狩野模型之精煉－三種改善方法與應用 Refining Kano’s Model－Three improvements

(1)

中華大學博士論文

狩野模型之精煉－三種改善方法與應用 Refining Kano’s Model－Three improvements

and applications

系所別：科技管理博士學位學程學號姓名：D09703008 王雅麗指導教授：李友錚教授林少斌副教授

中華民國 101 年 7 月

(2)

此頁空白

(3)

摘要

傳統的狩野模型(Kano’s model)已被各產業及許多研究者廣泛使用，但其品質屬

性的分類方法仍存在一些爭議，而其主要的爭議在於狩野模型的評價表忽略考量 25

個可能結果的屬性強度差異，導致企業無法利用狩野評價表精準的評估品質屬性對產品的影響。再者，狩野模型的分類結果並不是來自真正的量化分析，而且狩野模型也未考量顧客對於確定的品質要項認知之重要性程度。因此，傳統的狩野模型不能準確地反映出客戶的滿意程度。此外，狩野模型無法直接分析改善效果及考量品質屬性改善的優先順序，所以狩野模型須要被精煉。

為解決傳統狩野模型的爭議與限制，本研究提出三個改善方法。首先，本研究針

對狩野評價表中25 個可能結果的屬性強度進行量測，進而發展出新的狩野評價表以

改善品質屬性分類方法的準確性。本研究基於「相似性」(similarity)的概念先針對評價表定義出典型與非典型兩種屬性判斷類別，並透過評價表中這兩種類別間的屬性反應頻率與距離之計算提出「相似性」公式，最後藉由「相似性」之計算結果發展出新的狩野評價表。本研究透過實際案例對於品質屬性強度進行探討，同時驗證新的狩野評價表在品質屬性類別的判定確實比傳統評價表更精準，實證結果也顯示新的狩野評價表之實用性。

其次，重要度與表現度分析(IPA, Importance-performance analysis)常被用來探討改善效果及考量品質屬性改善的優先順序。然而，當品質屬性與客戶滿意度之間有非線性關係存在時，重要度與表現度分析將無法準確的分析重要度及建議品質屬性改善的優先順序，從而可能導致企業做出錯誤的決策。為改善狩野模型的分類結果非來自真正的量化分析，且補強狩野模型無法直接分析改善效果及考量品質屬性改善優先順序之缺點，同時解決重要度與表現度分析在理論假設上的限制，本研究結合修改後的分析性狩野模型(analytical Kano model)與傳統的重要度與表現度分析進而提出新的分析性狩野－重要度與表現度分析模型(analytical Kano-IPA model)以解決狩野模型與重要度與表現度分析的爭議與限制，然後透過量化的方式精準的鑑定改善之優先順序。本研究同時運用修改後的分析性狩野模型來探討品質屬性與客戶滿意度之間的非線性關係。最後，修改後的分析性狩野模型被用來修正品質屬性的重要度與表現度以

(4)

現度分析模型進行探討，同時驗證透過新的分析模型可以獲得較精準的改善優先順序，且企業在品質決策上可以收集到更多且更有價值的資訊。實證結果也顯示分析性狩野－重要度與表現度分析模型之實用性。

最後，為簡化分析性狩野－重要度與表現度分析模型之實際應用步驟，且考量顧客對於確定的品質要項認知之重要性程度，本研究利用重要度與表現度分析來精煉狩野模型，以改善傳統狩野模型且解決重要度與表現度分析的問題。本研究先將重要度的概念整合至狩野模型以提出新的品質屬性分類方法。之後利用迴歸分析的虛擬變數來估計品質表現度對於整體滿意度之影響以判斷品質屬性的類別。最後將表現度的概念整合至新的品質屬性分類方法，進而以重要度與表現度分析精煉狩野模型。藉由實際案例證明，透過重要度與表現度分析精煉後的狩野模型可以使企業在品質決策上收集到更有價值的資訊。

根據前述三個實際案例之驗證，本研究所提出的三個方法確實具可行性與有效性。這三個改善方法不僅是產業中有用的實務工具，也是學術研究的理論方法。

關鍵字：狩野模型、狩野評價表、品質屬性、客戶滿意度、重要度與表現度分析

(5)

ABSTRACT

The traditional Kano’s model is widely used by industries and researchers, but some controversy still exists surrounding the classification of quality attributes. The main controversy is that the model neglects variations in attribute strengths of the 25 possible outcomes in Kano’s evaluation sheet, thus preventing enterprises from using it to evaluate precisely the influences of quality attributes on products. Furthermore, the resultant Kano category is not a true quantitative measure. The Kano’s model does not consider the degree of importance accorded to certain quality elements by customers. Consequently, the traditional Kano’s model cannot precisely reflect the extent to which the customers are satisfied. Additionally, the Kano’s model cannot directly analyze improvement effects and prioritize quality attributes, and consequently must be refined.

To mitigate the disputes and limitations of the traditional Kano’s model, this study proposes three improvement methods. Firstly, this study is to measure the quality attribute strength of 25 possible outcomes in the evaluation sheet to develop a new Kano’s evaluation sheet to improve the accuracy of the classification of the quality attributes. This study defines the canonical and non-canonical judgment of the evaluation sheet based on a novel “similarity” calculation which calculates the response frequency and the distance between canonical judgment and non-canonical judgment. Through the case study, the quality attribute strength is probed and compared with the traditional Kano’s evaluation sheet, the new Kano’s evaluation sheet is more practical because it supports a precise judgment of the category of quality attributes. Empirical results also demonstrate the new Kano’s evaluation sheet is practical.

Secondly, Importance-performance analysis (IPA) is frequently used to discuss improvement effects and prioritize quality attributes. However, when nonlinear effects exist, the IPA model cannot accurately analyze the importance and prioritize improvements, potentially leading to inaccurate decision making. Several problems must be resolved. The resultant Kano category is not a true quantitative measurement, nor is Kano’s model able to analyze improvement effects and prioritize quality attributes directly. Considering the limitations on the assumptions of IPA, this study integrated the revised analytical Kano model and traditional IPA to present a novel analytical Kano-IPA model to mitigate the disputes and limitations of Kano’s model and IPA. Quantified methods were used to

(6)

identify the priority of improvements precisely. This study also discussed the nonlinear relationship between quality attributes and customer satisfaction by employing the revised analytical Kano model. Finally, the revised analytical Kano model was used to adjust the importance and performance of quality attributes to improve the analysis of traditional IPA.

Through the case study, the analytical Kano-IPA model is probed and compared with the traditional IPA, the analytical Kano-IPA model is more practical because the improvement order is gained more precisely and enterprises can gather even more valuable information on quality decisions. Empirical results also demonstrate the analytical Kano-IPA model is practical.

