The Foundation of Data-Based Decision Making
One of the defining features of applied behavior analysis is its commitment to using data to guide every clinical decision. Rather than relying on intuition, anecdotal reports, or predetermined timelines, BCBAs collect ongoing data and use visual analysis to determine whether an intervention is working. Knowing when and how to make these decisions is a core competency tested on the BCBA exam.
Data-based decision making involves collecting data consistently, graphing the data for visual analysis, comparing current performance to baseline levels and intervention goals, and using decision rules to determine the appropriate course of action.
Visual Analysis: The Primary Decision-Making Tool
BCBAs primarily rely on visual analysis of graphed data rather than statistical tests. Visual analysis involves examining several features of the data:
Level
Level refers to the average value of the data within a phase. Comparing the level across baseline and intervention phases tells you whether the intervention has changed the overall magnitude of the behavior. A clear change in level from baseline to intervention suggests the intervention is having an effect.
Trend
Trend describes the direction and rate of change within a phase. An increasing trend in a skill acquisition program indicates progress. A decreasing trend in a behavior reduction program indicates the intervention is working. A flat or decelerating trend in a skill acquisition program may signal the need for a change.
Variability
Variability is the degree to which data points fluctuate around the level and trend lines. Low variability means the data are consistent and predictions are reliable. High variability makes it harder to draw conclusions and may indicate that uncontrolled variables are influencing the behavior.
Overlap
Overlap refers to how much the data in adjacent phases share common values. Low overlap between baseline and intervention phases indicates a strong treatment effect. High overlap suggests the intervention may not be producing meaningful change.
Establishing Decision Rules Before You Start
Best practice involves setting explicit decision rules before an intervention begins. Decision rules specify exactly what data patterns will trigger specific actions. For example:
- "If the learner scores below 60% correct for three consecutive sessions, modify the teaching procedure"
- "If the behavior remains at or above baseline levels after 10 sessions of intervention, conduct a new functional assessment"
- "If the learner reaches 90% accuracy across three consecutive sessions, begin generalization probes"
- "If data show an increasing trend in the target behavior over five sessions despite intervention, consult with the treatment team"
Having these rules in place before data collection begins removes subjectivity from the decision-making process and ensures timely responses to data patterns.
When to Continue the Current Intervention
Continue the intervention as designed when the data show a clear improving trend in the desired direction, when the data are moving toward the established goal at an acceptable rate, and when variability is decreasing or remaining stable. Even if progress seems slow, a consistent improving trend usually warrants continuing the current approach before making changes.
When to Modify the Intervention
Modification is warranted when the data show no improvement after a reasonable number of sessions, when progress has plateaued well below the mastery criterion, when data are highly variable with no clear trend, or when the rate of progress is too slow to meet the learner's goals within a reasonable timeframe. Modifications might include changing the reinforcement schedule, adding or adjusting prompts, simplifying the task, changing the instructional format, or addressing potential motivating operations.
When to Discontinue and Reassess
Sometimes the data indicate that the entire intervention approach needs to be reconsidered. This is the case when the behavior is worsening despite intervention, when multiple modifications have failed to produce improvement, when new information suggests the original assessment was incomplete, or when the function of the behavior may have changed. In these situations, returning to the assessment phase is the most responsible course of action.
Common Exam Scenarios
BCBA exam questions about data-based decision making often present a graph and ask what the practitioner should do next. Practice reading graphs quickly and identifying level, trend, variability, and overlap. Questions may also describe a data pattern in words and ask you to choose the best course of action. Remember that the correct answer is always supported by the data, never by convenience, assumption, or preference.