An experimental analysis of reading interventions: generalization across instructional strategies, time, and passages.

The team tested three reading packages on one elementary student.

They used an alternating-treatments design. Each day the child got either a performance-based, skill-based, or combined fluency lesson.

Sessions lasted one school period. The goal was to see which package lifted oral reading speed and if the gain moved to new passages.

All three packages worked equally well. The child’s words-correct-per-minute rose in every condition.

Gains held one week later and showed up on brand-new stories. Picking one package over the others did not matter for this learner.

Stancliffe et al. (2007) and Matson et al. (2011) extend these results. They used the same brief-experimental logic but handed the lessons to parents and peers. Fluency still generalized, showing the method travels beyond trained teachers.

Wilson et al. (1973) is an earlier cousin. They used vibrotactile cues plus tokens to boost rate in hard-of-hearing children. Both studies find generalization to untrained passages, proving the outcome is durable across very different tactics.

Stricker et al. (2024) looks like a contradiction at first. Their large RCT found big standardized-score jumps after relational-frame training, while DeRoma et al. (2004) found equal but modest gains with direct instruction. The gap is in measurement: Stricker used broad achievement tests; M used brief oral-reading probes. Both can be true—drills polish daily fluency, relational training may lift overall comprehension.

You do not need to hunt for the “best” fluency package. Pick the one that fits your classroom routine and run it. Brief probes will tell you quickly if the child speeds up and carries the skill to new stories. If not, swap in parents or peers—the literature shows the method still works.