Functional Behavior Assessment: A Practitioner's Guide to FBA, FA, IISCA, and Trial-Based Methods

What an FBA actually is in 2026 practice

For most BCBAs, "FBA" is shorthand for a layered process: (1) records review and open-ended caregiver/teacher interview, (2) direct ABC observation in the natural environment, and (3) — when needed — an experimental analysis that manipulates antecedents and consequences to test the maintaining contingency (Pollack et al., 2024). A PRISMA-guided systematic review of 34 intervention studies for K-12 students with emotional/behavioral disorders found that 85.6% of cases reported a hypothesized or confirmed function before treatment, and the assessments that produced those functions almost always bundled record review, indirect tools, interviews, direct observation, and — in roughly a third of cases — a true functional analysis (Pollack et al., 2024). Attention was the most common identified function (41%), followed by multiple reinforcers (28.9%) and escape (25.3%), with tangible and automatic functions less frequent in this population (Pollack et al., 2024). The takeaway for practice is not the percentages themselves but the architecture: function gets identified by stacking methods, not by picking one.

The functional analysis is no longer a 2-hour multi-element party

The classic Iwata multi-element FA is still in the literature, but day-to-day school and clinic practice has moved heavily toward briefer formats. A systematic review of FAs conducted in U.S. public schools found that brief and trial-based variants dominated, that attention and escape were the most common identified functions, and that school-friendly FA formats reliably produced interpretable functions within typical school logistics Nesselrode et al. (2022). Saini and colleagues' quantitative-and-qualitative review of FA efficiency provides the operational rationale: pre-selecting test conditions from caregiver and teacher interview data — and dropping conditions for which no reinforcer is hypothesized (most often the tangible test) — meaningfully shortens FA duration without sacrificing the ability to detect function Saini et al. (2020). The same review notes that controlling for ambient stereotypy and competing stimulation during control conditions reduces false-positive interpretations of automatic reinforcement, which matters because false-positive automatic functions waste weeks of treatment time Saini et al. (2020).

IISCA: one synthesized test, one matched control

The interview-informed synthesized contingency analysis (IISCA) packages an open-ended caregiver interview into a single synthesized test condition compared against a matched control. The standardization paper by Jessel, Hanley, and Ghaemmaghami argues that this format makes assessment faster, safer, and more directly translatable to function-based intervention than multi-condition alternation, particularly when severe behavior makes prolonged exposure to evocative conditions ethically uncomfortable Jessel et al. (2020). A two-step validation of the performance-based IISCA across three children with developmental disabilities confirmed that the model can identify socially mediated functions within roughly two clinical visits and reliably hand off to skill-based, function-matched treatment (Jessel et al., 2024). The trade-off the corpus is honest about: when Rajaraman and colleagues evaluated inter-rater reliability of the open-ended PFA interview and IISCA across four children, categorical agreement on function was high (~97%), but specific-feature agreement on the exact establishing operation (EO), exact response form, and exact reinforcer detail was substantially lower and variable case-to-case Rajaraman et al. (2022). Two clinicians can agree the case is "escape" and still run different-looking IISCAs unless the EO and reinforcer are scripted in writing.

Trial-based FA: function identified inside the day's instruction

Trial-based FA (TBFA) embeds short test/control trials into ongoing instruction or care routines, so a function can be identified without pulling a learner into an analog clinic suite Nesselrode et al. (2022). The school-FA review found TBFA among the dominant formats used in classrooms Nesselrode et al. (2022). Two implementation studies show how to scale this approach. Griffith and colleagues used a 23-page self-instruction manual plus a single 2-hour group webinar to teach four entry-level community technicians (no prior FA experience) to administer TBFA at 93% procedural integrity within one week, with gains maintained at a 4-week in-vivo follow-up on real clients Griffith et al. (2020). Togashi extended the same logic to telehealth: asynchronous computer-based instruction (CBI) raised TBFA knowledge but only the addition of brief telehealth behavioral skills training (BST) — rehearsal plus feedback — produced mastery-level procedural integrity in four Japanese practitioners working in resource-scarce regions (Togashi, 2025). Both studies converge on the same training architecture: didactic modules first, then a short coached rehearsal, with a documented integrity check before clients are involved Griffith et al. (2020) (Togashi, 2025).

Do you actually need an experimental FA at all?

The corpus complicates the assumption that every FBA must culminate in a multi-element FA. Call and colleagues' comparative effectiveness trial across 57 children with autism contrasted FBAs that included a functional analysis component with FBAs that did not, before delivering a unified functional communication training (FCT) package; treatment outcomes were similar across the two FBA paths, though correspondence between the two assessment formats was only modest, indicating some individuals would still have benefited from the additional precision an experimental FA provides Call et al. (2024). Tereshko and colleagues pushed this further with a single-case component analysis showing that conditional probabilities computed from antecedent data inside a standard ABC descriptive assessment matched the function later confirmed by FA and successfully guided function-based treatment under reversal control Tereshko et al. (2024). Read together, these papers argue that descriptive data — when collected carefully and quantified — can sometimes carry the FBA on their own, and that the experimental FA's role in 2026 is closer to a confirmatory or escalation step than a default first move.

