Advanced Topics For A First Service Course In Statistics

Statistical software has made traditional statistical calculations accessible to almost anyone, but it has also stimulated new methods that are usually reserved for advanced courses. In this paper I argue for inclusion in the first course of non-parametric smoothing, density estimation, coplots, simulation, the bootstrap, time series forecasting, and plots of multivariate data. It is argued that the logic underlying these techniques is simpler and more useful than the logic underlying the inference usually included in first service courses in statistics.