1999 年 3 巻 1 号 p. 21-29
To develop and provide customer-oriented merchandise of attractive features, it is important for us to incorporate the feeling and words of customers into the process of new product development, that is, to implement the customer science. In recent years, Amasaka seeing the importance of SQC(＊1) as a behavioral science discusses its effectiveness affirmatively by proposing "Science SQC" for the quality improvement of business process of all the divisions (Amasaka et al., 1996; Amasaka, 1997a 1997b 1998; Amasaka & Osaki, 1998).
The subject is a study that constitutes the nucleus of the divisions of higher order (Amasaka, 1996b). It is important for the design strategy to study on "what style of vehicles will sell in the future?". The present study proposes a methodology that will help us improve the quality of business process involved in the designing and discusses its effectiveness by quoting the examples of actual application. To be concrete, this study applies the "SQC technical methods" (Amasaka, 1996a, 1996b) which are a part of core technologies of "Science SQC" to enhance the quality of the design planning job (Nunogaki et al., 1996).
The study first proceeds to materializing the vehicle images as desired by customers followed by the analysis of relevancy between the customer satisfaction assessment and the vehicle appearance review factors on the basis of the appearance style review. Moreover, on the basis of subsequently acquired knowledge, the relationship between the vehicle images and the proportion data will be grasped.
(＊1) Statistical Quality Control
To provide the customer-oriented merchandise of attractive features, it is required for all the divisions concerned in the planning, development, design, manufacturing and marketing to recognize the importance of SQC as a behavioral science, understand the fundamentals of manufacture and conduct verification studies that would raise the engineering level in every step of their business processes.
The task for us today is to implement the customer science that converts customers words correctly into the engineering words. It is thus important for all the divisions to share the objective-consciousness and clearly establish the tacit knowledge on the business processing through the integrated collaboration activities. As a scientific methodology to achieve this, Fig. 1 shows the "Science SQC", a systematic and organizational application of SQC under the new concept and application method where four cores mutually help develop others.
(1) conception and analysis in design
A concept of kansei engineering (see e.g., Shinohara, et al., 1996) has been in recognition for long time. For example, Amasaka et al have developed variable data out of sensory elements such as the digitization of sensory inspection of vehicles and/or mechanization of hunch and knack work, which are being applied to actual development (Amasaka, 1972, 1976, 1983; Shimizu & Amasaka, 1975). On the other hand, examples of study (see e.g., Amasaka, 1996b; Mori, 1991, 1996) are introduced where it is possible to apply a statistical method (analysis) to the development business of design, which can be interpreted as sense itself. When it comes to the application to actual business, however, there are few concrete examples of analysis worthy of introduction to a design development story for the presentation of new vehicles model.
This is attributable to the graphics that the design business (hereinafter referred to as "Designing") often ends up paying more importance to the end result, and that proposing good designs in the design planning or development process is based on the conception, which is not closely related with the analysis. It is apparent that the higher analysis develops, the more important the conception becomes. But the key point to the matter is how to yield a new conception, and the process of developing conception is important. In the advanced information society where identical conditions and/or data are shared, it hardly occurs to have remarkable difference in the environment.
Under the circumstances, the method for proposing a leading conception cannot be the same as before the advancement of the information society. We may be prone to use human power for what a machine can handle and may commit an error of misunderstanding that a conception has been gained. The refore, high-quality creative activities can be carried out by determining and studying the fields assigned to the conception adapted to the times.
Schematic Drawing of "Science SQC"
(2) positioning of "Design SQC"
Thus it is the objective of this study to find the guideline for establishing a method for scientifically supporting the designing so as to establish it expressly as a more creative activity from the state of tacit knowledge. It is considered that the very analysis process for establishing it as an activity would be the key to the successful conception making. It is necessary for us to create a live particular solution that catches the liking of the next generation, not the object teleology that seeks a general solution as the automotive design science. In this connection, "Science SQC" is applied to the flow of designing to actually enhance the quality of the designer's job. This is defined as "Design SQC" and applied as the activity guideline for "ADS" (＊3) project.
In this report, "SQC Technical Methods" will be applied as the scientific method of mountain-climbing for the problem-solving, and by analyzing the bridge portion between analysis (research) and conception (creation of contour images), "Design SQC" will be applied as the concrete conception support tool.
(＊3) Advanced Design by Utilizing "Science SQC" in Toyota
(1) guidelines
Success of designing directly affects the sale of enterprises. Therefore, design business is established as a marketing strategy and its significance lies in the quality of the proposal. True market-in should be in proposing a desirable thing before it is desired.
