BCBA common mistakes
BCBA · Understand the traps and how to avoid them.
On items asking why a behavior occurred, selecting the option that names a private state or trait ('he was frustrated,' 'she lacks motivation,' 'he enjoys attention') instead of an option that names an observable antecedent or consequence.
Everyday mentalistic language reads as a complete and even compassionate explanation, so those options feel correct unless you deliberately filter for observable, environmental variables.
What to do instead
Before looking at the answer options, write down the observable antecedent and consequence in the stem. Then eliminate any option in which a private event is offered as the cause of behavior rather than as something to be explained. Private events are behavior, governed by the same laws; they do not initiate action.
Calling a contingency 'negative reinforcement' simply because a stimulus was removed after the response, and calling a contingency positive merely because something was added.
Candidates memorize the shorthand 'positive = add, negative = remove' and skip the half of the term that refers to the effect on future frequency of the behavior.
What to do instead
Use two scratch lines before answering: first ask did the behavior increase or decrease (reinforcement vs. punishment), then ask was a stimulus added or removed after the response (positive vs. negative). Name the contingency only from those two answers, then find the single option that matches both.
Choosing a type of interobserver agreement statistic by which option sounds most rigorous, rather than by the data type in the stem - e.g., selecting exact count IOA for duration data, or total count IOA for interval data.
The four or five IOA statistics are memorized as an undifferentiated list instead of being tied to the data sheet that produces them.
What to do instead
Use a decision rule driven by how the data were recorded: frequency/count only → total count or exact count per interval; duration → total duration IOA; partial- or whole-interval → interval-by-interval; low-rate behavior with sparse occurrence → scored-interval; high-rate behavior → unscored-interval. Also remember IOA indexes agreement between observers, not the accuracy of either observer.
Judging a condition by its last data point or by comparing condition means, while ignoring within-condition trend and variability.
The final point is the most visually salient feature of a graph, and means are easy to compute, so both crowd out a structural description of the data.
What to do instead
Describe each condition in three words before comparing anything: level, trend, variability. Compare like with like - the last three to five baseline sessions against the first three to five treatment sessions, and the full condition against the full baseline - and state whether a change occurred before evaluating whether the intervention was responsible for it.
Endorsing a design option that does not control the confound described in the stem, e.g., choosing a reversal design for a behavior that cannot ethically or practically be reversed, or a multiple baseline across behaviors that are not independent.
Design names are memorized as templates, so candidates match the word 'baseline' or 'withdrawal' in the stem to the same word in an option without checking what the design actually rules out or what it assumes.
What to do instead
State the alternative explanation in the stem first. Then for each candidate design ask two questions: what competing explanation does this design rule out, and what assumption must hold? Multiple baseline requires independent tiers, a stable baseline, and staggered introduction with no change in untreated tiers; reversal requires reversibility and ethical acceptability; changing criterion requires stepwise, attainable criteria.
On ethics items, defaulting to the most drastic or self-protective option - refusing or terminating services, breaching confidentiality, or reporting to the BACB - when the Code requires a less intrusive, collaborative step first.
Candidates equate 'ethical' with 'cautious,' and distractors are written to sound principled, so the most severe action reads as the safest answer.
What to do instead
Rank the options in this order: act in the client's interest within the existing relationship (valid consent from the right party, assent, least restrictive option), then consult or obtain supervision/training, then document, and only then escalate. Report when the Code specifically mandates it (risk of harm, fraud, sanctions) and check consent-to-disclose provisions before sharing information.
Misidentifying the function of problem behavior by its topography or by the condition label - e.g., reading high responding in the alone condition as attention maintained, or treating a descriptive scatterplot as if it were a functional analysis.
Topography (hitting, screaming) is the most salient feature of a case description, and functional analysis conditions are memorized by name rather than by the contingency each one arranges.
What to do instead
For every condition, write the three-term contingency: motivating operation, response, consequence. Alone → automatic; attention → social positive; demand → social negative (escape); play/control → noncontingent attention and no demands. Then check whether the described outcome (e.g., persistence in alone with no social consequence) actually implicates that reinforcer, and distinguish indirect, descriptive, and experimental methods by what was manipulated.
Selecting among DRA, DRO, DRL, and NCR by the procedure name rather than by the contingency written in the option - e.g., choosing DRA when the option reinforces the absence of behavior, or choosing DRO when the option reinforces a specific alternative response.
The procedure labels are memorized as a list of 'reinforce something else' strategies, and the options are deliberately written so that all of them sound like differential reinforcement.
What to do instead
Translate each option into 'when [condition], [consequence] is delivered.' Then check two things: the schedule (does reinforcement depend on absence of behavior, a lower rate, or a specific alternative response?) and the function (does the consequence match the maintaining reinforcer?). Also check that extinction is not offered alone for behavior maintained by automatic reinforcement, where an enrichment or competing-stimulation component is needed.
Selecting an intervention on effectiveness alone, without checking social validity, feasibility of treatment integrity, maintenance and generalization planning, or the acceptability of the procedure to the client and caregivers - including endorsing a more restrictive procedure when an equally effective less restrictive one is available.
Stems emphasize behavior reduction, so effectiveness dominates the decision and the contextual criteria are treated as afterthoughts rather than as part of the correct answer.
What to do instead
Run a short checklist on every intervention item: function match, then least restrictive alternative, then acceptability and social validity, then whether the implementers can carry it out with integrity, then the explicit plan for generalization and maintenance. The option that satisfies the whole checklist beats the option that is merely the most powerful.
Treating supervision as administrative - e.g., concluding a supervisee is competent because they passed a written quiz, or endorsing an option that omits components of behavioral skills training (instruction, modeling, rehearsal, feedback) or omits direct observation of the supervisee's performance.
Supervision is often studied last and least, so BST is recalled as a list of words rather than as a sequence that must include in-vivo rehearsal and feedback against measured performance.
What to do instead
For supervision items, first identify the performance deficit, then check whether the option names the full BST sequence and includes measurement of the supervisee's actual performance in the natural setting. Competency is established by observed, measured performance with documented feedback, not by verbal knowledge alone; look for the option that includes a supervision contract, ongoing performance monitoring, and documentation.