Forcing square pegs into round holes: some comments on "an analysis-of-variance model for the intrasubject replication design".

Shimp (1974) wrote a short attack paper.

It said stop cramming single-subject learning curves into one-way ANOVA.

The author claimed ANOVA hides each learner’s unique path.

The paper found no new data.

Instead it showed that fixed-effect ANOVA treats each child like a clone.

It urged analysts to plot each subject’s own line and look for repeats across phases.

Bell (1999) picked up the torch. It told BCBAs to skip all p-values and trust visual analysis plus steady replication.

Falligant et al. (2022) gave you modern tools—conservative dual-criteria and fail-safe k—to make that visual check sharper without stats.

Cashon et al. (2013) seems to disagree. They added hierarchical linear modeling to single-case data and said it can strengthen, not replace, visual reads. The clash is only surface: Shimp (1974) hated old-school ANOVA, not careful modern modeling that still honors each subject’s curve.

Next time you open Excel after a reversal or multiple-baseline plot, do not click “ANOVA.”

Split the graph by participant, draw phase lines, and look for the same jump twice.

If you need extra eyes, Falligant’s aids are free online. Your decision stays visual, your kid stays unique, and your evidence stays clean.