Experimental functional analyses for challenging behavior: a study of validity and reliability.

The team ran experimental functional analyses on adults with intellectual disability. They wanted to know if different ways of reading the same data would give the same answer.

They also ran the same FA again two weeks to three months later. They checked if the second round gave the same function as the first.

Different experts looking at the same graphs picked different functions. One viewer might say escape, another might say attention.

When they repeated the FA weeks later, the function often flipped. The same person who once showed escape now showed attention.

Matson et al. (2013) seems to disagree. They found FA results stayed the same when they swapped therapists or moved from clinic to home. The key difference is time. Same-day changes in staff or room did not shake the function, but weeks of real life did.

Kahng et al. (1999) gives a same-year fix. They showed that watching minute-by-minute shifts inside one session can clear up muddy data. If T et al. had used this within-session rule, their second test might have matched the first.

Tassé et al. (2013) later cataloged hundreds of tweaks clinicians add when the standard FA is unclear. Their list of idiosyncratic antecedents and consequences is a direct answer to the poor reliability T et al. uncovered.

Your FA snapshot can age fast. Before you write a behavior plan, rerun the analysis if the last one is more than a couple of weeks old. Add within-session checks or idiosyncratic conditions if the data look fuzzy. Treat the first FA as a draft, not a verdict.