GHABA Proficiency II cheat sheet

Operant behavior is selected by consequences; respondent behavior is elicited by antecedent stimuli. The three-term contingency is antecedent (SD) → behavior → consequence (SR).

Reinforcement increases the future probability of behavior; punishment decreases it. Positive means a stimulus is added; negative means a stimulus is removed. Magnitude of effect depends on contingency, immediacy, and establishing operations.

Reinforcer identification: a stimulus is a reinforcer only if it demonstrably increases the behavior it follows. Preference assessments (paired-choice, MSWO, free-operant) rank stimuli but do not prove reinforcement.

Schedules: FR, VR, FI, VI. Ratio schedules typically produce higher response rates than interval schedules; VR and FI commonly produce post-reinforcement pauses and characteristic break-and-run or scalloped patterns.

Extinction is withholding the reinforcer that previously maintained behavior; it produces an extinction burst, response variability, and spontaneous recovery. Extinction of one member of a response class may not generalize to others.

Stimulus control: discrimination training produces an SD (reinforced) and S-delta (not reinforced). Generalization gradients are flatter after single-SD training and sharper after discrimination training.

Motivating operations: establishing operations increase the value of a consequence and evoke behavior; abolishing operations decrease value and abate behavior. MOs alter both reinforcer value and the momentary frequency of the relevant behavior class.

Verbal behavior taxonomy (Skinner): mand (request under MO), tact (label under nonverbal SD), intraverbal (verbal SD), echoic, textual, transcription. Mand is the only operant defined by its consequence.

Indirect methods (interviews, rating scales, FAST, MAS) are low-effort but have weak reliability and should be corroborated by direct observation.

Descriptive assessment records antecedents, behavior, and consequences in natural settings; it identifies correlations, not causal variables.

Experimental functional analysis (Iwata-style) tests attention, escape, tangible, and alone/automatic conditions against a play/control condition; it is the only method that demonstrates functional relations.

A functional analysis with a single test condition and no control cannot isolate the maintaining variable; a control condition (or pairwise design) is required for valid inference.

Brief functional analysis and trial-based FA reduce time and setting constraints but raise the risk of false negatives from limited exposure to contingencies.

Behavior definition must be operational: observable, measurable, and with onset and offset criteria. Vague labels (e.g., "aggressive") are not measurable and cannot be scored reliably.

Preference assessment hierarchy: free-operant (fast, least rigorous), single stimulus, paired choice (most rigorous ranking), MSWO (efficient ranking of many items).

Indirect measures of baseline (e.g., record review) may under- or overestimate rate; always establish baseline with direct observation before intervention.

Function-based interventions outperform non-function-based ones. Match the intervention to the maintaining contingency: attention → differential attention; escape → demand fading or break; tangible → contingency-based access; automatic → competing stimulation or response interruption.

Antecedent interventions (NCR, high-p, priming, choice, environmental modification) reduce the establishing operation before behavior occurs; consequence interventions act after.

Differential reinforcement: DRA (reinforce alternative), DRI (incompatible), DRO (reinforce omission), DRL (reinforce low rate). DRO intervals must be shorter than the baseline inter-response time.

Extinction must be paired with reinforcement of a functionally equivalent alternative to avoid resurgence and to reduce extinction bursts.

Token economies require a defined token, a schedule of exchange, backup reinforcers, and a plan to fade to natural contingencies.

Response cost and time-out are punishment-based; both require a rich reinforcement schedule for appropriate behavior and clear procedural integrity checks.

Behavior intervention plans must specify target behaviors, operational definitions, baseline data, function hypothesis, procedures, personnel responsible, data collection method, and criteria for fading or revision.

Treatment integrity (procedural fidelity) is measured as the percentage of steps implemented correctly; low integrity invalidates conclusions about treatment effectiveness.

Continuous measures: frequency/rate (count per unit time), duration (total or per occurrence), latency (SD to response), IRT (time between responses). Rate = count ÷ observation time.

