Single-case research designs are widely used in educational fields. But the analysis of single-case data has been debated for sometime. Recently, the statistical tests are proposed as a supplement to, or a replacement for, visual inspection in evaluating single-case data. Randomization tests and the C statistic are statistically valid methods for analyzing single-case data. These methods are also recommended to use. In this study, the statistical powers of these two statistical tests are estimated for comparison. The Monte Carlo method was used to estimate the power. The power was studied for six standardized effect size (0.0, 0.2, 0.5, 0.8, 1.1, 1.4) and four levels of autocorrelation (1st order autocorrelation coefficients of-0.3, 0.0, 0.3, 0.6). The AB design consisting of 35 observations was studied. The results showed that the C statistic can not control the type l error rate when the data have a positive autocorrelation and has very little power when the data show a negative autocorrelation. Therefore, the C statistic is inappropriate for evaluating single-case data.
View full abstract