This article was downloaded by: [134.117.10.

200]
On: 30 November 2014, At: 00:41
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Theoretical Issues in Ergonomics

Science
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/ttie20

Concepts, methods and tools in

Kansei engineering
a a a
Simon T. W. Schütte , Jörgen Eklund , Jan R. C. Axelsson
b
& Mitsuo Nagamachi
a
Quality and Human-Systems Engineering , Department of
Mechanical Engineering, Linköping University , SE-58183
Linköping, Sweden
b
College of Human and Social Environment , Hiroshima
International University , 555-36 Gakuendai, Kurosecho,
Hiroshima 724-0695, Japan
c
Quality and Human-Systems Engineering , Department of
Mechanical Engineering, Linköping University , SE-58183
Linköping, Sweden E-mail:
Published online: 23 Feb 2007.

To cite this article: Simon T. W. Schütte , Jörgen Eklund , Jan R. C. Axelsson & Mitsuo
Nagamachi (2004) Concepts, methods and tools in Kansei engineering, Theoretical Issues in
Ergonomics Science, 5:3, 214-231, DOI: 10.1080/1463922021000049980

To link to this article: http://dx.doi.org/10.1080/1463922021000049980

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever
as to the accuracy, completeness, or suitability for any purpose of the Content. Any
opinions and views expressed in this publication are the opinions and views of the
authors, and are not the views of or endorsed by Taylor & Francis. The accuracy
of the Content should not be relied upon and should be independently verified
with primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages, and other
liabilities whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
 This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.
Terms & Conditions of access and use can be found at http://www.tandfonline.com/
page/terms-and-conditions
Downloaded by [134.117.10.200] at 00:41 30 November 2014
 Theor. Issues in Ergon. Sci.
May–June 2004, vol. 5, no. 3, 214–231

Concepts, methods and tools in Kansei Engineering

Simon T. W. Schˇtte{*, J˛rgen Eklund{, Jan R. C. Axelsson{ and Mitsuo

Nagamachi{

{ Quality and Human-Systems Engineering, Department of Mechanical Engineering,

Linköping University, SE-58183 Linköping, Sweden
{ College of Human and Social Environment, Hiroshima International University, 555-36
Gakuendai, Kurosecho, Hiroshima 724-0695, Japan

Keywords: Kansei Engineering; proposed framework; semantic differential method; product

development.
Downloaded by [134.117.10.200] at 00:41 30 November 2014

Trends in product development today indicate that customers will find it hard to
distinguish between many products due to functional equivalency. Customers
will, therefore, base their decisions on more subjective factors. Moreover, in the
future, products will consist, to a higher grade, of a combination of a tangible and
intangible part. Kansei Engineering is a tool translating customer’s feelings into
concrete product parameters and provides support for future product design.
Presently, a total of six different types of Kansei Engineering are in use. The
aim of this paper is to propose a framework in Kansei Engineering to facilitate
the understanding of the different types of Kansei Engineering and to open Kan-
sei Engineering for the integration of new tools. The new structure includes the
choice of a product domain, which can be described from a physical and a
semantic perspective as building a vector space in each. For the latter mentioned
space, the Semantic Differential Method is used. In the next step, the two spaces
are merged and a prediction model is built, connecting the Semantic Space and
the Space of Product Properties together. The resulting prediction model has to
be validated using different types of post-hoc tests.

1. Introduction
Kansei Engineering is a proactive product development methodology which trans-
lates customers’ impressions, feelings and demands on existing products or concepts
into design solutions and concrete design parameters. As portrayed in figure 1, the
psychological impression intended for a future product is put into a Kansei Engin-
eering System (KES), which in turn delivers the required product design parameters
evoking the impression being aimed for.
Kansei Engineering is mainly a catalyst for a systematical development of new
and innovative solutions, but can also be used as an improvement tool for existing
products and concepts. Kansei Engineering is based on subjective estimations of
product and concept properties and gives expression to the demands on the products
which customers are not aware of, by using semantic tools developed by Osgood
(1969).
Several success stories contribute to the sound record that Kansei Engineering
nowadays has in Japanese companies. Mazda used Kansei Engineering in the devel-
opment of its model Miyata (in Europe: MX 5). More than 10 years after its first

* Author for correspondence. e-mail: simsc@ikp.liu.se

Theoretical Issues in Ergonomics Science ISSN 1463–922X print/ISSN 1464–536X online # 2004 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI 10.1080/1463922021000049980
 Kansei Engineering 215




Figure 1. Kansei Engineering System (KES), adapted from Nagamachi (1997a).

launch, the Miyata has become the best sold sports coupé in the world (The Guinness
Book of Records 2001). When Sharp introduced a newly developed video camcorder
with a LCD-display instead of a conventional ocular, they increased their market
share in this segment from 3 to 24%. Even in this case, Kansei Engineering identified
the customer’s demands on the new product resulting in a new concept. A third
example showing the wide product range Kansei Engineering can deal with is
Wacaol. This underwear manufacturing company collected Kansei data about the
Downloaded by [134.117.10.200] at 00:41 30 November 2014

