or behaviors prior to launching large-scale trials.

Comparing Single-Subject Design and Group Design

• Shared Conceptual Rationale:

  • Both design frameworks share the foundational goal of evaluating the functional effects of independent variables on dependent variables.

• Group Design Characteristics:

  • Requires a large sample of randomly assigned participants split into distinct groups.
  • Compares average outcomes between separate groups (e.g., control group vs. experimental group) exposed to different conditions.
  • Evaluates findings through descriptive and inferential statistical tests.

• Single-Subject Design Characteristics:
  • Allows rigorous experimental analysis with individual participants over time across different conditions.
  • Relies primarily on detailed visual analysis of graphed data rather than inferential statistics.
  • Complements group research by highlighting individual participant variations that are critical for tailoring clinical protocols.

Common Misconceptions About Single-Subject Design

• Misconception 1: SSD and Case Studies are Identical:
  • Correction: Case studies rely on unsystematic observations and subjective measures, whereas SSDs employ rigorous experimental control and systematic measurement.
• Misconception 2: Limited Scope:
  • Correction: SSDs are not restricted solely to traditional behavior modification techniques; they are used across diverse scientific disciplines.
• Misconception 3: Weak Causality:
  • Correction: SSDs permit strong causal statements through systematic baseline logic and experimental manipulation.
• Misconception 4: Lack of Generalizability:
  • Correction: SSD conclusions achieve external validity and population generalizability through systematic study replication.
• Misconception 5: Methodological Flawlessness:
  • Correction: SSD is a complementary research approach rather than a flawless substitute for group designs.

Core Research Questions Addressed by SSD

• Treatment Efficacy:
  • Does the treatment work better than nothing at all?
• Comparative Efficacy:
  • Does one treatment work better than another alternative treatment?
• Component Analysis:
  • Which specific components of a multi-part treatment program are directly responsible for behavior change?
• Parametric Analysis:
  • What is the optimal dosage or level of intervention required for behavior change?

Basic Treatment Manipulation and Graph Construction

• Standard Implementation Protocol:

  1. Operationally define target behaviors ($DV$) and treatment conditions ($IV$).
  2. Establish a stable behavioral baseline ($Phase\,A$).
  3. Introduce the treatment/intervention ($Phase\,B$).
  4. Return to baseline ($Phase\,A$) by withdrawing treatment.
  5. Maintain continuous, repeated recording of behavior across all phases, securing a minimum of $3$ data points per phase.

• Data Terminology:

  • Datum: A single, individual data point representing one distinct measurement of participant behavior.
  • Data: Multiple sequential measurements of participant behavior gathered over time, forming a series of data points.

• Essential Components of a Single-Subject Graph:

  • Vertical axis (Y\n\n-axis) and Horizontal axis (X\n\n-axis).
  • Explicit labels for both axes (e.g., "Frequency of Talking Out Behavior in Class" on $Y$, "Sessions" on $X$).
  • Numerical scales for both axes.
  • Individual data points representing distinct measurements.
  • Vertical phase lines indicating transitions between baseline and intervention conditions.
  • Phase labels identifying specific experimental conditions (e.g., "Baseline", "Intervention").

Visual Analysis of Data

• Primary Evaluation Method:

  • Visual inspection of graphed data serves as the primary standard for evaluating changes in single-subject methodology, though statistical analyses are occasionally integrated.

• Core Dimensions of Visual Analysis:

  • Level: The average or general magnitude/value of the behavioral measure within a phase.
  • Trend: The overall direction of behavioral change from the beginning to the end of a series of data points (e.g., ascending, descending, zero trend).
  • Variability: The extent of bounce or spread of individual data points around the mean level and trend line.

• Visual Patterns in Graphed Data:

  • Sudden Change in Level: Immediate vertical shift in data values between adjacent phases.
  • Gradual Change in Level: Slow, incremental vertical shift in data values across sessions.
  • Sudden Change in Level and Trend: Immediate shift in average value combined with a distinct change in data trajectory.
  • Sudden Change in Variability: Immediate change in the fluctuation range or stability of data points upon phase transition.
  • Delayed Change in Variability: A lag occurs before data points display altered stability following condition change.
  • Change in Level without Change in Trend: Data shifts vertically in magnitude while maintaining its ongoing direction.
  • Gradual Change in Level and Trend: Progressive adjustment in both data value average and trajectory.
  • Delayed Change in Level and Trend: Lagged transition in both vertical magnitude and trajectory following phase alteration.
  • Sudden but Inconsistent Change in Level: An immediate vertical shift that fails to maintain consistent values across sessions.
  • Delayed Change in Level: A lagged vertical shift in overall data magnitude following condition implementation.
  • Gradual Change in Level, Trend, and Variability: Simultaneous progressive transformation across all three visual dimensions.

Single-Subject Research Design Variations

• 1. Simple Phase Change ($AB$ Design):
  • Consists of one baseline phase ($A$) and one treatment phase ($B$).
  • Categorized as a quasi-experimental design rather than a true research design.
  • Exhibits severe threats to internal and external validity due to a lack of planned replication.

*   *Graph Example Data:* Baseline Mean = $12.8$ talking-out events; Intervention Mean = $3.6$ talking-out events.

