Treffer: The Use of the Unknown in Teaching Statistics.

A technique is described for teaching a second course in statistics at the undergraduate level, but is equally applicable to a first course. The procedure involves the use of individualized data sets, computer prepared to student requests. A package of five programs is presented, which generates the data. They include programs for sampling from bivariate and multivariate normal distribution from independent normal distributions for T-tests and one-way analysis of variance cases, generation of data for rank order correlation, and for analysis of variance designs with two or three independent variables. A separate set of hand-outs to the paper illustrates the use of the program for generating data. (Author/LH)