usage of the common brassiere and based on this information designed a new model.
Their market share topped Japanese markets at 42% (Nagamachi, Ishihara and
Nishino 2001, Personal communication).
The method was developed by Professor Misuto Nagamachi in the early 1970s in
Japan and has been used in many Japanese companies. In the middle of the 1990s,
the method spread to the US and Europe. Over the 30 years of its existence, Kansei
Engineering has been developed substantially. In total, six different Kansei
Engineering procedures have now been tested and are available. Table 1 presents
the types of Kansei Engineering currently available in the order of their introduction.
Interest in creating quantifying links between product properties and user
impressions has existed for a long time. Research on this question has been done
before in different areas, e.g. QFD (Cohen 1995) and Conjoint Analysis (Green and
Rao 1971) in TQM, Semantic Environment Description (SMB) in Architecture
(Küller 1975) or Perceptual Vector Spaces for positioning of competing products
and Means-End Analysis (Reynolds and Olson 2001) in economics and marketing.
However, a deeper look reveals that the user’s perception is a very complex
formation alluding to many different scientific fields, namely Mechanical
Engineering, Quality, Mathematics, Psychology and Ergonomics, etc. (Schütte
2002). Hence, the role of Kansei Engineering in this context is to tunnel through
the borders between the different scientific fields, identifying suitable tools and

Table 1. Types of Kansei Engineering (Nagamachi 2001).

Type I Category classification—Identifying the design elements of the product to be

developed, translated from consumer’s feelings and image.
Type II Kansei Engineering System—A computer aided system with an so called
interference engine and Kansei databases
Type III Hybrid Kansei Engieering System—The combined computer system or Forward
ansei, which goes from the user’s impressions to design specifications and vice
versa.
Type IV Kansei Engeering modelling—Mathematical modelling with an interference engine
and databases
Type V Virtual Kansei Engineering—An integration of virtual reality technology and
Kansei Engineering in a computer system
Type VI Collaborative Kansei Engineering designing—Group work design system utilizing
intelligent software and databases over the internet.
 216 S. T. W. Schütte et al.

reassembling them into new methods for Kansei Engineering. In fact, Kansei
Engineering does not develop new theories or tools in the different areas at all.
Rather, it is an all-embracing methodology containing rules for how different
tools can interact with each other in order to quantify the impact a certain product
trait has on the users’ perception.
Future growth of Kansei Engineering and the application to new areas make it
necessary to allow the integration of more tools and methods from other areas. This
might be essential for the success of Kansei Engineering. As mentioned, already the
methodology contains six different types, making it very complex to handle. A gen-
eral concept of Kansei Engineering working procedure is therefore desirable, facil-
itating the understanding and expansion of the methodology. Nagamachi (1996,
1997a, b) has already presented an overall ‘schema of the procedure of Kansei
Engineering System’, which in a broad way presents the basic Kansei Engineering
structure.
Downloaded by [134.117.10.200] at 00:41 30 November 2014

The aims of this paper are to give an overview of different existing approaches in
Kansei Engineering, and to introduce—based on the existing Kansei Engineering
schema—a conceptual model of the Kansei Engineering process, in order to demon-
strate the particular use of the different tools and to facilitate the use of the Kansei
Engineering concept.
In the following article, essential definitions of expressions and terms in Kansei
Engineering are given and a conceptual model is introduced, followed by a discus-
sion and conclusions.

2. Expressions and terms in Kansei Engineering context

2.1. What is the ‘Kansei’?
Kansei is a Japanese word. When translated into English it might mean ‘consumer’s
psychological feeling and image’ (Nagamachi 1997a). This means that the Kansei is
the impression somebody gets from a certain artefact, environment or situation using
all the senses of sight, hearing, feeling, smell, taste, as well as cognition. Sub-
sequently, a complex mind pattern is built up and stored in the brain, containing
all the impressions experienced and, thereby, building the foundation for human
behaviour.
Lagerlöv (1984) describes the main actor’s Kansei in her famous book ‘Nils
Holgersson’ after he began his journey through Sweden on the back of a tame goose:
‘Nils had grown so giddy that it was a long time before he came to himself. The winds
howled and beat against him, and the rustle of feathers and swaying of wings sounded like
a great storm. Thirteen geese flew around him, flapping their wings and honking. They
danced before his eyes and they buzzed in his ears. He didn’t know whether they were
flying high or low or in what direction they were travelling’.
Although every reader of the book gets their own highly individual Kansei of this
situation depending on their own experiences and mood, there are common patterns
visible. Probably everybody agrees that Nils got into a very unpleasant situation and
was in full panic. However, even these common patterns can change by changing the
cognition. The situation described may not appear as violent to today’s reader as it
affected contemporary readers. Therefore, when one measures people’s Kansei, one
should be aware of the fact that the values measured are short lived and should be
used as quickly as possible.
 Kansei Engineering 217

2.2. Product
Since Kansei Engineering deals mainly with product development, the perspective of
Kansei Engineering on products has to be clarified.
The word ‘product’ originates from the Latin word productum, which means
result or gain. During the industrial revolution it became synonymous with indust-
rially manufactured artefacts. Nowadays, the expression also includes services
(Röstlinger and Goldkuhl 1999). Originally Kansei Engineering only focused on
artefacts, but recent studies conducted on internet-services proved that Kansei
Engineering has a much wider applicability (Nishino et al. 1999). According to
Röstlinger and Goldkuhl (1999), artefacts can be connected with certain services,
e.g. delivery and installation of a washing machine. Since the number of this type of
products will increase in the future (IVA 1999), Kansei Engineering has to be capable
of conducting examinations of both the services and artificial parts of the products in
a single study. Figure 2 portrays the three different types of products Kansei
Downloaded by [134.117.10.200] at 00:41 30 November 2014

Engineering can evaluate.

