Modern Experimental Design Methods and How to Use Them

RAND Statistics Seminar Series

Modern Experimental Design Methods and How to Use Them

Dr. Douglas C. Montgomery—Arizona State University

Wednesday, April 27th, 2011
10:30 a.m. – 12:00 p.m. PT / 1:30pm – 3:00pm ET
Conference Room 1226/1228
RAND Corporation, Santa Monica, CA

Please contact Denise Miller if you would like to attend this seminar.

Abstract

The ease with which optimal experimental design can be constructed has ushered in a new phase of research and development in the field. This talk presents two recent applications of optimal design methodology to construct designs for previously unsolved problems. A new algorithm for generating near G-optimal designs for second-order models over cuboidal regions is introduced. Designs created using this new method either match or exceed the G-efficiency of previously reported designs. We also introduce a new graphical tool for comparison of the prediction variance for competing designs over a given region of interest. Using this tool to compare G-optimal designs to I-optimal designs shows that the G-optimal designs have higher prediction variance over the vast majority of the design region. This suggests that, for many response surface studies, G-optimal designs may not be the best choice. The second problem involves creating alternatives to the resolution IV regular fractional factorial designs in 16 runs for six, seven, and eight factors. These designs are in widespread use because they are economical and provide clear estimates of main effects when three-factor and higher-order interactions are negligible. However, because the two-factor interactions are completely confounded, experimenters are frequently required to augment the original fraction with new runs to resolve ambiguities in interpretation. We identify non-regular orthogonal fractions in 16 runs for these situations that are D-optimal and have no complete confounding of two-factor interactions. These designs allow for the unambiguous estimation of models containing both main effects and a few two-factor interactions. We present the rationale behind the selection of these designs from the non-isomorphic 16-run fractions and illustrate how to use them.

Speaker Bio

Dr. Douglas C. Montgomery is Regents’ Professor of industrial engineering and statistics and Foundation Professor of Engineering at Arizona State University. He holds BSIE, MS and Ph.D. degrees from Virginia Tech. His research and teaching interests are in industrial statistics. He is an author of 12 books and over 225 archival publications. Professor Montgomery is a Fellow of the ASA, a Fellow of the ASQ, a Fellow of the RSS, a Fellow of IIE, a Member of the ISI, and an Academician of the IAQ. He received the Deming Lecture Award from the ASA, the Greenfield Medal from the RSS, the Shewhart Medal, the William G. Hunter Award, the Brumbaugh Award, the Lloyd S. Nelson Award, and the Shewell Award (twice) from the ASQ, the George Box Medal from ENBIS, and the Ellis Ott Award. Professor Montgomery has won several teaching awards and is an ASU Outstanding Doctoral Mentor. He is a past Editor of the Journal of Quality Technology and a Chief Editor of Quality and Reliability Engineering International.

Attending a Seminar

Other Locations/Times:
Washington, D.C. Conf. Rm. 7730: 1:30 p.m. ET
Pittsburgh Conf. Rm. 6202: 1:30 p.m. ET

RAND visitors are welcome to attend and must RSVP at least one day prior to the seminar. To ensure your attendance please contact Denise Miller at dmiller@rand.org with your name, company (or university) affiliation, and national citizenship (for security purposes).

For parking and directions to RAND's Santa Monica office, please see: http://www.rand.org/about/locations/santa-monica.html.

For parking and directions to RAND's Pittsburgh office, please see: http://www.rand.org/about/locations/pittsburgh.html.

For further information and to be added to the mailing list contact Denise Miller at dmiller@rand.org.