Confounding in Regular, Single Replicate Factorial Experiments

Abstract

Substantial progress has been made over a four year period, with the conjecture having been proven in various special cases. However, a general proof of the conjecture has not been obtained, and progress during the last two reporting periods has been disappointing. Hence, no further support of this research has been requested. Progress made during the research period includes establishment of the following results. The strongest result known to date is that the conjecture is true if the orders of at most three factors are divisible by a common prime. That the use of pseudofactors of prime order is sometimes beneficial has also been shown. Specifically, for experiments in which three or more factors have orders divisible by a common prime, p say, with at least one of the orders divisible by p(2), there exist pseudofactor design for which there is no degrees-of-freedom equivalent generalized cyclic design. The result is somewhat surprising. If for each prime p the orders of at most two factors are divisible by p, then the two classes of designs are degrees-of-freedom equivalent.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 19, 1993
Accession Number
ADA268887

Entities

People

  • Daniel Voss

Organizations

  • Wright State University

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Bits
  • Combinatorial Analysis
  • Construction
  • Experimental Design
  • Factorial Design
  • Generators
  • Mathematical Analysis
  • Mathematics
  • Numbers
  • Prime Numbers
  • Statistical Analysis
  • Statistics
  • Universities

Fields of Study

  • Mathematics

Readers

  • Mathematical Modeling and Probability Theory.
  • Regression Analysis.
  • Systems Analysis and Design