Finally, to simplify the analytical Kano-IPA model in practical applications, and to consider the degree of importance given by customers to certain quality elements, this study used IPA to refine Kano’s model, improving the traditional model and addressing the concerns surrounding IPA. First, this study proposes a novel classification of quality attributes by incorporating the concept of importance into the Kano’s model. Second, the categories of quality attributes are determined based on regression analysis using dummy variables to estimate the impact of attribute performance on the overall satisfaction. Finally, the concept of performance is incorporated into the novel classification of quality attributes, thereby refining the Kano’s model using IPA. According to the case study, using the refined Kano’s model with IPA allows enterprises to gather even more valuable information on quality decisions.

The three preceding case studies verified the feasibility and effectiveness of these methods. The three improvement methods are not only practical tools for industries but are also theoretical methods for academic research.

Keywords: Kano’s model, Kano’s evaluation sheet, quality attribute, customer satisfaction, Importance-performance analysis.

(7)

誌謝辭

終於！終於到了寫致謝辭的這一刻，這意味著研究所生涯即將正式的畫上句點。

四年前不知哪來的想法與勇氣，竟然在離開校園多年後毫不猶豫的決定再度重拾書本從事學術研究。回首博士班的日子，真是一段艱苦且煎熬的歷程，幸虧有指導教授的引領才能順利完成。感謝恩師李博專業的指導及嚴格的敦促，讓學生學習到學術研究應有的精準與態度，而老師亦師亦友的互動、耐心與關心的付出是學生整體研究動力的來源，讓這份研究能一點一滴的堆砌岀今日的成果。此外，老師捍衛學生權益的心意與態度，更著實令學生備感溫馨！另外，感謝阿斌老師於論文研究期間的鼓勵與督促，使學生的論文研究得以順利進行，在此，要向您們說聲：「老師，感謝您們，辛苦了！」

論文口試期間，承蒙黃廷合教授、馬恆教授、蔣德煊副教授、邱紹一副教授及王明郎副教授在百忙之中，給予詳細的審閱與批示，並提供許多寶貴的意見，使得論文的內容更加詳實與嚴謹，在此致上最誠摯的謝意。

雖然博士班的日子充滿挑戰及挫折，但感謝老天讓我遇見一群可愛、風趣又志同道合的同學，不但在學習上相互幫忙且彼此打氣，更在沉悶的日子裡相約出遊舒展壓力，王大哥、雅萍、宜芳、靜宜一路上有你們相挺，讓博士班生涯也充滿了美麗的回憶，謝謝你們！另外，謝謝家族的所有學長姊與學弟妹於修業期間所提供的所有支持與協助。諸如：肖琳學姊、雲瀚學長、美蘭學姊、秋月學姊、嘉蕙學妹等。有你們的陪伴，使我博士班的生涯，一點都不孤獨！

最後，除了要感謝我的家人及朋友外，我最想感謝的是「萬能小助理」小帥Sam，

感謝你包容我所有任性的決定，感謝你支持我所有的夢想並一路相隨，感謝你在我幾度幾乎放棄的時刻堅定我的信念，因為有你的陪伴、協助、鼓勵及相挺，我才得以堅持到最後並完成博士學位，這份榮耀應該是屬於你的！

博士學位的完成，雖是個人的小成就，但卻是來自許多人的關懷與祝福，謝謝大家！謹以此論文獻給所有關心我、幫助我的至親家人、師長與所有貴人。

(8)

CONTENT

Abstract (in Chinese)...………...i

Abstract...………...iii

Acknowledgements (in Chinese)………...v

Content………...vi

List of Tables………...vii

List of Figures………...viii

Chapter 1 Introduction………..…….1

Section 1 Background and motivation……….1

Section 2 Research purpose and procedure……….4

Chapter 2 Literature review………...7

Section 1 The Kano’s model………...7

Section 2 The analytical Kano model………...13

Section 3 The refined Kano’s model………...17

Section 4 Importance-performance analysis………...20

Chapter 3 A new Kano’s evaluation sheet………...23

Section 1 Background and research purpose………...23

Section 2 Methodology………...24

Section 3 Practical example………...33

Section 4 Discussion………...36

Chapter 4 The analytical Kano-IPA model………...38

Chapter 5 Refining Kano’s model using IPA………...48

Chapter 6 Conclusion………...58

References………...61

(9)

LIST OF TABLES

Table 1. Evaluation sheet of Kano’s model………10 Table 2. Scores for functional/dysfunctional features………16 Table 3. Redefinitions of the quality attribute categories of the refined Kano’s model…18 Table 4. Canonical and non-canonical judgments in evaluation sheet………25 Table 5. Modified evaluation sheet of Kano’s model………32 Table 6. Customer requirements of PLM system………34 Table 7. Comparison results between Kano’s evaluation sheet and modified sheet…………35 Table 8. Comparison of number of items for Kano’s category………34 Table 9. Service quality elements of telecommunication industry………42 Table 10. The survey results of service quality elements in telecommunications………43 Table 11. Analysis of analytical Kano-IPA model………44 Table 12. Summarized results of service quality elements in telecommunications………53 Table 13. Dummy variable regression results………53

(10)

LIST OF FIGURES

Figure 1. Kano’s model and five categories of quality attributes………8

Figure 2. The Kano methodology: classification process………9

Figure 3. Two-Dimensional Representation of Kano Quality Categories………13

Figure 4. The analytical Kano model………14

Figure 5. Value function………17

Figure 6. Importance-performance analysis matrix………21

Figure 7. Partially derived curves from Kano’s evaluation sheet (1)………24

Figure 14. A revised analytical Kano model………40

Figure 15. Quality attributes classification of Analytical Kano-IPA model………44

Figure 16. Traditional IPA map of service quality attributes………45

Figure 17. Analytical Kano-IPA map of service quality attributes………45

Figure 18. The Kano’s model refined by IPA………49

Figure 19. The display of the refined Kano’s model using IPA………54

(11)

CHAPTER 1 INTRODUCTION

Section 1 Background and motivation

Kano, Seraku, Takahashi, and Tsuji (1984) proposed a two-dimensional quality model, which has been extensively accepted and applied. Kano’s model isusually used by product manufacturers and companies to evaluate product design and performance. Companies could design customer surveys about their products based on the Kano’s model to find out which product qualities most influence purchase decisions and satisfaction and which qualities do not. Results from those surveys might be used to tweak existing product design and influence the design of future products. Kano’s model was developed by adapting the Motivation-Hygiene theory of Herzberg, Bernard, and Snyderman (1959), also known as the theory of attractive quality.

