Writing operational definitions

Every graph you will ever interpret is only as good as the definition behind it. Before anyone counts, times, or samples a behavior, someone has to say exactly what that behavior is.

Why it matters

A vague target like "aggression" or "being off task" invites each observer to use a private rule. One staff member counts a shove in line, another does not; one parent counts a whine as a tantrum, another waits for screaming. The data then reflect the observers as much as the client, and you cannot tell whether a change on the graph came from the intervention or from shifting judgments.

A good operational definition protects three things at once: the validity of your measurement (you are measuring the behavior you care about), the agreement between observers, and your ability to communicate the target to caregivers, teachers, RBTs, and other professionals. Practice items on this task ask you to spot the weakest definition, pick the best revision, or decide between a function-based and a topography-based definition.

Key ideas

• Describe what the person does, not who they are. Labels ("disrespectful"), inner states ("frustrated", "anxious"), and guesses about causes ("to get attention") are not observable. Replace them with actions an observer can see or hear.
• Use the three classic quality checks. A definition should be:
  • Objective: it refers only to observable features of behavior and the environment.
  • Clear: it is unambiguous enough that a new observer could read it, restate it accurately, and use it without extra coaching.
  • Complete: it marks the boundaries of the response class, saying what counts and what does not, so observers do not have to make judgment calls.
• Include examples and non-examples. Near misses (a high five vs. a hit, humming a song in music class vs. vocal stereotypy during work) are where observers disagree most, so name them.
• Say when an instance starts and ends. You need an onset rule to measure latency, and both onset and offset rules to measure duration. They also tell observers when one episode ends and the next begins.
• Pick the right type of definition.
  • A function-based definition groups responses by their common effect on the environment, whatever form they take.
  • A topography-based definition describes the shape or form of the response.
• Test it, then revise it. Have two people observe with the draft definition and compare records. Low agreement often points to a definition problem, not just an observer problem.

Definitions

• Operational definition: a description of a target behavior in observable, measurable terms, written so that different observers identify and record each instance the same way.
• Function-based definition: a definition that includes every response producing the same effect on the environment, such as "any movement that takes the student more than 3 meters from the group without permission".
• Topography-based definition: a definition based on the form of the response, such as the hand position for a sign or the steps of a tennis serve.
• Onset and offset criteria: rules stating when an occurrence begins and when it is over, such as "the episode ends after 30 seconds with no crying or screaming".

Function-based definitions are usually preferred in applied work, because the effect of the behavior is what matters to the person and the people around them, and one definition can cover many forms. A topography-based definition makes more sense when the form itself is the skill (correct signs, safe lifting technique) or when the intended outcome does not reliably follow each correct response (or you cannot observe that outcome), so scoring the outcome alone would mislead you.

Examples from a session

Common mistakes / exam traps

• Building in a cause. "Hits peers to escape work" mixes a hypothesis about function into the definition. Define the hit; test the function separately.
• Relying on observer judgment. Phrases like "as determined by the teacher" or "inappropriately" move the definition back into someone's head.
• Leaving out the offset. Without an end rule, two observers can agree that a tantrum happened and still disagree widely about its duration.
• Blaming observers first. When agreement is low, check the definition before you schedule more training.

Quick check

Related topics

• Behavior, response, and response class (B.1)
• Direct, indirect, and permanent-product measures (C.2)
• Judging the validity and reliability of measurement (C.8)
• Choosing a measure that captures the critical dimension within real-world limits (C.9)
• Writing intervention goals in observable, measurable terms (H.1)