Barry L. Nelson
- Box, G. E.P. 1954. Some theorems on quadratic forms apphed in the study of analysis of variance problems, I. Effect of inequality of variance m the one-way classification. An. Math. Stat. 25, 290 302.Google Scholar
- GRAYBILL, F.A. 1969. Matrices with Apphcatlons in Stutzstzcs. Wadsworth, Belmont, Calif.Google Scholar
- GRIEVE, A. P., AND AG, C.-G. 1984 Tests of sphericity of normal distributions and the analys~s of repeated measures designs Psychometrzka 49, 2, 257-267Google Scholar
- HOCHBERG, Y., AND TAMHANE, A.C. 1987. Multiple Comparison Procedures. John Wiley, New York. Google Scholar
- HSU~ J. C., AND NELSON, B L 1988. Optimization over a finite number of system designs with one-stage samphng and multiple comparisons with the best. In Proceedings of the 1988 Winter Szmulatmn Conference, M. A. Abrams, P. L. Haigh, J. C Comfort, Eds. ACM/IEEE, New York, 451 457. Google Scholar
- MARSAGL1A, G., AND OLKIN, I. 1984 Generating correlation matmces. SIAM. Scz Star. Cornput 5, 470 475.Google Scholar
- NELSON, B. L. 1992. Common random numbers and multiple comparisons in s~mulation analysis. In Proceed~n~'s of the 1st Industrlal Engineering Research Conference. G Klutke. D. A. Mitta, B. O. Nnaji, L. M. Seiford, Eds IIE, 463 466.Google Scholar
- NELSON, B.L. 1990. Control variate remedies. Oper. Res. 38, 6, 974 992. Google Scholar
- NELSON, B. L., AND HSU, J C. 1993. Contro}-varlate models of common random numbers for multiple comparisons with the best. Manage Sci To be pubhshed. Google Scholar
- NOZARI, A., ARNOLD, S., AND PEGDEN, C. 1987. Statistical analys~s for use with the Schruben and Margohn correlation induction strategy. Oper. Res. 35, 1, 127-139 Google Scholar
- RAO, C. R 1973. Linear Stati~tzcal Inference and zts Applicatzons John Wiley, New York.Google Scholar
- SCHRUBEN, L., AND MARGOLIN, B. 1978. Pseudorandom number assignment in statistically designed simulation and distribution sampling experiments, d. Am. Star. Assoc. 73, 504-525.Google Scholar
- TEW, J. D., AND WILSON, J. R. 1993. Estimating simulation metamodels using combined correlation-based variance reduction techniques, lie Trans. To be published.Google Scholar
- TONG, Y. 1980. Probability llzequalztzes i~ Mu{tzvar~ate Distributtons. Academic Press, New York.Google Scholar
- WANG, H.-L. 1992. Multiple comparisons procedures in repeated measures designs. Ph.D. dissertation, Dept. of Statistics, The Ohio State Univ., Columbus, Ohio.Google Scholar
- YANG, W., AND NELSON, B.L. 1991. Using common random numbers and control variates in multiple-comparison procedures. Oper. Res. 39, 4, 583 591.Google Scholar
Index Terms
- Robust multiple comparisons under common random numbers
Recommendations
Using Common Random Numbers for Indifference-Zone Selection and Multiple Comparisons in Simulation
<P>We present a general recipe for constructing experiment design and analysis procedures that simultaneously provide indifference-zone selection and multiple-comparison inference for choosing the best among k simulated systems. We then exhibit two such ...
Some insights of using common random numbers in selection procedures
Indifference-zone selection procedures have been widely studied and applied to determine the sample sizes for selecting a good system or a subset of good systems among k alternative systems. It is known that using common random numbers can increase ...
Multiple comparisons with the best for steady-state simulation
Multiple-comparison procedures are useful for comparing the performance of competing systems via simulation. In this paper we extend a particular multiple-comparison procedure, multiple comparisons with the best, to steady-state simulation by using an ...
Reviews
R. Sambasiva Rao
One of the objectives of simulation experiments is to compare the expected performance of different systems. In this context, induced positive dependence across the responses (simulation outputs) and control-variate models have been in vogue. As a result of the inadequacies of these procedures, analysis of repeated-measures (replicate) experiments is proposed. The outcome of this analysis is useful in the investigation of simulation studies that employ common random numbers (CRN). CRN is a variance reduction technique that reduces the variance of estimators of the differences between the expected performance of two or more systems. Compound symmetry, a specific case of sphericity, is induced in the covariance structure through CRN. Nelson uses the multiple comparisons with the best (MCB) method, which is a simultaneous statistical inference procedure. The example discussed in detail is the minimization of expected cost per period over a planning horizon of 30 periods for five inventory policies. In this simulated experiment, the standardized average is the control variable, and the experiment is repeated with 30 replications for each of the 30 periods. Based on these results, 95 percent MCB is calculated. The entire experiment was repeated 1000 times to estimate the probability of the coverage. As the system has no control variables, the control-variate model reduces to one-way analysis of the variance model. The advantage of MCB intervals obtained under the sphericity assumption is that they are robust to departures from sphericity. The inferences drawn hold equally well for other control variate systems.
Access critical reviews of Computing literature here
Become a reviewer for Computing Reviews.
Comments