The theory of attractive quality is useful for understanding different aspects of how customers evaluate a product or an offering. Over the past two decades, this theory has gained its exposure and acceptance through articles in various marketing, quality, and operations management journals (Löfgren & Witell, 2005). Applied in strategic thinking, business planning, and product development, the theory has provided a respectful guidance for innovation, competitiveness, and product compliance (Watson, 2003). Several studies (e.g., Lee & Huang, 2009; Lee & Newcomb, 1997; Lee, Sheu, & Tsou, 2008b; Tan & Shen, 2000; Witell & Löfgren, 2007; Yang, 2005) also noted that a number of benefits are derived from employing the Kano methodology.

Analyzing customer need information is an important task with focus on the interpretation of the voice of customers and subsequently derivation of explicit requirements that can be understood by marketing and engineering (Jiao & Chen, 2006).

(12)

Recently, the two-dimensional quality model developed by Dr. Kano was proven to be an effective instrument for analyzing customer requirements (Lee & Huang, 2009). Kano, et al. (1984) considered two aspects of any given quality attribute, namely, an objective aspect involving the fulfillment of quality and a subjective aspect involving the customers’

perception of satisfaction. Kano’s model determines the customers’ affection toward products and services using a questionnaire, and the information is then used as a reference to improve the customers’ satisfaction. However, the classification of quality attributes by the traditional Kano’s model still encounters some controversy (Lee, Lin, & Wang, 2011;

Lee & Chen, 2009). The Kano methodology is originally a tool for surveying customer satisfaction with quality attributes based on a dysfunctional and functional questionnaire, and then categorizing the results of the survey using an evaluation sheet. The five-level Kano classification scheme yields 25 possible outcomes, which are spread over five quality dimensions, (1) Attractive quality, (2) One-dimensional quality, (3) Must-be quality, (4) Indifferent quality, and (5) Reverse quality. In Kano’s traditional evaluation sheet, all quality attribute strengths are unequal, it is unreasonable and not precise that to sum up equivalently each response frequency of every quality attribute to evaluate the influences of quality attributes. For instance, the quality attribute strength of “attractive” is different between attractive quality attribute of (Neutral, Like) and (Must-be, Like) which were obtained by Kano’s questionnaire, though the both quality attributes are categorized into attractive quality attribute.

Furthermore, the traditional Kano’s model focuses on the qualitative descriptions of various relationship curves; thus, only limited quantitative analysis or measurement of the relationships is discussed (Wang & Ji, 2010). Although Xu, Jiao, Yang, and Helander (2009) proposed the analytical Kano model to mitigate the disputes and limitations of the traditional Kano’s model, the model still has room for improvements regarding the scoring

(13)

scheme which is against the basic assumption of prospect theory. In addition, the disadvantage of the traditional Kano’s model is that it fails to consider the degree of importance given by customers to certain quality elements (Yang, 2005). Therefore, the traditional Kano’s model cannot precisely reflect the extent to which the customers are satisfied. Moreover, the Kano’s model cannot directly analyze improvement effects and prioritize quality attributes, and consequently must be refined.

Since Martilla and James (1977) first used the market strategy developed and organized by Importance-performance analysis (IPA), the method has been widely used as the primary tool for customer satisfaction management. IPA is frequently used to discuss improvement effects and prioritize quality attributes. Though IPA is equipped with easy application and interpretation features, some hidden problems exist. Two implicit assumptions underlie IPA: (1) Attribute performance and attribute importance are two independent variables, and (2) the relationship between quality attribute performance and overall performance is linear and symmetric (Matzler, Bailom, Hinterhuber, Renzl, &

Pichler, 2004a). Under the assumptions of IPA, when practical information does not accommodate these assumptions, the traditional IPA model cannot accurately analyze the importance of quality attributes and prioritize improvements. To address this situation, numerous researchers have integrated the Kano’s model into IPA (e.g., Eskildsen &

Kristensen, 2006; Lee, Cheng, & Yen, 2009; Yang, 2003).

Although Yang (2005) proposed a new and improved model to manage these issues, the model still has room for improvements regarding practical applications due to of its relative complexity. From a comprehensive survey of past research concerning the Kano’s model integrated with IPA (e.g., Eskildsen & Kristensen, 2006; Lee, et al., 2009; Matzler, et al., 2004a; Yang, 2005; Yang, 2003), numerous studies were found to primarily employ Kano’s categories to explain the asymmetrical and nonlinear effects of IPA before using the

(14)

categorized results of quality attributes to modify the importance of IPA. A few studies also further considered the impact of the categories of quality attributes on the performance of IPA. In these studies, though traditional IPA was improved by integrating the Kano’s model, the limitations and problems of the Kano’s model were not considered.

Section 2 Research purpose and procedure

To help fill the gap of previous research, and to mitigate the disputes and limitations of the traditional Kano’s model, this study proposes three improvement methods. Through these three methods, this study attempts to improve and refine the traditional Kano’s model.

At the same time, three case studies are employed to verify the feasibility and effectiveness of these respective methods in practice.

First, regarding the controversy of classifying quality attributes, the traditional Kano’s model lacks precise evaluation of the influences of quality attributes, and neglects variations of attribute strength of the 25 possible outcomes in Kano’s evaluation sheet.

Therefore, the purpose of this study is to present a new Kano’s evaluation sheet with 25 judgments which are used to review and redefine the quality attribute strength of the 25 possible outcomes in the Kano’s traditional evaluation sheet. Meanwhile, the customer satisfaction index (SII and DDI ), developed by Kuo (2004), is utilized by enterprises as valuable information in making quality-related decisions. The new evaluation sheet and the customer satisfaction index (SII and DDI ) are demonstrated in a case study.

Second, several problems should be resolved. The resultant Kano category is not a true quantitative measurement, nor is Kano’s model able to analyze improvement effects and prioritize quality attributes directly. Considering the limitations on the assumptions of IPA, this study presents a novel analytical Kano-IPA model to integrate the analytical Kano model and traditional IPA to mitigate the disputes and limitations of Kano’s model and IPA.

(15)

First, the study exploits the analytical Kano model to establish non-linear relationship between quality attributes and customer satisfaction. Considering that the scoring scheme of the original analytical Kano model is against the basic assumption of the prospect theory, this study revises the original Kano indices at the initial stage, and then suggests improvements for classifying quality attributes. Second, the revised Kano indices were used to adjust the importance and performance of quality attributes to improve the analysis of traditional IPA. Finally, the analytical Kano-IPA model is demonstrated in a case study.

Finally, to simplify the analytical Kano-IPA model in practical applications, and to overcome the traditional Kano model’s failure to consider the degree of importance given by customers to certain quality elements, this study refined Kano’s model with IPA, using the benefits of integrating Kano’s model and IPA to resolve the related difficulties.

