Comparing masked and traditional visual analysis of multiple‐baseline designs

Wolfe et al. (2026) asked experts to look at multiple-baseline graphs two ways. First, they did the usual visual check. Then they did the same check again, but the names of the phases were hidden.

The team counted how often raters agreed with themselves and with each other under each method.

Both masked and traditional checks gave moderate in-method agreement. On two-thirds of the graphs, the hidden-phase check reached the same yes-or-no decision as the open-label check at least 75 percent of the time.

In short, masking did not wreck reliability; it stayed in the same ballpark as the usual way.

Ferron et al. (2017) ran computer simulations and said masking keeps false alarms under five percent while still catching real effects. Wolfe et al. now show live experts behave the same way, so the simulation holds up in practice.

Lanovaz et al. (2021) found a machine-learning script beats naked-eye calls. Wolfe et al. do not contradict this; they simply show that if you stay with human eyes, adding a mask is a cheap safety step.

Adams et al. (2024) pushed computer scripts for functional-analysis graphs. Wolfe et al. keep the focus on plain visual rules, giving BCBAs a low-tech option before jumping to code.

You can add masked review to any multiple-baseline study without new software or stats classes. It takes five minutes to cover phase labels before you ask a colleague to check your graph. That small step guards against seeing an effect that is not there, keeping your single-case conclusions clean.