Conjoint analysis is an effective method to examine consumers' choice behavior, in which it is important to classify samples(subjects)homogeneous within the group in terms of preference. However, the statistical validity of the classification has not been considered in most of conventional analysis. This paper proposes a method for testing the homogeneity of the partworthvectors of the conjoint model between groups, based on their rank ordered data. This model, including the stochastic error term, is described as a stochastic utility model. RANKLOGIT is used as an algorithm to estimate the partworths, which enables us to define a log likelihood function and utilize the procedure of testing hypothesis in the multinomial logit model. To illustrate how the proposed test can be applied to real world, a set of preference data for telephones is provided. Furthermore, a numerical experiment is conducted for the evaluation of Type I errors and powers of the test.

View full abstract

For memory products, it is very important to develop a new production line as soon as possible. However, it is said that no sensor is more sensitive than the product itself to monitor the process state. Though it is difficult to inspect product in process, all products are inspected to get rid of defected products at the last testing stage. Those inspection data are called FCM(Failure Category Map). Yield decrease can be caused by various kinds of faults, that can be categorized into two types. One is parametric fault caused by specification error of design/manufacturing parameters, and the other is random fault caused by operation error of manufacturing. In this paper, based on the FCM data, Area Usage Factor(AUF)analyzing method will be proposed. In general, AUF analysis is perfomed during the low yield term at the early constructing stage of mass-production line, and AUF analysis manifests the ratio of parametric fault and random fault. Therefore, process engineers can make their diagnosing plan and can make up their mind to which direction they should concentrate their limited analyzing power. Finally, AUF analysis system based on the proposed method will be reported, and an evaluation result is shown.

View full abstract

Management by Policy(MBP, Policy management)has been very popular in many companies because it is a useful management tool by which top management can deploy its policy to managers. This study investigated Deming Prize Winners(1987〜1996), and showed that MBP has several kinds of effects due to size of corporate organization, and the following difficulties sxist in applying MBP : how to make original policy of top management is no clear, policy is not always integrated with corporate vision, and MBP based on business strategy is needed. A new management methodology, Strategic Management by Policy(SMBP), which is 'enlarged MBP' based on Strategic Planning methodology, is already proposed by H. Osada et al.to solve those MBP's problems and make MBP more powerful. SMBP includes seven strategic tools : environment analysis, product analysis, market analysis, product-market analysis, product portfolio analysis, strategic element analysis, and resouce allocation analysis. Based on the results of the above bibliographic study, this paper also showed that the concept and framework of SMBP are effective in improving existing MBP and integrating MBP with strategic planning.

View full abstract