Forensic Applied Behavior Analysis

So you want to do FABA. Here is how it actually works.

Forensic Applied Behavior Analysis is what happens when you apply ABA principles inside a legal proceeding. It is not a separate discipline with its own textbook. It is Board Certified Behavior Analysts testifying, producing reports, and conducting assessments that are scrutinized under rules of evidence. The BACB has a Behavior Analyst Certification Board Ethics Code for this, but the real work is figuring out how to make your data hold up when a defense attorney is looking for reasons to exclude it. The first thing most people get wrong is thinking you need fancy software or a specialized degree. You do not. You need a valid BCBA or BCaBA credential, knowledge of the relevant jurisdiction's rules of evidence, and the ability to explain operational definitions to a jury that has never seen a scatterplot in their life. I have seen analysts with 15 years of clinic experience shut down in deposition because they could not define "topography" without using jargon. Start there. You will need to learn the difference between opinion testimony and factual testimony. Under Daubert and Frye standards depending on your state your methodology has to be testable and generally accepted. ABA is generally accepted. Your specific assessment tool might not be. That distinction matters when you are laying the groundwork.

Most forensic cases involve one of three things: child custody evaluations where parental behavior is in question, competency assessments in civil commitment proceedings, or workplace misconduct investigations where behavioral patterns are the central evidence. There are others but those cover about eighty percent of what I see. Here is a practical workflow. You start by identifying the referral question and mapping it to observable, measurable behavior. Vague questions like "is this person a good parent" get you nowhere. Specific questions like "does the respondent engage in physical aggression toward the child more than once per week during supervised visits" you can answer with data. I had a custody case where the referring attorney asked whether the father was "neglectful." That is not a behavioral question. It is a value judgment. I rewrote the referral question to focus on observable neglect indicators: failure to provide meals within reasonable timeframes, missed medical appointments, lack of supervision leading to identifiable risk events. The resulting report was admissible. The original one would have been stricken.

Assessment Methods That Actually Hold Up

Standardized tools like the Functional Assessment Interview or the Questions About Behavioral Function scale can be used forensically but they come with caveats. Self-report data is unreliable when someone has something to gain or lose. In a custody evaluation I worked on, both parents completed the QABF and produced nearly identical scores despite making completely contradictory claims about the child's behavior. The discrepancy was not in the scoring. It was in the honesty of the respondents. I triangulated self-report with direct observation and archival records before including it in my report. Without that triangulation the QABF scores alone would have been weak evidence. Direct observation is your strongest tool but it is also the most resource-intensive. You need interobserver agreement checks, preferably at ninety percent or above. If you are doing single-observer data collection in a forensic context you are leaving yourself open to a challenge. I build IOA into my protocols from day one. It adds about twenty minutes per observation session but it saves you three hours in cross-examination prep. Archival data is where most people stumble. School records, medical records, prior incident reports. These are gold if they are properly authenticated and they are inadmissible if they are not. I use a chain-of-custody log for every document. I track who produced it, when, and through what channel. A defense attorney will attack the foundation before they attack the content. Make sure your foundation is solid.

Here is something I wish more analysts understood: functional analysis in a forensic setting is different from clinical functional analysis. In the clinic you might run a full FA with attention, escape, alone, and tangible conditions. In a forensic case you are often limited by access, by ethics, and by the fact that you cannot intentionally evoke dangerous behavior for the sake of data. I worked on a case involving a juvenile in a group home where the attorney wanted a full functional analysis to support a placement decision. The risk of inducing severe aggression was too high. I adapted by using a truncated contingency-based assessment focused on the two most probable maintaining variables. It gave us actionable results without crossing the line into unethical procedure. The report still held up in hearing.

Report Writing and Testimony

Your report is your primary exhibit. Everything else flows from it. I structure mine around the referral question, the methods used, the data collected, the analysis, the conclusions, and the limitations. Every claim is tied to a specific data point. No hedges. No "it appears that." Either the data supports the conclusion or it does not. Depositions are where most analysts panic. They over-explain, they volunteer information, they apologize for their methodology. I give myself three rules: answer only the question asked, do not speculate beyond your data, and never concede a point unless it is actually true. I had a case where the opposing attorney tried to get me to admit that ABA is "not a hard science." I corrected the record. ABA is an empirical discipline. It may not have the same level of objectivity as physics but calling it "not a hard science" in a forensic context is a deliberate mischaracterization designed to undermine credibility. I did not take the bait. When you testify, speak slowly. Juries and judges process information at a different pace than you do. If you are rushing through technical language you will lose them and the opposing counsel will exploit that confusion. I once watched a particularly effective cross-examination where the attorney simply asked the analyst to repeat three different technical terms and then pointed out that the analyst could not define any of them clearly under oath. The analyst had used those terms freely in the report. It was damaging.

Where This Work Breaks Down

Forensic ABA has real limitations. It cannot establish causation with the same certainty as a laboratory study. Behavioral data shows correlation and functional relations not proof beyond a reasonable doubt. When a case requires that standard forensic ABA is not the right tool. I have declined cases where the legal question demanded a level of certainty that behavioral assessment simply cannot provide. It is better to say no upfront than to overreach and destroy your credibility. Another limitation is the scarcity of forensic-trained behavior analysts. Most BCBAs have no trial experience. If you are getting into this you need to invest in training outside of your standard CE requirements. I take courses on evidence law, deposition technique, and expert witness preparation. It is not part of the BACB task list but it is essential for doing this work competently. Fee structures are another issue. Forensic work pays differently than clinical work. Some attorneys expect steep rates. Others try to lowball you. Know your market value and do not undersell yourself. At the same time, do not price yourself out of cases where you can do meaningful work. I charge a flat fee per case that covers evaluation, report writing, and up to four hours of testimony. Anything beyond that is billed hourly. It keeps things predictable for everyone involved.

If you are looking to get started the best move is to find a mentor who has actually testified. Read published forensic ABA reports. Study the ones that survived Daubert challenges and the ones that did not. The difference is usually in the methodology section, not the conclusions. Pay attention to that.