ENBIS Session @ ISBIS 2012
Author: Irena Ograjenšek, Created 2012/08/03 13:01
Fabrizio Ruggeri, Past President of ENBIS (2005-2006), organised an ENBIS session at the recent ISBIS World Meeting held in Bangkok on June, 17-21, 2012.
The first speaker, Kristina Lurz (Ph.D. student at the University of Wuerzburg, Germany) presented a joint paper with Rainer Goeb on recent developments in statistical audit sampling. The talk reviewed the implications of the position of ISA 500 and 530 for stochastic modelling and for the design of statistical audit sampling. The evolution of audit sampling has been accompanied and reflected by national and international auditing standards like SAS 39 (1981), SAS 47 (1983), SAS 106 (2006), SAS 107 (2006), SAS 111 (2006). For long, the standards have been strengthening the position of judgmental selection in the auditing community. However, the recent international standards ISA 500 and 530 adopt a markedly different position: judgmental selection is not considered as audit sampling, the position of statistical sampling is considerably strengthened. The talk illustrated the use of recent results on two-sided confidence limits for proportions and for the mean of zero-inflated populations in audit sampling.
Tashi Erdmann, a Ph.D. student from the Institute for Business and Industrial Statistics in Amsterdam, the Netherlands, gave a presentation about the research he and Jeroen de Mast have done on measurement system analysis (MSA) for binary measurements. Recently, a flow of literature has appeared regarding MSA for binary measurements. Despite the fact that binary measurements abound in industry, Erdmann showed that in many situations it is all but clear what method should be used to assess their reliability. In his talk he discussed merits and fallacies of the most common methods, and gave guidelines in deciding which method to use. For example, he identified a number of common errors in agreement studies using Cohen's kappa index, a method widely applied in industry. Highly relevant factors in choosing a method are whether the value of the underlying property, the measurand, is known, and whether it is continuous or truly dichotomous. Erdmann argued that many guidelines offered in industry are in conflict with his conclusion that random sampling is in many cases crucial.
Finally, Stelios Psarakis, from the Athens University of Economics and Business, Greece, gave a talk on monitoring nonlinear profiles. This is a joint work with Athanasios Yannacopoulos from the same university and two researchers (Javier Cano and Javier Moguerza) from Rey Juan Carlos University in Madrid, Spain. The quality of a process or product can be characterized by a functional relationship between a response variable and some explanatory variables. In his talk, Psarakis discussed the nonlinear profiles monitoring problem and developed a novel hybrid nonparametric-parametric procedure for the study of nonlinear profiles, that is, nonlinear noisy functional relationships between variables. In particular, he focused on the 'shape' property of profiles and, starting from a nonparametric reference curve, he selected a model from a universe of parametric deformations of the reference curve with the property of preserving certain important shape characteristics. To this aim, a metric based on the solution of an optimization problem was designed. In addition, it was shown that the problem is well-posed from a theoretical point of view. Finally, the performance of the proposal with numerical examples from simulated and real environments was illustrated.