• 2. Simple Phase Change with Follow-up ($AB/\text{Follow-up}$ Design):

  • Includes baseline ($A$), treatment ($B$), and a dedicated follow-up phase after treatment termination or clinician withdrawal.
  • Primary objective is assessing the long-term maintenance of behavior change.

• 3. Return to Baseline ($ABA$ Design):

  • Includes baseline ($A$), treatment ($B$), and a withdrawal phase returning to baseline ($A$).
  • The second baseline phase is technically an experimental withdrawal rather than a pure baseline due to prior treatment exposure.

• 4. Withdrawal Design ($ABAB$ Design):

  • Sequences through initial baseline ($A$), initial treatment ($B$), withdrawal/baseline ($A$), and reintroduction of treatment ($B$).
  • Demonstrates high internal validity by confirming that behavior changes systematically depend on treatment presence.

• 5. Combined Simple Phase Change ($ABAC$ Design):
  • Sequences baseline ($A$), Treatment 1 ($B$), withdrawal/baseline ($A$), and Treatment 2 ($C$).
  • Compares each unique treatment against baseline (no-treatment) and allows direct comparison between the two interventions ($B$ vs. $C$).
  • Examples include comparing Attention ($B$) vs. Escape ($C$) functions, or Time-out ($B$) vs. Medication ($C$) for ADHD management.

• 6. Parametric Design ($BB'B''$ Design):
  • Features a baseline ($A$) phase followed by continuous treatment phases where specific parameters/dosages of the same intervention are systematically manipulated.
  • Examples include evaluating Time-out durations of $1\,\text{minute}$ ($B$), $3\,\text{minutes}$ ($B'$), and $5\,\text{minutes}$ ($B''$); or Ritalin dosages of $5\,\text{mg}$ ($B$), $10\,\text{mg}$ ($B'$), and $15\,\text{mg}$ ($B''$).

• 7. Changing Criterion Element Design:
  • Involves an initial baseline ($A$) phase followed by a treatment ($B$) phase that contains step-wise, progressive performance criteria.
  • Systematically increments target goals across criteria steps (e.g., Criterion 1 = $8\,\text{points}$, Criterion 2 = $12\,\text{points}$, Criterion 3 = $14\,\text{points}$).

• 8. Alternating-Treatments Design (ATD):
  • Rapidly alternates two or more distinct conditions within the exact same experimental phase.
  • Evaluates efficacy by tracking data path divergence between treatment conditions (e.g., Repeated Readings vs. Listening Passage Previewing).

• 9. Multiple Baseline Design (MBL):
  • Features staggered implementation of intervention across separate baseline tracks to control for threats to internal validity without requiring treatment withdrawal.
  • MBL Across Participants: Baseline and treatment applied to the same behavior across two or more individuals.
  • MBL Across Behaviors: Baseline and treatment applied to two or more distinct behaviors in a single subject.
  • MBL Across Settings: Baseline and treatment applied to the same behavior in a single subject across two or more distinct settings.

Reliability, Validity, and Treatment Integrity

• Reliability (Inter-Observer Agreement / IOA):

  • Measures whether two independent observers simultaneously monitoring the target behavior agree on its occurrence or nonoccurrence.
  • Standard Criteria: Evaluated during $25\%$ to $33\%$ of all observational sessions, requiring an agreement threshold of $80\%$ or higher.

• Validity Metrics and Threats:

  • Internal Validity: The degree to which behavior changes can be attributed confidently to the independent variable rather than confounding variables.
    • Threats: History, maturation, testing, instrumentation, regression to the mean, selection biases, participant attrition, and diffusion of treatment.
  • External Validity: The extent to which empirical findings generalize to other subjects, settings, or target behaviors.
    • Threats: Generality across subjects, settings, behaviors, or behavior change agents; reactive experimental arrangements; reactive assessment; pretest sensitization; and multiple treatment interference.
  • Mitigation Strategy: Systematic experimental replication is the primary mechanism to eliminate validity threats.

• Treatment Integrity (Procedural Fidelity):

  • Definition: Verifies whether the intervention is implemented precisely as designed.
  • Assessment Methods: Systematic implementation checklists and direct behavioral observations.
  • Standard Criteria: Evaluated during $25\%$ to $33\%$ of intervention sessions, expecting an integrity level of $80\%$ or higher.

Standards for Evaluating Intervention Effectiveness

• Critical Evaluation Questions:

  • Were target behaviors ($DV$) and intervention conditions ($IV$) clearly and operationally defined?
  • Is there distinct visual separation in data between baseline and intervention conditions (evaluating level, variability, trend, and divergence)?
  • Did researchers maintain standard, uniform graphical scaling across all figures?
  • Were follow-up data systematically collected to evaluate behavioral maintenance?
  • Were formal empirical reports provided for inter-observer agreement ($IOA$) and treatment integrity?

• Key Indicators of Intervention Efficacy:

  • Sufficient Data Points: A minimum threshold of $3$ to $5$ stable baseline data points before initiating phase changes.
  • Immediacy of Effect: An immediate behavioral shift following intervention introduction indicates a stronger causal relationship.
  • Data Overlap: Minimal data point overlap between adjacent phases reflects higher intervention potency.
  • Replication of Effect: Demonstration of consistent behavioral shifts upon repeating phases or executing withdrawal-reintroduction sequences.