Indirect assessment is not interchangeable with descriptive or experimental data

Even a brief FA cannot rescue a thin interview. Fryling and Baires' practical analysis of indirect FBA tools argues that question phrasing — open-ended versus closed-ended — changes the kind of information you can later test Fryling & Baires (2016). Closed-ended checklists are efficient but tend to miss idiosyncratic antecedents and reinforcers; open-ended probes ("describe what usually happens just before," "what do you find yourself doing afterward to settle things down") generate the candidate maintaining variables that an FA or IISCA can then test Fryling & Baires (2016). Gatzunis and colleagues built BST around exactly this skill, training graduate-level students to deliver culturally responsive, empathic, open-ended functional assessment interviews; post-training role-play probes showed reliable use of open-ended antecedent and consequence questions and explicit attention to family-centered goals Gatzunis et al. (2023). Both papers argue that interview quality is itself a procedural variable — not a soft skill — and that you can train it.

Demand assessments and antecedent isolation

When a caregiver or teacher reports that "math sets him off," the FBA can be sharpened before any FA condition runs. Avery and Akers' practical guide to demand assessments lays out brief, low-risk procedures that vary task difficulty, amount, and novelty to determine whether academic or task demands reliably evoke problem behavior, and to identify the specific demand context (math vs. reading vs. transition) that should be tested in any subsequent FA or treatment analysis Avery & Akers (2021). The point is not just efficiency. By running a 5-10 minute demand-only condition with prompting hierarchies and brief escape contingencies, you reduce risk during the eventual FA — you've already calibrated the EO before you ask it to evoke behavior at full strength Avery & Akers (2021).

Low-rate, automatic, and otherwise difficult presentations

The classic FA was designed for behavior that occurs reliably enough to fill a 10-minute condition. Real caseloads include pica that occurs once a shift, self-injury that emerges only when the parent is in the room, and severe property destruction that no clinic wants to evoke twice. Brown and colleagues' decision-making review summarizes the procedural tweaks that keep FA interpretable in these cases: caregiver-implemented FAs with remote coaching, baited rooms with safe items for ingestion-related topographies, and systematic extension or replication of conditions across caregivers and settings before abandoning the functional hypothesis when initial outcomes are undifferentiated Brown et al. (2025). For automatically reinforced self-injury specifically, Hagopian and colleagues identified within-session predictive behavioral markers — most importantly level-of-differentiation in the automatic-reinforcement condition and resistance-to-extinction indices — that signal functional control even when the topography looks similar across conditions Hagopian et al. (2018). Falligant and Hagopian's "precision medicine" extension quantified this further across 103 retrospectively analyzed FA/FCT data sets: a continuous level-of-differentiation (LOD) value computed from play-versus-alone rates, with an optimal cut-off near 0.64, predicted whether reinforcement-alone FCT would succeed for automatically maintained self-injury, offering a reproducible decision aid for one of the field's hardest treatment-selection problems Falligant & Hagopian (2020).

Sensitivity and bias as quality control

A test condition that produces high rates of behavior is not, by itself, evidence of function (Allen et al., 2026). Allen and colleagues compared methods of evaluating sensitivity to common establishing operations alongside observer bias toward challenging behavior across repeated single-subject FAs, showing that derived sensitivity metrics and bias estimates flag when a test/control contrast is detecting function versus simply being confounded by extraneous variables (Allen et al., 2026). The practical implication for any BCBA running FAs is that documenting these quantitative checks in the FA report — not just the visual differentiation — strengthens the functional interpretation when a multidisciplinary team or payer second-guesses it later (Allen et al., 2026).

Risk screening before any experimental condition

FAs intentionally evoke the behavior they are trying to understand. Deochand, Eldridge, and Peterson built a 15-item risk-assessment checklist from a Delphi-style expert process, then validated it across 37 BCBAs scoring 20 written vignettes; the tool produced strong inter-rater agreement (κ = .88) and corresponded substantially with expert ratings of FA safety, allowing teams to triage cases into low-risk on-site FA, supervised/protected-setting FA, or alternative descriptive assessment before the first session Deochand et al. (2020). The authors argue — correctly — that operationalizing the BACB's "risk-benefit" language with a scoreable instrument also creates a defensible documentation trail for peer review or insurance audits Deochand et al. (2020).

The classroom is itself an independent variable

Kestner and colleagues' conceptual paper makes a point that is easy to forget when the FA is the main event: the baseline classroom — seating arrangement, instructional format, transition structure, teacher attention rate — is already a contingency operating on the student before you arrive Kestner et al. (2019). An FBA that doesn't describe these conditions can reach a function that is technically correct but practically untreatable, because the BIP gets implemented inside an ecology that was already evoking or suppressing the behavior in ways no one measured. Their recommendation is concrete: collect a brief eco-behavioral scan of seating, instructional format, transition timings, and teacher attention rates before the interview phase, and document baseline reinforcement schedules already in the room Kestner et al. (2019).

Verbal behavior and ACT need ongoing — not one-shot — functional assessment

For language-based interventions, the FBA does not stop after the assessment phase. Sandoz, Gould, and DuFrene argue that Acceptance and Commitment Therapy, when delivered as behavior analysis, requires moment-to-moment functional assessment of verbal behavior because the function of a self-rule, defusion statement, or values-clarification utterance cannot be inferred from its form Sandoz et al. (2022). The recommended pattern — evoke a contextual stimulus, observe the resulting behavior shift, manipulate the next stimulus accordingly — embeds FA logic inside the treatment session itself, and is essentially a continuous version of the same evoke-and-observe cycle that drives a full FA Sandoz et al. (2022). The same general principle is implicit in the field's broader push, articulated by Kodak and Halbur, to extend FA-style test/control comparisons into skill-acquisition decisions, not only behavior-reduction ones Kodak & Halbur (2021).

