When comparing system alternatives, simulationists will often use analysis of variance, a parametric statistical procedure. Using an example in which three different M/M/s queuing system con figurations are simulated, the authors demonstrate the impor tance of using the common random number streams as blocks. In addition, Friedman's Two-way Analysis of Variance by Ranks test, a distribution-free statistical procedure, is used to analyze the output.
CACI SIMSCRIPT 11.5 Programming Language. CACI, Los Angeles, California (1983 ).
2.
Conway, R.W. "Some tactical problems in digital simulation ." Management Science20 (October 1963), 47-61.
3.
Emshoff, James R. and Sisson, Roger L.Design and Use of Computer Simulation ModelsMacmillan, New York, (1970).
4.
Friedman, Linda W. and Friedman, Hershey H. "Statistical Considerations in Computer Simulation: The State of the Art." Journal of Statistical Computation and Simulation19 (Spring 1984), 237-263.
5.
Friedman, Milton'The use of ranks to avoid the assumption of normality implicit in the analysis of variance." Journal of the American Statistical Association32 (1937), 675-701.
6.
Gibbons, Jean D.Nonparametric Statistical Inference2d ed. Marcel Dekker, New York (1985).
7.
Heikes, Russell G.; Montgomery, Douglas C., and Rardin , Ronald L. "Using common random numbers in simulation experiments - An approach to statistical analysis .' Simulation27 (September 1976), 81-85.
8.
Ignall, Edward J. "On experimental designs of computer simulation experiments." Management Science18 (March 1972), 384-388.
9.
Jucker, James V and Gomez, Jorge Garcia "Policy-comparing simulation experiments: Design and Analysis." Decision Sciences6 (1975), 631-645.
10.
Kleijnen, Jack Pc. "Comparing means and variances of two simulations :' simulation26 (March 1976), 87-88.
11.
Kleijnen, Jack Pc. "Design and analysis of simulations. Practical statistical techniques.' Simulaoon28 (March 1977), 81-90.
12.
Miller, Rupert G., Jr. Simultaneous Statistical Inference2d ed. Springer-Verlag, New York (1981).