Subsequently, this study provides a simpler, quantitative, and more accurate method to easily enable companies and firms to make quality decisions with greater precision. First, this study proposes the novel classification of quality attributes, using the analytical Kano-IPA model as a basis, and referencing the defined eight categories of quality attributes from the refined Kano’s model (Yang, 2005). Second, regression analysis with dummy variables is used to estimate the impact of attribute performance on overall satisfaction and then to determine the categories of quality attributes. Then, this study considers the impact of the quality attribute categories, the attribute performance, and the asymmetric impact of attribute performance on overall satisfaction before suggesting definite priorities for improvement. Finally, the refined Kano’s model using IPA is demonstrated in a case study.

The subsequent sections proceed as follows. Chapter 2 is divided into four major parts.

Section 1 presents a critical review of the Kano’s model. Section 2 presents a review of the analytical Kano model, and a review of the refined Kano model is presented in Section 3.

(16)

Section 4 reviews the importance-performance analysis, and a new Kano’s evaluation sheet is proposed in Chapter 3, including a practical example and discussion. Chapter 4 presents the analytical Kano-IPA model and discusses the application of the model in the case study.

The refined Kano’s model using IPA is proposed in Chapter 5 by implementing the proposed method that addresses fundamental concerns. Finally, Chapter 6 provides some concluding remarks and recommendations.

(17)

CHAPTER 2 LITERATURE REVIEW

Section 1 The Kano’s model

Since Kano, et al. (1984) proposed the two-dimensional quality model, it has been widely accepted and applied. The two-dimensional quality model (Kano, et al., 1984) was developed by adapting the Motivation-Hygiene theory of Herzberg, et al. (1959). The M-H theory posited that the factors which cause job satisfaction were different from those cause job dissatisfaction. Inspired by Herzberg’s M-H theory in behavioral science, Professor Kano and his co-workers developed the concept of an “M-H property of quality,” positing that quality is not a one-dimensional construct. Kano, et al. (1984) named their theory

“attractive quality” and “must-be quality” to differentiate it from M-H theory.

Professor Kano first pointed out Customer satisfaction was not only included with linear function for quality attributes (i.e. one‐dimensional), but also with a nonlinear function of quality performance depended on its attribute (i.e. must‐be, attractive). Several scholars (Anderson & Sullivan, 1993; Lee, Hu, Yen, & Tsai, 2008a; Lee, et al., 2009;

Oliva, Oliver, & Bearden, 1995) and practitioners (Coyne, 1989) have agreed with Kano on this assertion. Matzler, Fuchs, and Schubert (2004b) further pointed out that the relationship is not only nonlinear but also asymmetric.

The theory of attractive quality explains how the relationship between the degree of sufficiency of a given quality attribute and customer satisfaction. To understand the role significance of quality attributes, Kano, et al. (1984) presented a model which evaluates patterns of quality on the basis of customers’ satisfaction with specific quality attribute and their degree of sufficiency, as shown in Figure 1.

The five categories of quality attributes proposed by Kano, et al. (1984) are：

(18)

z Attractive：Attractive quality attributes can be described as surprise and delight attributes, and provide satisfaction when achieved fully but do not cause dissatisfaction when not fulfilled.

z One-dimensional：customer satisfaction is proportional to the level of fulfillment -- the higher the level of fulfillment, the higher customer satisfaction, and vice versa.

z Must-be：Must-be quality attributes are taken for granted when fulfilled but result in dissatisfaction when not fulfilled.

z Indifferent：customer satisfaction will not be affected no matter whether this quality is provided or not.

z Reverse：customers will be dissatisfied if this quality attributes is provided; otherwise, they will be satisfied. (Kano, et al., 1984; Matzler & Hinterhuber, 1998; Tan & Shen, 2000).

Note. From “Attractive quality and must-be quality,” by Kano, et al., The Journal of the Japanese Society for Quality Control, 14(2), p.39-48.

In the original version of the theory of attractive quality (Kano, et al., 1984), the classification process is based on utilizing Kano questionnaire. This questionnaire consists of pairs of questions about customer requirement and each question has two parts: how do you feel if that feature is present in the product (functional form of the question), and how

Customer Satisfied Attractive Quality

One-dimensional Quality

Indifferent Quality

Quality attribute Dysfunctional Quality attribute Fully Functional

Reverse Quality Must-be Quality

Customer Dissatisfied

Figure1 Kano’s model and five categories of quality attributes

(19)

do you feel if that feature is not present in the product (dysfunctional form of the question) (Kano, et al., 1984). For each part of the questions, customers select one out of five alternative answers. These five alternatives are described as “Like”; “Must-be”; “Neutral”;

“Live-with”; and “Dislike” (Matzler & Hinterhuber, 1998). The customers’ perceptions are then evaluated into quality dimensions on the basis of how the respondents perceived the functional and dysfunctional form of a quality attribute (as displayed in Figure 2). The five-level Kano classification scheme thus has 25 possible outcomes (see Table 1), which are spread over five quality dimensions, which are (1) Attractive quality (A), (2) One-dimensional quality (O), (3) Must-be quality (M), (4) Indifferent quality (I), and (5) Reverse quality (R).

Figure2 The Kano methodology: classification process

(20)

Table1

Evaluation sheet of Kano’s model

Dysfunctional Customer requirements Like Must-be Neutral Live-with Dislike

Functional Like Q A1 A2 A3 O1

Must-be R1 I1 I2 I3 M1

Neutral R2 I4 I5 I6 M2

Live-with R3 I7 I8 I9 M3

Dislike R4 R5 R6 R7 Q

Notes: A: Attractive, O: One-dimensional, M: Must-be, I: Indifferent, R: Reverse, Q: Questionable Note. From “How to make product development projects more successful by integrating Kano’s model of customer satisfaction into quality function deployment,” by K. Matzler and H. H. Hinterhuber, Technovation, 18(1), p.25-38.

The category of the highest response frequency determines each placement of the quality attribute in the classification scheme. If two or more of Kano’s categories are tied for a given quality attribute, the selected category would have the greatest impact on the products and services. It is determined based on the following order: M＞O＞A＞I (Berger, et al., 1993).

The evaluation sheet of Kano’s model can be easily used to define the category of quality attributes that influence customer satisfaction. However, some controversies about Kano’s evaluation sheet remain, and some scholars have proposed modified methods, which have some practical deficiencies when applied. In a review of empirical studies of the Kano methodology, Löfgren and Witell (2008) assessed different approaches to the classification of the quality attributes identified in 28 studies. Most related studies concern the quality dimension and wording, for example, Lee and Newcomb (1997) modified the traditional methodology of Kano by classifying five combinations of 25 options as

(21)

questionable, and proposed category strength, (defined as the percentage difference between the highest category and second-highest category) and total strength (defined as the total percentage of attractive, one-dimensional, and must-be responses). Tontini (2000) enlarged the 5 × 5 evaluation sheet to a 7 × 7 evaluation sheet and changed the wording.