Variability is the field's open wound

A mega-review of 17 prior systematic and narrative reviews covering 437 primary studies of school-based FCT found that only 39–61% of students received an experimental FA before FCT, 11–45% had multi-method FBAs, and the rest had indirect-only assessment or no FBA reported at all; roughly a third of cases involved multiply controlled behavior Corr et al. (2025). The implication for practitioners is uncomfortable: when you replicate a published school-based FCT protocol, you cannot assume the underlying FBA matches the rigor of the FCT itself. Plan for multiply maintained behavior in roughly a third of cases, and do not skip the experimental layer just because the original report did Corr et al. (2025).

Implementation fidelity decays after mastery

Even a well-derived, function-matched plan erodes if consequence delivery drifts. Jones and colleagues' translational analogue work showed that intermittent, accidental consequence-fidelity errors after a behavior had reached mastery rapidly weakened previously accurate responding, often within a small number of trial blocks (Jones et al., 2026). The clinical implication is that FBA-derived behavior plans need consequence-fidelity checks — not just antecedent and procedural-step checks — built into ongoing supervision, particularly during the maintenance phase when teams stop watching closely (Jones et al., 2026).

Performance management keeps FBA volume up

Sellers and colleagues evaluated a remote performance-management package (goal-setting, self-monitoring, weekly graphical feedback, and remote supervision) across six clinical supervisors in a multi-state residential agency; TBFA completion rose from a baseline of 0–4 per month to a stable 20+ per month per supervisor Sellers et al. (2019). The package required no on-site travel and used a defined mastery criterion (one full TBFA per working day). For agency leaders, the study reads as an existence proof that FBA throughput is a manageable variable, not a fixed property of staff motivation Sellers et al. (2019).

Who actually runs FBAs in schools

A statewide field-of-practice survey in Vermont found that 83% of practicing behavior analysts endorsed competence in functional assessment of problem behavior, while only 43% endorsed competence with classic functional analysis, and a parallel interprofessional survey found that BCBAs reported significantly higher FBA frequency than school psychologists (Mayo & Hoffmann, 2024) (Snyder et al., 2024). The gap matters operationally. Districts that funnel FBA referrals through school psychology may be relying on staff with lower self-reported competence in the procedure, and the more efficient FA variants (brief, TBFA, IISCA) better fit what most BCBAs say they can actually deliver (Mayo & Hoffmann, 2024). LaMarca and LaMarca's ADDIE-based assessment-selection guide complements this by reminding teams to vet whether the broader assessment battery in use (VB-MAPP, ABLLS-R, PEAK, etc.) explicitly samples functions of behavior, so the FBA does not get orphaned from the rest of the program (LaMarca et al., 2024).

A foundation page should be honest about where evidence comes from (Pollack et al., 2024). The FBA literature lives mostly at the systematic-review and single-subject-experimental-design (SCED) layers, with one comparative effectiveness trial and a band of theoretical and survey work around the edges Call et al. (2024) (Jessel et al., 2024).

Systematic and narrative reviews. Pollack and colleagues' PRISMA review of 34 EBD intervention studies anchors the multi-method FBA picture and the function distribution (attention 41%, multiple 28.9%, escape 25.3%) (Pollack et al., 2024). Nesselrode and colleagues' systematic review of FAs in U.S. public schools establishes that brief and trial-based formats now dominate that setting and reliably identify function inside school logistics Nesselrode et al. (2022). Saini and colleagues' efficiency review quantifies which procedural tweaks shorten FA without sacrificing validity Saini et al. (2020). Brown and colleagues review FA decision-making for low-rate and undifferentiated cases Brown et al. (2025). Avery and Akers review demand assessments as a sharpening layer Avery & Akers (2021). Corr and colleagues' mega-review of 17 reviews documents the wide variability in FBA rigor before classroom FCT Corr et al. (2025).

Comparative / quasi-experimental group studies. Call and colleagues' comparative effectiveness trial across 57 children with ASD is the closest the corpus comes to a randomized head-to-head test of FBA-with-FA against FBA-without-FA, and reports broadly equivalent FCT outcomes with only modest concordance between methods Call et al. (2024).

Single-subject experimental designs. Most FBA evidence is SCED (Jessel et al., 2024). Jessel and colleagues' two-step validation of the performance-based IISCA across three children grounds the IISCA as a trauma-informed format (Jessel et al., 2024). Rajaraman and colleagues' dual-scorer reliability work across four children quantifies the categorical/specific agreement gap Rajaraman et al. (2022). Tereshko and colleagues demonstrate antecedent-conditional-probability-driven function identification in a single case under reversal control Tereshko et al. (2024). Falligant and Hagopian's level-of-differentiation work pools 103 prior cases Falligant & Hagopian (2020); Hagopian and colleagues' predictive behavioral markers paper is grounded in a 4-case design Hagopian et al. (2018). Allen and colleagues develop sensitivity and bias measures across repeated single-subject FAs (Allen et al., 2026). On the training side, Griffith and colleagues' multiple-probe TBFA study (n=4) and Togashi's replication blending asynchronous instruction with telehealth coaching (n=4) are the foundational SCED demonstrations that FBA skills can be installed efficiently Griffith et al. (2020) (Togashi, 2025). Sellers and colleagues' multiple-baseline performance-management study (n=6 supervisors) sits at the agency-throughput layer Sellers et al. (2019). Gatzunis and colleagues' BST study for empathic, culturally responsive interviewing (n=6) belongs here as well Gatzunis et al. (2023). Jones and colleagues' translational analogue on post-mastery fidelity decay (n=4 undergraduates) is single-subject but needs further clinical generalization (Jones et al., 2026).

