Some simple statistical tests for exploring single-case time-series data

Statistical analysis of single-case time-series data has received considerable attention. Many of the techniques discussed in the literature require long series of data or sophisticated statistical techniques. The present paper describes a number of 'quick and dirty' non-parametric tests which can be applied to relatively short time series. These tests require minimal calculations and can be carried out by inspecting graphical plots of the data. Each test is illustrated and tables for significance testing have been worked out. The importance of these tests lies in their potential to facilitate the accurate description of time series data. They are recommended as techniques for exploratory data analysis, the philosophy of which parallels Sidman's (1960) 'data led' approach to experimentation.