Identifying Cyclical Patterns of Behavior Using a Moving-Average, Data-Smoothing Manipulation.

Retzlaff et al. (2024) looked at daily counts of destructive behavior in two cases. They ran simple 3-, 5-, and 7-day moving averages across the raw data.

The team then compared the smoothed lines to what standard visual inspection had flagged. They wanted to see if the smoothing uncovered any hidden cycles.

The moving-average lines showed clear up-and-down cycles that the naked eye had missed. Standard graphs looked flat; smoothed graphs revealed repeating weekly swings.

Once the cycles were visible, the clinician could plan interventions for the high-risk days instead of treating every day the same.

Annable et al. (1979) already warned that visual inspection is shaky—reviewers only agreed 61 % of the time. Retzlaff’s smoothing trick gives you a simple way to cut that noise.

Cramm et al. (2009) also hunted for time patterns, using scatter plots in 2-hour blocks. Retzlaff keeps the same case-series spirit but swaps in moving averages for whole-day data.

Friedel et al. (2019) used Monte Carlo smoothing on relapse data. Retzlaff shows you don’t need fancy software; a basic Excel average works for daily cycles.

If you graph behavior each day and see only a messy blur, add a 5-day moving-average line before you call it ‘stable.’ The smoothed line can expose weekly, bi-weekly, or monthly swings that drive your data. Once you spot the rhythm, you can time booster sessions, staff training, or environmental changes for the days just before the usual uptick. Five extra minutes of Excel work can save weeks of guessing.