Decision-tools and case-series. Deochand and colleagues' FA risk-assessment checklist study (n=37 BCBAs scoring 20 vignettes) provides a usable triage instrument with strong agreement, while still relying on written scenarios rather than live FA outcomes Deochand et al. (2020).

Theoretical and conceptual. Kestner and colleagues on baseline classroom ecology, Sandoz and colleagues on continuous functional assessment in ACT, Fryling and Baires on open- versus closed-ended interview phrasing, Jessel, Hanley, and Ghaemmaghami on FA standardization, Kodak and Halbur on FA logic in skill acquisition, and LaMarca and LaMarca on assessment-selection are conceptual papers without experimental design — useful as procedural anchors, weaker as outcome evidence Kestner et al. (2019) Sandoz et al. (2022) Fryling & Baires (2016) Jessel et al. (2020) Kodak & Halbur (2021) (LaMarca et al., 2024).

Survey and field-of-practice. Mayo and Hoffmann's Vermont survey and Snyder and colleagues' interprofessional comparison sit at the descriptive layer and should be read as practice-pattern data, not outcome evidence (Mayo & Hoffmann, 2024) (Snyder et al., 2024).

Bottom line. The convergent picture across these tiers is strong for the operational claims this page makes — that brief FA, TBFA, and IISCA produce interpretable functions efficiently in real settings; that risk and ecology need to be measured; that interview phrasing and consequence fidelity are procedural variables; and that training and performance management can scale FBA throughput Nesselrode et al. (2022) Jessel et al. (2020) Deochand et al. (2020) Kestner et al. (2019). It is weaker for any claim that one FA format produces durably better client outcomes than another in head-to-head trials Call et al. (2024). There, the field has one quasi-experimental study and a lot of SCED Call et al. (2024).

The FBA decisions a senior practitioner makes are not "FA or no FA" so much as "how much assessment, in what format, given this risk profile and these resources." A defensible logic, drawn directly from the corpus:

1. New referral, externalizing problem behavior. Run records review and an open-ended caregiver/teacher interview first — closed-ended checklists alone tend to miss idiosyncratic antecedents and reinforcers Fryling & Baires (2016). In parallel, complete a 15-item FA risk screen Deochand et al. (2020).
2. Risk screen low and ABC data point clearly to a single function. Proceed directly to function-based intervention; reserve a brief FA probe as a second-line check if the BIP fails after a defined trial period Call et al. (2024). Conditional probabilities computed from antecedents in your descriptive ABC data can carry more weight than practitioners typically give them Tereshko et al. (2024).
3. Risk screen low but ABC data ambiguous, or multiple topographies / suspected multiple control. Run an IISCA: open interview → single synthesized test condition versus matched control Jessel et al. (2020) (Jessel et al., 2024). Drop the tangible test by default unless the interview gives a tangible hypothesis — that's the single biggest FA-shortening lever in the literature Saini et al. (2020).
4. School setting with classroom-embedded behavior. Use TBFA: trial-based test/control conditions inserted into ongoing instruction Nesselrode et al. (2022). Pair this with an eco-behavioral scan of seating, instructional format, transitions, and teacher attention rates so the resulting BIP fits the ecology that actually exists Kestner et al. (2019).
5. Suspected escape from demands, but unclear which demands. Insert a brief demand assessment that varies task difficulty, amount, and novelty before any FA condition; this calibrates the EO and pre-selects the demand context to test Avery & Akers (2021).
6. Behavior is low-rate or initially undifferentiated. Move to caregiver-implemented FA with remote coaching, extended/replicated conditions, latency-based metrics, and — for ingestion topographies — baited rooms with safe items Brown et al. (2025).
7. Suspected automatic reinforcement, especially self-injury. Compute level-of-differentiation from your alone/play data; an LOD around or above 0.64 supports proceeding with reinforcement-alone FCT, while lower values signal you likely need matched stimulation, sensory enrichment, or extended conditions before treatment selection Falligant & Hagopian (2020). Track resistance-to-extinction and within-session differentiation as ongoing markers Hagopian et al. (2018).
8. Risk screen elevated or topography severe. Move to a protected setting, caregiver-implemented FA with coaching, or — at the top end — refer for consultation with a specialized FA team or inpatient behavior unit Deochand et al. (2020) Brown et al. (2025).
9. FA results undifferentiated after one round. Replicate across caregivers or settings, extend session lengths, control for ambient stereotypy in the control condition, and document sensitivity/bias metrics to flag confounding before abandoning the functional hypothesis Saini et al. (2020) Brown et al. (2025) (Allen et al., 2026).
10. Language-based or ACT-style intervention. Build moment-to-moment functional assessment into the treatment session itself — evoke contextual stimulus, observe shift, adjust — rather than relying on one-shot pre-treatment assessment Sandoz et al. (2022).

