14th Meeting of the Decision Based Design Open Workshop
Open Forum Discussion:
This is a summary of the question and answer session between the audience and panelists at the 14th meeting.
Q: What do we do when there is not enough information to derive model coefficients?
- KS: May need to investigate a more subjective means of modeling. There are non-parametric techniques for measuring shapes of functions without imposing any particular assumption (e.g. normal or lognormal).
- HC: Measure and compare to market research data.
- JC: Consider interval analysis, fuzzy mathematics - bounds the space - tough to do on new designs. May be able to map to a perception curve or use historical market information.
Q: How can this be applied to other factors, like the emotional connection a user has with a product?
- PP: Need to think about the curve of satisfaction vs. novelty - too little novelty leads to little satisfaction, too much novelty can have the same effect.
- KS: This is a question of psychology and the use of Discrete Choice Theory can incorporate these subjective measures. This type of work has been looked at by McFadden and Luce.
- HC: Direct Value model is useful here. May be useful to look compare value changes between "base" product model and models with wild aesthetic changes. This is something that should be modeled and validated.
- JC: There is a model with an "appeal" variable that may involve novelty, bandwagon effect, etc.
- JC: I use 8 components to represent appeal. These would be more critical in manufacturers of durable goods.
- HC: Design and manufacturing should be more closely tied. Really need to look at entire process.
- HC: The demand model should be the basis for decision making. No matter what, you have to deal with demand in a NPV approach. Engineer couples the customer, manufacturing and design studio - the engineer must understand the business side as well.
- PP: Can only consider rational behavior models - can't account for the potential irrationality of customers and manangement.
- JC: There is design for the sake of design. When we tie design to demand we are doing so in order to make decisions on changes or selection between another. Demand model cannot provide the choice to accept or reject a design but it can provide answers to variability to guide detailed design.
- KS: Collect a lot of data (1000 individuals) and you can handle attributes individually in the model. The use of factor analysis on the attributes can help determine which attributes have and independent effect on the model.
- HC: Major off-road equipment supplier used Direct Value Method to find which of a set of attributes to consider. It worked out well for the supplier. Also the choice to put a second sliding door on a the mini-van - Chrysler did it base on focus groups eyes when a model was presented - Ford decided not to - Chrysler leads the minivan market.
- PP: Wouldn't be able to find examples. Companies would claim that the design content decisions were viewed at a very high level of abstraction.
- PP: Students still need to think like designers. They should be creative and have common sense - perhaps the tools will change but the basics remain the same.
- JC: Management is more open to science and the idea of "specialists" within a company or outside consultant groups.
- JC: Need to teach students some economics.
- KS: IT changes the models and can be incorporated into framework - new characteristics of products, draws attention to certain features. There is a role for the human economist there to provide potential scenarios for consideration.
- KS: Saari's work applies here. The models proposed apply to individual rankings that can be aggregated.
- KS: Paradoxes occur when you ask individuals to make a decision.
- Kemper Lewis: Keeney said you CAN aggregate preferences by rating not ranking.
- HC: We do it all the time, the question is never considered.
 |