Los contrastes de aleatorización en el análisis de datos en los diseños experimentales de caso únicoun estudio de simulación

Abstract

Single case experimental designs allow to test the efficacy of the psychological treatments, with total scientific guarantees, even with a single subject. Nevertheless, the existence of autocorrelation in the data does not permit the use of conventional statistical tests, searching alternative analysis strategies. For this reason, several tests have been developed, including non parametric tests or interrupted time series analysis, without agreement in which is the best one to be used. In literature, several authors have concluded that the randomization tests could solve the problem derived from the autocorrelation of the data. In this work, the analysis of the data, with randomization tests, was carried out by a Montecarlo simulation study, including two factors: the number of observations by phase (10, 20 and 30) and the autocorrelation in the data (0, .2, .4, .6 and .8).