Kano (2001) introduced a three-level Kano questionnaire to improve the classification process, but there was no empirical investigation of this methodology. Jané and Dominguez (2003) simplified the 5 × 5 evaluation sheet to a 3 × 3 evaluation sheet, which they claimed was more suitable for use with a long questionnaire. Yang (2005) incorporated the concept of importance, and provided eight quality dimensions. Numerous studies have suggested different approaches for the Kano model. (e.g., Lee, Ho, & Liang, 2006; Lee, Ho, Liang, & Lin, 2007; Lee & Chen, 2009; Matzler & Hinterhuber, 1998;

Schvaneveldt, Takao, & Masami, 1991)

A literature review of research on the theory of attractive quality by Löfgren and Witell (2008) highlighted the notion of a life cycle of quality attributes as one of the most interesting and fruitful developments of the theory of attractive quality. Moreover, several studies within the marketing field have suggested that quality attributes are dynamic (Nilsson-Witell & Fundin, 2005). According to Kano (2001), the roles of quality attributes conceptually change over time. Therefore, a successful attribute follows a life cycle from being indifferent, to being attractive, to being one-dimensional and, ultimately to being a must-be item.

Berger, et al. (1993) proposed the Better and Worse value, and then Kuo (2004) named it the customer satisfaction index, which indicates whether satisfaction can be increased by providing quality attributes, or whether fulfilling quality attributes only prevented the customer from being dissatisfied. Based on a satisfaction increment index (SII) and dissatisfaction decrement index (DDI), quality attributes can be selected to

(22)

optimize outcome. This method can also be used to differentiate priority for the same attribute category. According to Kuo (2004), the customer satisfaction index is as follows.

SII = (A + O) / (A + O + M + I) DDI = - (O + M) / (A + O + M + I)

where A denotes response frequency of Attractive attributes; O denotes response frequency of one-dimensional attributes, M denotes response frequency of must-be attributes, and I denotes response frequency of Indifferent attributes.

A minus sign precedes the DDI to emphasize its negative influence on customer satisfaction when this quality attribute is not fulfilled. The SII index can suggests attributes that can increase customer satisfaction. The DDI index can be used to indicate attributes whose absence decreases customer satisfaction.

In terms of SII and DDI, when the value is close to 0, the effect of the attribute is low.

When it is close to 1, the attribute has a positive effect on increasing customer satisfaction;

when it is close to -1, the attribute can decrease customer dissatisfaction. Pairs of SII and DDI points for each quality attribute can be plotted on a two-dimensional diagram, as shown in Figure 3 (the minus sign in front of DDI has been ignored in this diagram for purposes of clarity) (Berger, et al., 1993). As specified by these two indexes, we can roughly estimate the category of quality attributes, and determine which attributes influence customer satisfaction. Attention should be paid to attributes with indices with large magnitudes (Kuo, 2004).

(23)

Note. From “Kano’s methods for understanding customer-defined quality,” by Berger, et al., The Center for Quality Management Journal, 2(4), p.3-36.

Section 2 The analytical Kano model

To summarize the preceding research, the Kano diagram is found to provide a rough sketch of customer satisfaction regarding the level of attribute performance. A convenient method of employing quantitative measures is to assign scales for levels of customer satisfaction/dissatisfaction (Matzler & Hinterhuber, 1998). However, the resulting Kano category is still qualitative in nature, which does not precisely reflect the extent to which customers are satisfied (Berger, et al., 1993). Hence, Xu, et al. (2009) proposed an analytical Kano model based on the Kano principles to incorporate quantitative measures into customer satisfaction. The classification of a quality attribute can be defined based on the corresponding location of the value pair (dysfunctional and functional) in the diagram, as shown in Figure 4.

Figure3 Two-Dimensional Representation of Kano Quality Categories 1.0

1.0

0.0

Must-be

Attractive One-

dimensional

Indifferent

DDI SII

(24)

Note. From “An analytical Kano model for customer need analysis,” by Xu, et al., Design Studies, 30(1), p.87-110.

From the customer’s perspective, the characteristics of a quality attribute can be represented as a vector rv , the magnitude of the vector denotes the overall importance of quality attribute to customers, and the angle α determines the relative level of satisfaction and dissatisfaction. Therefore, the magnitude of the vector rv is called the importance index; and the angle α is called the satisfaction index. Both 0≤ r≤ 2 and 0≤α ≤π 2

are collectively called the Kano indices (Xu, et al., 2009). According to Xu, et al. (2009), the Kano indices are as follows.

2 2

i i

i X Y

rv = + , 0≤ rr_i ≤ 2 where X_i ,Y is the average level of satisfaction of the quality element_i ⁱ for the

dysfunctional/functional form question.

(

i i

)

i =tan⁻¹Y X

α , 0≤α_i ≤π 2 Figure4 The analytical Kano model

Attractive

One-dimensional

Must-be

Indifferent α

r

0 1

1

r0

αH

αL

(25)

whenα_i =0means that quality element ⁱ is an ideal must-be attribute. Conversely, π 2

α_i = means that quality elementⁱ is an ideal attractive attribute.

Based on the above formulation, the quality attributes can be classified into four categories, i.e., indifferent, must-be, attractive and one-dimensional (see Figure 4). In Figure 4, a threshold value of the importance indexr is used to differentiate important 0

quality attributes from less important ones. If the radius is smaller thanr , it would be 0

considered as the indifferent region. Hence r is called an indifference threshold. 0

Likewise, a lower threshold value of the satisfaction index is defined asαL, such that for quality attributeⁱ , ifr_i >r0 and αi≤αL, it is considered as a must-be quality attribute. A

higher threshold value of the satisfaction index is defined asα , such that for quality H

attributeⁱ , ifr_i > r0 and αi >αH, it is considered as an attractive quality attribute. If

r0

r_i > and αL<αi≤αH, quality attributeⁱ is considered as a one-dimensional quality attribute. The set of thresholds, r ,0 αL_{, and}α are collectively called Kano classifiers _H

and denoted as k=

(

r₀,αL,αH

)

(Xu, et al., 2009). Determining appropriate values of Kano classifiers is challenging in that these threshold values may be problem-specific and context-aware for different applications (Xu, et al., 2009).

Xu, et al. (2009) adopts a scoring scheme that defines customer’s satisfaction and dissatisfaction as shown in Table 2. The scale is designed to be asymmetric because Xu, et al.argued thatpositive answers are considered to be stronger responses than negative ones.

In other words, the scaling has the effect of diminishing the influence of negative evaluations. This Argument of Xu, et al. is, however, against the prospect theory.

(26)

Table2

Scores for functional/dysfunctional features

Answers to the Kano question Functional form of the question

Dysfunctional form of the question

I like it that way (like) 1 -0.5

It must be that way (must-be) 0.5 -0.25

I am neutral (neutral) 0 0

I can live with it that way (live with) -0.25 0.5

I dislike it that way (dislike) -0.5 1

Note. From “An analytical Kano model for customer need analysis,” by Xu, et al., Design Studies, 30(1), p.87-110.

