Goals of Behavior Analysis as a Science

Why "Goals of a Science" Is Domain A

The BACB Test Content Outline places the goals of behavior analysis as a science in Domain A (Behaviorism and Philosophical Foundations), weighted about 5% - roughly 8 to 9 of the 185 questions on the exam. The weight is small, but the logic shows up everywhere: measurement (Domain C), experimental design (Domain D), and intervention selection (Domain H) all rest on it.

A science is a systematic approach to understanding phenomena that relies on objective observation and experimentation. Behavior analysis qualifies because it studies lawful relations between behavior and environmental events rather than relying on opinion, tradition, or authority. Cooper, Heron, and Heward identify three goals of any science, and behavior analysis adds a fourth aspiration once those are met.

The three classic goals are description, prediction, and control. Many textbooks add understanding as the product of control: when you can reliably control a phenomenon and rule out competing explanations, you understand it. The exam rarely tests "understanding" directly, so anchor on the three core goals and the order in which they build.

The Three Goals, In Order

The goals are cumulative - each later goal subsumes the earlier ones. You cannot predict what you have not described, and you cannot demonstrate control without first being able to predict.

Description is the foundation. You operationally define the behavior, choose a measurement dimension (frequency, duration, latency, IRT), and record what happens. Description alone makes no causal claim.

Prediction rests on a correlation: when two events covary repeatedly, observing one lets you anticipate the other. Crucially, correlation is not causation. A baseline that shows tantrums rise every afternoon predicts afternoon tantrums - but does not tell you why, and does not justify saying the afternoon causes them.

Control is the highest goal and the only one that establishes cause. You demonstrate a functional relation when you systematically manipulate an independent variable (IV) and the dependent variable (behavior) changes reliably, while other plausible explanations are held constant or ruled out through replication. This is influence over behavior, achieved by arranging environmental conditions - not domination of a person.

How Items Hide the Goal (and the "Control Is Not Coercion" Trap)

The most common Domain A error is treating control as if it meant making someone do something against their will. On the exam, control = demonstrated functional influence over behavior. The BCBA controls variables; the data then show those variables are responsible for the change.

A second high-frequency trap is mistaking prediction for control. If a stem describes only an observed correlation (no manipulation), the science has reached prediction, not control. Look for the verb: "we noticed behavior tends to follow X" = prediction; "we introduced/removed X and behavior changed, then we replicated it" = control.

Use this decision aid to grade options quickly:

• If the option labels the person ("he's just defiant"), it is not scientific - reject it.
• If it relies on staff opinion alone, it is not empirical.
• If it asks for more precise measurement before interpreting, it usually reflects description and is often the strongest first step.
• If it claims causation from a single observation, it overreaches (no replication, no comparison condition).
• If it ties behavior change to a systematic, replicated manipulation, it reflects control.

A strong scientific answer almost always moves toward better data and reserves causal language for designs that can support it.

Worked Example: Sorting the Three Goals

Scenario. A team reports that a client, Maya, "calls out for attention" during independent seatwork. They want to remove attention immediately because "that's obviously why she does it."

Description step. The BCBA operationally defines calling out ("any vocalization above conversational volume not directed at the assigned task") and records its frequency across the period. No explanation yet - just measured facts.

Prediction step. Data across a week show calling out reliably covaries with periods when the teacher is helping other students. That correlation lets the BCBA anticipate high rates when the teacher is busy. It does not prove attention maintains the behavior - low task interest or task difficulty could covary with the same periods.

Control step. The BCBA runs brief test conditions, alternately delivering and withholding attention contingent on calling out and replicating the effect. When calling out rises only in the attention condition and the pattern repeats, a functional relation is demonstrated. Now - and only now - is "attention-maintained" a defensible causal claim.

Notice the discipline: the team's label ("obviously attention") is a hypothesis, not a finding. Description and prediction set it up; only control confirms it. This sequence protects the client from an intervention aimed at the wrong function and keeps decisions tethered to evidence rather than confident-sounding assertions.