FA Efficiency vs Validity: Does a Faster Functional Analysis Lose Accuracy?

The first thing people say when they hear about a 15-minute FA is that shorter must mean sloppier, and Dr. Joshua Jessel's answer is that the thing you are trying to close in a functional analysis is not a clock but an interpretive leap, the gap between what you do not yet know about a kid's problem behavior and what you need to know to treat it. More time does not mean more control. A BCBA (board certified behavior analyst) running a 3-hour standard FA (functional analysis, the test that proves what is driving problem behavior) can end up with the same level of confidence as one who ran a clean IISCA (interview-informed synthesized contingency analysis, the short version Jessel helped build) in 15 minutes. What closes the gap is the pattern in the data, not the duration of the test. This page walks through where that idea comes from, what the literature actually shows, and how to defend a short FA when a supervisor or a parent asks why you wrapped up so fast.

The old assumption: longer FA means stronger control#

Most BCBAs were taught the standard FA from Iwata 1982 and the 1994 reprint. Five components. Multiple test conditions. Each function tested in its own block. A play control to compare each one against. That model worked. It also took time. Jessel's team pulled every standard FA published in the last ten years and added up the session minutes. The bare-bones average came out to about three hours per case. That number does not include setup, staff training, or rerunning a condition because the first pass came back undifferentiated. So the question stopped being "is this method valid" and started being "is this method usable in a clinic with 30 kids on caseload and no spare staff."

The old assumption was that the long version was the safe version. That if you were going to cut anything, you were cutting accuracy. Jessel pushed on that assumption and asked a more useful question. What is the analysis actually doing, and can we get the same answer with fewer minutes of exposure to a kid being upset?

What Jessel's 10-year literature review actually showed#

Here is where the data flips the default belief. Jessel's team scored the level of control in every ISCA and every standard FA from the last ten years of the published record. Level of control means how cleanly the test condition pulled problem behavior and how cleanly the control condition stopped it. Pristine differentiation is the top tier: no overlap and no problem behavior during the control. That is the result every BCBA wants on the graph before moving to treatment.

we evaluated the control obtained during every ISCA and every standard in the past 10 years of the literature and found that the majority of these ISCA applications had pristine differentiation From the talk — Dr. Joshua Jessel

So the short version did not pay a tax for being short. In the published record, it more often hit pristine differentiation than the long version did. That is the headline a lot of clinicians have not heard yet. Jessel's read of his own data is blunt:

something faster can be more accurate and there are really no benefits to taking a longer time and drawing out the assessment period From the talk — Dr. Joshua Jessel

The 'interpretive leap': what really closes the uncertainty gap#

The reframe Jessel uses is the interpretive leap. Picture two dots. One dot is what you know at intake, before you have seen any data. The other dot is what you need to know to write a treatment plan. The work of an FA is closing the gap between those two dots. The question is what closes that gap. A long session, or a clean pattern.

Closing the gap or getting from here to here is not dependent on some standardized measure of time. Getting from the large gap, essentially when you first meet the child and know little about their behavior, to the small gap where you have an empirical demonstration of control, does not require long sessions with extended exposure to problem behavior. From the talk — Dr. Joshua Jessel

If you can turn problem behavior on by removing the reinforcer and turn it off by giving it back, and you can do that across a handful of repeated cycles, you have the same proof a long FA gives you. The mechanism is not minutes on a clock. It is replicated change. That is the heart of why a 5-minute or 3-minute session can carry the same evidence as 25 minutes.

When efficiency and validity move in the same direction#

Most pages frame efficiency and validity as a tradeoff. Jessel's data says they usually move together. Here is why. The longer an FA runs, the more time a kid spends being evoked into problem behavior on purpose. That raises three real risks. First, the kid escalates into harder topographies because the test keeps running. Second, the BCBA waits longer to start the part that actually helps, which is treatment. Third, staff burn out and start cutting corners on procedure, which is where most validity threats live anyway.

A short, well-run IISCA flips all three. Less evoked behavior. Faster path to treatment. Cleaner procedural fidelity because the staff are fresh. None of that is a soft sell. The literature review backs it up. Shorter FAs in the published record showed higher rates of pristine differentiation, not lower. So when you choose efficiency in this case, you are not paying with rigor. You are paying with the right currency, which is exposure time.

When a longer FA actually IS the right call#

The honest version of this story is that the short FA is not always the answer. Jessel says efficiency is a construct, and what counts as efficient depends on the case. Three situations push you back toward a longer or more structured FA.

One. You ran an IISCA and got weak or no control on the rubric. That means the test condition and the control overlapped, or problem behavior kept showing up during the control. You go back to the drawing board. Maybe the synthesized contingency was wrong. Maybe the caregiver interview missed a piece. Add a session. Try a different synthesis.

Two. The kid has more than one likely function, and the family or team needs them parsed out for a school plan or a placement decision. A single test condition does not give you that. A multi-element FA does.

Three. The target behavior is so severe that even a 30-minute test is too much exposure. In that case, the answer is not a longer FA. It is a performance-based ISCA that ends as soon as you have five clean cycles of evoked and eliminated behavior, often in 10 minutes or less, with non-dangerous precursors used in the open contingency class so the kid never has to escalate to the worst form.

The rule is not "always go short." The rule is "let the level-of-control rubric, not the clock, tell you when you are done."

How to defend a short FA in supervision or a clinical review#

Most pushback on a short FA shows up in three places. A supervisor reviewing a case. A parent asking why this was so fast. An insurance reviewer asking why you billed for an assessment that took 15 minutes. Here is what to say in each one.

For supervision. Bring the within-session graph. Point to the reinforcer-absent intervals where problem behavior spiked and the reinforcer-present intervals where it dropped to zero. Show the level-of-control rating. If it is strong, the discussion is over. Strong means no overlap and no problem behavior in the control. There is nothing a longer session would add.

For the parent. Explain that the goal of the test is to make their kid upset on purpose, and you want to do that for the shortest amount of time that still proves what is going on. Most parents lean in once they hear that framing. The short FA is not a corner cut. It is a kindness.

For the insurance reviewer. The CEU and the literature both back the IISCA up. You can cite Jessel's 10-year review and note that the published record shows shorter analyses hitting pristine differentiation at higher rates than long ones. The billable activity is not the minutes of the test. It is the time spent in the interview, the observation, the analysis, and the write-up. The test is the middle step. Short test, full clinical work.

Frequently asked questions#

Does a shorter FA mean I'm sacrificing scientific rigor?

No. Rigor in an FA comes from replicated change across the test and control conditions, not from how long the test ran. Jessel's 10-year review found that ISCAs hit pristine differentiation at higher rates than standard FAs did. If your within-session graph shows clean turning on and turning off across repeated cycles, you have the same evidence a 3-hour standard FA gives you.

What is the 'interpretive leap' Jessel talks about?

The interpretive leap is the gap between what you know at intake and what you need to know to treat. The FA closes that gap. Jessel's point is that closing the gap is about pattern, not time. A few repeated cycles of behavior turning on when the reinforcer is removed and turning off when it is given back is the proof. A long session with no pattern does not close anything.

When should I run a longer FA instead of an IISCA?

Three cases. One, your IISCA came back with weak or no control on the rubric. Two, you need to parse multiple functions for a school or placement decision. Three, you tried an IISCA and the kid escalated past what your staff can safely manage. In that last case, the answer is usually a performance-based ISCA with non-dangerous precursors, not a longer standard FA.

How do I explain my FA choice to a parent or supervisor?

For a parent, say the test is designed to make their kid upset on purpose, and your job is to do that for the shortest amount of time that still proves what is going on. For a supervisor, bring the within-session graph and the level-of-control rating. If the rating is strong, no overlap and no problem behavior in the control, the case is closed.

Watch the full talk#

If this lined up with the FA cases you are sitting on right now, the full CEU walks through Jessel's level-of-control rubric, the single-session ISCA, and the performance-based ISCA with the paper data sheet you can download. It is free and counts for one general learning credit.

Watch the full CEU on FA efficiency and the interpretive leap