DRS Common Ground Conference

1
Quantifying the Unquantifiable ?
The development of empirical techniques for the
investigation of design perception
ABSTRACT This project proposes a scientific
approach to the investigation of design
perception in order to develop concepts, theories
and techniques for improving the effectiveness of
design. This research shows that
participant-based perception experiments can be
used to support intuitive understanding of design
by allowing formal hypothesis to be formulated
and tested under scientific conditions. This
study then, seeks to build on our intuitive
knowing and ensure the effective mapping of
design intent onto consumer perception by
investigating the nature of consumer
interpretation.
2
CONTENTS Abstract Aims Introduction Why Assess
Design Character? How can Design Character be
Measured? General Method Experimental
Protocols Experiment 1 Sample Stimulus
Set Experiment 2 Making Sense of the
Data Experiment 1 Hypothesis Discussion of
Results Experiment 2 Hypothesis Discussion of
Results
Determinants of Product Character Opportunities
for Design Practitioners Gender Perception in
Product Design The Role of Prototypicality in
Design Practical Implications Anecdotal
Evidence Significant Findings Further
Study Comments About the Authors References
3
The purpose of a product is not what the
engineer explicitly says it is but what the
consumer implicitly demands that it shall be.
INTRODUCTION According to Klaus Krippendorff,
design is the creation of meaning as well as the
creation of a physical artefact. If so, then
designers must look beyond the visual and
investigate how designed artefacts are
interpreted by the intended audience. As
Theodore Levitt put it, "The purpose of a product
is not what the engineer explicitly says it is,
but what the consumer implicitly demands that it
shall be. This would suggest the need to
develop an empirical method for assessing
perceptual response to design a sort of
perceptual ergonomics to complement intuitive
knowledge. This is not to say that intuition has
not been successful - it has guided product
development since Neanderthal man fashioned the
first crude axe - or that scientific enquiry is
inherently superior, just that intuitive ability
varies from person-to-person and as such can be
unreliable.
Intuitive ability varies from person-to-person
and as such can be unreliable. This would
suggest the need to develop an empirical method
for assessing perceptual response to design a
sort of perceptual ergonomics to complement
intuitive knowledge
4
AIMS OBJECTIVES Develop empirical techniques
to quantitatively assess the perception of design
character Show that participant-based
experiments can be used to complement traditional
intuitive methods. Improve communication in
design by investigating how consumers interpret
design character and exploring what factors
affect product perception Improve the
effectiveness of design though better
understanding of consumer interpretation of
design character
5
IMAGE CREDIT Argos Catalogue Spring 2002
WHY ASSESS DESIGN CHARACTER?
  Facing the intense changes in the market, a
well-designed product should not only satisfy
consumers physical requirements but should also
satisfy their psychological needs. (Hsiao and
Chen, 1997) Only by understanding how shape and
form affect perceived character can the designer
hope to satisfy not just physical requirements
but emotional and psychological requirements. It
has long been recognised that shape and form can
be used to communicate the perception of
elegance, durability or ease-of-use.
Designers have always tried to exploit the
connection between visual appearance and
perceived character (Janlert, 1997). For
instance, many small domestic products e.g. power
tools attempt to communicate the perception of
robustness through design. It could even be
argued that perception is more important than
reality, for few would buy a chair which looked
fragile regardless of the actual strength or
durability of the product. Physical strength is
not enough, the perception of strength must be
communicated though design. Consider the two
exercise cycles shown here, although the cycle on
the right may look more robust but both products
actually have the same maximum load strength of
100kg  
Which products appear most robust?
 
WHY?
6
People often use adjective image words to
describe their perception of a object or product.
It is not unreasonable then to use pre-defined
keywords as a means of assessing perceived
character
HOW CAN DESIGN CHARACTER BE MEASURED?   The
study of product semantics assumes that shapes
certain shapes evoke certain emotions and
associations in the eye of the beholder (Hsiao
and Chen, 1997 Janlert, 1997). By
systematically manipulating the shape of
different products and assessing the
corresponding change in perceived character it
may be possible to better understand how design
affects consumer interpretation.   But is it
possible to measure the perception of design
character?   People regularly talk of products
as having character as a means of understanding
and describing them (Janlert, 1997) and it is not
unusual for people to use adjectival image words
to describe the feelings engendered by an object.
Someone might describe the chairs (shown left)
as luxurious or modern. Given that people
are accustomed to evaluating products in this way
there is no reason why participant groups could
not be asked to assess perceived character on the
basis of pre-defined keywords.
IMAGE CREDIT Argos Catalogue Spring 2002
7
Product variables were systematically manipulated
to explore the affect of specific design changes
on design character
GENERAL METHOD Each study is based on a
convenience sample of undergraduate students
drawn from a range of courses in the Faculties of
Engineering and Arts at the University of Ulster.

