Level: Advanced
【Presenter】
Interuniversity Research Institute Corporation Research Organization of Information and Systems Institute of Statistical Mathematics
Risk Analysis Strategy Research Center Visiting Professor Takenori Takahashi
[Presentation summary]
Most of the cases of questionnaire surveys (questionnaire surveys) to date have ended at the consideration stage, and there are some cases in which proposals are made based on the consideration, but in such cases the proposals are only based on hypothetical ideas (thoughts and ideas). However, there are almost no presentations of cases in which hypothesis verification (verification experiments), which is the objective basis for the proposal, have been carried out. In order to make concrete proposals linked to problem solving and task accomplishment with evidence, it is necessary to conduct experiments to provide the basis for such proposals. Conjoint analysis is an option for this. However, most conventional conjoint analyzes are independent efforts, and their analyzes also have three characteristics: (1) All factors are treated qualitatively, (2) Interactions and quadratic terms are ignored, and (3) Experiments The design is a factorial arrangement or an orthogonal experiment), which is an extremely restrictive (limited) experiment.
This study introduces a method of collecting the resulting data and an experiment using a structural model table and an optimal design (referred to as a virtual experiment in this study). It is mathematically clear that it is desirable for the result data to be on a continuous scale in virtual experiments. However, abrupt continuous scale evaluations place a burden on respondents and provide unreliable data. In this study, we propose an adjusted difference apportioning method for this purpose.
[Presenter profile]
After serving as a professor at Tokyo University of Science's Faculty of Engineering, Keio University's Graduate School, and then Mejiro University's Faculty of Business Administration, he currently continues his research and data science education as a visiting professor at the Institute of Statistical Mathematics. In terms of research, he proposes a new design paradigm, superdesign, in the field of engineering, and new analysis methods, two-sided causal analysis and virtual experiments, in the field of social science.
On the other hand, while I was a graduate student, I developed the theory of competitive skiing (mainly slalom and giant slalom) and taught practical skills to a certain university's ski club for several years, and published the contents as a book on competitive skiing theory, in parallel with my specialized research. Based on this experience, we have been implementing active data science education, which is statistical education based on practical training, at universities, graduate schools, companies, and various seminars for about half a century.
During his sabbatical, he advanced his research on super-design at Yale University in the United States and developed an assembling biplane helicopter as a real teaching material. The former is the main theme of the author's research, and the latter is a full-fledged actual teaching material that can take more than a dozen variables (factors) and can also handle quadratic models. In addition, we have also developed a paper glider as an extremely advanced real-life teaching material, and coin shooting, a general-purpose real-life teaching material aimed at everyone from beginners to experts.
*Poster display is paper only.
