The Randomized Multiple Baseline Experimental Design: Its Power and a Clinical Application to the Cognitive Modification of Delusions

The dissertation describes the first reported application of a small-N experimental design, the randomized baseline experimental design across subjects and behaviours. It is applied to a small scale clinical psychological experiment on the cognitive modification of delusional ideation in four people with a diagnosis of schizophrenia. The data were analysed by a form of randomization test which does not depend on the classical parametric assumptions. The randomization test based on random data permutation gave statistically significant results for the effect of the independent variable (cognitive modification of delusions) on two dependent variables (strength of conviction, and preoccupation) but not on a third dependent variable, amount of distress. Estimates of effect size are provided based on Cohen's d and on the Common Language Effect Size. It presents data on the statistical power of the procedure derived from Monte Carlo power analysis. It provides reviews of the concept of statistical power in applied psychological research, of the concept of effect size, of the use of cognitive modification of delusional ideation and of randomization tests. The results support the feasibility of small-N clinical experiments using the randomized baseline experimental design, analysing the data graphically and by use of randomization tests and designing experiments with the aid of Monte Carlo power analysis.