Why Graphing Matters in ABA
Graphing is not merely a way to display data; it is a fundamental tool for clinical decision-making in applied behavior analysis. Well-constructed graphs allow behavior analysts to identify patterns, evaluate treatment effects, communicate results to stakeholders, and make data-based decisions about intervention modifications. The BCBA exam tests your knowledge of graphing conventions, and practicing these skills will serve you throughout your career.
Behavioral data are almost always displayed in time-series graphs, where the x-axis represents time (sessions, days, or trials) and the y-axis represents the level of behavior. This format allows for visual analysis of trends, level changes, and variability across conditions.
Types of Graphs in ABA
Line Graphs
Line graphs are the most commonly used graphs in ABA. They plot data points connected by lines, with each point representing a single measurement occasion (usually a session or day). Line graphs are ideal for showing changes in behavior over time and are the standard format for single-subject research.
When multiple data series appear on the same graph (such as in an alternating treatments design), each series should use a distinct line style, marker shape, or both. Data paths should never be connected across phase change lines.
Bar Graphs
Bar graphs are used less frequently in ABA but are appropriate for displaying summary data, such as the mean of each condition or the percentage of trials correct across different skill domains. They are useful for comparing discrete categories rather than showing change over time.
Bar graphs are sometimes used to display results of preference assessments or to compare performance across multiple targets at a single point in time.
Cumulative Records
Cumulative records plot the running total of responses over time. Each data point represents the sum of all responses up to that point. The slope of the line indicates the rate of responding: a steeper slope means a higher rate, and a flat line means no responding. Cumulative records never decrease because they represent accumulated totals.
Cumulative records are particularly useful for displaying rate of acquisition, monitoring fluency building, and comparing response rates across conditions.
Essential Graph Components
Axes and Labels
The x-axis (horizontal) should be labeled with the unit of time or measurement occasion (for example, "Sessions," "Days," or "Trials"). The y-axis (vertical) should be labeled with the dependent variable and its unit of measurement (for example, "Percentage of Intervals with On-Task Behavior" or "Rate of Correct Responses per Minute").
Both axes should include tick marks and numerical values at regular intervals. The y-axis should begin at zero (or have a clear break if starting at another value) and extend slightly beyond the highest data point. The x-axis should extend to accommodate all planned data collection.
Phase Change Lines
Vertical dashed or solid lines separate different phases or conditions on the graph. These lines indicate when the independent variable was introduced, withdrawn, or changed. Data paths should not be connected across phase change lines because this could imply a continuous trend through a condition change.
Condition Labels
Each phase should be clearly labeled at the top of the graph. Common labels include "Baseline," "Treatment," "Intervention," "Follow-up," or the specific name of the condition (such as "DRA" or "Extinction"). Labels should be centered above the relevant phase.
Data Points and Data Paths
Individual data points should be clearly visible, using filled or open circles, squares, triangles, or other distinct markers. Connected data points within a phase form the data path. Different conditions or behaviors should use different marker styles for clear differentiation.
Common Graphing Mistakes
- Connecting data across phase change lines: This is one of the most common errors. Data paths should break at every phase change line to avoid implying continuity across conditions.
- Inappropriate y-axis scaling: If the y-axis range is too large, meaningful changes in behavior may appear trivially small. If too small, minor fluctuations may appear dramatically large. Scale the y-axis to represent the data accurately.
- Missing axis labels: Every graph must have clearly labeled axes. Omitting labels makes the graph uninterpretable.
- Inconsistent data point spacing: If sessions are not conducted on consecutive days, the spacing between data points should reflect the actual passage of time. Gaps in data collection should be shown as breaks in the data path.
- Cluttered graphs: Plotting too many data series on a single graph can make it impossible to read. If more than three or four data series are needed, consider using multiple panels (stacked graphs).
- No zero line on the y-axis: The y-axis should typically begin at zero to avoid distorting the visual representation of the data.
Multi-Panel Graphs
Multiple baseline designs and other complex designs are typically displayed using multi-panel (stacked) graphs. Each panel shares the same x-axis but has its own y-axis. The panels are aligned vertically so that visual comparison across tiers is straightforward.
Each panel should be clearly labeled to indicate which baseline, behavior, participant, or setting it represents. Phase change lines should extend across all panels when the same condition change applies to multiple tiers.
Graphing for Stakeholders
While behavioral graphs follow specific conventions for research and professional communication, graphs shared with parents, teachers, or other non-behavioral stakeholders may need modifications for clarity. Consider adding brief annotations to explain notable changes, using color coding to distinguish conditions, and providing a legend or key for multiple data series. The fundamental accuracy and honesty of the graph must be maintained regardless of the audience.
BCBA Exam Preparation
For the exam, you should be able to identify correct and incorrect graphing practices, interpret graphed data using visual analysis, and recognize different graph types and their appropriate uses. Practice creating and reading graphs regularly so that these skills become second nature.