Accession Number : ADA292627

Title :   Some New Saturated Two-Level Designs.

Descriptive Note : Final rept. Apr-Oct 94,

Corporate Author : EDGEWOOD RESEARCH DEVELOPMENT AND ENGINEERING CENTER ABERDEEN PROVING GROUND MD

Personal Author(s) : Crosier, Ronald B.

PDF Url : ADA292627

Report Date : JAN 1995

Pagination or Media Count : 19

Abstract : First-order saturated designs can have the two desirable properties of being an orthogonal design and having two levels only if n=O mod 4, where n is the number of design points. For other cases, researchers have developed D-optimal two level designs. Recently, two-level saturated designs that are efficient for submodels containing only a few of the factors were developed. These p-efficient designs (p is the number of parameters in a submodel) have two properties not possessed by the D-optimal designs: (near) equal-occurrence of -1's and 1's in each column and near orthogonality of each pair of columns. In this report saturated, two-level designs that estimate the effects of all factors with equal precision are developed. These equal-precision designs are generally better than the p-efficient designs by the three criteria D-effidency, G-efficiency, and maximum variance inflation factor. Equal-precision designs are given for n=3-30, 38, 42, and 50.

Descriptors :   *EXPERIMENTAL DESIGN, *VARIABLES, PARAMETERS, COEFFICIENTS, POLYNOMIALS, ORTHOGONALITY.

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE