Regression to the mean in pretreatment measures of stuttering

Abstract

Pre-post treatment evaluation designs are common in stuttering research. Their propriety depends on the assumption that spontaneous remission is not likely. There are six studies in the literature in which stutterers have been measured on two occasions some months apart. In all studies there was a trend to less stuttering on the second assessment, but in no study was the difference between scores significant. One hundred and thirty-two stutterers awaiting treatment were assessed when they were first seen and then at the beginning of treatment 1--23 months later. There was a small but significant improvement between the two assessments. The size of the improvement was comparable to those reported in the six published studies. This spontaneous improvement occurs mainly in the three months following the first assessment, and there is little change thereafter. It is concluded that pre-post studies of subjects who waited more than three months for treatment are valid and that the observed improvement can be due solely to the effects of treatment. Studies that assess improvement from the time subjects are first seen should allow for spontaneous remission to determine the improvement due to treatment.