摘要:

Experimental designs used in industry often allow no degrees of freedom for the estimation of error. Nevertheless, analysis of variance results can (if used properly) be used to determine which factors are significant. We give a back-of-the-envelope calculation for making this determination, intended for practitioners desiring a convenient, anova-based analysis. Power comparisons show that competing methods, regardless of computational complexity, have remarkably similar operating characteristics.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979297

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

The problem of determining the minimum sample size for two-sidedβ-content tolerance intervals when the population size is finite is investigated. It is assumed that a simple random sample is selected from the population. An expression for the tolerance coefficient is derived. We also determine the minimum population size in order for this population size to be considered infinite. Tables of the minimum sample sizenand the corresponding minimum population size are provided.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979298

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

This paper is an addendum to a previous paper of mine, which discussed modified XVERT mixture screening designs (MXMSDs) for mixture constraint regions defined by lower and upper bounds on the component proportions. MXMSDs consist of a center point and a subset of the extreme vertices derived from a classical two-level screening design (e.g., Plackett-Burman or fractional factorial) using a modified XVERT algorithm.The previous paper showed that the properties of MXMSDs can be nearly as good as or equal to the properties of designs in which the subset of extreme vertices is selected using optimal experimental design software. In this addendum, it is shown that the properties of MXMSDs can also be considerably worse than those of optimal designs for some mixture constraint regions. Simple functions of the component lower and upper bounds provide a basis for determining whether or not MXMSDs will have properties close to those of optimal experimental designs. Two versions of a nine-component herbicide example are used to illustrate the discussion.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979299

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

Nonparametric methods are presented for testing the null hypothesisHo:ξ1=…=ξk=ξo, whereξiis the median of theithpopulation andξois specified, against one-sided alternatives. Several simple tests are described along with a multiple comparisons procedure. These methods are nonparametric analogues to the one-sided parametric methods proposed by Mee, Shah, and Lefante in an earlier issue of this Journal. The parametric and nonparametric methods are illustrated using the same data set.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979300

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

Taguchi developed his catalog of orthogonal arrays from mathematical procedures published in well-known English language journals. These arrays evolved as extensions of factorial plans and latin squares. This paper illustrates the fractional factorial nature of Taguchi's two-, three-, four-, and five-element orthogonal arrays. Similarly Taguchi's mixed-element orthogonal arrays define fractionated multi-level factorial plans. Thus Taguchi's orthogonal arrays are an element in the continuum of the development and use of statistically planned experiments.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979301

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

A control chart technique that considers exponentially weighted moving averages (EWMA) of the ranks of the observations is proposed. This nonparametric technique is outlier-resistant and performs well if one is concerned about sudden shifts in the process mean. The procedure is illustrated on a data set taken from the literature.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979302

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

A process must be considered free from variation due to assignable cause before its capability can be determined. In those situations where variation due to an assignable cause occurs and is tolerated, process capability cannot currently be assessed. By allowing the process capability to be considered dynamic, a procedure for assessing capability is proposed and an estimator developed for those situations where a systematic assignable cause occurs. Some statistical properties of the estimator are examined and a technique for managing a process exhibiting a systematic assignable cause is discussed. Examples from a particular class of systematic assignable cause known as tool wear are included.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979303

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

Statistical process control procedures typically entail monitoring the process by selecting rational subgroups of equal size at equal time intervals. A generalization of this standard paradigm removes the restriction of equal waiting times between subgroups. We consider process control procedures where the time until the next subgroup is shortened when there is evidence that the process is off target, with a compensating lengthening of the waiting time otherwise. This adapting of the waiting time significantly improves the performance of the process control procedure. Using the ShewhartX̄procedure as an example, the adaptive sampling strategy is introduced. A simple, dual waiting time procedure is described and shown to be optimal and easy to implement in practice. Performance comparisons to classical procedures are provided.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979304

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

摘要:

We describe a FORTRAN program that will enable users to determine the control limit parameters and sample size of anX̄andRchart designed jointly under a statistical criterion. This criterion can be stated in terms of Average Run Length, Type I and Type II error probabilities, or average times to signal for up to 10 specified shifts in the process distribution.

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979305

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1991.11979306

出版商:Taylor&Francis    

年代:1991

数据来源: Taylor