ABSTRACT
A computer can be a useful tool for developing a better understanding of complex genetic and statistical problems. The ability to simulate sample data for a specified genetic model, or to simulate data from a specified multivariate normal distribution, and the ability to subsequently analyze the simulated data, can often extend one’s understanding of a complex problem. Being able to specify the underlying nature of a population, and to then observe the results of analysis of a sample of data from that population, is a type of experimentation that will often be more convincing than any theoretical argument. The program will generate the specified sample and write the estimated genotypic and phenotypic variances and covariances on a diskette file for subsequent use. Many commercial statistical software packages include programs for analyzing multivariate data. However, not all of them are capable of providing estimates of mean squares and mean cross-products as required for developing selection indices.