The prospect theory of Kahneman and Tversky (1979) represents an alternative theory of decision making under uncertainty. Based on a series of experimental observations, Kahneman and Tversky(1979) propose a value function(sketched in the Figure 5) defined on the gains or losses relative to a reference point, instead of the absolute level of consumption or wealth. The value function which passes through the reference point is s-shaped and, as its asymmetry implies, given the same variation in absolute value, there is a bigger impact of losses than of gains. In other words, the displeasure associated with loss is greater than the pleasure associated with the same amount of gains. Therefore, the revision of the scoring scheme used to evaluate customer satisfaction and dissatisfaction by Xu, et al. (2009) is necessary. Otherwise, it would influence the precision of the Kano indices.

(27)

Note. From “Prospect theory: An analysis of decision under risk,” by D. Kahneman and A.

Tversky, Econometrica, 47, p.263-291.

Section 3 The refined Kano’s model

Kano’s model is widely used by industries and researchers. However, the model has a deficiency that prevents enterprises from precisely evaluating the influences of quality attributes. The disadvantage of the traditional Kano’s model is that it fails to consider the degree of importance given by customers to certain quality elements. Therefore, Yang (2005) proposed the refined Kano’s model, which incorporates the concept of importance.

The refined model divides Kano’s first four categories into eight categories, namely, highly attractive and less attractive, high value-added and low value-added, critical and necessary, and potential and care-free. Table 3 lists the redefined quality attribute categories obtained by refining Kano’s model (Yang, 2005).

The redefinitions of the quality attribute categories according to the refined Kano’s model allow enterprises to make quality decisions with more precision. However, for

Value

Gains Losses

Reference Point

Figure5 Value function

(28)

Table3

Redefinitions of the quality attribute categories of the refined Kano’s model Categories of

quality attributes in the Kano’s

model

Categories of quality attributes with high

importance in the refined model

Categories of quality attributes with low

importance in the refined model

Descriptions of redefinition

Attractive Highly attractive Less attractive

Quality attributes with high importance can be classified as highly attractive quality attributes, whereas those of lesser importance can be classified as less attractive quality attributes.

One-dimensional High value-added Low value-added

Increasing such attributes will enhance customer satisfaction. A one-dimensional quality attribute is, therefore, a value-added quality attribute. Thus, some one-dimensional quality attributes with high importance can be defined as high value-added quality attributes, whereas others can be classed as low value-added attributes.

Must-be Critical Necessary

If such a quality attribute is found to be of high importance to the customers, this quality attribute is not only a necessary quality requirement but a critical quality requirement. By contrast, it can be defined as a necessary quality requirement, but without being considered critical.

Indifferent Potential Care-free

If an indifferent quality attribute does possess greater importance than another, it can be defined as a potential quality attribute because it has some potential to attract customers. The indifferent quality attributes can be classed as carefree or potential depending on their degree of importance.

(29)

practical applications three separate questionnaires must be designed to facilitate the survey, complicating the process. The three questionnaire topics include the importance of quality attributes, the satisfaction of quality attributes, and the categorization of attributes according to Kano’s model. Additionally, direct measurers of attribute importance tend to be ambiguous and unreliable (e.g. Oliver, 1997).

Some empirical evidence indicates that attribute importance changes during different phases of the purchase and consumption process. For example, Gardial, Clemons, Woodruff, Schumann, and Burns (1994) found significant differences in attribute importance during the pre-purchase phase and post-purchase evaluation using consumers’

inductive retrospective verbalizations recalling their pre- and post-purchase product experience. Though determinant attributes in the pre-purchase phase were more specific and goal-directed, post-purchase evaluations of attributes tended to be at an aggregate level and significantly different from the attributes mentioned in the pre-purchase phase.

Therefore, if the researcher and the consumer do not relate importance weights to the same situation, the attributes are ambiguous and difficult to interpret (Matzler & Sauerwein, 2002).

Matzler and Sauerwein (2002) indicated that when using a form of self-stated importance, customers may not consider the current level of attribute satisfaction, whereas a regression analysis determines the weight of importance at the current level of performance. Simultaneously, Matzler and Sauerwein believe that attribute importance is a function of performance. Matzler, et al. (2004a) further confirmed the asymmetric relationship between attribute performance and overall satisfaction using regression analysis with dummy variables. Additionally, through the results of regression analysis, they identified the asymmetric impact of attribute performance on overall satisfaction, and subsequently selected the category of quality attributes. Matzler, et al. (2004a) suggested that if the impact on overall satisfaction with quality attributes is high when the performance is low, but satisfaction when performance is high is not affected, then these can be classified as must-be quality attributes. If quality attributes have a higher impact on overall satisfaction when performance is high, but do not affect satisfaction when performance is low, then these can be classified as attractive quality attributes. If the impact of attribute performance (whether high or low) on overall satisfaction is high, they can be classified as one-dimensional quality attributes.

(30)

Section 4 Importance-performance analysis

Quality management systems developed in the 1980s used customer satisfaction as the primary measurement indicator of organizational performance. This resulted in a great number of studies of service quality using importance and satisfaction. Therefore, numerous researchers use importance–performance analysis (IPA) to determine the advantages and disadvantages directly from the organization’s market information analysis.

IPA, originally proposed by Martilla and James (1977), provides insights into product or service attributes, which firms should address to enhance customer satisfaction. The fundamental concept of IPA is to appreciate the importance of customer cognition toward quality attributes through market surveys used to measure the degree of the customer satisfaction. IPA analyzes quality attributes on two dimensions: their performance level (satisfaction), and their importance to the customer. Evaluations of quality attributes in these two dimensions are then combined into a matrix that allows a firm to identify the key drivers of satisfaction, and to form improvement priorities. Therefore, IPA is very helpful in deciding how to best allocate scarce resources to maximize satisfaction.

The means of performance and importance divide the matrix into four quadrants (see Figure 6): (I) Keep Up the Good Work: Attributes in this quadrant, evaluated as high for both satisfaction and importance, represent opportunities for gaining or sustaining competitive advantage. The quality attributes in this area should be maintained. (II) Possible Overkill: This means that customers pay less attention to the quality attributes, but are still satisfied with the performance of quality attributes, suggesting that the supply of quality attributes in this quadrant exceeds demand. This result implies that resources committed to these attributes would be better employed elsewhere. (III) Low Priority:

Although customers pay less attention to these quality attributes, they are not satisfied with the listed quality attributes. The quality attributes should be improved, but are not the first priority. (IV)Concentrate Here: This means that customers think highly of the quality attributes, but they are not satisfied with the quality. Ignorance of these attributes poses a serious threat to the firm. Therefore, quality attributes in this quadrant should be listed first on the improvement list. From the division, managers can organize and implement improvements with limited resources. Plotting the results on an x-y axis can help managers identify which strategies to adopt first regarding customer satisfaction improvements.