ROBUST
A
B
C
D
E
G
F
H
Participants were exposed to a set of images and
asked to evaluate them on the basis of
pre-defined keywords
I
L
J
K
PRODUCT DESIGN RESEARCH STUDY
INSTRUCTIONS
PERSONAL INFORMATION
INSTRUCTIONS
SECTION A
SECTION B
THANK YOU FOR YOUR CONTRIBUTION
IMAGE CREDIT Shane Crothers
8
Participants received both verbal and written
instructions on the task and were instructed to
determine which design is most representative,
moderately representative and least
representative of each keyword.
EXPERIMENTAL PROTOCOLS Participants received both
verbal and written instructions on the task to be
undertaken and were instructed to determine which
design is most representative, moderately
representative and least representative of each
keyword. The keywords were not printed on the
response sheet so that evaluation would reflect
the perceived character of the product rather
than previous responses. Also, keywords were
strategically presented to avoid the introduction
of systematic bias i.e. terms with obvious
associations such as, masculine and feminine
were not presented sequentially. Keywords were
presented, using the same size and type of font
(Arial) and because exposure time was not
considered to be of critical importance no limit
was imposed on response time.
IMAGE CREDIT Shane Crothers
9
There are many aspects of product appearance such
as colour, surface highlights and reflections
that may affect the perception of design
character. So in both experiments the kettles
were presented as black and white silhouettes.
EXPERIMENTAL OBJECTIVES The study consisted of
two experiments. The same procedures, sample
group and keywords were used in both experiments
only the stimulus set changed. Experiment 1
investigated the affect of changes in form on the
perception of design character while Experiment 2
investigated the affect of the addition of
graphic elements on perceived character.
Unfortunately, there are many aspects of product
appearance such as colour, surface highlights and
reflections that may affect the perception of
design character. So in both experiments the
kettles were presented as black and white
silhouettes.
IMAGE CREDIT Shane Crothers
10
Modifications were made to the spout, handle and
base of each primary type in order to generate
the stimulus set
EXPERIMENT 1 In Experiment 1 participants were
exposed to a set of 12 black and white kettle
silhouettes derived from 3 primary types
designated Alpha, Beta and Gamma. It
should be noted that these primary silhouettes
were based on 3 actual products. Modifications
were then made to the spout, handle and base of
each primary type in order to generate the set
of 12 i.e. each primary silhouette was
complemented by 3 variations.
PRIMARY
HANDLE
BODY
SPOUT
ALPHA
BETA
GAMMA
IMAGE CREDIT Shane Crothers
11
IMAGE CREDIT Shane Crothers
These images were then arranged so that no two
variations of each primary kettle type were
positioned adjacent to each other.
SAMPLE STIMULUS SET
12
IMAGE CREDIT Shane Crothers
EXPERIMENT 2 In Experiment 2 the shape/form did
not change but graphic elements were added to the
side of the kettle. It is reasonable to assume
that any change in perceived character would be
attributable to the addition of these graphic
elements. A new set of 8 kettles (see below)
designated KA through KH was used as the stimulus
and again design character was assessed using
pre-defined keywords.
IMAGE CREDIT Shane Crothers
13

• MAKING SENSE OF THE DATA
• Evidence supporting the formal hypothesis is
  provided in a number of ways
• Comments by the participants often provide
  useful insight into the evaluative process.
• Data is examined in order to identify
  agreement (consensus) on how design modifications
  affect the perception of design character.
•  
• Correlation coefficients are calculated using
  the Spearman Method.
•  
• Chi Squared values (written as C2) are
  computed which describe the degree of mismatch
  between expected values and observed values.

14
EXPERIMENT 1 HYPOTHESES The objective of this
first experiment is to provide evidence that
changes in form influence the perception of
design. This intention is captured in these
formal hypotheses. Ho Changes in form do not
alter the perceived character of a product H1
Changes in form do alter the perceived character
of a product STATISTICAL REASONING   If the null
hypothesis Ho is correct and changes in form do
not alter perceived character then there is an
equal probability of each kettle being seen as
representative of a given keyword. That is, each
kettle has an equal chance of being considered
representative of a given keyword. BUT If
there is not an equal distribution this would
suggest that Ho is false and changes in form do
alter the perceived character of the product
(suppH1). By calculating Chi Squared values
(written as C2) it is possible to quantify the
deviation from the expected equal distribution
and assess the significance of this
deviation.   Essentially then, the C2 statistic
measures the degree of disagreement or mismatch
between the data and the null hypothesis. The
further the observed responses (O) are from the
expected value (E) greater the mismatch and the
larger C2 will become. Therefore, large values
of C2 imply that Ho is false and that changes in
form do alter perceived character.
15
 