Product



  




Figure 2. Product definition used by Kansei Engineering.

3. Semantic differential method

Osgood (1969) developed a method of measuring the emotional content of a word
more objectively, called the ‘Semantic Differential Technique’, which more than 30
years later became one of the foundations of Kansei Engineering. He was influenced
by the different political ideologies which became evident in World War II and the
following Cold War.
His assumption was to make a separation between the object and an object-
representing sign:

. The object, ‘which is a pattern of stimulation which evokes reactions on the

part of an organism’, and
. The sign, ‘which is any pattern of stimulation which is not the object but yet
evokes reactions relevant to ‘object’-conditions under which this holds lying
the problem for theory’ (Osgood 1969).

This can be shown using a hammer as an example. The spoken word ‘hammer’ is not
the same stimulus as the object hammer. The former is a pattern of sound waves and
the latter a combination of visual, olfactory and tactual sensations. The word ‘ham-
mer’ elicits a type of behaviour which is in some manner relevant to the object
‘hammer’. This means that the spoken or read word ‘hammer’ is the sign for the
object ‘hammer’. Osgood’s research resulted, in the simplest terms, into the question
‘Under what conditions does something which is not an object become a sign of that
object?’ (Osgood 1969).
To answer this question, Stagner and Osgood (1946) conducted questionnaire
studies. The subjects chosen were supposed to rate signs (words) of objects like
 218 S. T. W. Schütte et al.


       
Figure 3. Example of a 7-point rating scale, originally used by Osgood (1969).

Pacifist, Russians, Germans, Dictator or Neutrality (remember that the experiments

were conducted under the impression of World War II) on bipolar scales. These
scales were defined with a number of contrasting adjectives at each end on which
the participants checked the position which best represented the direction and
intensity according to their point of view. An example of the scale type used is
shown in figure 3.
The analysis which applies multivariate analysis tools led to the discovery of the
existence of a common pattern (Carroll 1959). It could be seen that all the word pairs
examined span a three dimensional orthogonal vector space, as it is presented in
figure 4. Osgood called this space the semantic space and the method Semantic
Downloaded by [134.117.10.200] at 00:41 30 November 2014

Differential Techniques. When the word pairs in the individual factors were con-
sidered, it was possible to identify a pattern and name these factors.

 

 


  

Figure 4. The Semantic space, adapted from Carroll (1959).

. Evaluation (E) usually contains word pairs like good–bad, timely–untimely,

kind–cruel, beautiful–ugly, successful–unsuccessful, important–unimportant,
true–false, wise–foolish, etc. All these word-pairs have in common a potential
of evolving into a better or worse stage.
. Potency (P) usually contains word-pairs like large–small, hard–soft, mascu-
line–feminine, strong–weak, etc. These pairs characterize a potential, a capa-
city for change.
. Activity (A) is characterized by word pairs like active–passive, fast–slow, hot–
cold, sharp–dull, angular–rounded, etc. This factor indicates the grade and
speed of change.
Applying these factors into the semantic space, as seen in figure 4, these factor names
become the names of the axes. Now it is possible to project every concept in the
semantic space and give it an individual position. For example, a ‘dictator’ scores
high on the potency axis (hard, strong, etc.), low on the evaluation axis (bad, cruel,
ugly, etc.) and receives low positive values on the activity axis (active, fast, etc.).

4. Proposing a framework for Kansei Engineering

Figure 5 portrays a framework for Kansei Engineering methodology based on the
literature review above. The basic idea is to describe—based on an earlier chosen
domain—the idea behind the product from two different perspectives:
 Kansei Engineering 219

Choice of
Domain

Span the Span the

Semantic Space of
Space Properties

Synthesis update
update
Downloaded by [134.117.10.200] at 00:41 30 November 2014

Test of Validity

Model building

Figure 5. A model of the Kansei Engineering concept.

(1) The semantic description; and

(2) The description of product properties.

These two descriptions span a vector space each. Subsequently, these spaces are
merged with each other in the synthesis phase, indicating which of the product
properties evokes which semantic impact. Only after these steps have been carried
out is it possible to conduct a validity test, including several types of post-hoc
analyses. As a result of this step, the two vector spaces are updated and the synthesis
step is run again. When the results from this iteration process are satisfactory, a
model can be built describing how the semantic and the space of application are
associated.

4.1. Choosing the product domain

Selecting the domain includes the selection of a target group and market-niche and
the specifications of the new product. Based on this information, product samples
are collected, representing the domain. The Kansei Domain can be understood as the
perfect idea behind a certain product.
Despite the fact that a circle can never be drawn perfectly round, everybody
knows what the perfect idea of a circle is. This is also the case with the Kansei
domain. It is an abstract super-ordinated mind structure, while the representative
products are either tangible or intangible samples from this domain. As a result,
a domain includes both existing products, concepts and as yet unknown design
solutions.
 220 S. T. W. Schütte et al.

The task in this first step is to define the domain and find representatives
(products, drawings, samples, etc) covering as much as possible of the domain.

