產品感性意象的品質模型初探 - 以手機為例

Abstract

本研究嘗試將產品造形所造成的「感性意象」，帶入「Kano品質模型」，賦予各「感性意象」特定的「品質屬性」，以更精確的了解消費者在造形意象上的喜好。透過本研究所建立的模型，在資源與時間有限的設計環境下，設計者將可策略性的針對目標族群，以及競爭者的策略，加強「提昇滿意係數」與「解除不滿意係數」，以達到企業獲利，消費者滿意的目標。 為了建構「感性意象品質模型」，本研究嘗試了三種不同的建構方法，並比較彼此的差異：「直接分類法」、「Kano品質調查法」、「感性意象調查法」。測驗樣本為「行動電話」，目的在於採用消費者最熟悉的產品類型，以提高問卷調查的準確度。並且針對不同「消費者族群」，比較其對於「感性意象」在「品質屬性」上的差異。 本研究將「感性意象」與品質管理中的「Kano品質模型」結合，賦予「感性意象」五種「品質屬性」，使其在應用於設計的過程中，能夠傳達更精確的概念。此五種關係包括：「魅力的」、「一元的」、「必要的」、「無差別的」、「反向的」。 從本研究的數據顯示，透過「Kano品質調查法」、「直接分類法」與「感性意象調查法」，調查「感性意象」的「品質屬性」，其結果雖然沒有完全相同，但整體而言仍然具有相當的一致性。 在「感性意象調查法」中所用的「迴歸模型」，雖然「線性」與「非線性」的模型皆有相當的解釋力，但「線性」的解釋力卻優於「非線性」的，此結果並不完全符合「Kano品質模型」的概念。在其定義中，「魅力的」的特性為：「非對稱」（asymmetric）與「非線性」（nonlinear）。本研究的結果僅能證明其「非對稱」，即：「品質充足與否」對於「滿意度」與「不滿意度」的影響比例並不一致。 本研究亦嘗試以「類神經網路」模擬「感性意象」與「喜好度」之間的關係，以判斷其「品質屬性」，但結果並不穩定。可能的原因在於：樣本數不足，且「感性意象評價」分布不夠寬廣，造成網路「廣義化」的能力不佳、模擬的可信度不足。另一方面，「感性意象」可能有更多不同的「品質屬性」，尚未被發現與定義。

In order to explore the customers’ “kansei” feelings on product form more thoroughly, this study tried to apply “Kano model” to classify “kansei images” into five kinds of “quality attribute”. With the model built in this study, designers could strategically strengthen “customer satisfaction coefficient” according to the target customers and the competitors’ strategies, so that the corporation would get the benefit and the customers would be satisfied at the same time. Three different approaches, “direct classification,” “questionnaire of Kano model,” and “questionnaire of kansei image,” were used to build the “quality model of kansei”, and were compared with each other. Mobile phones, which are familiar products to the customers, were used as test samples to raise the accuracy of the questionnaire. The difference of the “quality attribute of kansei” classification among different customer groups was also discussed. The result of this study revealed that the “quality attributes of kansei” classified by the three different questionnaires are not exactly the same but are similar in general. In the approach of “questionnaire of kansei image”, both linear and logarithm regression models were analyzed. The result has proved the robustness of both models, but the linear model is slightly more reliable. This result is not exactly consistent with the general concept of “Kano model”. According to “Kano model”, “attractive quality” should present asymmetric and nonlinear features. In this study, only “asymmetry” feature has presented in “attractive quality”, which means the negative and positive performances affect the overall satisfaction asymmetrically. This study also tried to simulate the relations between “kansei image” and “preference” with neural network to classify “quality attributes of kansei images”. But the result is not stable enough to effective generalize the simulation. Insufficient test samples and the narrow range of its scores of “kansei” image may be the reason.. Nevertheless, the result imply that there may be more kinds of “quality attributes of kansei image” to be discovered and defined.