Schools (K-12)

Public schools are where most FBAs happen and where the corpus is densest. The school-FA review documents that brief and trial-based formats dominate the published classroom literature, that attention and escape are the most commonly identified functions, and that school-friendly FA variants reliably pinpoint function inside typical school logistics Nesselrode et al. (2022). The practical pattern that emerges across the studies is consistent: an open-ended teacher and caregiver interview, an ABC observation tied to specific instructional contexts, an eco-behavioral scan of the classroom, and — when needed — TBFA trials embedded in regular instruction Kestner et al. (2019) Nesselrode et al. (2022). Kestner and colleagues are explicit that ignoring the classroom's existing reinforcement schedules will produce hypotheses that are technically correct and practically untreatable Kestner et al. (2019). Demand assessments add precision when escape is suspected but the specific evocative demand is unclear Avery & Akers (2021). Two structural realities to plan around: BCBAs in U.S. schools conduct FBAs at significantly higher rates than school psychologists, so referral pathways should flow toward BCBAs by default rather than being shared evenly across roles (Snyder et al., 2024); and rigor before FCT in classrooms is not yet standard, so when replicating a school FCT protocol, do not assume the underlying FBA was airtight Corr et al. (2025).

Outpatient and university clinics

Outpatient and university clinics are where the IISCA and brief-FA validation work mostly lives. The two-step validation of the performance-based IISCA shows the model can produce a clear function and a defensible function-based treatment within roughly two clinic visits in a trauma-informed format (Jessel et al., 2024). The standardization paper makes the case that this efficiency does not come from cutting corners but from collapsing idiosyncratic contingencies into a single synthesized test/control comparison after a careful interview Jessel et al. (2020). Saini and colleagues' efficiency review applies cleanly to clinic operations: pre-selecting test conditions from the interview, dropping unsupported tangible conditions, and controlling stereotypy in the control condition meaningfully reduce assessment time without weakening interpretation Saini et al. (2020). Risk screening before any experimental session is the operational hinge — a 15-item checklist with κ = .88 reliability gives clinics a defensible decision rule for when to keep a case in-house versus escalate Deochand et al. (2020).

Home and telehealth

Home-based and telehealth-delivered FBAs are increasingly central, particularly for low-rate behavior, behavior that primarily occurs in the family environment, and underserved regions. Brown and colleagues describe caregiver-implemented FAs with remote coaching as a default move when behavior is too infrequent to evoke in clinic, and explicitly endorse baited environments with safe items for ingestion-related topographies Brown et al. (2025). Togashi's training package combining asynchronous instruction with brief telehealth coaching shows that even procedural-integrity training can be done remotely: asynchronous content gets practitioners to mastery-level knowledge, and a small number of brief telehealth BST sessions take them the rest of the way to ≥80% procedural integrity on TBFA, including in settings where in-person expert access is limited (Togashi, 2025). The cultural-responsiveness and empathic-interview BST work fits home-based practice especially well, where caregivers are central data sources and interview quality directly determines what the FA can later test Gatzunis et al. (2023).

Residential and adult disability services

Residential settings concentrate three problems: severe topographies, dispersed staff, and inconsistent FBA throughput. The corpus addresses each. Brown and colleagues' decision-making review covers the kinds of low-rate and undifferentiated severe behavior typical in adult ID services, including caregiver-implemented FAs and protocol modifications that increase observation opportunities without escalating risk Brown et al. (2025). Sellers and colleagues' multiple-baseline study demonstrates that a remote performance-management package — goal-setting, self-monitoring, weekly emailed feedback graphs, and remote supervision — can lift TBFA completion from near-zero to 20+ per supervisor per month across a six-supervisor multi-state agency, with no on-site travel Sellers et al. (2019). Griffith and colleagues' self-instruction-plus-webinar package gives the staffing layer underneath: entry-level technicians can reach high TBFA procedural integrity in roughly 2 hours of training Griffith et al. (2020). The risk-screening tool is particularly valuable here because residential populations sit disproportionately in the higher-risk bands Deochand et al. (2020).

• Treating closed-ended checklists as a sufficient indirect assessment. Open-ended probes generate the idiosyncratic antecedents and reinforcers that closed-ended items quietly hide, and the FA you build downstream can only test what your interview surfaced Fryling & Baires (2016).
• Agreeing on the function category and stopping there. Categorical agreement on PFA-style assessments is high (~97%); specific-feature agreement on the exact EO, response form, and reinforcer is lower and inconsistent. Two clinicians can both say "escape" and run different IISCAs unless those features are scripted in writing Rajaraman et al. (2022).
• Running a tangible condition by default. When the interview gives no tangible hypothesis, including the tangible test inflates assessment time, increases risk exposure, and adds noise; efficiency reviews specifically identify it as the highest-yield condition to drop Saini et al. (2020).
• Skipping the eco-behavioral scan in school FBA. Failing to capture seating, instructional format, transition timing, and existing teacher-attention rates produces hypotheses the resulting BIP cannot survive in the actual classroom Kestner et al. (2019).
• Accepting an undifferentiated FA as definitive. Brown and colleagues are explicit: extend, replicate across caregivers and settings, and use latency-based metrics before concluding the behavior is automatically reinforced or "not function-based." Undifferentiated outcomes are a cue to extend assessment, not to abandon it Brown et al. (2025).
• Ignoring sensitivity and bias. Visual differentiation is necessary but not sufficient; documented sensitivity-to-EO and observer-bias estimates protect the functional interpretation when teams or payers push back (Allen et al., 2026).
• Treating consequence fidelity as a "training phase" issue only. Post-mastery fidelity decay is real and fast; supervision plans need ongoing consequence-fidelity probes, not just antecedent-step checks (Jones et al., 2026).
• Inferring verbal-behavior function from topography. Especially in ACT-style or language-based interventions, the function of a self-rule or defusion utterance has to be tested in real time; assuming form predicts function is the same mistake the field stopped making about overt problem behavior decades ago Sandoz et al. (2022).
• Deploying FA without a written risk screen. A documented risk score creates both a clinical safeguard and the audit trail teams will want when something unexpected happens during a session Deochand et al. (2020).