(31)

Note. From “Importance-performance analyses,” by J. A. Martilla and J. C. James, Journal of Marketing, 41(1), p.77-79.

There are two implicit assumptions underlie the IPA: (1) Attribute performance and attribute importance are two independent variables. (2) The relationship between quality attribute performance and overall performance is linear and symmetric (Matzler, et al., 2004a).

Since Martilla and James (1977) initially employed the market strategy developed and organized by IPA in a practical application, the IPA method has been widely applied in various industries for almost 30 years. For instance, the latest study Matzler, Sauerwein, and Heischmidt (2003) used IPA to improve bank service quality and development strategy.

Zhang and Chow (2004) exploited IPA to improve tourism guide service quality. Matzler, et al. (2004b) used IPA in the automobile industry. Lee, Yen, and Tsai (2008c) employed IPA in supplier performance evaluation. Lee, Yen, and Tsai (2008d) used DEMATEL and IPA to improve organization performance in computer industry. In recent years, numerous researchers have attempted to modify conventional IPA to increase its rationality. In the study by Yavas and Shernwell (2001), the performance and competitor’s difference were multiplied by the relative importance to modify the IPA model. Tarrant and Smith (2002) used the average and standard error to modify the IPA model. Although these studies have Figure6 Importance-performance analysis matrix

Importance High Low

Performance High

Low

Quadrant I Keep Up the Good

Work

Quadrant IV Concentrate Here Quadrant II

Possible Overkill

Quadrant III Low Priority

(32)

made significant contributions, they are still limited by the assumptions that the importance and performance of quality attributes are independent in the conventional IPA model.

Therefore, a number of researchers believe that some relationship must exist between the importance and the performance of quality attributes; thus, they consider customer satisfaction to induce a functional relationship between importance and performance.

Matzler, et al. (2004a), Oh (2001), and Ryan and Huyton (2002) have indicated that the importance and satisfaction regarding quality attributes are related. Moreover, some studies (e.g., Matzler & Sauerwein, 2002; Sampson & Showalter, 1999) not only presented a correlation between the importance and the performance of quality attributes, but also established a linear relationship among importance, performance, and satisfaction.

In recent years, research (e.g., Matzler & Renzl, 2007; Matzler, et al., 2004a; Ting &

Chen, 2002) has shown that the relationship between attribute performance and overall satisfaction is asymmetric.To address this issue, numerous researchers have integrated Kano’s two-dimensional quality model into IPA. For instances, Eskildsen and Kristensen (2006) integrated the loss function of Taguchi, the two-dimensional quality model of Kano and the regression analysis into IPA to enhance IPA model. Yang (2003) integrated Kano’s two-dimensional quality model and the crucial customer interview into IPA to improve the electrical appliance maintenance service quality for the purpose of enhancing organizational competition. Yang (2005) put forth the revised two-dimensional quality model by Kano and exploited IPA to collect more valuable decision making information on quality. Lee, et al. (2009)integrate Kano’s model and IPA to improve order-winner criteria and win orders. Others adopted a three-factor theory: basic factor, performance factor and excitement factor, to provide insights into the satisfaction issues. However, the majority of research primarily focused on the asymmetrical and nonlinear effects. Although a small number of studies further utilize the categorized results of quality attributes by the Kano's model to modify the importance of IPA, few studies have considered the impact of the quality attribute categories on the performance of IPA.

The above researchers have made numerous significant contributions to IPA methodology. Their research not only demonstrates the linear relationships between quality attributes and satisfaction, but also confirms that quality attributes still have nonlinear effects on customer satisfaction. However, when nonlinear effects exist, the IPA model cannot accurately analyze the importance and prioritize improvements, potentially leading to inaccurate decision making.

(33)

CHAPTER 3 A NEW KANO’S EVALUATION SHEET

Section 1 Background and research purpose

Kano’s model offers some insight into the product or service attributes which are perceived to be important to customers. It is a useful technique for differentiating product or service features, but some controversy still exists surrounding the classification of quality attributes. Many scholars have modified Kano’s evaluation sheet, but Witell and Löfgren, (2007) recommended that practitioners use the five-level Kano methodology.

However, there seems to be an inconsistency between the Kano diagram and Kano’s traditional evaluation sheet. For example, how is the quality attribute of (Dislike, Must-be) judged to be “M”(Must-be), rather than “O”(One-dimensional) or a combination of “M”

and “O”？ How is the quality attribute of (Live-with, Like) judged to be “A”(Attractive), rather than “O”(One-dimensional) or a combination of “A” and “O”？ Furthermore, the model locks precise evaluation on the influences of quality attributes, and neglects variations of attribute strength of the 25 possible outcomes in Kano’s evaluation sheet. The aforementioned literature review revealed that the five-level Kano classification scheme has 25 possible outcomes (Table 1), which could be identified in five quality dimensions - (1) Attractive quality (A) : three outcomes; (2) One-dimensional quality; (O) : one outcome; (3) Must-be quality (M) : three outcomes; (4) Indifferent quality (I) : nine outcomes; (5) Reverse quality (R) : seven outcomes. In Kano’s traditional evaluation sheet, the attribute strength of all outcomes are treated equally (A1= A2= A3; M1= M2= M3) in classification of quality attributes. However, this method results in the categorization of too many attributes as indifferent, relative to the numbers in the must-be and attractive categories, and too few attributes in the one-dimensional category. Hence, the category with less impact on satisfaction would be selected as a target for improvement. This result of categorization using Kano’s traditional evaluation sheet is unreasonable in practice.

In fact, the attribute strength of all outcomes are unequal. With reference to Table 1, consider the attractive quality attributes. If A2 is the baseline, then the “attractive” attribute strength would be: A2＞A3 and A2＞A1, and the attribute strength of A3 would be

(34)

affected by O1 (one-dimensional quality attribute). Likewise, the “must-be” attribute strength: M2＞M3 and M2＞M1, and the attribute strength of M1 would be affected by O1. Accordingly, every quality attribute could be further divided into the category to which it belongs. Therefore, the purpose of this study is to review the 25 possible outcomes of its attribute strength and then make 25 judgments about the 25 possible outcomes using a new Kano’s evaluation sheet, to improve the accuracy of the classification of the quality attributes.

Section 2 Methodology

This study develops a new Kano’s evaluation sheet with 25 judgments which are used to review and redefine the quality attribute strength of the 25 possible outcomes in Kano’s traditional evaluation sheet. Meanwhile, the customer satisfaction index (SII and DDI) that were developed by Kuo (2004) are used to provide enterprises with valuable information for making decisions regarding the quality of products and services.