From Fig. 2 it is important for "Design SQC" to contribute to enhancing individual designer's proposing capability.
(2) approaches
Conventionally, designing is generally developed directly to the profile design after analyzing the research itself (event analysis).
In this study, bridging will be attempted to span the research-oriented analysis as the event analysis to the profile design in three steps of researches from Step 1 to 3 as Fig. 3 shows. The following state, by step, the application process of the profile design support tool that will be the guidelines for establishing a scientific methodology of "Designing" and determine the effectiveness of the "Design SQC" by quoting an application case.
Step 1 analyzes relationship between images of vehicles desirable to customers and those actually selected to search for and actualize apparent relevancy whereby a vehicle type can be specified by the desirable image.
Step 2 grasps what part of a vehicle customers observe to evaluate it. By coming down from the overall assessment, partial assessment and to detailed assessment, this report clarifies which design factor should better be given priority to satisfy customers. By thinking that true customer-in will be to propose a desirable thing before it is desired.
Step 3 classifies the vehicle proportion with the time axis (year) and high class grade (price) to grasp relevancy between the vehicle images and proportion data. This will help improve the designing process. By accumulating the improvement processes, this report intends to improve the quality of "Designing" as the dotted line indicates in Fig. 3.
Desirable relationship between “Designing” and “Design SQC”
(1) Understanding correlationship between customers' desirable images of the vehicles and the most favored vehicles (Step 1)
Questionnaire was used with 157 customers (domestic panel) various in personality. First their desirable images are determined for the vehicles in the panel without indicating the appearance (as shown in Fig. 4-1). Then, four representative vehicle models, consisting of domestic and imported models (BMW850i/1990 model, Benz300-24/1989 model, Legend Coupe/1991 model and Soarer 4.0GT/1991 model) by using photograph without vehicles' brand and name are indicated for them to answer their most favorite vehicle (as shown in Fig. 4-2). Using the desirable images and the most favored vehicle data from the questionnaire, we have conducted four-group discriminant analysis. As a result, it has been found that all four models roughly discriminated with a discrimination ratio of around 70% and that the desirable panel images selected for individual models are almost similar. On the basis of the obtained data, the panel groups belonging to BMW, Benz, Legend and Soarer are specifically extracted. Fig. 5 shows a scatter diagram of individual-score obtained as the result of the analysis using the quantification method of the third type to more clearly understand the correlationship between the desirable images and the most favored vehicle. It is understood from the figure that a panel group in favor of Soarer has desirable vehicle images such as sophistication and/or sportiness as embodied with Soarer. Since similar result has been verified with other vehicle models, it is possible to conclude that customers' liking is consistent. It was found that the customers' words can be materialized as the concrete contour of the vehicle.
(2) Method for exploring important factors customers place in vehicle appearance evaluation (Step 2)
The task taken up here is to objectify the above established theories (implicit knowledge) as qualitative, empirical rules for professional designers who plan automobile profile designs.
In established theories, Japanese users tend to lay greater stress on the front design while North American users lay stress on overall outlook including the front, side and rear views in evaluating a vehicle appearance. As far as the author know, there is no example of objectification based on investigation and analysis.
Therefore, quantitative evaluation on the sections of vehicle appearance customers are interested in will enable to advance a customer-in design strategy. Here, the same 157 customers who took part in Step 1 evaluation make an overall evaluation on the appearance of the four vehicle models. At the same time, they evaluate three vehicle appearance factors, front, side and rear views for their liking, and their cause and effect relationships are checked with multiple regression analysis as analysis I. The three appearance factors are further divided into the design balance (profile) and detailed elements (4, 9, and 5 sections respectively) for a similar study on their causal relationships as analysis II.
A preliminary cluster analysis shows that the customers can be stratified in terms of the overall liking of the vehicle appearance into a group lower in age and annual income and a group higher in age and annual income in their personalities for all four models. Fig. 6 shows an example of analytical results on a vehicle model specified to the group lower in age and annual income. Values in the figure are standard partial regression coefficients B representing the degrees of influence. In analysis I, the contributory factor adjusted for the degree of freedom (R＊2) representing the degree of influence to the overall evaluation of vehicle appearance (XI) is 0.74, indicating a high causal relationship. The breakdown is as follows: The influence of front view (X2) is fairly high at Bfv = 0.46 while those of the side and rear views (X7 and Xl7) are at Bsv = 0.30 and Brv = 0.29, showing their positive influences. In analysis II, the head lamp and grille (X3) have a high degree of influence on the front view (X2) while the overall side view and design (X16) and tail lamp (X20) and rear bumper design (X21) exert much influences on the side view (X7) and rear view (X17) respectively.