4.2. Spanning the semantic space

4.2.1. Measuring the Kansei: Since the Kansei is an internal sensation, the ques-
tion arising is how the Kansei can be grasped and measured. Unfortunately, all
the presently available measuring methods are external methods interpreting differ-
ent body expressions. A series of measurement methods have been developed,
interpreting:
. Physiological responses (e.g. heart rate, EMG, EEG),
. People’s behaviour and actions,
. Factual and body expressions, and
. Words (spoken)
Downloaded by [134.117.10.200] at 00:41 30 November 2014

(Nagamachi 2001b).
The list above is sorted according to the complexity of behavioural patterns. Since
the Kansei is a multifaceted phenomenon, Kansei Engineering mainly uses the evalu-
ation of words and their emotional impact on the human mind. This guarantees
detailed descriptions of the Kansei, but as a result those parts of the Kansei which
cannot explicitly be expressed in words are latent or in the worst case excluded.
Moreover, there is a risk that words which do not belong to the domain are collected
as well. These ‘impurities’ cannot be detected until a post-hoc test is conducted,
causing a certain amount of extra work.
Kansei Engineering is based on subjective estimations of products and concept
properties and it helps users to express their demands on the products—even those,
which they are not aware of.
Therefore, semantic tools, e.g. Semantic Differential Method developed by
Osgood et al. (1969), are used. In this way it is possible to quantify such complex
emotions as spatial perception (Bergqvist and Domeij 2001) or the impression of the
sound of vehicles (Nagamachi 1994).

4.2.2. The procedure of spanning the semantic space: In practice, the step ‘Span
the semantic space’ in figure 5 is carried out in three steps, as portrayed in figure
6. Using the desired domain as a starting point, Kansei words describing the con-
sidered product are collected. In a second step, the number of words is reduced to
a more practical number. This can be done by using different tools, as described
below. In the last part, the data is compiled in a standardized way in order to fa-
cilitate the following synthesis phase. If important Kansei words are missed in this
step, the result can become practically unusable. Hence, it is better to select a few
more words than necessary.

4.2.3. Collection of Kansei words: A Kansei word is a word describing the pro-
duct domain. Often these words are adjectives, but other grammatical forms are
possible, e.g. when describing the domain ‘fork-lift truck’, adjectives like effective,
robust, quick, etc. but also verbs and nouns (acceleration) can occur (Schütte and
Eklund 2001). In order to get a complete selection of words, all available sources
have to be used, even if the words emerging seem to be similar or the same.
Suitable sources can be:
 Kansei Engineering 221



 
  


  
  

  
  
Downloaded by [134.117.10.200] at 00:41 30 November 2014

 







Figure 6. Spanning the Semantic Space broken down into three steps (insprired by
Nagamachi (1997) and Osgood and Suci (1969)).

. Magazines;
. Pertinent literature;
. Manuals;
. Experts;
. Experienced users;
. Relating Kansei studies; and
. Ideas, visions.
An important point is to translate ideas and visions into Kansei words because non-
existing solutions should also be considered. Only in this way can Kansei Engin-
eering be used as a creative product development tool, which generates new and
revolutionary solutions. The task is to describe the domain, not the existing
products.
Depending on the domain considered, the number of existing Kansei words
generally varies between 50–600 (Nagamachi 1997a). Since it is of great importance
to collect all existing words, the word collection is continued until no new words
occur. The data gathered will critically influence the validity of the results if import-
ant words are missing.

4.2.4. Selection of Kansei words using data reduction methods: Generally, it is

considered advantageous to use the original number of words, because a selection
 222 S. T. W. Schütte et al.

always causes a loss of information. On the other hand, if the number of words
collected exceeds a critical size it can be difficult to find volunteers to fill in ques-
tionnaire forms, due to the amount of time needed. Therefore, the statistical
power may suffer from a low number of participants (Körner and Wahlgren
2000). Also, the quality of the gathered data will also be relatively poor due to ef-
fects of fatigue on the participants (SCB 2001).
Hence, the data quality is appreciably affected by the number of Kansei words or
if the evaluation time of the questionnaire reaches critical dimensions, and so a
reasonable data reduction must be carried out. Two empirically tested possibilities
are portrayed in figure 7 (Arnold and Burkhard 2001).
One of the possible word reduction methods presented in figure 7 is a pilot study
using Osgood’s Semantic Differentials and factor analysis. The participants are sup-
posed to think about the domain per se and answer the question ‘How do you think
this Kansei Word corresponds to the Domain?’ Subsequent factor and/or cluster
Downloaded by [134.117.10.200] at 00:41 30 November 2014

analyses are applied in order to reveal the connections between the words and to
allow the choice of representatives for each factor or cluster which become the new
set of Kansei words.
Another possibility is to use a focus group and/or expert group to gather the
words together according to their affinity and choose representatives for every group
(Card-System). Since the outcome of this will be a very condensed number of words,
the validity of these words must be tested. That can be done in two steps:

. Manual inspection if the selected words represent the semantic space suffi-
ciently directly after the clustering; and
. Executing a post-hoc factor analysis after finishingthe main study.

Which method is used depends on the context, but no study has yet compared the
two different methods with each other.

4.2.5. Compiling data: After the relevant words have been collected and rated on
the semantic scales, the number of words selected is reduced in a way that the re-
maining words represent the semantic space properly. The outcome from this step
is a list containing the rankings of the selected words against the artefacts used for
each participant. This can be stacked in a three dimensional matrix, as presented
in figure 8.

 




 


 
 
 


Figure 7. Selection of Kansei words using data reduction methods.

