摘要:

Recent developments in quality engineering methods have led to considerable interest in the analysis of dispersion effects from designed experiments. A commonly used method for identifying important dispersion effects from replicated experiments is based on least squares analysis of the logarithm of the within-replication variance (Bartlett and Kendall 1946). Box and Meyer (1986) introduced a pooling technique for unreplicated two-level experiments. We extend this to replicated two-level experiments and compare its performance with the least squares analysis. We show that both of these methods can be obtained as special cases of maximum likelihood estimation under normal theory. The pooling technique is generally biased and is not recommended for model identification. The least squares analysis performs well as a model identification tool, but the estimators can be inefficient. In such cases we recommend that the parameters of the identified submodel be estimated by maximum likelihood. We derive some properties of the maximum likelihood estimator in balanced designs. An experiment for the robust design of leaf springs for trucks is used to illustrate the results.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488398

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

Time trends may affect the results of experiments that are conducted sequentially. A simple, yet powerful, way to model such an experiment is to represent the trend by an autoregressive integrated moving average time series model. I show how such models can be used to jointly estimate factorial and time-order effects and how they can be used as a diagnostic device to detect time trends in complex experiments.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488399

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

A graphical method is proposed to display a number of point estimates while allowing for their differing standard errors. More generally, it can be viewed as a representation of interval estimates by points on a bivariate plot. The method exploits a familiar connection between standardized estimates and regression through the origin and has several advantages over some alternative plots used in the literature. It is particularly useful when there may be a mixture of parameters, as illustrated by the problem of “mixed ages” in fission track dating.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488400

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

Safety studies of a nuclear reactor often center their interest on the probable number, of component failures in a time span of durationTo. Using the asymptotic normality of the estimator from Poisson regression, we develop approximate upper tolerance bounds for the distribution of the number of failures. Tables are given for easy computation of such bounds when the bound itself is less than 50. An example consisting of 90 failure records for nuclearreactor valve types provides illustration of the tolerance bound computations and dataanalytic techniques for validating a Poisson regression model.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488401

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

This article presents the design procedures and average run lengths for two mulativariater cumulative sum (CUSUM) quality-control procedures. The first CUSUM procedure reduces each multivariate observation to a scalar and then forms a CUSUM of the scalars. The second CUSUM procedure forms a CUSUM vector directly from the observations. These two procedures are compared with each other and with the multivariate Shewhart chart. Other multivariate quality-control procedures are mentioned. Robustness, the fast initial response feature for CUSUM schemes, and combined Shewhart-CUSUM schemes are discussed.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488402

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

The cumulative sum (CUSUM) is a basic diagnostic tool in the analysis of change-point data. It is shown that Fourier analysis of the CUSUM can be a useful supplementary tool in such analyses. The technique is applied to three data sets that have appeared previously in the statistical literature.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488403

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

The evidence for transformation of the response in a regression model may sometimes depend crucially on one or a few observations. Diagnostic methods based on the deletion of single cases are well established. Multiple deletion methods are likewise well known, but are little applied because of combinatorial problems. But sometimes the pattern of multiple outliers and influential cases cannot be revealed by the sequential use of single deletion methods. In such instances. masking is said to occur. The method of unmasking used in this article is least-median-of-squares regression, calculated at several values of the transformation parameter. The structure of the residuals from this robust analysis serves as an exploratory method for the identification of outliers. The confirmatory least squares analysis uses multiple-deletion diagnostic methods. Addition diagnostics are also used. These are developed for the score test and estimated transformation parameter.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488404

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

Minimum bias estimation was suggested as an alternative to least squares for polynomial tits by Karson. Manson, and Hader (1969). To use this alternative, a combination of least squares estimates of coefficients of the model order fitted, and of the additional model order whose bias is being guarded against, needs to be calculated. Individual estimation of the higher-order coefhcients is usually not necessary, and the question of finding parsimonious designs that provide only those combinations of estimated coefficients that are needed is explored. Some specific types of designs are suggested, including a new type of second-order rotatable design consisting of combinations (inkdimensions) of two-dimensional equiradial point sets.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488405

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

Several parameter estimation methods for dealing with heteroscedasticity in nonlinear regression are described. These include variations on ordinary, weighted, iteratively reweighted, extended. and generalized least squares. Some of these variations are new, and one of them in particular,modified extended iteratively reweighted least squares(MEIRLS), allows parameters of an assumed heteroscedastic variance model to be estimated with an adjustment for bias due to estimation of the regression parameters. The context of the discussion is primarily that of pharmacokinetic-type data, although an example is given involving chemical-reaction data. Using simulated data from 21 heteroscedastic pharmacokinetic-type models, some of the methods are compared in terms of mean absolute error and 95% confidence-interval coverage. From these comparisons, MEIRLS and the variations on generalized least squares emerge as the methods of choice.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488406

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor

摘要:

When fitting a model to a contingency table, a significant lack of fit can sometimes be caused by a few outlier cells, with the model fitting the remaining cells well. These cells can be identified by using deleted residuals (the residual from the expected count with the cell deleted) and tested using the drop in likelihood ratio goodness-of-fit statistic (from the model with the cell included to the model with the cell deleted), with the cells being tested from least extreme to most extreme (“backwards-stepping”). This article shows that using a Bonferroni bound for the outlier test at each step results in a conservative test with good power to detect multiple outliers; backwards-stepping and the use of deleted residuals results in the limiting of both masking and swamping effects. The procedure generalizes easily to complicated probability models.

ISSN:0040-1706    

DOI:10.1080/00401706.1988.10488407

出版商:Taylor & Francis Group    

年代:1988

数据来源: Taylor