A review of Kano’s five quality categories reveals that only nine of 25 judgments in response to pairs of questions coincide completely with the definitions of the categories to which they belong, as shown in Figure 7. In the following figures and equations, l stands for “like”; m stands for “must-be”; n stands for “neutral”; lw stands for “live-with” and d stands for “dislike”.

Customer Satisfied (l, l)

(l, d) (n, l)

(m, m)

(d, l)

(n, n)

Quality Element Dysfunctional Quality Element Fully Functional

(lw, lw)

(d, n)

(d, d)

Customer Dissatisfied

Figure7 Partially derived curves from Kano’s evaluation sheet (1)

(35)

In Figure 7, because the shape of judgment (m, m) is similar to that of judgments (l, l), (n, n) and (d, d), and judgments (l, l) and (n, n) are the two nearest neighboring judgments to (m, m), judgment (m, m) is represented as a linear combination of judgments (l, l) and (n, n), and so on. Accordingly, the other seven judgments in response to pairs of dysfunctional and functional questions, which are (l, l), (l, d), (n, l), (n, n), (d, l), (d, n) and (d, d) represent delight-indifferent (I1), reverse, attractive, neutral-indifferent (I2), one-dimensional, must-be and dissatisfaction-indifferent (I3) canonical attributes, respectively, and are called canonical judgments in this study.

The other 16 judgments coincide partially with the definitions of one of the categories.

In this study , these 16 judgments and the judgments (m, m) and (lw, lw) are called non-canonical judgments, which should be classified neither using the evaluation sheet proposed by Kano nor by directly applying the impact rule proposed by Berger, et al.

(1993), but by their similarity with their neighboring canonical judgments (Table 4).

Table4

Canonical and non-canonical judgments in evaluation sheet

Dysfunctional Customer requirements Like Must-be Neutral Live-with Dislike

Functional Like I1 Q A Q O

Must-be Q Q Q Q Q

Neutral Q Q I2 Q M

Live-with Q Q Q Q Q

Dislike R Q Q Q I3

Notes: A: Attractive, O: One-dimensional, M: Must-be, I: Indifferent, R: Reverse, Q:

Questionable

Based on the assumption that the similarity of a given non-canonical judgment to its certain neighboring canonical judgments is inversely proportional to the distance between them and directly proportional to the response frequency of the neighboring canonical judgment, for each non-canonical judgment, the proportion of similarity to the ith neighboring canonical judgment, S_NJ(CJi), is

(36)

where fNJ(CJi) is the response frequency of the ith neighboring canonical judgment of the non-canonical judgment NJ.

d(NJ, CJi) is the distance between non-canonical judgment NJ and its ith

neighboring canonical judgment.

In Table 4, for instance, fNJ(CJ_i) of the non-canonical judgment (m, l) contains the response frequency f(I1) of the canonical judgment (l, l) and that f(A) of the canonical judgment (n, l). Furthermore, in Table 4, each row and each column corresponds to a unit of distance. Therefore, the distance between the non-canonical judgment (m, l) and the canonical judgment (l, l) is one; the distance between the non-canonical judgment (m, l) and the canonical judgment (n, l) is one; the distance between the non-canonical judgment (l, m) and the canonical judgment (l, d) is three.

For simplicity and practical application, smaller proportion of similarity can be ignored, which implies that only the nearer and similar neighboring canonical judgments should be considered in the determination of the proportion of similarity, farther and dissimilar neighboring canonical judgments of the non-canonical judgment can be neglected.

Consider the non-canonical judgment (m, l), the proportion of similarity to (l, l) and (n, l), which are the nearer and similar neighboring canonical judgments, will be

That is

( ) ( ) ( )

( ) ( )

⁽¹⁾

=

∑

i

i i

NJ

i i

i NJ NJ

NJ,CJ d

CJ f

NJ,CJ d

CJ CJ f

S

( )

( ) ( )

( )

( ) ( )

( )

^n,^l ^is^the ^response ^frequency ^of^the ^canonical ^judgment

( )

^n,^l

l,l is the response frequency of the canonical judgment l,l where

l n, l

l, l l n,

n,

l n, l

l, l l l,

l,

l m,

f f

S f

f f

S f

= +

( )

( ) ( )

( )

( ) ( )

Â îs^the ^response ^frequencyôf^the^canonical âttractive âttribute

I is theresponse frequency of the canonical delight -indifferent attribute where

A I

A A

A I

I I

1 1 l

m,

1 1 1

l m,

f

f f

S f

f f

S f

= +

(2) (3)

(5) (4)

(37)

All of the eighteen non-canonical judgments and their nearer and similar neighboring canonical judgments are figured as shown in Figure 8 to Figure 13.

In Figure 9, for example, (l, n) is the non-canonical judgment, which is between the canonical judgment (l, l) and (l, d). The proportion of similarity to the canonical judgment (l, l) and (l, d) will be:

S(l, n)(R) = f(R) / [f(R) + f(I1)] (6) S(l, n)(I1) = f(I1) / [f(R) + f(I1)] (7)

Considering the non-canonical judgment (l, m) is also between the canonical judgments (l, l) and (l, d), but it’s nearer to (l, l ) than ( l, d). The proportion of similarity to the canonical judgment (l, l) and (l, d) will be:

S(l, m)(R) = f(R) / [f(R) + 3f(I1)] (8) S(l, m) (I1) = 3f(I1) / [f(R) + 3f(I1)] (9)

Satisfied (l, l)

(m, l)

(d, m) (n, l)

(n, n)

Dysfunctional Functional

(d, n) (lw, l)

(d, lw) (d, l)

(d, d) Dissatisfied

(38)

In Figure 10, for example, the non-canonical judgment (n, m) is between the canonical judgment (n, n) and (n, l). The proportion of similarity to the canonical judgment (n, n) and (n, l) will be:

S(n, m)(I2) = f(I2) / [f(I2) + f(A)] (10) S(n, m) (A) = f(A) / [f(I₂) + f(A)] (11) Considering the non-canonical judgment (lw, n) is between the canonical judgments (n, n) and (d, n). The proportion of similarity to the canonical judgment (n, n) and (d, n) will be:

S(lw, n)(I₂) = f(I₂) / [f(I₂) + f(M)] (12) S(lw, n) (M) = f(M) / [f(I2) + f(M)] (13)

Satisfied

(l, l)

(l, m) (m, d)

(l, n)

(n, d) (l, lw)

(lw, d)

(l, d) Dissatisfied

Satisfied

(n, l)

(n, m)

(n, n)

(lw, n)

(d, n)

Dissatisfied

狩野模型之精煉－三種改善方法與應用 Refining Kano’s Model－Three improvements

中 華 大 學 博 士 論 文