 Kansei Engineering 223








Figure 8. Raw store data matrix, obtained when a group of subjects (x-axis) judges a sample
of concepts (y-axis) against a set of semantic scales (z-axis). Each cell contains a number
from 1–7, representing the judgement of a particular concept on a particular scale by a single
subject (adapted from Osgood and Suci (1969)).
Downloaded by [134.117.10.200] at 00:41 30 November 2014

4.3. Spanning the space of product properties

The spanning of the space of product properties is similar to the semantic space. One
underlying idea is that there is an existing vector space depicting the domain’s
properties. The task in this part is to collect all the attributes representing the domain
chosen and select those which apparently have the largest impact on the user’s
Kansei, and choose products representing the chosen product properties before
the data is compiled for the following synthesis phase. The particular step shown
in figure 5 is presented in more detail in figure 9.
The possible tools are multifaceted and can both integrate several or only one
step of the proposed concept. Figure 10 portrays the tools currently available. In
most studies, these tools are not solely used, but they can be combined.
The spanning of the space of product properties has some sensitive points, where
limitations are made. In order to make this discussion visible, it has to be carefully
documented. One of these points is the selection process. The result may vary,
depending on which technique is used. Subjective estimations, especially in the
first case, can play an important role. When using an affinity diagram or a pareto
diagram, the correct choice of the evaluation group members is important. As men-
tioned in the choice of the domain, the group members should ideally be chosen
randomly from the intended target group in the domain.

4.3.1. Manual collection of product properties and selection: The most common
variant in every designing process is the manual collection and selection of the
product attributes made by the product designer alone. In complex matters, tech-
nical aids can be used, like fish-bone diagrams, etc. However, in the end it is the
designer’s experience and intuition that decide which product parameters will be
taken into account. The quality of the result depends on factors like the designer’s
experience, the company structure or the product’s maturity.

4.3.2. Using quality tools in product development: When working in the early
stages of a project, the product specifications are usually not fixed. So, designers
work more holistically and in teams. All their opinions could contain a potential
benefit, and everybody’s knowledge should be included in the future product.
 224 S. T. W. Schütte et al.



 
  


 
    

   


  
  
Downloaded by [134.117.10.200] at 00:41 30 November 2014

 

 







Figure 9. Spanning the space of product properties, broken down into three steps
(inspired by Nagamachi (1997) and Osgood and Suci (1969)).

Figure 10. Possible tools for spanning the space of product properties.

Quality techniques can provide a variety of different tools, which can be used to
achieve this.
In this case, an affinity diagram (Klefsjö et al. 1999) can be applied, including a
brainstorming phase when collecting the existing parameters and a joint decision
phase containing a ranking and selection of the different solutions.
 Kansei Engineering 225

4.3.3. Collection and selection made by using focus group data: In the third alter-
native, the collection and selection processes are separated and supported by two
different tools. In the first step, all the existing product parameters are collected
from different sources like:

. Technical documents,
. Comparisons of competing products,
. Magazines,
. Pertinent literature,
. Manuals,
. Experts,
. Experienced users, and
. Related Kansei studies.

By only using these sources, it would hardly be possible to develop innovative prod-
Downloaded by [134.117.10.200] at 00:41 30 November 2014

ucts. Hence, it is necessary to think abductively. It is more important to look for

things which are not present and to search for the domain properties; this means all
the imaginable properties the domain could have. Therefore, more sources have to
be considered:

. Ideas,
. Visions,
. Concept studies,
. Analysis of the usage of existing products and conclusions drawn, and
. Related product groups.

In the second step, a focus group, consisting of potential users, is asked to choose
desired product properties from a list containing the collected domain properties
(Domain properties are all the properties the product in question could have). A
frequency analysis or pareto diagram over the product traits mentioned reveals their
importance and rank. Figure 11 shows the pareto diagram made for wrist watches. It
can be seen that the type of dial, the material and the colour represent together more
than 80% of all votes.

160 100
140 90
80
120
70
100




60





80 50
60 40
30
40
20
20 10
0 0
Dial Material Colour Strap Accuracy





Figure 11. Pareto diagram, importance of product properties of watches.
 226 S. T. W. Schütte et al.

The following products have to possess these parameters. These products will be
used in the synthesis in order to represent the new product’s properties (according to
Nagamachi 1995: items). In many cases, it is not possible to find products corre-
sponding to any item; especially when the item originates from a concept or idea. In
these cases, it is possible to use computer-made images or video-clips. The products
have to be chosen carefully according to the rules of Design of Experiments, other-
wise any ensuring statistical treatment can cause problems.

4.3.4. Compiling: As mentioned before, distributing the data in a standardized

way facilitates the choice of the evaluation tool in the synthesis phase. Moreover,
multiple tools can be used on the same data in order to compare and validate the
results, since the input data required is the same and does not have to be adjusted
before using a different tool.
Downloaded by [134.117.10.200] at 00:41 30 November 2014

4.4. Synthesis
In the synthesis step, the Semantic Space and the Space of Properties are linked
together, as displayed in figure 12. For every Kansei word, a number of product
properties are found, affecting the Kansei word. Ishihara et al. (1998) conducted a
study on beer can design. Their results showed that the score of the Kansei Word
‘bitter’ is most affected by the colour of the can and the shape of the logo. In fact, a
black colour in combination with a non-oval logo evoked a strong bitter Kansei,
whereas a white can with an oval logo involved the opposite Kansei.
The research into establishing these links has been one of the core parts of
Nagamachi’s work with Kansei Engineering in the last few years. At present, a
number of different qualitative and quantitative tools are available. Since the incom-
ing data is stacked in a standardized way, every tool can be used, and it is even
possible to use different tools and compare the results afterwards in order to reveal
the best suitable tool.

4.4.1. Qualitative treatments: People working with design are usually aware of
the links between peoples’ impressions and the product traits. They know the tar-
get groups well from different sources and have a ‘sixth sense’ or intuition about
how the products should be designed. This latent knowledge cannot be expressed
and is very difficult to communicate. By providing data from the previous steps,
the experts become able to share their knowledge with their colleagues and in that
way create a new awareness. QFD uses almost the same principles when linking

Item2
Kansei Word 2 Item 1
Kansei Word 1 Item 3 Item 4

Figure 12. The synthesis phase, KW = Kansei word.

 Kansei Engineering 227

the customers needs to the metrics in the house of quality (Nagamachi, Ishihara
and Nishino 2001, Personal communication). In a Kansei Engineering context,
this procedure is called Kansei Engineering Type I.

4.4.2. Statistical treatment: In many cases, experts are more aware of the user’s
demands than the users themselves. On the other hand, users can easily assess
whether a product is suitable in a certain respect or not. Nagamachi (2001) and
his research group have developed a number of different statistical procedures
using different mathematical implements to capture the user’s impression and
make the synthesis independent of expert knowledge. Those are.
. Linear regression (Ishihara 2001);
. General Linear Model (GLM) (Arnold and Burkhard 2001);
. QT1 (Komazawa and Hayashi 1976);
Downloaded by [134.117.10.200] at 00:41 30 November 2014

. Neural Networks (Ishihara et al. 1996);

. Genetic Algorithm (Nishino et al. 1999); and
. Rough set analysis (Nishino et al. 2001).
The use of different tools depends on the context. Emotions and feelings do not
follow mathematical laws. Sometimes it is possible to use linear methods, which are
easiest to handle, sometimes more complex models have to be applied. The outcome
of all the tools presented describes only in which way the Kansei words are corre-
lated to the product properties.

4.4.3. Test of validity and iterations: At this point of the Kansei procedure, a
model of the Kansei is available, but nothing is said about the validity of this
model. Using Weinreich’s (1958) ideas about the Semantic Space, it is possible to
conduct a factor analysis from the data gathered and compare the results with the
Kansei words delivered from the Semantic Space. As mentioned in section 4.2.1,
the number of output words was deliberately too large. By comparing the result
from the first (after selecting the Kansei words) and the second factor analysis
(after the completed synthesis), it is now possible to spot the words which have no
effect on the Kansei. This is fed back to the Semantic Space and, if an iteration
process is necessary only, when the new words are used.
Theoretically, this procedure can also be used for determining which of the prod-
uct properties is obsolete, but this has not been tested yet. Newly developed models
can be tested with other different products possessing certain product items. By
comparing the predicted value with the data acquired from a new test questionnaire,
the model quality can be determined.

4.4.4. Model building: When the validity tests give a satisfactory result, the data
gathered from the synthesis can be presented as a model. These models are a func-
tion depending on the product properties and predict the Kansei score for a cer-
tain word:

yKansei ¼ f ðproduct propertiesÞ

Depending on the tool chosen, the function can be qualitative, linear or non-linear.
 228 S. T. W. Schütte et al.

5. Discussion
One of the advantages of Kansei Engineering is the fact that the methods and tools
used are collected from different areas of research. This enables Kansei Engineering
to solve the task in hand using the most suitable methods and makes it flexible for
solving very different problems. However, does this mean that Kansei Engineering is
just a re-combination of already known methods? The authors consider Kansei
Engineering rather as an independent methodology, integrating and complementing
known methods and tools into new units.
In fact, Kansei Engineering relates the research from different areas with each
other and combines the methods to make a new whole. In that way, e.g. Hayashi’s
Quantification Theory (Komazawa and Hayashi 1976) could be connected to
Osgood’s (1969) semantical scales to a indispensable cycle within the Kansei
Engineering process.
Even in practical applications, Kansei Engineering is often seen as an indepen-
Downloaded by [134.117.10.200] at 00:41 30 November 2014

dent methodology. However, in product development Kansei Engineering is some-

times used as a method leading a whole process. Sometimes, it is used as a
complementary method within a bigger process. In these cases, Kansei
Engineering is been used in combination with other methods in product development
like QFD, Conjoint Analysis or SMB (compare Schütte 2002).
The authors developed the Key-Figure (figure 5) based on an in-depth analysis of
the available literature and experience gathered from different studies conducted at
Linköping University. Contemplating Kansei Engineering from this perspective
reveals the ideas on which Kansei Engineering is based, giving the opportunity for
a better understanding of the existing types of Kansei Engineering and giving scope
for future expansion.
The clear distinction between the Kansei Engineering concept and the possible
tools for the different steps allows the identification of both the well-developed areas
and has enabled the authors to identify further development opportunities, and to
propose future research areas and new tools.
Beginning from the top of figure 5, it can be remarked that little work has been
reported on the ‘choice of domain’. Usually, the domain and the target groups are
pre-chosen, based on the product manufacturer’s experience.
Proceeding downward in figure 5, the domain is split into the Semantic Space and
the Space of Product Properties. Whereas the Semantic Space is supported by
Osgood’s work, only a few tools are available for the Space of Product Properties.
The authors present a number of possible tools, e.g. from the field of quality, but
more research on this field is required.
A lot of attention is paid to the synthesis step, since the synthesis step is one of
the core parts in Kansei Engineering and Nagamachi and colleagues are constantly
working to develop and adapt new tools in this area.
Subsequently, the ‘test of validity’ is carried out. Although a number of advanced
mathematical tools are being developed, giving feedback to the Semantic Space and
confirming the validity of the Kansei words delivered, no tool is available to do the
same with the Space of Product Properties. Would is be possible to determine the
validity of the choice of product properties in a similar way?
It is the goal of Kansei Engineering to provide a model for the relationship
between the product properties and the Kansei. In almost all studies, a model is
built, but it is still difficult to say within which limits this model is valid. Such limits
 Kansei Engineering 229