「profile design support tool」 using “SQC Technical Methods”
Questionnaire on 「desirable images」 and the 「most favored vehicle」
Correlationship between 「desirable images」 and the 「most favored vehicle」
Causal relationship between customer satisfaction assessment and vehicle appearance assessment factors by multiple regression analysis
In the group higher in age and annual income, though not illustrated, the influence of the front view is even higher at Bfv = 0.59 while the influence of the side and rear views (X7 and X17) are relatively low at Bsv = Brv = 0.18 in analysis I. In analysis II, the analytical results are similar to those for the group with lower age and annual income but the influence of bonnet (X4) is high on the front view (X2). It is verified that the vehicle appearance is evaluated in a wider range; for example, the influence of the line (X19) from the rear to the trunk and the design balance (X22) of the rear as a whole are high on the side view (X7). This analytical trend also applies to other three models.
A similar survey and analysis are conducted in North American market. While the weight of evaluation of the front view is generally high in Japan, it is known that the weight of evaluation of side and rear views are equivalent to that of the front view in North America.
From the above, the conventional theories have analytically been tested.
Through this analytical study approach, designers have understood the need for the customer-in design strategy that gives con sideration to the characteristics of each country (Amasaka, 1996a, 1996b).
(3) Study on the customer-orientedness of profile design (Step 3)
Here, analysis will be made with the following approaches so as to estimate what would be the proportion of a vehicle that would sell in future. First, if we study why customers buy merchandise, the following two factors may be listed; They buy because they get what they pay for. They buy because the product is newer than what they have now.
By supposing that these two factors represent class feeling = price and newness = model year, relationships between these and vehicle proportion (hood ratio: hood length / overall length, trunk ratio: trunk length / overall length, cabin Rio: cabin skirt length / overall length, roof ratio: roof length / overall length, front overhang ratio: front overhang length / overall length, rear overhang ratio: rear overhang length / overall length, wheel base ratio: wheel base length / overall length, roof / cabin ratio: roof length / skirt length, overall height ratio: overall height / overall length, and overall width ratio: overall width / overall length) have been analyzed. The data used for the analysis was measured with the autograph (or measurement diagram). In the case of the "Aristo" ('94 model) for example, the hood ratio is 0.26 ( = hood length of 1,280 mm/overall length of 4,950 mm). A total of 62 vehicle models, domestic and imported (sedans) are selected to measure their proportions (ratios) and a scatter diagram (principal component score) is obtained as Fig.7.
Fig. 7-1 and Fig. 7-2 show the result of stratified classification by the class and the model year respectively. In Fig. 7-1, the class drops toward the right of the scatter diagram and in Fig. 7-2 the model year is younger toward the right. For example, if one of these two sheets of scatter diagrams is placed on top the other, it is understood that it is hard to realize a combination of the highest class and the latest model. In other words, it can be quantitatively verified that the class and the newness are contrary to each other, which is a newly found knowledge. From this, we have been able to obtain a desirable direction for the study so as to forecast the liking of future users, not of current users, by incorporating into analysis a factor containing the concept of time axis that the customers' liking can change in terms of newness. This enables us to propose a desirable thing to customers before they want it, thereby obtaining a guideline that helps us improve the processing of design business by proposing for design process improvement.
Finally, we would like to add that the result of this study has been incorporated in the design of new "ARISTO" (Motor Fan, 1997) that has been produced in a large quantity for sales.
Classification of vehicle model by the 「class degree」
Classification of vehicle model by the 「year mode」
(1) We have analytically indicated the causal relationship of steps that convert the customers' sensory factor which is a direct decision-making factor for purchasing, into a concrete contour as "Design SQC" by utilizing" Science SQC".
(2) By noting that the processing of analysis itself can be part of conception and judgement of the "Designing", "SQC Technical Methods" which are a part of core technologies of "Science SQC", is applied as an analytical method. It is proposed and verified with the examples of application as the profile design support tool.
(3) In the current study, the following items have been verified:
In step 1, customers' liking is consistent and that it is possible to materialize customers' words into contour. In step 2, customers' interest in particular elements of vehicle appearance can be stratified by their age, annual income and country. In step 3, it is possible to forecast the liking of the future users by incorporating the time axis factor of newness for the change of liking.
(4) We would like to add that the result of this study has been incorporated in a newly developed mass production model named "ARISTO" introduced to the market since 1997.
In the future, we intend to propagate and establish design SQC by developing "Science SQC" for the enhancement and expansion of the studies on the area of bridging that leads to conception.
