By Matt Harrington, BCBA · Behaviorist Book Club · Research-backed answers for behavior analysts
There is no universally established threshold in the behavior analytic literature, though 80% is frequently cited as a floor in research and training contexts. In clinical practice, the appropriate standard depends on the nature of the procedure. For highly structured discrete trial instruction, 90% or higher is reasonable. For procedures involving safety protocols or restrictive practices, any implementation error is significant and should trigger immediate retraining rather than waiting for aggregate scores to fall below a threshold. The more important question is not what the floor is, but what specific steps are being missed and whether those errors are systematic or random — each pattern has different implications for intervention.
IOA measures agreement between two observers recording the same behavioral events — it is an index of measurement quality. Treatment fidelity measures whether the intervention procedure is being implemented as designed — it is an index of implementation quality. Both matter because errors in either domain corrupt your clinical data. Poor IOA means you cannot trust the outcome data you are collecting. Poor fidelity means the treatment you think is operating on the client's behavior may bear little resemblance to what is actually happening in sessions. A program with excellent IOA but poor fidelity tells you precisely that a different intervention than the one you designed is consistently being implemented. You need both metrics to have confidence in your clinical conclusions.
Use exact agreement when the sequence or timing of behavioral occurrences matters — for interval recording, partial interval, momentary time sampling, or trial-by-trial data. Exact agreement requires that observers agree on each specific interval or trial, making it the most conservative and most informative method for detecting actual disagreements. Total count IOA is appropriate only for frequency data where the absolute count matters more than the precise moment of occurrence, such as counting total vocalizations during a play session. In most clinical ABA contexts, exact agreement provides more actionable information about where and why observers are diverging, which is what you need to correct measurement drift.
BACB Ethics Code 2.19 requires BCBAs to design, implement, and evaluate systems to ensure supervisees are implementing behavior-change programs as designed. This is not aspirational — it is a professional obligation. Code 2.01 further requires competent service delivery, which presupposes that someone is verifying implementation quality. Code 2.09 requires treatment decisions to be based on data, and that data is only interpretable if you have confidence in its fidelity. Practically speaking, a supervision structure that does not include systematic fidelity monitoring violates multiple ethics code sections simultaneously, regardless of whether it produces good client outcomes, because you cannot know whether the outcomes are attributable to the treatment without fidelity data.
Observer reactivity — staff performing differently when they know they are being observed — is a well-documented phenomenon that inflates fidelity scores relative to unobserved performance. Practical countermeasures include increasing observation frequency so that being observed becomes routine rather than exceptional, using brief unannounced observations rather than scheduled formal assessments, reviewing session recordings rather than conducting only live observations, and calibrating observers periodically against criterion recordings of expert implementation. If your fidelity data consistently shows near-perfect scores across all staff and programs, that uniformity itself is a signal worth investigating — it may reflect reactivity rather than genuine program-wide implementation quality.
When fidelity data is absent, BCBAs lose the ability to distinguish between treatment failure and implementation failure. A flat or declining data path might reflect a poorly designed intervention, but it might equally reflect a correctly designed intervention that is not being implemented. Without fidelity data, both possibilities remain on the table, and the BCBA is making modification decisions under conditions of incomplete information. The consequence is often unnecessary treatment complexity: adding components, changing reinforcement schedules, or increasing session frequency when the actual problem was that existing components were not being implemented correctly. This wastes resources and delays client progress.
Fidelity and IOA data should be documented in a format that links each observation to the specific technician, session date, program being assessed, fidelity score or IOA percentage, the specific steps or intervals where discrepancies occurred, and the supervisory action taken in response. Aggregate percentages alone are insufficient. The documentation should allow a reviewer — whether an internal quality assurance auditor, a licensing board, or a legal proceeding — to reconstruct the implementation history of any given program. Supervision logs that record only that fidelity was checked without specifying outcomes or follow-up actions do not meet the standard of meaningful data-driven supervision required by the Ethics Code.
The most common sources of procedural drift include ambiguous or overly complex procedure descriptions that allow for multiple valid interpretations, insufficient initial training that results in staff beginning program implementation before mastery is demonstrated, feedback delays that allow errors to consolidate through practice before they are identified, high staff-to-supervisor ratios that reduce observation frequency to the point where drift goes undetected for extended periods, and organizational cultures where errors are associated with punitive consequences rather than supportive training. Ambiguity in written procedures is particularly problematic because it produces inconsistency that appears as implementation error when it actually reflects genuine procedural ambiguity that the BCBA should resolve through clearer writing.
Effective self-monitoring training begins with ensuring technicians have fluent knowledge of the procedure steps — not just exposure to written instructions but demonstrated mastery through practice and feedback. Once procedural knowledge is solid, introduce self-monitoring checklists that technicians complete after sessions, initially with immediate supervisor verification to calibrate accuracy. Gradually fade supervisor verification while maintaining periodic spot-checks to assess whether self-monitoring scores correlate with observed performance. Technicians whose self-monitoring scores consistently match supervisor observations can be trusted with lower-frequency direct observation. Discrepant self-monitors — whether systematically over- or under-estimating their fidelity — need additional calibration training before self-monitoring serves as reliable data.
Yes, and the core measurement logic remains unchanged — what shifts is the observation mechanism. For remote services, video review of recorded sessions replaces live observation as the primary fidelity assessment tool. This can actually improve the quality of fidelity data because recordings allow for repeated viewing, slow-motion review of ambiguous moments, and calibration sessions using shared recordings across supervisors. For hybrid models where a technician is in the home while the BCBA supervises remotely, real-time video observation provides similar information to in-person observation with the added benefit of not requiring physical co-location. Ensure that data privacy and confidentiality requirements under Ethics Code 2.04 are addressed in the telehealth consent and service agreement before implementing video-based fidelity systems.
The ABA Clubhouse has 60+ on-demand CEUs including ethics, supervision, and clinical topics like this one. Plus a new live CEU every Wednesday.
Ready to go deeper? This course covers this topic with structured learning objectives and CEU credit.
Precision Playbook: Building Excellence in ABA Treatment — Melanie Shank · 1.5 BACB Supervision CEUs · $10
Take This Course →1.5 BACB Supervision CEUs · $10 · BehaviorLive
Research-backed educational guide with practice recommendations
Side-by-side comparison with clinical decision framework
All behavior-analytic intervention is individualized. The information on this page is for educational purposes and does not constitute clinical advice. Treatment decisions should be informed by the best available published research, individualized assessment, and obtained with the informed consent of the client or their legal guardian. Behavior analysts are responsible for practicing within the boundaries of their competence and adhering to the BACB Ethics Code for Behavior Analysts.