can be the duration of validity of the chosen Kansei words, the target groups, prod-
uct groups, etc.

6. Conclusions
The Kansei Engineering concept introduced in this paper is a synopsis of the
methods and tools used in Kansei Engineering in order to evaluate the relationship
between the individual’s psychological experience of a certain product and its design.
Seen from this perspective, this paper can be understood in a wider sense as a
definition of contemporary Kansei Engineering, allowing the opportunity for further
development. The Key-Figure (figure 5) identifies the different areas in Kansei
Engineering and links the available tools from the different areas to the different
steps.
The adaptation of the different tools to Kansei Engineering can in some cases be
difficult, since many tools used in Kansei Engineering are borrowed from other
Downloaded by [134.117.10.200] at 00:41 30 November 2014

scientific or technological fields. This results in a restricted validity of the tools

used and this has to be taken in account when drawing conclusions from a Kansei
Engineering study.
Some of these tools require considerable resources and time, affecting the dura-
tion of the validity and involving high costs for the user. So, it is desirable to strive
after less time consuming methods or to complement Kansei Engineering with a type
evaluating more quantitative data.

Acknowledgements
The Swedish authors wish to thank Professors Nagamachi, Ishihara and Nishino
from Hiroshima International University for their help and support on many occa-
sions. Moreover, we want to express our thanks to BT Industries for financial sup-
port and providing us with the opportunity to test our theories in practice.

References
Arnold, K. and Burkhard, D. 2001, Kansei Engineering—From the Customers Point of
View, Special Report LiTH-IKP-R-1226, IKP, Linköping.
Bergqvist, I. and Domeij, O. 2001, Hemkänsla i en virtuell verklighet, EKI, Linköping,
LiTH-EKI-Ex-2001: 84.
Carroll, J. B. 1959, Review of the measurement of meaning, in J. G. Snider and C. E.
Osgood (eds), Semantic Differential Technique (Chicago: Aldine Publishing), 96–115.
Cohen, L. 1995, Quality Function Deployment (Corporate & Professional Publishing Group
Addison-Wesley Publishing Company).
Fr
ga Rtt, S. C. B. 2001, Special Report Statistiska Centralbyrå.
Green, E. P. and Rao, V. 1971, Conjoint measurement for quantifying judgemental data,
Journal of Marketing Research, 61–68.
Ishihara, S. 2001, Kansei Engineering Procedure and Statistical Analysis, Workshop at
International Conference on Affective Human Factors Design, Singapore.
Ishihara, S., Ishihara, K. and Nagamachi, M. 1998, Hierarchical Kansei analysis of beer
can using neural network, Proceedings of Human Factors in Organizational Design and
Management - VI (Amsterdam: Elsevier), 421–425.
Ishihara, S., Ishihara, K., Nagamachi, M. and Matsubara, Y. 1996, Neural network
approach for Kansei analysis on milk carton design, Proceedings of Human Factors in
Organizational Design and Management—V (Breckenridge, USA: Elsevier Science B.V.),
7–12.
IVA 1999, Styrkor, svagheter, möjligheter och hot, Special Report 91-7082-657-9.
Klefsj˛, B., Eliasson, H., Kennerfalk, L., Lundbck, A. and Sandstr˛m, M. 1999, De
sju ledningsverktygen (Lund: Studentlitteratur).
 230 S. T. W. Schütte et al.

Komazawa, T. and Hayashi, C. 1976, A statistical method for quantification of categorical

data and its applications to medical science, in F. T. de Dombal and F. Gremy (eds)
(North-Holland Publishing Company).
K˛rner, S. and Wahlgren, L. 2000, Statistisk dataanalys (Lund: Studentlitteratur), 195–
196.
Kˇller, R. 1975, Semantisk Miljö Beskrivning (SMB) (Stockholm: Psychologyförlaget AB
Liber Tryck Stockholm).
Lagerl˛v, S. 1984, The wonderful adventures of Nils (London: J.M. Dent & Sons).
Nagamachi, M. 1994, Kansei Engineering: an ergonomic technology for a product develop-
ment, Proceedings of IEA ‘94, 1220-122.
Nagamachi, M. 1996, Kansei Engineering and implementation of human-oriented product
design, in R. J. Koubek and W. Karwowski (eds), Manufacturing Agility and Hybrid
Automation - 1 (Louisville, Kentucky, USA: IEA Press), 77–80.
Nagamachi, M. 1997a, Kansei Engineering: the framework and methods, in M. Nagamachi
(ed.), Kansei Engineering 1 (Kure: Kaibundo Publishing co. Ltd), 1–9.
Nagamachi, M. 1997b, Kansei Engineering as consumer-oriented ergonomic technology of
product development, Proceedings of IEA ‘97 (Tampere, Finland: Finnish Institute of
Downloaded by [134.117.10.200] at 00:41 30 November 2014

Occupational Heath), 228–230.

