Impact of treatment integrity on intervention effectiveness.

Rojahn et al. (2012) wrote a narrative review. They looked at past behavior-analytic studies. The authors asked: Are researchers checking if the intervention was done as planned?

The paper argues we must measure treatment integrity. Without it, we cannot trust that the intervention caused any change.

The review found a gap. Many studies test an intervention but skip integrity checks. The authors say this weakens every conclusion. If you do not know the dose was right, you cannot say the drug worked.

Laraway et al. (2019) extends the same worry. They add pre-registration and data sharing to the wish list. Both papers push for tighter methods, just with different labels.

LaPoint et al. (2025) also extends the call. They ask autism journals to force trial registration. The 2012 paper asked for integrity logs; the 2025 team asks for public protocols. The goal is identical: stop hidden slop.

Jonsson et al. (2016) looks at autism social-skills RCTs. They find external-validity details are missing. This pairs with Rojahn et al. (2012): if you skip integrity and you skip context, the evidence is doubly blind.

Next time you run a session, tape two minutes. Score one integrity item, such as ‘delivered prompt as written.’ If integrity is low, adjust before you graph the child’s data. One quick check turns a shaky case study into something you can trust.