Blockinq in simular experimental designs

Abstract

The repetitive use of generated and/or recorded sequences in the model encounters comprising simular experimental designs has been a suggested technique of simulation methodology. The application of the technique has been deemed compatible with the experimental statistician's concept of blocking. The present paper verifies this compatibility, though notes that the resulting block effect must properly be interpreted as a variance component, or random block effect. A corollary to this result is a delineated procedure for selecting the random number seeds for successive encounters with a dynamic, stochastic simulation model, the procedure providing a proper measure of experimental error. The applicability of the established methodology of the design and analysis of experiments to systemic science via simulation modelling is thereby assured.