From Percentages to Precision: Using Response Rates to Advance Analyses of Procedural Fidelity

St. D'Incal et al. (2025) wrote a how-to paper, not an experiment. They looked at how BCBAs usually track fidelity and said, 'Add speed counts.'

The authors explain why percent-correct scores can hide drift. They give formulas for turns-per-minute and responses-per-minute so you catch slow, sloppy delivery.

The paper shows two made-up graphs. One uses only percent fidelity and looks perfect at 100%. The same data with a rate line shows the teacher waited ten seconds between prompts.

Their point: rate metrics spotlight timing errors that percentages miss.

Abuin et al. (2026) ran an experiment that backs this up. They found 50% fidelity delivered faster beat 100% slow fidelity. The rate lens the target paper pushes would have spotted that speed edge.

Faso et al. (2016) offered a success-rate rule for choosing interventions. St. Peter adds a rate rule for judging how well you carry those interventions out.

Stolz (1977) warned that old studies rarely checked reliability each phase. The new paper widens the same worry to fidelity: a flat percent agreement can still mask slow, uneven delivery.

Next time you score fidelity, add one quick column: count how many learning trials occur in each five-minute block. If the count drops but percent correct stays high, you have silent drift to fix on the spot.