Technometrics Highlights
TECH Features Industrial Statistics Panel
T
he lead article in the May 2008 issue of Technometrics is a used to monitor the stability
panel discussion about the future of industrial statistics. The of the covariance matrix. The
panel covers a wide range of topics, including Six Sigma and new chart can be used with
the omnipresence of statistical methods in industry, emerging the multivariate exponen-
trends, hot new areas and missed opportunities, the democratization tially weighted moving aver-
of statistics through the wide availability of easy-to-use software, and age (MEWMA) chart, one of
the decline of strong statistical research groups in industry. The 10 the best tools to detect small
panelists bring a wealth of experience and wisdom: Søren Bisgaard, changes in the process mean.
Necip Doganaksoy, Nick Fisher, Bert Gunter, Gerald Hahn, The authors show that the
Sallie Keller-McNulty, Jon Kettenring, William Meeker, Douglas new chart generally outper-
Montgomery, and C. F. Jeff Wu. We see this article not as the end forms current control charts
of the story, but the take-off point for further discussion. To that for detecting changes in the
end, we are initiating an open discussion forum. You can take part covariance matrix.
by going to www.asq.org/pub/techno and clicking on Networking Tirthankar Dasgupta
and Events. and Abhyuday Mandal pres-
Following the panel discussion, the May issue includes nine ent “Estimation of P Process
research articles, beginning with “Analysis of Window-Observation Parameters To Determine the
Recurrence Data” by Jianying Zuo, William Q. Meeker, and Optimum Diagnosis Interval for Control of Defective Items.” This
Huaiqing Wu. Recurrent event data are common in reliability article relates to an online quality-monitoring procedure for attri-
assessment and can be used to analyze quantities of interest, such butes proposed by Taguchi. That procedure requires determination
as the mean cumulative number of events. Due to practical con- of the optimum diagnosis interval, which depends on parameters
straints, recurrence data are sometimes recorded only in windows, related to the process failure mechanism. Improper estimates of
with unobservable gaps between the windows. This paper extends these parameters may lead to the incorrect choice of the diagno-
existing statistical methods, both nonparametric and parametric, to sis interval and, consequently, huge economic penalties. Dasgupta
window-observation recurrence data. The nonparametric estimator and Mandal propose a Bayesian approach to estimate the process
requires minimum assumptions, but will be biased if the size of the parameters. They discuss a systematic way to use available engi-
risk set is not positive over the entire period of interest. There is no neering knowledge to elicit the prior for the parameters. The per-
such difficulty when using a parametric model for the recurrence formance of the proposed method is demonstrated using extensive
data. Two alternative hybrid estimators are proposed and compared simulations and a case study from a hot rolling mill.
for cases in which the size of the risk set is zero for some periods of Multivariate ordinal data arise in many applications. Earl
time. The methods are illustrated with two applications. Lawrence, Derek Bingham, Chuanhai Liu, and Vijayan N. Nair
Debasis Kundu also addresses a problem in reliability in the focus on the popular multivariate probit model and propos a new
article “Bayesian Inference and Life Testing Plan for Weibull efficient method for Bayesian inference using Markov chain Monte
Distribution in Presence of Progressive Censoring.” This paper Carlo techniques in “Bayesian Inference for Multivariate Ordinal
deals with Bayesian inference of unknown parameters of the pro- Data Using Parameter Expansion.” The key idea is the novel use of
gressively censored Weibull distribution. Progressive censoring parameter expansion to sample correlation matrices. A nice feature
is a testing scheme in which some currently working units may of the approach is that inference is performed via straightforward
be removed, along with a failed unit, at the time the latter unit Gibbs sampling. Bayesian methods for model selection also are dis-
fails. When the Weibull shape parameter is unknown, closed-form cussed. The approach is motivated by a study to understand how
expressions of the Bayes estimators cannot be obtained. Instead, women make decisions about treatment to reduce the risk of breast
Lindley’s approximation is used to compute the Bayes estimates cancer. The authors compare the performance of their approach
and Gibbs sampling to calculate credible intervals. For given pri- with other methods.
ors, a methodology is proposed to compare two censoring schemes The next two articles address problems related to experiments
and, hence, to find the optimal Bayesian censoring scheme. The run on computer simulators. Peter Z. G. Qian and C. F. Jeff Wu
proposed methods are studied via simulations illustrated with an consider appropriate models when data are available from both low-
actual application. and high-accuracy experiments. This setting occurs, for example,
Douglas M. Hawkins and Edgard M. Maboudou-Tchao pro- when the same physical system can be simulated by either a fast,
pose a new process-monitoring chart in their article, “Multivariate but low-accuracy, computer code or a high-accuracy code that has
Exponentially Weighted Moving Covariance Matrix.” This article long run times. In their article, “Bayesian Hierarchical Modeling
extends the notion of exponentially weighted moving average charts for Integrating Low-Accuracy and High-Accuracy Experiments,”
to monitoring the covariance matrix when the process-monitor- Qian and Wu show that an integrated analysis of all the data can
ing scheme collects multivariate data. The resulting chart can be produce better results than the standard practice of analyzing each
MAY 2008 AMSTAT NEWS 19
Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56  |  Page 57  |  Page 58  |  Page 59  |  Page 60  |  Page 61  |  Page 62  |  Page 63  |  Page 64  |  Page 65  |  Page 66  |  Page 67  |  Page 68  |  Page 69  |  Page 70  |  Page 71  |  Page 72  |  Page 73  |  Page 74  |  Page 75  |  Page 76  |  Page 77  |  Page 78  |  Page 79  |  Page 80  |  Page 81  |  Page 82  |  Page 83  |  Page 84  |  Page 85  |  Page 86  |  Page 87  |  Page 88  |  Page 89  |  Page 90  |  Page 91  |  Page 92