• Suspected medical or biological substrate. Behavior with possible pain, sleep, GI, or seizure involvement; pica with non-food items that could cause injury; or any topography producing tissue damage during prior assessment attempts. Refer for medical evaluation before any experimental FA and document the consult.
• High-risk score on the FA risk-assessment checklist that does not drop after environmental safeguards. Move to a specialist BCBA team with FA-safety training or an inpatient behavior unit Deochand et al. (2020).
• Persistently undifferentiated FA across replications and extensions. When extension across caregivers, settings, and condition lengths still does not produce interpretable differentiation, refer for external peer review or specialist consultation rather than committing to a function-based plan that the data don't support Brown et al. (2025).
• Automatically reinforced self-injury with low LOD that does not resolve with sensory enrichment trials. Refer to a setting with capacity for extended alone sessions, latency analyses, and matched-stimulation programming Falligant & Hagopian (2020) Hagopian et al. (2018).
• Resource ceiling. When neither in-person nor telehealth training brings staff to ≥80% TBFA procedural integrity after two cycles, refer the case to a regional consultation team rather than running an underpowered FA in-house (Togashi, 2025) Griffith et al. (2020).
• Active psychiatric crisis or imminent safety concern that overwhelms behavioral assessment timelines. Behavioral assessment is not the right first move; refer to licensed mental-health crisis services and resume FBA after stabilization.

The honest read of the corpus is that the operational, practitioner-facing claims on this page — about format selection, training, ecology, and risk — sit on solid systematic-review and SCED evidence, while the comparative-effectiveness layer is thin Call et al. (2024) Corr et al. (2025). Call and colleagues' study is the closest existing test of FBA-with-FA versus FBA-without-FA at the group level, and it found broadly equivalent outcomes with only modest concordance, which leaves open the question of which subgroups actually require the FA layer Call et al. (2024). Prospective head-to-head trials of brief FA versus IISCA versus descriptive-only FBA, paired with common downstream treatment packages, would clarify this — and would tell us whether the LOD biomarker for automatic reinforcement holds up prospectively beyond Falligant and Hagopian's retrospective 103-case re-analysis Falligant & Hagopian (2020).

The reliability picture also needs deeper work. Specific-feature agreement during PFA-style interviews and IISCAs is materially lower than categorical agreement, and we do not yet know how much that drift actually degrades treatment outcomes when multiple staff implement the resulting plan; a study explicitly linking specific-feature agreement to treatment effect sizes would change how we write behavior plans Rajaraman et al. (2022). Sensitivity and bias metrics derived during FA are promising but largely procedural — embedding them in standard FA software with usability testing would convert a methodological recommendation into a routine quality-control layer (Allen et al., 2026).

Telehealth FA training has been demonstrated with small samples in resource-scarce regions, including Togashi's four-practitioner Japanese cohort; larger replications across U.S. and international contexts are needed to confirm that the asynchronous-plus-coaching architecture scales (Togashi, 2025). Similarly, the BST work for empathic and culturally responsive interviewing has only been validated through role-play probes with graduate students; field generalization to in-home interviews with diverse caregivers is the obvious next step Gatzunis et al. (2023). The classroom-ecology argument needs a standardized brief eco-behavioral inventory; right now, every team improvises one Kestner et al. (2019). And the post-mastery fidelity-decay finding — convincing inside a translational analogue — needs a longitudinal field replication that tracks consequence delivery accuracy across months in actual programs (Jones et al., 2026). Each of these gaps is tractable; none require the field to invent new methods.

1. Lead with an open-ended interview, not a checklist. Closed-ended forms are efficient but quietly miss idiosyncratic antecedents and reinforcers; open-ended probes generate the candidate variables your FA or IISCA can later test Fryling & Baires (2016).
2. Always document baseline classroom ecology in school FBAs. Seating, instructional format, transition timing, and teacher attention rates are existing contingencies; missing them produces functions you can't actually treat in the room they live in Kestner et al. (2019).
3. Run a 15-item FA risk screen before any experimental session. Use the score to triage cases into on-site FA, caregiver-implemented FA with remote coaching, or referral Deochand et al. (2020).
4. Default to brief or trial-based FA in schools. They produce interpretable functions inside typical school logistics; the evidence base for them is now substantial enough to make the multi-element FA the exception rather than the rule Nesselrode et al. (2022).
5. Use the IISCA when the case warrants a single, trauma-informed test/control comparison. Two clinic visits is realistic; the format compresses idiosyncratic contingencies into one synthesized test Jessel et al. (2020) (Jessel et al., 2024).
6. Drop the tangible test by default. Include it only when the interview gives a tangible hypothesis. This is the single highest-yield FA-shortening lever in the literature Saini et al. (2020).
7. Add a brief demand assessment when escape is suspected. Varying difficulty, amount, and novelty calibrates the EO before any FA condition runs and reduces unnecessary test conditions later Avery & Akers (2021).
8. Mine your descriptive ABC data for antecedent conditional probabilities. Done carefully, they can match FA-derived function in a single case; at minimum they sharpen which conditions to test next Tereshko et al. (2024).
9. Script the exact EO, response form, and reinforcer in the BIP. Categorical agreement masks specific-feature drift; written specifics are how you keep the second clinician's IISCA looking like the first Rajaraman et al. (2022).
10. For automatically reinforced self-injury, compute LOD before selecting treatment. Values around or above 0.64 support reinforcement-alone FCT; lower values mean you likely need matched stimulation or extended conditions first Falligant & Hagopian (2020) Hagopian et al. (2018).
11. Do not accept an undifferentiated FA as definitive. Extend conditions, replicate across caregivers and settings, control ambient stereotypy in the control condition, and document sensitivity/bias before changing the hypothesis Brown et al. (2025) Saini et al. (2020) (Allen et al., 2026).
12. For low-rate or hard-to-evoke behavior, move the FA to the caregiver and the natural environment. Remote coaching, baited rooms with safe items, and latency metrics produce interpretable data without the clinic-evocation problem Brown et al. (2025).
13. Train TBFA the cheap, scalable way. A self-instruction package plus one short group webinar — or asynchronous CBI plus brief telehealth BST — gets entry-level staff to roughly 90% procedural integrity in about 2 hours, with maintained gains at follow-up Griffith et al. (2020) (Togashi, 2025).
14. Treat FBA throughput as a managed variable. Goal-setting, self-monitoring, weekly graphical feedback, and remote supervision can move agency TBFA volume from near-zero to 20+ per supervisor per month Sellers et al. (2019).
15. Build consequence-fidelity probes into maintenance. Post-mastery drift erodes accurate responding fast; antecedent-only fidelity checks miss it (Jones et al., 2026).
16. For ACT and language-based work, embed FA logic inside the session. Evoke a contextual stimulus, observe the verbal behavior shift, manipulate the next stimulus, and adjust — do not assume topography reveals function Sandoz et al. (2022).

