Prevalence of multiply controlled problem behavior.

The team read 521 published functional analyses. They counted how many showed “multiple control,” meaning behavior happened in two or more test conditions. They also looked at how authors grouped, or “collapsed,” different forms of behavior before they ran the stats.

About one in six cases looked like several reinforcers kept the problem going. Yet almost 90 % of those files had squashed all topographies into one number first. The pattern hints that the math, not the child, created the “mixed” answer.

Hernández et al. (2018) watched escape sessions day-by-day. They saw most aggressive forms drop out while one stayed strong—proof that uncollapsed tracking changes the story.

Fernandez et al. (2024) ran 116 new FAs and found true acquisition is rare. Their low numbers back the idea that many “multi-function” charts are measurement noise, not learning.

Peters et al. (2013) give a fix: a five-minute alone probe safely rules out automatic reinforcement first, so you can shorten assessment before any collapsing temptation starts.

Before you write “dual control” in the report, list each response form separately. Run short, clean conditions like Peters et al. (2013) to screen out obvious single functions. Then graph each topography alone; if only one survives, you just saved hours of needless treatment splitting.