摘要:

For the analysis of designed experiments, Taguchi uses performance criteria that he callssignal-to-noise (SN) ratios. Three such criteria are here denoted by SNT, SNL, and SNS. The criterion SNTwas to be used in preference to the standard deviation for the problem of achieving, for some quality characteristicy, the smallest mean squared error about an operating target value. León, Shoemaker, and Kacker (1987) showed how SNTwas appropriate to solve this problem only when σywas proportional to μy. On that assumption, the same result could be obtained more simply by conducting the analysis in terms of logyrather thany. A more general transformation approach is here introduced for other, commonly met kinds of dependence between σyand μy(including no dependence), and alambda plotis presented that uses the data to suggest an appropriate transformation. The criteria SNLand SNSwere for problems in which the objective was to make the response as large or as small as possible. It is argued here that these predecided “portmanteau” criteria can provide an inadequate summary of data and that, regarded as measures of location, they can be extremely inefficient. In preference to such performance criteria, the merits of simple methods of data analysis that can uncover information both expected and unexpected are urged. A reanalysis of an interesting experiment due to Quinlan (1985) illustrates the value of this approach and its contribution to the art of discovery. It is argued that improvement of quality will best be catalyzed by engineers using elementary data analysis with computer graphics rather than by those trained only to employ more rigid predecided criteria.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488313

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488314

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488315

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488316

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488317

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488318

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488319

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488320

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

Industrial problems frequently require estimates from various sources of information. For example, one may need to predict the tensile strength of a future bar from a particular casting based on limited data on other bars from that casting and extensive data on bars from other castings. Or one may wish to estimate the true viscosity of a batch of material based on a single measurement for the current batch, subject to appreciable measurement error, and similar readings on a large number of other batches. Simple weighting functions that use all of the data provide point estimates for these two problems, and a Bayesian framework yields associated interval estimates. Other applications and possible generalizations are also suggested.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488321

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

An analysis of means (ANOM) procedure is suggested for testing the significance of two-factor interactions in fixed-effects balanced designs. When at least one factor is at only two levels, the suggested procedure uses the exact critical points for the ANOM. New critical points are given for the case in which both factors are at more than two but fewer than six levels. The procedure can be extended to higher-order interactions, but the quantities being plotted become so complex that they are of little use in assessing practical significance.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488322

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor