A similar problem can occur in research studies, where consistency of all but the treatment variable is essential. If participants with different severities of aphasia are given the same script, there may be great variability in their pre-treatment level of performance. We confronted this problem in our own studies. Although participants who were less severe received longer scripts, their initial performance accuracy was still considerably higher than those who were more severe (64% versus 26.2%), making it difficult to compare absolute gains across participants. In addition, participants who were milder and started the intervention at a baseline accuracy greater than 80% quickly reached ceiling.