Graphical methods for the design and analysis of simulation experiments

Intuitive graphical methods are developed for designing, running, and analyzing simulation experiments. Output plots from a single run are analyzed. The concept of a standardized simulation output plot is presented and its use illustrated. The design of experiments that may involve several (sequential) simulation runs is considered. Experimental design discussions in simulation often focus on special topics unique to the field. Control over random variables permits design techniques such as antithetic variates and common random number streams. Yet many of the general topics of experiment design including confounding, fractional designs, and sample size determination are important in a simulation setting. These general topics are addressed, and graphical tools for generating experiment designs, namely, causal diagrams and multidimensional point plots, are presented.