以品質機能展開與狩野模式輔助感性工學研究之設計參數挑選：以頭戴式耳機為例

過往消費者對於一樣產品的要求大多是希望實用且好用，亦即能夠滿足理性上的需求，但是隨著經濟的發展以及生活品質的上升，人們開始追求感性上的滿足，產品的開發也就必須同時符合消費者感性的期望。因此，「了解消費者如何看待一樣產品並根據這些感性反應設計產品」便成為了現代的設計者急需解決的課題，感性工學也就應運而生。感性工學為了將感性與設計連結，提出了一系列的資料蒐集流程以及分析方法，通常會歷經五個步驟：（1）選定目標產品；（2）製作語意區辨量表；（3）定義設計參數；（4）執行互動實驗；（5）分析與推論。雖然感性工學的研究大多依照這五個步驟執行，但是由於每個步驟中可以執行的方式並非只有一種，各自的使用時機也未盡明確，導致感性工學的研究產生了一些瓶頸與限制。為了解決這些問題，本研究透過文獻的蒐集與探討，發現在「定義設計參數」的步驟中尚沒有一套公認的標準方法，且每個方法都有一定的限制，如果定義出的設計參數因此而不具代表性或是過多，將可能造成互動實驗的有效性不足或效率不彰。因此，本研究以頭戴式耳機為標的產品，期望透過品質管理的方法預先篩選設計參數，以協助感性工學的互動實驗能夠有效地直接鎖定對於使用者的感性期望較有貢獻性的設計參數進行更詳細的探討。本研究將透過感性工學、品質機能展開以及狩野模式等三種手法分別決定影響頭戴式耳機舒適度的最佳設計參數組合及重要性排序，再予以比較成效。在感性工學的部分，一開始會從社群網站、官方網站等管道蒐集與頭戴式耳機舒適度有關的形容詞彙，並透過使用者以及專家觀點篩選，製作成雙極的語意區辨量表；設計參數的選定則透過與頭戴式耳機設計專家討論、決定五種與舒適度有關的設計參數，最後透過10副耳機的互動實驗與數據分析得出這五種設計參數對舒適度的影響與排序。在品質機能展開的部分，本研究將以感性工學方法所蒐集而來的形容詞彙當作品質機能展開中的顧客聲音，並邀請專家評量前述之五種設計參數與顧客聲音的關係，最後計算出各設計參數的加權分數以及重要程度排序。狩野模式則將透過網路問卷調查頭戴式耳機使用者對前述之五種設計參數的存在意義進行評價，進而歸類為魅力型、一維型、必須型、無差異型、或反向型之品質特性，然後轉換為量化的顧客滿意係數與不滿意係數、予以排序。最後分別比較品質機能展開、狩野模式等兩個替代方法與感性工學的排序結果，根據吻合程度發現狩野模式的不滿意係數較有助於加速感性工學的定義設計參數步驟並維持分析的正確性。此外，本研究根據較佳的替代方法─即狩野模式的不滿意係數─篩選出較為少量的設計參數，並挑出對應的6副頭戴式耳機產品樣本、再次執行感性工學之互動實驗，以驗證篩選過後的分析結果與原始分析結果的一致程度；結果發現，分析所得最具貢獻的設計參數與原始的分析結果相近，即確認「使用狩野模式的不滿意係數篩選設計參數是一套可以接受的替代方法」，且狩野模式的不滿意係數可以節省36.7%的時間成本以及39.3%的金錢成本，對於產品成本較高、互動時間較長的感性工學研究尤其有幫助。

關鍵字

品質機能展開；狩野模式；滿意係數；不滿意係數；定義設計參數

並列摘要

In the past, customers’ demands on products were utility and usability which can fulfill rational requirements. However, along with the improvement of economics and quality of life, people starts to pursue emotional satisfaction on products. That is to say, product development must fulfill rational and emotional expectation of customers at the same time. Therefore, how to understand the emotional needs of customers and develop products based on them become crucial lessons for modern designers. For this reason, kansei engineering emerged to meet this requirement. In kansei engineering, a series of data collection processes and analysis methods were proposed in order to link users’ emotion with design parameters. There are five steps in kansei engineering: (1) choosing the target product, (2) making the semantic differential scale, (3) defining design parameters, (4) carrying out interaction experiment, and (5) analyzing and inferring the results. Most kansei engineering researchers followed these five steps, but there are still some limitations and bottlenecks. In each step, there could be multiple alternatives to choose from, but there is no clear guideline about the appropriate option under a specific condition. Through literature reviews, this study found out that there is no standard method in “defining design parameters,” and each method has its own limitation. If the design parameters are hence not representative or excessive, it will reduce the effectiveness and efficiency of the experiment. Therefore, taking headsets as an example, this study used quality management methods to help filter design parameters in advance. In this way, it will be allowed to explore deeply on the more important parameters and hence make the kansei evaluation more effective. In this study, kansei engineering, quality function deployment (QFD), and Kano model were adopted to determine the ranking of design parameters according to how each influences the users’ wearing comfort of headsets. By comparing the ranking results of these three methods, the better alternative for kansei engineering can be identified. In kansei engineering, adjectives related to wearing comfort were first collected from social media and official websites of manufactures. After being filtered from views of users and experts, the semantic differential scales were developed. Five design parameters related to wearing comfort were determined through discussion with experts of headset design. At last, optimal combinations and importance ranking of design parameters are inferred based on the results of interaction experiment with 10 headsets. In QFD, the adjectives collected in kansei engineering were considered as voice of customer (VOC), while experts were invited to rate the relationship between VOC and design parameters. By calculating the weighted score of each design parameter, the importance ranking was defined. As for Kano model, design parameters were evaluated through asking about users' feelings in terms of the extent of fulfilment and non-fulfilment of a specific design parameter. Each design parameter was then categorized as an attractive (A), one-dimensional (O), must-be (M), indifference (I) or reverse (R) quality. The percentages of these five categories can be further calculated as customers’ coefficients of satisfaction and dissatisfaction to get the importance ranking of design parameters. Finally, by comparing the results of kansei engineering with the result of QFD and the result of Kano model, the coefficient of dissatisfaction in Kano model was found to be able to better help filter design parameters efficiently, as well as ensuring the accuracy of data analysis. In addition, in order to validate the effectiveness of the filtering process, the 3 design parameters with higher importance according to the coefficient of dissatisfaction in Kano model were considered to run a validation experiment with the corresponding 6 headset samples. Results showed that ranking of the three design parameters is similar with that obtained in the original experiment. It is hence inferred that the use of coefficient of dissatisfaction in Kano model is an acceptable method for filtering design parameters. Moreover, it helped save 36.7% of time and 39.3% of budget for the kansei evaluation. More specifically, it will be more helpful for kansei evaluation with a higher cost of product collection and longer duration of product interaction.