EXPERIMENT 1 DISCUSSION OF RESULTS   The
results indicate that, for this experiment, the
null hypothesis is invalid. The Observed values
(O) do not equal the expected values (E)
suggesting that Ho is false and that changes in
form do alter the character of a product.
Furthermore, all of the computed C2 values
(summarised left) achieved significance at the a
0.1 level or better. It is on this basis that
we reject the null hypothesis and assert that
changes in form do alter perceived character.  

  Table 1 Experiment 1 - C2 values for each
kettle suggesting that Ho is false
1 Veryzer, R., W., Hutchinson, J., W., (1998),
The influence of unity and prototypicality on
aesthetic responses to new product designs,
Journal of Consumer Research, Vol.24, March 2
McClave, J., T., Sincich, T., Statistics, London,
Prentice-Hall, 2000
16
EXPERIMENT 2 HYPOTHESES The objective of this
second experiment is to provide evidence that the
addition of graphic elements affects the
perception of design character. This is captured
in these formal hypotheses. Ho The addition
of graphic elements does not alter perceived
character H1 The addition of graphic elements
does affect the perceived character of a
product. STATISTICAL REASONING   The statistical
reasoning used here is exactly the same as in
Experiment 1. If the null hypothesis Ho is
correct and changes and the addition of graphic
elements does not alter perceived character then
there is an equal probability of each kettle
being seen as representative of a given keyword.
BUT If there is not an equal distribution
this would suggest that Ho is false and changes
in form do alter the perceived character of the
product (i.e. supporting H1). Again, it is
possible to quantify the deviation from the
expected equal distribution and assess the
significance of this deviation using the Chi
Squared method.   So as before, C2 statistic
measures the degree of disagreement or mismatch
between the data and the null hypothesis. Again,
large values of C2 imply that Ho is false. In
this case suggesting that the addition of graphic
elements does affect perceived character.    
17
EXPERIMENT 2 DISCUSSION OF RESULTS   Observed
values (O) do not equal the expected values (E)
suggesting that Ho is false and that the addition
of features does affect product perception.
Again, Chi Squared values describing the degree
of mismatch between the expected mean and the
observed values were calculated for each kettle.
Most of these values (summarised left) achieved
significance at the a 0.1 level or better
suggesting that Ho is false. It is on this basis
that we reject the null hypothesis and assert
that the addition of graphic elements does affect
product perceived character (H1).
  Table 5 Experiment 2 - C2 values for each
kettle suggesting that Ho is false
1 The statistical reasoning used to test the
formal hypothesis is the same as in Experiment 1
18
The evidence suggests that the character of a
kettle is affected more by modification of the
handle than modification of the body or the spout.
DETERMINANTS OF DESIGN CHARACTER Upon examination
of the results, it is interesting to note the
difference between the Chi Squared values for
each primary type and the modified
silhouettes. For each primary type there would
appear to be a large difference between Chi
Squared values for unmodified primary silhouette
and silhouettes with modified handles. Also, for
each primary kettle type there is only a small
difference between Chi Squared values for
silhouettes with modifications to the body and
the spout. Significantly, this suggests that the
character of a kettle is affected more by
modification of the handle than modification of
the body or the spout.
MODIFIED SILHOUETTES
PRIMARY
HANDLE
BODY
SPOUT
C2 115 (a .005)
C2 37 (a .005)
C2 70 (a 0.005)
C2 70 (a .005)
C2 127 (a .005)
C2 33 (a .005)
C2 31 (a 0.005)
C2 30 (a .005)
C2 296 (a .005)
C2 34 (a .005)
C2 53 (a .005)
C2 30 (a .005)
IMAGE CREDIT Shane Crothers
What are the practical implications?
19
IMAGE CREDIT www.autotrader.co.uk
Product character may be affected more by the
modification of certain design elements
OPPORTUNITIES FOR DESIGN PRACTITIONERS   Visual
re-styling is often used as a means of prolonging
the product life cycle. This practice is
particularly common with mature products - the
automobile being a notable example (see below).
But these facelifts cost money in terms of
design, re-tooling, etc. so clearly the goal
would be to cause the greatest (positive) impact
on consumer perception as the result of the
minimum number of design changes. This research
suggests that product character may be affected
more by the modification of certain design
elements. Using this empirical method to assess
perceived character it may be possible to
determine which design elements most strongly
affect product perception and focus creative
effort on those features thereby improving the
effectiveness of design.
This could help improve design effectiveness by
focusing creative effort on those features which
significantly affect product perception
IMAGE CREDIT www.autotrader.co.uk
20
IMAGE CREDIT Argos Catalogue Spring 2002
What exactly constitutes a masculine design and
what constitutes a feminine design?
GENDER PERCEPTION IN PRODUCT DESIGN Design is
often used to communicate gender and there are
many examples of products may be functionally
identical but are designed with a masculine or
feminine appearance, a good example being the
Remington electric razor (shown right). But what
exactly constitutes a masculine design and what