Primary research synthesized in this guide. DOIs link to the original source.

• Pollack, M. S., Lloyd, B. P., Doyle, L. E., Santini, M. A., & Crowell, G. E. (2024). Are function‑based interventions for students with emotional/behavioral disorders trauma informed? A systematic review. Behavior Analysis in Practice, 17, 709–726. https://doi.org/10.1007/s40617-023-00893-y https://doi.org/10.1007/s40617-023-00893-y
• Rajaraman, A., Hanley, G. P., Gover, H. C., Ruppel, K. W., & Landa, R. K. (2022). On the Reliability and Treatment Utility of the Practical Functional Assessment Process. Behavior Analysis in Practice, 15(3), 815-837. https://doi.org/10.1007/s40617-021-00665-6 https://doi.org/10.1007/s40617-021-00665-6
• Togashi, K. (2025). Training in trial-based functional analysis via computer-based instruction and behavioral skills training. Behavior Analysis in Practice. https://doi.org/10.1007/s40617-025-01136-y https://doi.org/10.1007/s40617-025-01136-y
• Nesselrode, R., Falcomata, T. S., Hills, L., & Erhard, P. (2022). Functional Analysis in Public School Settings: A Systematic Review of the Literature. Behavior Analysis in Practice, 15(3), 958-970. https://doi.org/10.1007/s40617-022-00679-8 https://doi.org/10.1007/s40617-022-00679-8
• Sandoz, E. K., Gould, E. R., & DuFrene, T. (2022). Ongoing, Explicit, and Direct Functional Assessment is a Necessary Component of ACT as Behavior Analysis: A Response to Tarbox et al. (2020). Behavior Analysis in Practice, 15(1), 33-42. https://doi.org/10.1007/s40617-021-00607-2 https://doi.org/10.1007/s40617-021-00607-2
• Mayo, M. R., & Hoffmann, A. N. (2024). A survey of the state of the field of applied behavior analysis in Vermont. Behavior Analysis in Practice, 17, 581–600. https://doi.org/10.1007/s40617-023-00901-1 https://doi.org/10.1007/s40617-023-00901-1
• Brown, K. R., Helvey, C. I., Kranak, M. P., & Lavin, A. (2025). Functional Analysis Decision-Making Considerations. Behavior Analysis in Practice, 18(4), 1237-1254. https://doi.org/10.1007/s40617-025-01057-w https://doi.org/10.1007/s40617-025-01057-w
• Saini, V., Fisher, W. W., Retzlaff, B. J., & Keevy, M. (2020). Efficiency in functional analysis of problem behavior: A quantitative and qualitative review. Journal of Applied Behavior Analysis, 53(1), 44-66. https://doi.org/10.1002/jaba.583 https://doi.org/10.1002/jaba.583
• Griffith, K. R., Price, J. N., & Penrod, B. (2020). The Effects of a Self-Instruction Package and Group Training on Trial-Based Functional Analysis Administration. Behavior Analysis in Practice, 13(1), 63-80. https://doi.org/10.1007/s40617-019-00388-9 https://doi.org/10.1007/s40617-019-00388-9
• Snyder, S. M., Huber, H., Hornsby, T., & Leventhal, B. (2024). Overlapping training and roles: An exploration of the state of interprofessional practice between behavior analysts and school psychologists. Behavior Analysis in Practice, 17, 880–892. https://doi.org/10.1007/s40617-023-00904-y https://doi.org/10.1007/s40617-023-00904-y
• Falligant, J. M. & Hagopian, L. P. (2020). Further extensions of precision medicine to behavior analysis: A demonstration using functional communication training. Journal of Applied Behavior Analysis, 53(4), 1961-1981. https://doi.org/10.1002/jaba.739 https://doi.org/10.1002/jaba.739
• Kestner, K. M., Peterson, S. M., Eldridge, R. R., & Peterson, L. D. (2019). Considerations of Baseline Classroom Conditions in Conducting Functional Behavior Assessments in School Settings. Behavior Analysis in Practice, 12(2), 452-465. https://doi.org/10.1007/s40617-018-0269-1 https://doi.org/10.1007/s40617-018-0269-1
• Tereshko, L. M., Weiss, M. J., Ross, R. K., Harper, J. M., & Keane, D. (2024). A component analysis of ABC assessments as demonstrated through function based interventions. Behavioral Interventions, 39(3). https://doi.org/10.1002/bin.2009 https://doi.org/10.1002/bin.2009
