Statistical analysis for single-case designs. Evaluating outcome of imaginal exposure treatment of chronic PTSD.

Mueser et al. (1991) tested a new way to score single-case data. They used the method on four Vietnam veterans who listened to taped war scenes to ease PTSD. Heart rate was tracked each session to see if fear dropped over time.

Two veterans got clearly better and kept the gains. One improved only a little. One got worse. The vets whose heart rate calmed fastest had the best outcome.

Baek et al. (2023) now offers a newer tool—multilevel modeling—to pool many single-case graphs. The 1991 paper is the grandparent; the 2023 paper is the updated grandchild.

Laureano et al. (2023) warns that change can look slow at first. In their big file review, about 3% of cases still acted like baseline for five sessions. The 1991 vets may have shown the same lag before their data turned.

Jamshidi et al. (2018) checked 30 years of single-case meta-analyses. They found most skipped bias tests and used wrong stats. Mueser et al. (1991) did not face those checks, so today you would add them before trusting the numbers.

You now have a quick way to spot real change in one client. Run the 1991 formula first, then cross-check with the 2023 multilevel steps if you plan to publish. Always watch the first five data points—transition states can fake failure. Add heart-rate or other biosensors; quick calm still predicts better PTSD response.