constitutes a feminine design? This research
provides experimental evidence which suggests
that certain shapes are considered to be
masculine while others are considered feminine.
Specifically, there is evidence which suggests
that graphic elements incorporating straight
lines are considered masculine, while graphic
elements incorporating curves and circles are
considered feminine.
Which of these products is masculine and which is
feminine? WHY?
21
PROTOTYPICALITY IN DESIGN   This research
indicates that perceived prototypicality might
have a positive affect on the formulation of
preference in product design. That is, consumers
prefer products that they perceive to be most
representative of a given product type or
category. Furthermore there is evidence to
suggest that these representations are not just
visual constructs they also carry gender
associations and meaning. WHAT IS A
PROTOTYPE? The concept of prototypicality
originated in computational cognitive psychology.
Essentially prototypes are mental constructs
that facilitate object recognition and can best
be described as an, idealised internalised
representation (Guenther, 1998). In this
context term prototype describes a product
which is considered to be representative of a
given product category or set. The term should
not be confused the more familiar interpretation
as being a visualisation or physical model of a
product concept. It should be also be noted that
a prototype does not necessarily correspond to
any physical object, it is an abstraction, a
composite of many examples of a given object.
BUT HOW CAN THIS BENEFIT DESIGN PRACTICE?
22
Successful design is often a balance between the
excitement derived from innovation and the
reassurance derived from familiarity
PRACTICAL IMPLICATIONS The notion that
prototypical (familiar) products may preferred
over atypical (novel designs) is not alien to
product design. Conventional wisdom states that
new products are adopted more rapidly if they are
compatible with existing values, product
knowledge and past experience. This principle is
often exploited in design. Early automobiles
were designed to look like horse drawn carriages
they replaced, 50s television sets were designed
to look like furniture, even early computers bore
more than a passing resemblance with the more
familiar typewriter (above right). Many new
products have failed largely because they
deviated too far from the consumers prototypical
preconceptions. In the 30s the Chrysler Airflow
(right) flopped because it was considered too
radical yet the VW (right) designed in 1937 by
Ferdinand Porsche incorporated much of the style
and technological innovation of the Airflow.
Successful design is often a balance between the
excitement derived from innovation and the
reassurance derived from familiarity. This
duality is reflected in Raymond Loews's MAYA
formula Most Advanced Yet Acceptable.  
Most Advanced Yet Acceptable.
Use of this method to assess perceived
prototypicality affords the opportunity to
explore prototypical preconceptions and provides
a means positioning a new product so that it is
perceived to be innovative but not excessively
radical.
IMAGE CREDIT www.webshots.com IMAGE CREDIT
Penny Sparke, Design for the 20th Century
23
Comments by the participants provide useful
insight into the evaluative process
This kettle was considered to be feminine by some
because it looked pregnant and masculine by
others because it looked as if it had a beer
belly
ANECDOTAL EVIDENCE   Comments by the participants
provide useful insight into the evaluative
process. Overheard comments suggest that, while
consensus and general agreement with respect to
perceived character certainly exists, it is
impossible insulate evaluation of design from
subjective influences. For instance, kettle E
(shown right) was considered to be feminine by
some because it looked pregnant and masculine
by others because it looked as if it had a beer
belly. This kind of comment illustrates that
evaluation may make reference to the social and
cultural context. One mechanism for this being
the use of analogy. It also demonstrates that
verbal language is often inadequate when
describing the physical and semantic properties
of a design. Therefore people often seek to
describe design through the use of an analogy
which describes the both shape and the character
of the object. The Ford KA (right) was once
described as, an egg with Grace Jones haircut.
The comment that kettle E had a beer belly was
an attempt to describe, not just the shape but
also the character (in this case the perceived
gender) of the product. Such comments represent
useful qualitative evidence that shape affects
design character.
The Ford KA was once described as, an egg with
Grace Jones haircut
People often seek to describe design through the
use of an analogy which describes the both shape
and character of the object.
IMAGE CREDIT Shane Crothers IMAGE CREDIT
www.autotrader.co.uk
24

• SIGNIFICANT FINDINGS
•  
• To date, this research shows that
• Empirical methods can be used to assess the
  perception of design character
• Quantitative techniques can be used to test
  formal hypothesis about design perception.
• Design changes, such as changes in form and
  the addition of features, affect the
• perceived character of a product
• Product perception may be affected more by the
  modification of certain design elements.
• Certain shapes are considered to be masculine
  and others feminine.
• Perceived prototypicality might have a
  positive affect on the formulation of preference
  in
• product design.

