Treatment Integrity Reporting in Behavior Analysis in Practice 2008–2019

Falakfarsa et al. (2022) read every article published in Behavior Analysis in Practice from 2008 through 2019. They counted how many papers told readers whether the intervention was done correctly. They also noted how many reported interobserver agreement, a separate check on data quality.

The team screened 334 articles. They kept any study that tested a behavioral intervention and could have measured treatment integrity.

Only 46.8 % of studies included treatment integrity data. That is less than half. In contrast, 94.2 % reported interobserver agreement. Nearly half of the articles gave no clue about how well the intervention was actually carried out.

When integrity was measured, the risk of implementation errors was high because details were missing.

Castañe et al. (1993) saw the same gap thirty years earlier. In 1980-1990 child studies in JABA, only 16 % checked treatment integrity. Falakfarsa et al. (2022) show the field has improved, but the jump is modest. We still publish without knowing if the treatment was done right.

Thillainathan et al. (2024) give a real-world example. Their residential program hit 84 % integrity and produced big behavior gains. That paper would meet Falakfarsa’s inclusion rules, yet many BAP studies fail to report such numbers.

Aherne et al. (2019) tracked staff integrity after training. Scores slid without self-monitoring. Their data point is exactly what Falakfarsa says is missing from most BAP papers: ongoing integrity checks.

If you submit a paper, add a short integrity section. State who watched, how often, and the percent of steps done correctly. It takes one paragraph and shields your results from the “maybe they didn’t really do the intervention” doubt. Your readers—and future meta-analysts—will thank you.