Опубликовано на портале: 03-10-2003Harry V. Roberts Statistical Science. 1990. Vol. 5. No. 4. P. 372-390.
Statistical methodology has great potential for useful application in business, but that potential is seldom realized. However, companies are increasingly exploiting simple statistical tools in quality and productivity improvement and developing "company cultures" congenial to effective use of statistics. Statistical and probabilistic thinking is essential for sound decision-making. Only with understanding of statistical variability can managers distinguish special from common causes of variation, intelligently direct efforts to improve processes, and avoid the tampering that can make processes worse. Statistics can be used most effectively in business when many employees--"parastatisticians"--have some grasp of statistical tools and thinking. Fortunately, there is evidence that very elementary tools suffice to make rough-and-ready studies that can illuminate most business problems and facilitate most decisions.
Опубликовано на портале: 19-07-2004David Freedman, Kenneth W. Wachter Statistical Science. 1994. Vol. 9. No. 4 . P. 476-485.
Current techniques for census adjustment involve the "synthetic assumption" that undercount rates are constant within "post-strata" across geographical areas. A poststratum is a subgroup of people with given demographic characteristics; poststrata are chosen to minimize heterogeneity in undercount rates. This paper will use 1990 census data to assess the synthetic assumption. We find that heterogeneity within poststrata is quite large, with a corresponding impact on local undercount rates estimated by the synthetic method. Thus, any comparison of error rates between the census and adjusted counts should take heterogeneity into account.
Опубликовано на портале: 19-10-2004David Banks Statistical Science. 1993. Vol. 8. No. 4.. P. 356-377.
This paper is a critique of the current direction of industrial statistics in the U.S. The sweep includes not just the toolkit of statistical methods most often employed in industry, but also international competitiveness, the corporate climate in which statistical solutions are sought and the educational process which trains applied statisticians. Much is found that is good, but bland endorsement does little to advance a field. Therefore most of the paper is deliberately iconoclastic. Almost all of the extreme viewpoints are rooted in the author's direct experience, working with statisticians and managers across a range of industries. Some of the perspective must be attributed to knowledgable gossip, from friends and students who are now employed by various companies. Incidentally, the paper reviews four new statistical textbooks for engineers and an edited volume of papers on experimental design for industry.