BUT MOST IMPORTANTLY.
25
MOST IMPORTANTLY This research demonstrates how
participant-based testing can be used to
complement traditional intuitive understanding of
how design is interpreted by the
consumer. AND Systematically manipulating
product variables and quantitatively assessing
the perception of design character represents a
scientific means of investigating design
perception.
26
IMAGE CREDIT www.webshots.com
FURTHER STUDY   To investigate the affect of
visual context and presentation format on the
perception of product character. Using this
method it should be possible to assess the affect
of visual context and presentation format (2D, 3D
and physical model) on perceived character.
Given that effective concept screening requires
accurate feedback from the consumer it is
important that the concept is presented in an
appropriate manner which is likely to yield an
realistic response.
Does the visual context affect the perception of
this Mercades?
27
IMAGE CREDIT www.webshots.com
Products are often presented in carefully chosen
context so that the product will be associated
with the positive attributes of the surroundings
FURTHER STUDY   Intuition tells us that visual
context affects product perception (just compare
the two Mercades shown below). This principle is
often used in advertising where a product is
presented in carefully chosen context. A sports
car may be presented in a highly dynamic,
exciting context as with the Golf GTI (shown
left). The hope being that the product will be
associated with the positive attributes of the
surroundings. Consider the new VW Beetle (shown
below left) which is presented in a fun,
fairground context beside the original Beetle.
Essentially, advertisers are trying to forge - or
to borrow a term from information theory - encode
conceptual relationships between the visual
context and product. Using this method it should
be possible to assess the affect of visual
context and presentation format (2D, 3D and
physical model) on perceived character. Given
that effective concept screening requires
accurate feedback from the consumer it is
important that the concept is presented in an
appropriate manner which is likely to yield an
realistic response.
A sports car may be presented in a highly
dynamic, exciting context as with this Golf GTI.
Here the new VW Beetle is presented in a fun,
fairground context beside the original
Beetle.advertisers are trying to forge - or to
borrow a term from information theory - encode
conceptual relationships between the visual
context and product.
IMAGE CREDIT www.webshots.com
28
COMMENTS Significantly, this research
demonstrates how participant-based testing can be
used to complement traditional intuitive methods
in design. It has been shown that quantitative
methods can be used to explore consumer
perception of design character and that these
methods could help improve design effectiveness
in a practical context. If you have any
comments or suggestions for the authors please
contact as your input would be greatly
appreciated and valued
29
ABOUT THE AUTHORS Shane Crothers BSc (Hons) is a
Research Associate in the School of Electrical
and Mechanical Engineering at the University of
Ulster. He is currently using an empirical
method for assessing design character to conduct
an investigation into design perception. Dr.
Robin Clarke BSc DPhil CEng MIMechE FIED is the
Head of the School of Electrical and Mechanical
Engineering and has research interests in product
design/development and engineering design
education. Dr Ian Montgomery BA DPhil MCSD
MSTDEd FRSA is Lecturer in Graphic Design and is
Course Director for BA(Hons) Visual Communication
in the School of Art Design. He has research
interests in design management and the
appropriateness of design.
EXHIBITION COMPLIED BY The Design Perceptions
Group at University of Ulster The Design
Perceptions Group is composed of researchers,
academics and students with a common interest in
studying design perception. If you are
interested in this research and wish to know more
please feel free to contact Shane Crothers,
Research Associate School of Electrical
Mechanical Engineering University of Ulster,
Newtownabbey Northern Ireland, BT37
0QB Telephone [phone] Fax [phone]
6804 Email se.crothers_at_ulst.ac.uk
30
REFERENCES Dahl, D., W., Chattopadhyay, A.,
Gorn, G., J., (1999), The use of visual mental
imagery in new product design, Journal of
Marketing Research, Vol.36, No.1 Guenther, R.,
K., Human Cognition, London, Prentice Hall,
1998 Hsiao, S-W., Chen, C-H., (1997), A semantic
and shape grammar based approach for product
design, Design Studies, Vol.18, No.3 Janlert,
L., E., (1997), The character of things, Design
Studies, Vol.18, No.3