Discontinuous measures: partial-interval (overestimates), whole-interval (underestimates), momentary time sampling (least biased for high-rate behavior). Choose based on behavior duration and observer capacity.

Percent of opportunities, trials-to-criterion, and permanent products are useful for discrete academic or self-care targets.

Interobserver agreement: total count, exact count, interval-by-interval, and occurrence/nonoccurrence methods. IOA = agreements ÷ (agreements + disagreements) × 100; target ≥ 80%, with 90%+ for high-stakes decisions.

Graphing: line graph with time on x-axis and behavior on y-axis; include condition change lines, phase labels, and axis labels. Visual analysis judges level, trend, and variability within and across phases.

Single-case designs: withdrawal (ABAB), multiple baseline across behaviors/settings/participants, alternating treatments, changing criterion. Multiple baseline avoids withdrawal but requires stable baselines and independent behaviors.

Functional relation requires replication of effect within and across designs; a single AB comparison cannot rule out confounding variables such as maturation or setting events.

Data-based decision rules: extend baseline if trend is unstable; change intervention if no improvement after a predetermined number of sessions or if trend is in the wrong direction.

Informed consent requires capacity, voluntariness, disclosure of procedures, risks, benefits, alternatives, and the right to withdraw at any time without penalty.

Least restrictive, least intrusive effective procedure: choose the intervention with the best risk-benefit ratio and the least aversive effective option first.

Reinforcement-based procedures are preferred over punishment; punishment requires documented failure of less intrusive options, oversight, and a plan to fade.

Confidentiality: disclose only with consent or legal mandate; store records securely; know mandatory reporting duties for suspected abuse or neglect.

Dual relationships that impair objectivity or exploit the client must be avoided; when unavoidable, disclose and manage them in writing.

Scope of competence: practice only within training and supervised experience; seek consultation or refer when outside competence.

Documentation must be objective, timely, and behaviorally descriptive; avoid opinion and jargon; correct errors without altering the original record.

Research and data use require IRB or equivalent review, informed consent, and protection of identifiable information; data sharing requires de-identification or consent.

Supervision must be behavior-analytic, individualized, and documented: specify goals, competencies, observation methods, and feedback schedules.

Behavioral skills training: instruction → modeling → rehearsal → feedback until criterion performance; role-play is required, not optional.

Feedback is most effective when immediate, specific, and contingent on observed performance; include both praise for correct steps and corrective feedback.

Competency checklists with operational definitions allow reliable scoring of trainee performance; interobserver agreement on scoring should be checked.

Supervision documentation includes dates, duration, activities, trainee performance data, and signatures; retain records per regulatory requirements.

Case management: assign a single point of coordination, track authorization dates, service units, and progress toward goals; review data at each team meeting.

Delegation requires verifying the trainee's competence for the specific task; responsibility for client outcomes remains with the supervising professional.

Fading supervision should be data-based: reduce frequency only after the trainee meets predetermined performance criteria across settings.

Interdisciplinary teams share assessment and intervention responsibilities; multidisciplinary teams work in parallel with less integration. Clarify roles and decision authority at the outset.

Behavior analysts contribute operational definitions, data systems, and functional hypotheses; avoid prescribing outside scope (e.g., medication, diagnosis).

Collaboration requires a shared language: translate behavioral terminology into plain language for families, educators, and medical staff.

Coordination of care: share only the minimum necessary information, with consent, and document what was shared and with whom.

Conflicts are resolved by returning to data and client goals, not authority; escalate through defined channels when consensus fails.

Caregiver training is an intervention: use BST, measure caregiver implementation fidelity, and evaluate child behavior change as the outcome.

Cultural responsiveness: assess whether goals and procedures fit the family's values and context; adapt procedures while preserving the active ingredients.

Transition planning across settings (school to home, clinic to community) requires generalization programming: train loosely, use natural contingencies, and plan for maintenance.