Nagamachi, M. 2001, Workshop 2 on Kansei Engineering, Proceedings of International
Conference on Affective Human Factors Design, Singapore.
Nishino, T., Nagamachi, M., Ishihara, K., Ishihara, S., Ichitsubo, M. and Komatsu, K.
1999, Internet Kansei Engineering system with basic Kansei database and genetic algo-
rithm, Proceedings of TQM and Human Factors (Linköping, Sweden: Centre for Studies
of Humans, Technology and Organization), 367–372.
Nishino, T., Nagamachi, M. and Ishihara, S. 2001, Rough set analysis on Kansei evalua-
tion of color, Proceedings of The International Conference on Affective Human Factors
Design (Singapore: Asean Academic Press).
Osgood, C. E. 1969, The nature and measurement of meaning, in C. E. Osgood and J. G.
Snider (eds), Semantic differential technique—a source book (Chicago: Aldine publishing
company), 3–41.
Osgood, C. E. and Suci, G. J. 1969, Factor analysis of meaning, in C. E. Osgood and J. G.
Snider (eds), Semantic differential technique—a source book (Chicago: Aldine publishing
company), 42–55.
Osgood, C. E., Suci, G. J. and Tannenbaum, P. H. 1969, Measurement of meaning, in C. E.
Osgood and J. G. Snider (eds), Semantic differential technique—a source book (Chicago:
Aldine publishing company), 56–82.
Reynolds, T. J. and Olson, J. C. 2001, Understanding Consumer Decision making—The
Means-End Approach to Marketing and Advertising Strategy (New Jersey: Lawrence
Erlbaum Associates, Publishers).
R˛stlinger, A. and Goldkuhl, G. 1999, Produktbegreppet- En praktikteoretisk innebör-
desbestämmning, Special Report 1999:07, Centre for Studies of Humans, Technology
and Organisation, Linköping.
Schˇtte, S. T. W. 2002, Designing Feeling into Products—Integrating Kansei Engineering
Methodology in Product Development, Institute of Technology, Linköping, LiU-Tek-
Lic 2002: 19.
Schˇtte, S. and Eklund, J. 2001, An approach to Kansei Engineering—methods and case
study on design identity, Proceedings of Conference on Human Affective Design
(Singapore: Asean Academic Press, London).
Stagner, R. and Osgood, C. E. 1946, Impact of war on a nationalistic frame of reference,
Journal of Social Psychology, 389–401.
The Guiness Book of Records, 2001, Guiness.
Weinreich, U. 1958, Travels through semantic space, in J. G. Snider and C. E. Osgood (eds),
Semantic Differential Technique (Chicago: Aldine Publishing Company), 116–129.

About the authors

Simon Schütte graduated with a Master’s Degree in Mechanical Engineering in 2000 from the
Technical University at Braunschweig, Germany. Currently, he is a PhD candidate in the area
of Quality and Human Systems Engineering at the department of Mechanical Engineering at
 Kansei Engineering 231

Linköping University, Sweden. His main field of research is Kansei Engineering in product
development. Together with Jörgen Eklund and Jan Axelsson, Schütte is a member of the
Kansei Engineering Research Group at Linköping University (www.ikp.liu.se/kansei).
Jörgen Eklund received his PhD in Industrial Ergonomics at Nottingham University, UK, in
1986. He is presently a Professor at the Division of Quality and Human–Systems Engineering,
Linköping University, Sweden. He has published over 150 publications in the field of applied
ergonomics, with a focus on industrial production and product design. His research interests
include the borderline between the disciplines quality technology and ergonomics, and also
cover methods for product design. He has long experience working and collaborating with
industry.
Jan Axelsson holds an Associated Professorship in quality and human–systems engineering at
Linköping University, Sweden. He is the current President of the Ergonomics Society of
Sweden (ESS) and Editor in Chief of the Journal of Nordic Ergonomics. He is also a board
member of the Nordic Ergonomics Society (NES) as well as the Swedish Association for
Quality (SFK). In the International Ergonomics Association, he represents NES in the council
and is a member of the technical and scientific committee on Quality Management. With over
Downloaded by [134.117.10.200] at 00:41 30 November 2014

10 years of interdisciplinary research on the integration of quality and ergonomics, he is the

author of more that 40 academic publications and a frequently invited speaker on the subject.
Mitsuo Nagamachi received a PhD in psychology from Hiroshima University in 1963. Since
then he has been employed in Hiroshima University to teach human factors and management.
He retired from it in 1996 and became President of Kure National Institute of Technology. In
2002, He moved to Hiroshima International University as the Dean of the College of Human
and Social Environment and established the Department of Kansei Ergonomics there. His
major research fields are manufacturing, TQM, safety, macro ergonomics and Kansei engi-
neering. He has published more than 80 books so far, several of which were translated into
Korean and Chinese. He has contributed to hundreds of Japanese large companies from the
viewpoints of his research fields for many years.