• Sellers, T. P., Clay, C. J., Hoffmann, A. N., & Collins, S. D. (2019). Evaluation of a Performance Management Intervention to Increase Use of Trial-Based Functional Analyses by Clinicians in a Residential Setting for Adults with Intellectual Disabilities. Behavior Analysis in Practice, 12(2), 412-417. https://doi.org/10.1007/s40617-018-00276-8 https://doi.org/10.1007/s40617-018-00276-8
• Fryling, M. J. & Baires, N. A. (2016). The Practical Importance of the Distinction Between Open and Closed-Ended Indirect Assessments. Behavior Analysis in Practice, 9(2), 146-151. https://doi.org/10.1007/s40617-016-0115-2 https://doi.org/10.1007/s40617-016-0115-2
• Gatzunis, K. S., Weiss, M. J., Ala’i-Rosales, S., Fahmie, T. A., & Syed, N. Y. (2023). Using Behavioral Skills Training to Teach Functional Assessment Interviewing, Cultural Responsiveness, and Empathic and Compassionate Care to Students of Applied Behavior Analysis. Behavior Analysis in Practice. https://doi.org/10.1007/s40617-023-00794-0 https://doi.org/10.1007/s40617-023-00794-0
• Jessel, J., Fruchtman, T., Raghunauth‑Zaman, N., Leyman, A., Lemos, F. M., Costa Val, H., Howard, M., & Hanley, G. P. (2024). A two step validation of the performance‑based IISCA: A trauma‑ informed functional analysis model. Behavior Analysis in Practice, 17, 727–745. https://doi.org/10.1007/s40617-023-00792-2 https://doi.org/10.1007/s40617-023-00792-2
• Call, N. A., Bernstein, A. M., O'Brien, M. J., Schieltz, K. M., Tsami, L., Lerman, D. C., Berg, W. K., Lindgren, S. D., Connelly, M. A., & Wacker, D. P. (2024). A comparative effectiveness trial of functional behavioral assessment methods. Journal of Applied Behavior Analysis, 57(1), 166-183. https://doi.org/10.1002/jaba.1045 https://doi.org/10.1002/jaba.1045
• Deochand, N., Eldridge, R. R., & Peterson, S. M. (2020). Toward the Development of a Functional Analysis Risk Assessment Decision Tool. Behavior Analysis in Practice, 13(4), 978-990. https://doi.org/10.1007/s40617-020-00433-y https://doi.org/10.1007/s40617-020-00433-y
• Corr, F., Rispoli, M., & Welker, N. P. (2025). A Mega-Review of Functional Communication Training for Students with Disabilities in Educational Settings. Journal of Behavioral Education. https://doi.org/10.1007/s10864-025-09598-4 https://doi.org/10.1007/s10864-025-09598-4
• Allen, A. E., Bridges, K. G., Pizarro, E. M., & Morris, S. L. (2026). Comparing methods of evaluating sensitivity to common establishing operations and bias toward challenging behavior. Journal of Applied Behavior Analysis, 59(1), e70046. https://doi.org/10.1002/jaba.70046 https://doi.org/10.1002/jaba.70046
• Hagopian, L. P., Rooker, G. W., & Yenokyan, G. (2018). Identifying predictive behavioral markers: A demonstration using automatically reinforced self‐injurious behavior. Journal of Applied Behavior Analysis, 51(3), 443-465. https://doi.org/10.1002/jaba.477 https://doi.org/10.1002/jaba.477
• LaMarca, V. J., & LaMarca, J. M. (2024). Using the ADDIE model of instructional design to create programming for comprehensive ABA treatment. Behavior Analysis in Practice, 17, 371–388. https://doi.org/10.1007/s40617-024-00908-2 https://doi.org/10.1007/s40617-024-00908-2
• Jones, L., Brand, D., Bensemann, J., Heinicke, M. R., Penrod, B., & Burlison, S. (2026). A translational approach to investigating the effects of consequence-based procedural fidelity errors postmastery. Journal of Applied Behavior Analysis, 59(1), e70042. https://doi.org/10.1002/jaba.70042 https://doi.org/10.1002/jaba.70042
• Avery, S. K. & Akers, J. S. (2021). The Use of Demand Assessments: A Brief Review and Practical Guide. Behavior Analysis in Practice, 14(2), 410-421. https://doi.org/10.1007/s40617-020-00542-8 https://doi.org/10.1007/s40617-020-00542-8
• Jessel, J., Hanley, G. P., & Ghaemmaghami, M. (2020). On the Standardization of the Functional Analysis. Behavior Analysis in Practice, 13(1), 205-216. https://doi.org/10.1007/s40617-019-00366-1 https://doi.org/10.1007/s40617-019-00366-1
• Kodak, T. & Halbur, M. (2021). A Tutorial for the Design and Use of Assessment-Based Instruction in Practice. Behavior Analysis in Practice, 14(1), 166-180. https://doi.org/10.1007/s40617-020-00497-w https://doi.org/10.1007/s40617-020-00497-w