Statistical methods for a single-case study approach to aphasia therapy research

Publisher Website
Google Scholar
Abstract
The approach of cognitive neuropsychology has started to have an influence on language therapy research. Besides the delineation of adequate treatment approaches the problem of methodologically sound designs for single-case efficacy studies has been discussed extensively. Less attention has been devoted to statistical and psychometric aspects. For single-case studies a caveat concerning the use of any inferential statistical procedure assuming independent replications is, however, necessary. The binomial test model in the context of criterion-referenced and mastery testing is considered the basic model for pass/fail responses. Using this model, operational definitions for classical, strong and trend (double) dissociations in characterizing a patient's pre-therapy pattern of impairment are presented. Easy-to-use exact versions of tests for four-fold contingency tables will be used for the comparison of tasks, the evaluation of improved competence after treatment, and a test for differential treatment effects in the context of the crossover treatment design. It is strongly recommended that (i) the statistical power for the proper choice of the number of test items be considered and (ii) specific treatment effects are demonstrated when possible. Additional normative data from aphasic patients are thought to be useful for determining possible differences in inherent task difficulties.

This publication has 28 references indexed in Scilit: