You are on page 1of 16

Christian Homburg, Martin Schwemmle, & Christina Kuehnl

New Product Design: Concept,


Measurement, and Consequences
Product design is a source of competitive advantage for companies and is an important driver of company
performance. Drawing on an extensive literature review and consumer interviews, the authors define product
design and its dimensions. Using data from three samples (6,418 U.S. consumers and 1,083 and 583 European
consumers), the authors develop and validate a new scale to measure product design along the dimensions of
aesthetics, functionality, and symbolism. In addition, they investigate the impact of these design dimensions on
purchase intention, word of mouth, and willingness to pay. The results indicate that the design dimensions
positively influence willingness to pay and also have a positive effect on purchase intention and word of mouth,
both directly and indirectly through brand attitude.

Keywords: consumer behavior, product design, scale development, aesthetics, functionality, symbolism

Online Supplement: http://dx.doi.org/10.1509/jm.14.0199

P
Fifteen years ago companies competed on price. Today Dahl 2011). However, although design is a research priority
it’s quality. Tomorrow it’s design. for the Marketing Science Institute (2010, 2012), it is not
—Robert Hayes, Harvard University, 1998 yet a well-established field of academic inquiry (Di
Benedetto 2011; Luchs and Swan 2011; Veryzer 1995). Par-
roduct design is a source of competitive advantage
ticularly surprising is that despite product design’s high
for companies (Gemser and Leenders 2001; Noble
practical relevance, both academics and practitioners lack
and Kumar 2010). Recognizing that product design
can improve customer retention and thus increase company (1) a widely accepted definition of product design and its
performance (Candi 2010; Hertenstein, Platt, and Veryzer dimensions that is applicable across a broad range of prod-
2005), practitioners have acknowledged the significance of uct categories, (2) an adequate measure of it that is indepen-
product design for a company’s success. This importance dent of product categories, and (3) a systematic investiga-
holds especially true in today’s marketplace because prod- tion of design outcomes distinguishing the dimensions.
uct design has become a principal means of making prod- A significant body of empirical research has measured
ucts differentiable. The former chairman of Sony, Norio product design in a very limited way, ranging from a one-item
Ogha, underscored the weight of product design by com- measure referring to aesthetic aspects (Landwehr, Wentzel,
menting, “At Sony, we assume that all products of our com- and Herrmann 2012, 2013) to measures of selected dimen-
petitors have basically the same technology, price, perfor- sions, such as the hedonic and utilitarian dimensions (e.g.,
mance, and features. Design is the only thing that Chitturi, Raghunathan, and Mahajan 2008; Dhar and Werten-
differentiates one product from another in the marketplace” broch 2000). However, recent conceptual research extends
(Peters 2005, p. 39). beyond a one- or two-dimensional measurement and con-
Given the relevance of product design in practice, siders product design a three-dimensional concept (e.g.,
research has begun to address its merits (e.g., Bloch 2011; Bloch 2011) composed of the aesthetic, functional, and
symbolic dimensions. To the best of our knowledge, only
one study has used this three-dimensional conceptualiza-
tion, employing slightly different terms for the various
aspects (Srinivasan et al. 2012). However, that study
Christian Homburg is Professor of Marketing, Chair of the Marketing

focuses on one product category (cars) and uses highly spe-


Department, and Director of the Institute for Market-Oriented Manage-
ment, University of Mannheim (e-mail: homburg@bwl.uni-mannheim.de).

cific items for measuring these dimensions, such as “front


Martin Schwemmle is a doctoral student, Marketing Department, Univer-

seat head room” and “driving range between fuel stops”


sity of Mannheim (e-mail: martin.schwemmle@bwl.uni-mannheim.de).

(Srinivasan et al. 2012, p. 17).


Christina Kuehnl is Assistant Professor, Marketing Department, University

The third of the aforementioned shortcomings of previ-


of Mannheim (e-mail: ckuehnl@bwl.uni-mannheim.de). The authors thank
the JM review team for their insightful suggestions during the review

ous research results from the other two inadequacies. No


process, which improved this article considerably. They are also indebted

study yet has systematically linked the three aspects of


to Thu Hoa Le, Dominik Leiner/SoSci Panel, and the student assistants

design to key outcomes of consumer behavior. As a result,


involved in this project. The helpful comments of Monika Imschloß and

the importance of the three dimensions is to date unknown.


Arnd Vomberg are also gratefully acknowledged. Rajkumar Venkatesan
served as area editor for this article.

© 2015, American Marketing Association Journal of Marketing


ISSN: 0022-2429 (print), 1547-7185 (electronic) 41 Vol. 79 (May 2015), 41–56
Our study is an initial step in addressing the outlined Gestalt theory helps clarify on which level product
gaps and makes two main contributions. First, we theoreti- design should be measured: either on the atomistic level, to
cally develop and empirically validate a product category– measure the design elements such as color and shape, or on
independent scale to measure product design as a three- the holistic level, to measure the design of the product
dimensional construct. To do so, we follow established through consumers’ holistic perception of it. We chose the
scale-development procedures, use three data sets from dif- second option and measured product design through con-
ferent countries, and show discriminant validity of our scale sumers’ holistic perception because “one cannot visually
from the hedonic/utilitarian (HED/UT) scale (Voss, Span- recognize objects without perceiving the whole although
genberg, and Grohmann 2003). Before describing the scale- one can still recognize objects without perceiving the parts”
development process, we theoretically derive the definition (Jia, Shiv, and Rao 2014, p. 343). This reasoning suggests
of product design and its dimensions through an extensive that holistic processing is mandatory, whereas atomistic
literature review, and we validate this definition through processing is optional. Thus, relating the measurement of a
qualitative interviews with consumers. Although we scale to the mandatory process is appropriate.
employ visual product stimuli for the scale-development
process, our scale is also applicable to situations in which Product Design Definition
consumers can experience products through other senses. To the best of our knowledge, the literature contains no
Second, we apply the scale and investigate how the three widely accepted and conceptually sound definition of prod-
dimensions of product design affect key outcomes of con- uct design and its dimensions that can be transferred to a
sumer behavior—purchase intention, word of mouth scale and applied across a broad range of product categories
(WOM), and willingness to pay (WTP). We also introduce (Bloch 2011; Luchs and Swan 2011; Noble and Kumar
brand attitude as a mediator. Overall, we provide a scale for 2010). Therefore, before developing a product design scale
product design measurement that can be applied in practice
we offer a definition for product design. For this purpose,
and that advances research on product design.
we follow the procedure of Luchs and Swan (2011). These
In addition to being theoretically relevant, our work is
authors identified a list of 168 articles on product design
highly useful for practitioners in that it fosters a common
published between January 1995 and December 2008 in
understanding of what product design is, thus avoiding
eight leading journals (Journal of the Academy of Market-
communication problems within companies. Our threefold
ing Science, Journal of Business Research, Journal of Con-
conceptualization emphasizes that product design is more
sumer Research, Journal of Marketing, Journal of Market-
than the product’s aesthetic aspects and shows that during
ing Research, Journal of Product Innovation Management,
the research-and-development (R&D) process, interdiscipli-
nary teams should include experts on the functional and Management Science, and Marketing Science). Applying
symbolic dimensions. Practitioners can easily use our scale, the same procedure for the literature search in the same
which is based on this conceptualization, in field and online selection of journals, we extended this list by another 103
studies. The scale does not require product experience and articles that were published between December 2008 and
is therefore also valuable in the early stages of product September 2014 (the original list is available online at
development, such as testing prototypes or product con- http://masonweb.wm.edu/productdesign; the list of addi-
cepts. Finally, the scale also helps benchmark products with tional articles appears in Web Appendix W1). Our final set
regard to product design, both within the company and with comprised 271 articles, of which 24 contained a definition
competitors’ products, providing managers with data to jus- of product design (the articles and definitions appear in
tify design expenditures. Web Appendix W2). In many of these definitions, design
refers to a product’s properties or characteristics and/or
holistic dimensions, with which atomistic properties can be
Conceptualization combined (e.g., Bloch 2011; Luchs and Swan 2011; Orth
of Product Design and Malkewitz 2008).
Conceptualization refers to both the theoretical meaning of However, the definitions differ with regard to the con-
product design in terms of its definition and the specifica- tent and number of dimensions. Bloch (2011, p. 378), for
tion of the construct and its dimensions (Mueller 2004). The example, speaks of “utilitarian, hedonic, and semiotic bene-
conceptualization of product design, which we discuss in fits,” whereas Luchs and Swan (2011, p. 338) mention
this section, provides the basis for its operationalization. “form,” “function,” and “holistic properties of the inte-
grated form and function.” We therefore reviewed the 271
Theoretical Foundation: Gestalt Theory articles again with a focus on possible dimensions of prod-
Gestalt theory deals with how people perceive objects and uct design. We extracted 43 articles that contained some
postulates that the whole is more than the sum of its parts. expression of the conceptualization of product design (a list
That is, people perceive the unitary whole of an object (the of the 43 articles, the dimensions, and their definitions
so-called gestalt) rather than analyzing its separate constitu- appears in Web Appendix W3). We allocated all dimensions
tive elements (the parts). These types of perceptions are to six categories labeled according to the dimensions they
called holistic or gestalt processing on the gestalt level and contained: (1) aesthetics, (2) functionality, (3) symbolism,
atomistic processing on the parts level (Koffka 1962; Rad- (4) shape, (5) ergonomics, and (6) others, which included
ford and Bloch 2011; Wertheimer 1938). dimensions that did not fit into the other five categories.

42 / Journal of Marketing, May 2015


The categories and their corresponding dimensions appear other product categories, we conducted interviews with an
in Web Appendix W4. additional 32 consumers (53.13% female; mean age = 38
In line with our aim of developing a parsimonious scale years). This time, we asked consumers to think of a product
(Churchill 1979; Ramani and Kumar 2008; Rossiter 2002), of their choice whose design they liked, to explain why they
we determined whether all six dimensions were needed for liked it, and to describe how the design of this product dif-
the conceptualization of product design, which constitutes fers from the design of other products within that product
the basis for our scale-development process. This examina- category. We provide the interview questions in Web Appen-
tion led to eliminating or summarizing some dimensions. dix W6. All interviews were recorded and then transcribed.
We excluded shape, as this category refers to atomistic pro- We analyzed the interviews following established pro-
cessing, which we do not consider for our product design cedures of qualitative research (Strauss and Corbin 1998),
measurement. Within the “others” category, only three conducting the procedures separately for the two sets of
dimensions were mentioned twice, and we therefore interviews. First, we gave the transcribed interviews to
excluded this category. Because the ergonomics category is three coders who were not familiar with our research and
closely related to functionality in that it “concerns the com- did not know the three design dimensions. We provided a
prehensibility and usability of a product, the suitability to short briefing on the coding procedures and asked the
perform and correctly to communicate its utilitarian func- coders to abstract the interviews to codes through line-by-
tions” (Creusen and Schoormans 2005, p. 67), we sub- line analysis. Second, we asked the coders to summarize the
sumed it under functionality. codes of the first step and allocate them to different cate-
The three resulting dimensions of product design are gories. These categories were comparable for both sets of
aesthetics, functionality, and symbolism. We refer to them interviews with minor differences in terminology. Third, we
as “design dimensions” in the following discussion. In line checked whether the categories the coders had identified
with previous scholars (e.g., Batra and Ahtola 1990), we could further be meaningfully summarized. Apart from
propose that products have characteristics of all three categories that referred to atomistic processing (e.g., color,
design dimensions. That is, a product is not aesthetic or size) and minor aspects that were not directly related to
functional or symbolic but manifests all of these dimen- product design (e.g., price, ethics), we could allocate all
sions simultaneously to different extents. remaining categories to our three design dimensions. Table 1
To examine whether consumers also conceive product gives examples of quotations; Web Appendix W7 contains
design along these three literature-based design dimensions, more details on codes and categories. In summary, the inter-
we conducted qualitative interviews with consumers. We view analyses clearly demonstrate that consumers’ percep-
randomly selected two images of smartphones from a set of tions of product design align with our literature-based
three and showed them to 28 consumers (57.8% female; design dimensions.
mean age = 24.54 years). We asked them to select one Combining gestalt theory—positing the holistic percep-
smartphone and to describe the reasons that led to the selec- tion of products—with the literature review of the defini-
tion. (More details on the stimulus selection, images of the tions and dimensions of product design and the consumer
stimulus products and interview questions appear in Web interviews, we derive the following product design defini-
Appendix W5.) To show that the results are also valid for tion (inspired mainly by Bloch 2011; Luchs and Swan

TABLE 1
Consumer Quotations for the Three Design Dimensions
Dimension Consumer Quotations
Aesthetic dimension •“I think the design is more beautiful.”
•“I find this somehow more appealing. It seems somehow wider and rounder. This one is more
massive. That one is a little more elegant.”
•“When I imagine having this one in the kitchen, it would truly be an eye-catcher.”
•“I prefer this … because of its design: It’s just more striking.”
Functional dimension •“The features are pretty important to me. I don’t need a camera that is super good but I want a
strong Wi-Fi connection.”
•“It seems to be well made and sturdy … and probably is also Wi-Fi and even LTE-enabled.”
•“It is light and practical.”
•“Falling toy blocks don’t cause any scratches or defects. This table is made of a special, hardened
glass.”
Symbolic dimension •“When you want to portray something to the outside world, you always want to show that you have
good taste, that you value yourself sufficiently to allow yourself some luxuries such as a fancy
phone. You portray parts of your lifestyle.”
•“If a person really cares about how a product is designed, then maybe this could imply something
about that person. However if the person does not care then there is nothing that can be implied
about him or her.”
•“I think that many people define themselves through, for example, their smartphones.”
•“It’s elegant and streamlined. And it doesn’t have the image of a show-off car … or a granny car.”
Notes: The interviews were conducted in German and were translated by someone other than the authors.

New Product Design / 43


2011): product design refers to a set of constitutive ele- Developing a
ments of a product that consumers perceive and organize as Product Design Scale
a multidimensional construct comprising the three dimen-
sions of aesthetics, functionality, and symbolism. Because Review of Existing Scales
these constitutive elements can be both visual and non-
visual, this definition of product design applies to both Before we operationalize the three dimensions of product
design, we provide a review of existing scales that are
visual and nonvisual examination of products. Next, we
linked with product design. These measures include the
describe the three design dimensions in more detail.
Centrality of Visual Product Aesthetics scale (Bloch,
Product Design Dimensions Brunel, and Arnold 2003), the Vehicle Quality Survey
(Srinivasan et al. 2012), and the HED/UT scale (Voss,
Aesthetics. The aesthetic dimension refers to the per- Spangenberg, and Grohmann 2003). Next, we briefly elabo-
ceived appearance and beauty of a product (Bloch 2011; rate on each scale and explain why it is not adequate to
Desmet and Hekkert 2007). Aesthetics can be (1) an attri- meet our objectives.
bute of the product itself, (2) created in the eye of the behol- The Centrality of Visual Product Aesthetics scale mea-
der, or (3) a combination of these two options (Reber, sures the “level of significance that visual aesthetics hold
Schwarz, and Winkielman 2004). Our definition of the aes- for a particular consumer in his/her relationship with prod-
thetic dimension corresponds with the third, integrative ucts” (Bloch, Brunel, and Arnold 2003, p. 551). It consists
alternative and thus links to our definition of product of three subdimensions—value, acumen, and response—
design: a product has attributes that cause a perception of and comprises 11 items. Because this scale focuses on char-
beauty for the beholder (Leder et al. 2004). acteristics inherent in consumers and not in products, it
Functionality. The functional dimension reflects the does not meet our objective of measuring product design.
consumer’s perceptions of a product’s ability to fulfill its The Vehicle Quality Survey is part of the Automotive
purpose (Bloch 2011; Boztepe 2007). We acknowledge that Performance, Execution and Layout study conducted annu-
ally by J.D. Power and Associates. The survey asks con-
for some products (e.g., cars), a proper assessment of func-
sumers who have bought a new car to state how satisfied
tionality is only possible upon consumption or usage. How-
they are with 95 attributes of the car. Consequently, items
ever, as Hoegg and Alba (2011) emphasize, in many cases,
are highly specific to cars and refer to characteristics such
consumers are able to assess the functionality of a product as “front seat head room,” “driving range between fuel
from merely seeing it. This view is also supported by stops,” “look and feel of steering wheel,” and “overall inte-
Hollins and Pugh (1990) as well as Radford and Bloch rior quietness” (Srinivasan et al. 2012, p. 17). Three major
(2011). Such perceived functionality is especially important issues arise with this scale. First, an evaluation comprising
for online shops, in which consumers do not have the 95 items is burdensome. Second, because the items are
opportunity to comprehensively experience the product highly product category specific, they cannot be transferred
(Spears and Yazdanparast 2014). to other product categories. Third, because this scale
Symbolism. The symbolic dimension refers to the per- focuses on customer satisfaction, the consumer can make
ceived message a product communicates regarding a con- the evaluation only after having significant experience with
sumer’s self-image to both the consumer and others on the the product, whereas our definitions of the design dimen-
basis of visual elements (Aaker 1999; Belk 1988; Bloch sions explicitly require no prior use of the product.
2011). The symbolic meaning of a product can evoke varied The HED/UT scale was developed to measure “hedonic
associations, including those linked with a certain place or and utilitarian dimensions of consumer attitudes toward
product categories” (Voss, Spangenberg, and Grohmann
time (Creusen and Schoormans 2005). It can also be used to
2003, p. 311). The scale consists of ten adjective pairs, such
express personal values and dispositions or to form one’s
as “dull/exciting” and “unnecessary/necessary.” From a
own identity (McCracken 1986). This third dimension
conceptual point of view, it differs from our scale in one
extends our conceptualization beyond many existing prod- important respect: the HED/UT scale does not include the
uct design conceptualizations (see Web Appendix W3). important symbolic dimension and thus does not capture all
Symbolism is an essential dimension because aesthetics and relevant aspects of product design (Creusen and Schoor-
functionality alone do not fully capture product design mans 2005; Verganti 2008). In a subsequent section, we
(Bloch 2011; Rindova and Petkova 2007). For example, the empirically show that our product design scale, which
aesthetic and symbolic dimensions may contain opposing includes a symbolic dimension, is superior to the HED/UT
perceptions, as when “someone who likes a colorful design scale.
[aesthetic dimension] may not buy it because it looks ‘too
childish’ [symbolic dimension]” (Creusen and Schoormans Summary and Overview of the Scale-Development
2005, p. 67). Furthermore, the symbolic dimension is “as Process
important as the utilitarian perspective” (Verganti 2008, As the prior discussion demonstrates, existing scales are not
p. 440) because products often reflect consumers’ desire to appropriate for measuring product design. We therefore
express their extended selves (Belk 1988; Holt 1997; develop a new scale, which covers the design dimensions
Kleine, Kleine, and Kernan 1993). we derived conceptually, is parsimonious with respect to

44 / Journal of Marketing, May 2015


the number of dimensions and items (Churchill 1979; 1959; Churchill 1979; Gerbing and Anderson 1988). Figure 1
Rossiter 2002; Voss, Spangenberg, and Grohmann 2003), shows the six steps in the scale-development process.
and does not refer to a particular product category. In line
with prior research (e.g., Brakus, Schmitt, and Zarantonello Generation of the Item Pool
2009; Ramani and Kumar 2008), we followed established At the beginning of the scale-development process, we gave
scale-development procedures (e.g., Campbell and Fiske the definitions of the three design dimensions to 45 con-

FIGURE 1
Overview of the Scale-Development Process
Process Steps Data and Methods Results

(1) Item  45 consumers Initial set of 29 items



Generation  Literature review (Web Appendix W8).

 30 consumer judges
(2) Item  Statistical procedures (EFA, CFA), Final scale with nine items

Reduction data of 6,418 U.S. consumers were used (Table 3).
(Study 1).

 Discriminant validity between the design
dimensions: Fornell–Larcker criterion Model with three
(3) Scale
 Comparison of different models with CFAs,  dimensions has the best
Dimensionality
selection of the best model based on the fit (Table 4, Figure 2).
lowest AIC (data from Study 1).

 Collection of further data (1,083
Results are stable—i.e.,
European consumers, Study 2;
(4) Scale they could be verified with
583 European consumers, Study 3). 
Validation two other data sets (see
 Calculation of Step 3 with the new data
Web Appendix W10).
(Studies 2 and 3).

Discriminant validity of the
 CFAs to check discriminant validity of
design dimensions from
(5) Discriminant product design from brand personality and
 brand personality and the
Validity the HED/UT scale.
HED/UT scale could be
 Data from Studies 2 and 3 were used.
shown.

The three design
dimensions have an
(6) Use of the
 Structural equation model impact on WTP as well as
Scale to Analyze
(with mediation analysis/bootstrapping).  on purchase intention and
Outcomes of
 Data from Studies 1 and 3 were used. WOM, both directly and
Product Design
indirectly via brand
attitude (Figure 3).

Notes: EFA: exploratory factor analysis, CFA: confirmatory factor analysis, AIC: Akaike’s Information Criterion; WOM: word of mouth; WTP:
willingness to pay.

New Product Design / 45


sumers and asked them to generate questions to measure the TABLE 2
dimensions. Surprisingly, for each dimension the proposed Sample Characteristics (Study 1)
items were remarkably similar. With regard to the aesthetic
Criterion Characteristics
dimension, two aspects were mentioned frequently: the
product’s beauty and the consumers’ liking of the product. Age (years) M 39.95
With regard to the dimension of functionality, consumers SD 14.63
Gender Female 63.04%
emphasized that functional product designs reveal what the Male 36.96%
products do and imply how to use them so that they fulfill Education Some grade school .06%
their purpose and “do their job.” With regard to the sym- Completed grade school .56%
bolic dimension, answers were more diverse, but the gist of Some high school 2.82%
the responses was that symbolic product designs “show Completed high school 26.71%
who I am.” These statements correspond with the results of Some college 33.44%
Completed college 25.65%
the consumer interviews we conducted to validate the three Postgraduate 10.77%
design dimensions (Table 1). Occupation Not employed 33.90%
In addition to the user-generated items, we included (hours worked per week) Less than 35 hours 18.98%
items that were adapted from or inspired by existing scales 35 hours or more 47.09%
(e.g., Babin, Darden, and Griffin 1994; Brakus, Schmitt, Refused to answer .03%
and Zarantonello 2009; Peck and Childers 2003; Ramani Household income Under $10,000 4.32%
$10,000–$24,999 14.76%
and Kumar 2008; Sprott, Czellar, and Spangenberg 2009; $25,000–$39,999 20.65%
Tian, Bearden, and Hunter 2001). In total, these two $40,000–$74,999 34.14%
sources—consumers and the literature—yielded an initial $75,000 and above 19.01%
set of 29 items (see Web Appendix W8). Refused to answer 7.12%
Notes: Sample size = 6,418.
Reduction of the Number of Items (Study 1)
Because a scale with 29 items is too lengthy to be usable in Because we intended to develop a product category–
practice, we had to reduce the initial item pool. Reducing a independent scale, we ran our analyses on an aggregate
scale to an acceptable number of items essentially relies on level for all products. An exploratory factor analysis (EFA)
two approaches: the use of consumer judgments and statis- with Varimax rotation revealed three factors with eigenval-
tical purification procedures (Churchill 1979; Rossiter ues greater than 1. All items loaded on the factor they were
2002; Voss, Spangenberg, and Grohmann 2003). We supposed to load on (see Web Appendix W8).
applied both approaches. Next, we conducted confirmatory factor analyses (CFAs)
First, we asked 30 consumers (different from the 45 with Stata 13 (Brakus, Schmitt, and Zarantonello 2009; Gerb-
consumers who generated the items) to rate the degree to ing and Anderson 1988). With regard to the definition of prod-
which each of the 29 items fit the definition of its respective uct design as having three dimensions, we first ran a model
design dimension. Second, we applied statistical reduction with the three latent factors—aesthetics, functionality, and
procedures to the 29 items. For the data collection, we symbolism. This model showed acceptable goodness-of-fit
cooperated with a large international market research insti- measures. One functional item (F1) had an indicator reliability
tute that allowed us to integrate our items into its regular below .4; thus, we excluded it (Bagozzi and Yi 1988, 2012).
online panel survey of U.S. consumers. The market To achieve a parsimonious scale, we selected the three
research institute collected the data and oversaw the data items with the highest indicator reliabilities for every dimen-
collection process. Within this panel, 10,657 households sion (Batra, Lenk, and Wedel 2010; Park et al. 2010): A3,
received online access to the questionnaire. In total, 6,418 A4, A7; F3, F4, F6; and S3, S4, S5 for the aesthetic, func-
consumers (63% female; mean age = 39.95 years; effective tional, and symbolic dimensions, respectively (see Table 3).
response rate = 60.22%) used five-point Likert scales to rate Consumer judges had also given these items very high rat-
one randomly assigned image among 15 typical household ings. This smaller model (shown in Figure 2) achieved
and multimedia products with respect to the three design excellent goodness-of-fit values1: comparative fit index
dimensions. The products were a television, an MP3 player, (CFI) = .99; Tucker–Lewis index (TLI) = .99; standardized
a camcorder, a DVD player, two digital cameras (one com- root mean square residual (SRMR) = .02; root mean square
pact and one SLR), a pocket camcorder, a Blu-ray Disc error of approximation (RMSEA) = .05. All indicator relia-
player, a coffee machine, two irons, an espresso machine, a bilities were above .4 (the lowest was .71). Composite relia-
hair dryer, a juicer, and a vacuum cleaner. These products bilities (CRs) and average variances extracted (AVEs) were
reflect the common product range of a consumer electronics above the required thresholds of .6 and .5 (Hu and Bentler
retailer and are familiar to, relevant for, and owned by many 1995): CRaesthetic = .92; CRfunctional = .95; CRsymbolic = .93;
consumers. Furthermore, they have been used in previous
studies (e.g., Hoegg, Alba, and Dahl 2010; Mugge and Dahl 1Chi-square is a direct function of the sample size. Given the
2013). The item order of the product design scale was ran- large sample of 6,418 consumers, this measure could not be used
domized across participants. Table 2 provides the sample meaningfully for judging the model’s fit to the data and is there-
characteristics in greater detail. fore not reported.

46 / Journal of Marketing, May 2015


TABLE 3
Items, Means, and Standard Deviations (Study 1)
Item No. The product… M SD
A3 ...is visually striking 2.26 1.13
A4 ...is good looking 2.08 1.07
A7 ...looks appealing 2.10 1.07
F3 ...is likely to perform well 1.86 .88
F4 ...seems to be capable of doing its job 1.76 .85
F5 ...seems to be functional 1.78 .86
S4 ...would help me in establishing a distinctive image 3.34 1.29
S5 ...would be helpful to distinguish myself from the mass 3.34 1.30
S6 ...would accurately symbolize or express my achievements 3.35 1.29
Notes: All items assessed on a five-point scale (1 = “strongly agree,” and 5 = “strongly disagree”). Only items included in the final scale are
reported. For items F3–5 and S4–6, the questionnaire indicated to the respondent to judge the product “only from looking at it.” Item
numbers are consistent with Web Appendix W8.

FIGURE 2
CFA: The Three-Factor Model

*p < .01.
Notes: Numbers in parentheses refer to the item numbers as indicated in Table 3 and Web Appendix W8. All coefficient values are standard-
ized and appear above the associated path. Dotted lines and numbers next to them represent correlations of factors. Circles represent
the measurement errors for each of the items.

New Product Design / 47


AVEaesthetic = .79; AVEfunctional = .86; AVEsymbolic = .83. Validation of the Product Design Scale
Table 3 provides means and standard deviations of the final (Studies 2 and 3)
item selection. We also ran CFAs with more than three The main purpose of Studies 2 and 3 is to validate the
items per dimension and compared their goodness-of-fit design scale in a different cultural background and to col-
values (Voss, Spangenberg, and Grohmann 2003). The lect WTP as another dependent variable as well as further
three-item measurement proved to be the superior alterna- variables for discriminant validity testing. Whereas the data
tive (for the model comparison, see Web Appendix W9). for Study 1 were collected in the United States, data for
Establishing Scale Dimensionality Studies 2 and 3 were collected from European consumers.
Product design items were translated and checked with
A first indication of the scale’s dimensionality is that the backward translation.
model shown in Figure 2 clearly meets Fornell and Larcker’s Study 2 used different products from those in Study 1:
(1981) criterion, which requires that for every pair of fac- water kettles, headphones, and sunglasses. This difference
tors, the squared estimated correlation should be smaller from Study 1 is important because we aim to develop a
than each factor’s AVE. Computing these squared correla- product design scale that can be applied to a broad range of
tions from Figure 2 and comparing them with the AVEs product categories. For each category, we selected two
reveals that discriminant validity exists for each pair of fac- products that were randomly assigned to consumers. Like
tors. This finding is the first indication of the adequacy of prior investigators of product design (e.g., Hoegg, Alba, and
the three design dimensions underlying the nine items. Dahl 2010; Mugge and Dahl 2013), we selected these prod-
To further clarify the dimensionality, we compared ucts for reasons of relevance and familiarity to the target
alternative models with confirmative factor analyses (see group. We collected data with the help of the provider of a
Table 4) using procedures from prior research (Bloch, European online consumer panel that gives scholars access
Brunel, and Arnold 2003; Brakus, Schmitt, and Zaran- to its convenience sample for scientific purposes. Every
tonello 2009; Tian, Bearden, and Hunter 2001). The models consumer rated one product in each of the three product
were (1) the null model assuming correlations of 0 between categories. Consistent with Study 1, design dimensions
all measured variables, (2) the aforementioned three-factor were assessed with five-point Likert scales.
model (aesthetic, functional, and symbolic dimensions), We invited 6,500 randomly selected panel members to
and (3) a one-factor model in which all items load on one participate in our survey. Of these, 6,421 were successfully
factor. We also included three two-factor models in which sent e-mail invitations, and 1,088 recipients completed
items of two dimensions were grouped into one factor: (4) a questionnaires. In addition to offering a lottery of ten Ama-
model with the factors aesthetics/functionality and symbol- zon.com vouchers (€15 each), for each completed question-
ism, (5) a model with the factors aesthetics/symbolism and naire we donated €.50 to a charity organization supporting
functionality, and (6) a model with the factors functionality/ children who suffer from cancer. We eliminated four par-
symbolism and aesthetics. ticipants because of nondifferentiation (Barge and Gehlbach
For Models 2, 4, 5, and 6, we investigated both a case 2012) and one participant because of a low answering time,
with correlated factors and a case with orthogonal factors resulting in 1,083 usable questionnaires (63% female; mean
(Models 1 and 3 have only one factor and therefore no cor- age = 34.78 years; effective response rate: 16.87%). Table 5
relation between factors). In all cases, the correlated models presents details on the sample characteristics.
had goodness-of-fit values either the same as or better than In Study 3, we used a subset of the products that were
the corresponding models with orthogonal factors. The three- included in Study 1. We selected two products from the
factor model achieved the best fit statistics (see Table 4). In household category (iron and coffee machine) and two
addition, the Akaike information criterion (AIC) is lowest products from the multimedia category (television and SLR
for the three-factor solution (Homburg 1991). This result camera) and collected data from students of a major Euro-
confirms the product design operationalization with the aes- pean university. The selected products and their brands are
thetic, functional, and symbolic dimensions. common in Europe, so students would know them well.

TABLE 4
Model Comparison: Scale Dimensionality
AIC
Number Model Chi-Squarea d.f. CFI TLI RMSEA SRMR Differenceb
1 Null 54,059.15 36 —
.
2 Three factors 341.06 24 .994 .991 .045 .019 0
3 One factor 25,127.26 27 .535 .381 .381 .158 24,780.20
4 Two factors (A/F combined) 10,223.40 26 .811 .739 .247 .128 9,878.34
5 Two factors (A/S combined) 12,938.03 26 .761 .669 .278 .125 12,592.97
6 Two factors (F/S combined) 16,107.38 26 .702 .588 .310 .189 15,762.33
aChi-square is a direct function of the sample size. Given the large sample of 6,418 consumers, this measure could not be used meaningfully
for judging the model’s fit to the data.
bDifferences are calculated with regard to Model 2, which is the best model with the lowest AIC.
Notes: A = aesthetic; F = functional; S = symbolic. The model in boldface (Model 2) is the selected model.

48 / Journal of Marketing, May 2015


TABLE 5 tor 2. The two-factor model was superior because it showed
Sample Characteristics (Study 2) a better model fit and also had a lower AIC value. The
AVEs of both factors were greater than their squared corre-
Criterion Characteristics
lation. Thus, consumer attitudes toward product categories,
Age (years) M 34.77 which are measured by the HED/UT scale, and product
SD 14.02 design, which is measured by our scale, are constructs with
discriminant validity.
Gender Female 63.25%
Male 36.75%
Educationa Completed grade school .09% Second, we compared the constructs on the dimension
Completed high school 8.32% level. We ran a CFA with the five factors of HED, UT, aes-
Completed college 39.13% thetic, functional, and symbolic. The Fornell–Larcker crite-
Postgraduate 47.92% rion was met for every pair of these five factors. This result
Other 4.53% implies that not only the scales themselves but also the
Occupation School/education 3.52%
Student 34.17% dimensions of both scales—for example, aesthetic and
Employee 36.76% hedonic or utilitarian and functional—show discriminant
Self-employed 7.78% validity and are thus distinct.
Other 17.78% Third, whereas the HED/UT scale measures “hedonic
Monthly income Under €1,000 40.88% and utilitarian dimensions of consumer attitudes toward
€1,000–€2,499 33.62%
€2,500–€3,999 11.05%
product categories” (Voss, Spangenberg, and Grohmann
€4,000 and above 4.91% 2003, p. 311), our new scale measures product design.
Refused to answer 9.54% Thus, our scale should be a better predictor of product
aThe school system in Europe differs from the system in the United design than the HED/UT scale. To verify this argument and
States. We tried to make results comparable by allocating the further show the validity of our product design scale, we set
European categories to the U.S. categorization. “Postgraduate” up two models. In Model 1, our three design dimensions
includes all academic degrees (including a bachelor’s degree).
(measured through the corresponding nine items) are the
independent construct, and the product design factor is the
Notes: Sample size = 1,083.

Every participant was randomly assigned to one product. In dependent construct. This dependent construct is introduced
total, we received 587 questionnaires. For reasons of homo- by the statement “How do you assess the design of the
geneity, we excluded four participants because they were shown [product]?” and measured by three overall items (1 =
senior students and older than 40 years, resulting in 583 “like very much,” and 5 = “like not at all”; 1 = “very good,”
usable questionnaires (69.87% female; mean age = 20.4 and 5 = “very bad”; 1 = “very appealing,” and 5 = “not at
years). all appealing”; inspired by MacInnis and Park 1991). In
To validate the product design scale, we applied the pro- Model 2, instead of our design dimensions, the factors HED
cedures of Study 1 to Studies 2 and 3. Web Appendix W10 and UT (measured with their ten items) are the independent
provides detailed results, and we summarize the findings construct explaining the product design construct (Voss,
here. First, for both studies, an EFA with Varimax rotation Spangenberg, and Grohmann 2003). The variance of prod-
with the nine design items confirmed the three factors as uct design explained by the design dimensions model
identified in Study 1. That is, every item loaded on the (69%) clearly exceeds the variance explained by the
intended dimension and yielded no large cross-loadings. HED/UT model (43%).
Second, the comparison of models (see Table 4 and Web
Appendix W10) showed that for both studies, the three-factor
solution was the superior model. Third, fit indices, compos-
Using the Scale to Analyze
ite and indicator reliabilities, and AVEs of the three-factor Outcomes of Product Design
models all exceeded the recommended thresholds. Fourth, Theoretical design frameworks suggest that product design
the Fornell–Larcker criterion was met for every pair of fac- influences consumer behavior (Bloch 1995; Luchs and
tors. In summary, these analyses of both samples confirm Swan 2011; Noble and Kumar 2010). However, these
the scale items and the scale dimensionality of aesthetic, frameworks are conceptual, and prior empirical work has
functional, and symbolic dimensions for other products in focused on single elements of product design. Thus, to date,
different regions (United States vs. Europe). no empirical research has linked product design in its
entirety (rather than merely parts of the customer design)
Product Design Scale and HED/UT Scale (Study 3) with consumer behavior. Our definition and scale allow us
Previously, we discussed the conceptual difference between to close this gap. In this section, we explicitly consider the
our scale and the HED/UT scale (Voss, Spangenberg, and aesthetic, functional, and symbolic dimensions separately.
Grohmann 2003). We next extend this reasoning with This discrete analysis enables us to investigate which
empirical analyses. dimension has a stronger impact and should therefore be the
First, we compared a one-factor model (with the items focus of managers’ attention. The empirical investigation of
of the HED/UT scale and the items of our scale loading on the three design dimensions, especially the separate exami-
the same factor) with a two-factor model in which HED/UT nation of effects for each dimension, is a central contribu-
items load on factor 1 and our product design items on fac- tion of this article.

New Product Design / 49


In line with the theoretical framework of Luchs and We further expect that the three design dimensions posi-
Swan (2011), we chose two prominent constructs of con- tively affect WOM. Consumers engage in WOM because of
sumer behavior as outcomes of product design. First, we social, emotional, or functional benefits—that is, to “send
considered purchase intention because it has been widely signals to others about one’s expertise [and] uniqueness,...
used in consumer research and because design researchers to share positive or negative feelings, and ... to provide and
also regard it as an important outcome of product design supply information” (Lovett, Peres, and Shachar 2013, p.
(e.g., Chang and Wildt 1994; Herbst et al. 2012; Raghubir 428). One of the risks associated with WOM is the sharing
and Greenleaf 2006). Second, we selected WOM as an out- of inappropriate advice (Gatignon and Robertson 1986).
come of product design in our model. Defined as “informal Product design can act as a cue or signal and reduce the risk
communications between private parties concerning evalua- of this kind of WOM (Akdeniz, Calantone, and Voorhees
tions of goods and services to other consumers” (Anderson 2013; Erdem, Swait, and Valenzuela 2006). We therefore
1998, p. 6), WOM is a common variable in marketing and hypothesize the following:
product design research (e.g., Cheema and Kaikati 2010;
H2: The (a) aesthetic, (b) functional, and (c) symbolic dimen-
Chitturi, Raghunathan, and Mahajan 2008; Lovett, Peres, sions of product design positively influence WOM.
and Shachar 2013). Because consumers can talk about the
design of a product solely on the basis of its visual evalua- In addition to these direct effects, we also expect an
tion, such as after seeing it in a shop window, and because indirect effect of product design on purchase intention and
the aforementioned definition of WOM refers only to com- WOM through brand attitude. Product design positively
munication and not to recommendations, which would influences brand attitude and can help build a strong brand
require product experience, WOM fulfills our requirement identity. In particular, it can reflect brand values, create con-
that the product design scale require no experience with sistency over the product range, and support the differentia-
using the product. tion of brands (Karjalainen and Snelders 2010; Schmitt and
Furthermore, prior research suggests a strong link Simonson 2009; Stompff 2003). Product design can serve
between product design and brand and calls for its empiri- as a signal and, as such, help consumers to overcome uncer-
cal investigation (Keller and Lehmann 2006; Luchs and tainty such as doubt about the quality of a product. This line
Swan 2011; Noble and Kumar 2010). Against this back- of reasoning leads to the following hypothesis:
ground and because this application is the first empirical H3: The (a) aesthetic, (b) functional, and (c) symbolic dimen-
use of our product design scale, we aimed to analyze the sions of product design positively influence brand attitude.
role of the brand for the effects of product design in a more
general way. We thus looked for an overall brand construct. In addition, we assume that brand attitude influences the
The literature suggests using brand attitude, which refers to dependent variables of purchase intention and WOM. When
an overall evaluation of the brand (e.g., Keller 2003) and consumers trust and positively think about a brand, they
thus perfectly matches our aim. Therefore, we selected increase their intention to purchase this brand (Herbst et al.
brand attitude, defined as “psychological tendencies to 2012). Analogously, a positive brand attitude should influ-
evaluate objects along a degree of favor or liking” (Schmitt ence purchase intention. Furthermore, a positive brand atti-
2012, p. 7), as a mediator for our model. tude also positively influences WOM (Lovett, Peres, and
Shachar 2013). Thus,
Hypothesis Development
H4: A positive brand attitude increases purchase intention.
Product design influences consumers’ purchase intention H5: A positive brand attitude increases WOM.
through several mechanisms. Product design (especially the
aesthetic dimension) affects consumers’ evaluation of the
product, their immediate desire to own the product, and Measures
their decision to acquire it (Reimann et al. 2010; Yeung and To test our hypotheses, we relied on the U.S. consumer
Wyer 2005). Moreover, product design (especially the func- sample used in Study 1. We measured product design with
tional dimension) can be “a reliable indicator of functional the items of our product design scale. We assessed WOM
performance” (Hoegg and Alba 2011, p. 346) and thus can and brand attitude as single-item constructs: “I would tell
increase the likelihood of purchase (Bao, Bao, and Sheng other people about that product” (1 = “strongly agree,” and
2011; Hoegg and Alba 2011; Richardson, Dick, and Jain 5 = “strongly disagree”), inspired by Chitturi, Raghunathan,
1994). The symbolic dimension in particular can help con- and Mahajan (2008), and “My attitude toward the brand is
sumers protect or maintain their self-concept (Belk 1988; very positive” (1 = “strongly agree,” and 5 = “strongly dis-
Sirgy 1982; Tian, Bearden, and Hunter 2001). Because con- agree”), inspired by Ferraro, Kirmani, and Matherly (2013).
sumers want to defend and foster the understanding of who Participants indicated their purchase intention with two
they are, products and product designs that are linked with items: “How do you feel about buying this product in the
this self-supporting function have a greater likelihood of near future?” (1 = “definitively would buy,” and 5 = “defin-
being purchased (Tian, Bearden, and Hunter 2001). Accord- itively would not buy”) and “When would you be most
ingly, we propose the following: willing to buy this product?” (1 = “within the next 3
H1: The (a) aesthetic, (b) functional, and (c) symbolic dimen- months,” and 10 = “never”; intermediate labels were also
sions of product design positively influence purchase shown), inspired by Coulter and Coulter (2005) and Herbst
intention. et al. (2012).

50 / Journal of Marketing, May 2015


The use of single-item measures is subject to discussion SRMR = .04; RMSEA = .07). These fit indices are in line
in marketing research because of potential reliability and with the established thresholds (Bagozzi and Yi 1988, 2012;
validity issues. We used single-item measures for three rea- Hu and Bentler 1995). All indicator reliabilities were above
sons. First, this use was a restriction imposed by the market .4, all AVEs were above .5, and all construct reliabilities
research institute, which would not allow us to add more were above .6. We found discriminant validity—checked
items for the dependent variables. Second, previous with the Fornell–Larcker criterion—for all factors.
research in consumer behavior and product design has used All path coefficients were positive and significant at the
single-item measures for brand attitude and WOM .01 level except the direct effect of the aesthetic dimension
(Bergkvist and Rossiter 2007; Chitturi, Raghunathan, and on purchase intention, which was not significant. This result
Mahajan 2008). Third, research has shown that single-item means that the functional and symbolic dimensions influ-
measures can be highly reliable and valid (Bergkvist and ence purchase intention directly. Furthermore, all three
Rossiter 2007; Fuchs and Diamantopoulos 2009; Wanous, dimensions influence WOM directly and influence both
Reichers, and Hudy 1997). In addition, our results are fairly dependent variables indirectly through brand attitude. Thus,
independent of the reliabilities of our single-item constructs apart from H1a, all hypotheses are confirmed. In the follow-
(e.g., whether we assign a value of 1 or .85; Jöreskog and ing subsections, we elaborate on these effects.
Sörbom 1982; Malhotra et al. 2012).
Direct effects. Regarding the standardized direct effects
Results and Discussion of product design on purchase intention, the effect of the
To check our hypotheses, we ran a structural equation symbolic dimension (.25) was more than three times the
model (Figure 3) with bootstrapping (10,000 repetitions; effect of the functional dimension (.08). This effect might
Preacher and Hayes 2008; details appear in Web Appendix be due to consumers’ strong will to express themselves with
W11) and a reliability of .85 for our single-item measures, the help of products. This result and the insignificant effect
which yielded excellent model fit2 (CFI = .97; TLI = .96; of aesthetics on purchase intention might also stem from the
influence of the product category, which we analyze subse-
2Chi-square is a direct function of the sample size. Given the quently. Regarding the direct effects of product design on
large sample of 6,418 consumers, this measure could not be used WOM, all three design dimensions showed positive and
meaningfully for judging the model’s fit to the data and is there- significant coefficients with similar effect sizes (aesthetic:
fore not reported. .17; functional: .18; symbolic: .21).

FIGURE 3
Analyzing Outcomes of Product Design

*p < .01.
Notes: All coefficients are standardized and appear near the associated path(s). Boldfaced arrows represent three paths: aesthetic (A), func-
tional (F), and symbolic (S).

New Product Design / 51


Indirect and total effects. Aside from their direct effects, effect on brand attitude (followed by functionality and then
the product design dimensions also had an effect on attitude symbolism), irrespective of the product category. These
toward the brand, which in turn had positive effects on the results further support the validity of our findings.
dependent variables. This effect is called an indirect effect, Post-hoc analysis with WTP.4 In addition, we investi-
and brand attitude is called a mediator. A mediator is a third gated WTP as an additional dependent variable. Therefore,
variable that “represents the generative mechanism through in the survey of the data collection for Study 3, we included
which the focal independent variable is able to influence the WTP with an open question: “How much would you be
dependent variable of interest” (Baron and Kenny 1986, willing to pay for this [product]?” (Miller et al. 2011). A
p. 1173). If only the indirect effect is significant, full medi- structural equation model with dummy variables for the
ation occurs. If both the direct and indirect paths are signifi- products (i.e., controlling for their different price levels);
cant, partial mediation occurs (Zhang, Wedel, and Pieters the three design dimensions of aesthetics, functionality, and
2009). All indirect effects are positive and significant for symbolism as independent variables; and WTP as the
both dependent variables in our model. Regarding the sig- dependent variable showed adequate goodness of fit
nificant direct effects, we can state that brand attitude par- (c2(48) = 101.51, p < .001; CFI = .98; TLI = .97; SRMR =
tially mediates the effects of product design. Because the .03; RMSEA = .05). The standardized path coefficients of
indirect effect, but not the direct effect, of the aesthetic the design dimensions on WTP were all positive and sig-
dimension on purchase intention (.12) was significant, the nificant (p < .05; aesthetic: .09, functional: .14, symbolic:
aesthetic dimension is fully mediated by brand attitude. .11). The result that the functional dimension shows the
For the functional dimension, the indirect effect through strongest effect on WTP is consistent with Okada’s (2005)
brand attitude on purchase intention was greater than the findings.
direct effect (.10 vs. .08). The indirect effect of the sym- In summary, the three design dimensions of aesthetics,
bolic dimension was smaller than the direct effect (.05 vs. functionality, and symbolism significantly influence three
.25) but is still positive. For the dependent variable WOM, relevant dependent variables of consumer behavior: pur-
indirect effects were .10 (aesthetics), .09 (functionality), chase intention, WOM, and WTP. The finding that the sym-
and .05 (symbolism). As for purchase intention, the indirect bolic dimension shows strong effects for all dependent
effect of the symbolic dimension was the smallest indirect variables emphasizes the need for its inclusion as a dimen-
effect. Nevertheless, all three indirect paths on WOM sion of product design.
through brand attitude increase the total effect sizes of the
design dimensions.
Adding the direct and indirect effects together leads to General Discussion
the total effects. The standardized total effects on purchase
intention are .12 (aesthetics), .18 (functionality), and .30 Theoretical Implications
(symbolism) and on WOM are .27 (aesthetics), .27 (func- Although product design is recognized as being important
tionality), and .26 (symbolism). Thus, all design dimensions for both research and practice, some gaps still exist with
positively influence the dependent variables of our model. regard to its conceptualization and operationalization as
Differences for product categories.3 Furthermore, to well as for the investigation of product design outcomes. In
analyze whether the product category exerts an influence, closing these gaps, we contribute to product design research
we estimated the model shown in Figure 3 for multimedia in several ways.
and household products separately. Notably, the effect of First, we have defined product design as a set of consti-
aesthetics on purchase intention, which had previously been tutive elements of a product that consumers perceive and
insignificant, became significant (p < .05) for multimedia organize as a multidimensional construct comprising the
products. This finding seems logical because aesthetics play three dimensions of aesthetics, functionality, and symbol-
a larger role for multimedia products than for household ism. The definition of product design as comprising these
products. Furthermore, all other effects of the design three dimensions builds on a large literature review and
dimensions (including the effects on purchase intention, extends the existing distinction between hedonic and utili-
WOM, and brand attitude) showed consistency across product tarian dimensions by adding a symbolic dimension.
categories. Moreover, our results indicate that the pattern of Second, our investigation is the first to operationalize the
effect sizes also remains stable across the three models— product design construct by developing a product category–
that is, including all products, household products only, and independent design scale along these three dimensions. The
multimedia products only. In particular, among the three scale is easy to administer and is adaptable to different
design dimensions, for all three models the symbolic research contexts, and it thus advances research on the
dimension constitutes the strongest driver of purchase effects and processes underlying product design. Most
intention, followed by functionality and then aesthetics. The importantly, we verified the scale’s applicability in three
three design dimensions influence WOM almost equally, studies using different products from different product cate-
consistently across product categories. Finally, the aesthetic gories in distinct cultural contexts. Moreover, the scale
dimension of product design always shows the strongest demonstrated discriminant validity from the HED/UT scale.

3We thank an anonymous reviewer for this helpful suggestion. 4We thank an anonymous reviewer for this helpful suggestion.

52 / Journal of Marketing, May 2015


The results of this study have several theoretical impli- design of fast-moving consumer goods. Future studies
cations. First, we develop a theoretically well-grounded and could additionally test and, if necessary, adapt or develop a
easily applicable conceptualization and scale of product new design scale for products in the service industries (e.g.,
design that provides researchers with an instrument for software) or business-to-business industries. A related limi-
product design research. Investigators can build on our tation refers to the extent of newness of the product cate-
scale and deepen the understanding of product design in gories tested in our study. Specifically, although marketers
marketing. Furthermore, a common operationalization could certainly employ our scale to test new product proto-
makes results comparable. Second, some previous studies types, we can only speculate on the degree to which our
in product design research have used only the aesthetic scale is applicable to radical innovations that evoke the
dimension or the hedonic and utilitarian dimensions to cap- development of a totally new product category.
ture product design. In our model, the total effects on pur- Third, our results show that the pattern of effect sizes
chase intention were greater for the functional (.18) and the remains stable for the three design dimensions irrespective
symbolic (.30) dimensions than for the aesthetic dimension of product category. This finding requires deeper investiga-
(.12). This result implies that understanding design solely in tion, as one might expect the importance of the symbolic
terms of aesthetics restricts the full potential that product dimension to vary across product categories. For example,
design could have and also that a broad understanding one could assume that consumers might find the symbolic
includes the functional and symbolic dimensions. Thus, an dimension to be more relevant for publicly consumed prod-
important aspect of this article is the introduction of the ucts than for privately consumed products. In a similar vein,
symbolic dimension and the evidence that all three dimen- the importance of the design dimensions may vary across
sions of product design have an impact on consumer behavior. purchase decisions, differing, for example, between buying
Third, we have shown that product design influences a product for oneself and purchasing a gift for someone
the outcome variables on both direct and indirect paths. else. Further research might therefore explore when and
This finding is especially relevant for further research on how the three product design dimensions vary in influenc-
ing distinct outcomes of consumer behavior.
product design, which should especially consider additional
mediators or moderators. Managerial Implications
Limitations and Future Research Directions Practitioners also acknowledge the importance of product
design; therefore, generating new and helpful insights for
Our conceptualization and operationalization of product
them is essential. Our research helps managers in four criti-
design lay the foundation for further research to better
cal ways. First, a sound definition of product design offers
understand how product design works. As with any study,
practitioners a common mental representation and under-
our work contains some limitations, which may provide a
standing of what constitutes design. A clear definition could
starting point for further research.
help managers avoid getting lost in personal convictions or
First, although our product design definition refers to both
subjective judgments of product design, and having a com-
visual and nonvisual elements of a product, we employed mon definition also helps circumvent communication prob-
product images as stimuli for the scale-development lems within companies.
process. Despite our examination of strictly visual stimuli, Second, we conceptualize and operationalize product
our scale and its items are also applicable to nonvisual stim- design as comprising three dimensions: aesthetics, func-
uli. Consumers’ perceptions of product design can include tionality, and symbolism. This expanded conceptualization
constitutive product elements that can be either visual or might be contrary to what many managers have in mind
nonvisual, and because the design dimensions are more when they think about design, because design often is
abstract representations of these constitutive elements, they restricted to the aesthetic dimension. However, our research
can refer to both visual and nonvisual stimuli. This refer- shows that this perspective is inadequate from both theo-
ence is apparent through a closer examination of our items. retical and empirical viewpoints. Although all three design
For example, the item “is likely to perform well” (F3 of the dimensions have a measureable impact on purchase inten-
functional dimension) or the item “would help me in estab- tion, WOM, and WTP, the aesthetic dimension does not
lishing a distinctive image” (S4 of the symbolic dimension) have the strongest effect. Consequently, companies should
can also be answered through a haptic (instead of visual) actively pay attention to all three dimensions. This finding
evaluation of a specific product design. With some slight implies that product design can benefit from contributions
modifications of the items, such as “feels striking” instead from interdisciplinary R&D or design teams consisting of,
of “looks striking” (A3), the aesthetic dimension might also for example, artists/designers (aesthetics), engineers (func-
work well for nonvisual stimuli. Further research might vali- tionality), and market researchers who know their cus-
date this claim by employing our scale with, for example, tomers’ lifestyle (symbolism).
haptic stimuli. Third, we provide a scale to measure product design.
Second, we demonstrate the generalizability of our With the help of this scale, practitioners can let consumers
scale for 16 product categories. We note that these cate- evaluate current or future products with respect to the three
gories refer solely to consumer durables; therefore, we call design dimensions. Our scale items are easy to understand
for further research to explore whether our product design and applicable to many product categories, and because the
scale is readily transferrable to, for example, the package scale consists of only nine items, it can be answered rela-

New Product Design / 53


tively quickly. In addition, because the scale requires nei- performance of competitors. This evaluation may also help
ther product experience nor handling of the product, it can justify product design expenditures.
be used in product tests for new products and in cases in Fourth, our model demonstrates that product design
which only product images are shown. This aspect makes it influences purchase intention and WOM both directly and
applicable to both field and online studies. Moreover, hav- indirectly through brand attitude. Thus, companies should
ing the ability to measure product design with our scale, align their product design decisions with branding deci-
companies may consider and measure product design as a sions. Practically, this recommendation implies that the
performance indicator and as a sound criterion for evaluat- R&D and marketing departments (especially all teams
ing the company’s design performance against the design responsible for branding) should work together closely.

REFERENCES
Aaker, Jennifer L. (1999), “The Malleable Self: The Role of Self- Boztepe, Suzan (2007), “User Value: Competing Theories and
Expression in Persuasion,” Journal of Marketing Research, 36 Models,” International Journal of Design, 1 (2), 55–63.
(February), 45–57. Brakus, J.J., Bernd H. Schmitt, and Lia Zarantonello (2009),
Akdeniz, M. Billur, Roger J. Calantone, and Clay M. Voorhees “Brand Experience: What Is It? How Is It Measured? Does It
(2013), “Signaling Quality: An Examination of the Effects of Affect Loyalty?” Journal of Marketing, 73 (May), 52–68.
Marketing- and Nonmarketing-Controlled Signals on Percep- Campbell, Donald T. and Donald W. Fiske (1959), “Convergent
tions of Automotive Brand Quality,” Journal of Product Inno- and Discriminant Validation by the Multitrait–Multimethod
vation Management, 728–43. Matrix,” Psychological Bulletin, 56 (2), 81–105.
Anderson, E.W. (1998), “Customer Satisfaction and Word of Candi, Marina (2010), “Benefits of Aesthetic Design as an Ele-
Mouth,” Journal of Service Research, 1 (1), 5–17. ment of New Service Development,” Journal of Product Inno-
Babin, Barry J., William R. Darden, and Mitch Griffin (1994), vation Management, 27 (7), 1047–64.
“Work and/or Fun: Measuring Hedonic and Utilitarian Shop- Chang, Tung-Zong and Albert R. Wildt (1994), “Price, Product
ping Value,” Journal of Consumer Research, 20 (4), 644–56. Information, and Purchase Intention: An Empirical Study,”
Bagozzi, Richard P. and Youjae Yi (1988), “On the Evaluation of Journal of the Academy of Marketing Science, 22 (1), 16–27.
Structural Equation Models,” Journal of the Academy of Mar- Cheema, Amar and Andrew M. Kaikati (2010), “The Effect of
keting Science, 16 (1), 74–94. Need for Uniqueness on Word of Mouth,” Journal of Market-
——— and ——— (2012), “Specification, Evaluation, and Interpre- ing Research, 47 (June), 553–63.
tation of Structural Equation Models,” Journal of the Academy Chitturi, Ravindra, Rajagopal Raghunathan, and Vijay Mahajan
of Marketing Science, 40 (1), 8–34. (2008), “Delight by Design: The Role of Hedonic Versus Utili-
Bao, Yongchuan, Yeqing Bao, and Shibin Sheng (2011), “Motivat- tarian Benefits,” Journal of Marketing, 72 (May), 48–63.
ing Purchase of Private Brands: Effects of Store Image, Prod- Churchill, Gilbert A., Jr. (1979), “A Paradigm for Developing Bet-
uct Signatureness, and Quality Variation,” Journal of Business ter Measures of Marketing Constructs,” Journal of Marketing
Research, 64 (2), 220–26. Research, 16 (February), 64–73.
Barge, Scott and Hunter Gehlbach (2012), “Using the Theory of Coulter, Keith S. and Robin A. Coulter (2005), “Size Does Matter:
Satisficing to Evaluate the Quality of Survey Data,” Research The Effects of Magnitude Representation Congruency on Price
in Higher Education, 53 (2), 182–200. Perceptions and Purchase Likelihood,” Journal of Consumer
Baron, Reuben M. and David A. Kenny (1986), “The Moderator– Psychology, 15 (1), 64–76.
Mediator Variable Distinction in Social Psychological Research: Creusen, Mariëlle E.H. and Jan P.L. Schoormans (2005), “The
Conceptual, Strategic, and Statistical Considerations,” Journal Different Roles of Product Appearance in Consumer Choice,”
of Personality and Social Psychology, 51 (6), 1173–82. Journal of Product Innovation Management, 22 (1), 63–81.
Batra, Rajeev and Olli T. Ahtola (1990), “Measuring the Hedonic Dahl, Darren W. (2011), “Clarity in Defining Product Design:
and Utilitarian Sources of Consumer Attitudes,” Marketing Inspiring Research Opportunities for the Design Process,”
Letters, 2 (2), 159–70. Journal of Product Innovation Management, 28 (3), 425–27.
———, Peter Lenk, and Michel Wedel (2010), “Brand Extension Desmet, Pieter and Paul Hekkert (2007), “Framework of Product
Strategy Planning: Empirical Estimation of Brand–Category Experience,” International Journal of Design, 1 (1), (accessed
Personality Fit and Atypicality,” Journal of Marketing March 12, 2015), [available at http://www.ijdesign.org/ojs/
Research, 47 (April), 335–47. index.php/IJDesign/article/view/66/15].
Belk, Russell W. (1988), “Possessions and the Extended Self,” Dhar, Ravi and Klaus Wertenbroch (2000), “Consumer Choice
Journal of Consumer Research, 15 (2), 139–68. Between Hedonic and Utilitarian Goods,” Journal of Market-
Bergkvist, Lars and John R. Rossiter (2007), “The Predictive ing Research, 37 (February), 60–71.
Validity of Multiple-Item Versus Single-Item Measures of the Di Benedetto, C.A. (2011), “From the Editor,” Journal of Product
Same Constructs,” Journal of Marketing Research, 44 (May), Innovation Management, 28 (3), 320.
175–84. Erdem, Tülin, Joffre Swait, and Ana Valenzuela (2006), “Brands
Bloch, Peter H. (1995), “Seeking the Ideal Form: Product Design as Signals: A Cross-Country Validation Study,” Journal of
and Consumer Response,” Journal of Marketing, 59 (July), Marketing, 70 (January), 34–49.
16–29. Ferraro, Rosellina, Amna Kirmani, and Ted Matherly (2013),
——— (2011), “Product Design and Marketing: Reflections After “Look at Me! Look at Me! Conspicuous Brand Usage, Self-
Fifteen Years,” Journal of Product Innovation Management, 28 Brand Connection, and Dilution,” Journal of Marketing
(3), 378–80. Research, 50 (August), 477–88.
———, Frédéric F. Brunel, and Todd J. Arnold (2003), “Individual Fornell, Claes and David F. Larcker (1981), “Evaluating Structural
Differences in the Centrality of Visual Product Aesthetics: Equation Models with Unobservable Variables and Measure-
Concept and Measurement,” Journal of Consumer Research, ment Error,” Journal of Marketing Research, 18 (February),
29 (4), 551–65. 39–50.

54 / Journal of Marketing, May 2015


Fuchs, Christoph and Adamantios Diamantopoulos (2009), “Using Landwehr, Jan R., Daniel Wentzel, and Andreas Herrmann (2012),
Single-Item Measures for Construct Measurement in Manage- “The Tipping Point of Design: How Product Design and
ment Research: Conceptual Issues and Application Guide- Brands Interact to Affect Consumers’ Preferences,” Psychology
lines,” Die Betriebswirtschaft, 69 (2), 195–210. and Marketing, 29 (6), 422–33.
Gatignon, Hubert and Thomas S. Robertson (1986), “An ———,———, and ——— (2013), “Product Design for the Long
Exchange Theory Model of Interpersonal Communication,” in Run: Consumer Responses to Typical and Atypical Designs at
Advances in Consumer Research, Vol. 13, Richard J. Lutz, ed. Different Stages of Exposure,” Journal of Marketing, 77 (Sep-
Provo, UT: Association for Consumer Research, 534–38. tember), 92–107.
Gemser, Gerda and Mark A.A.M. Leenders (2001), “How Inte- Leder, Helmut, Benno Belke, Andries Oeberst, and Dorothee
grating Industrial Design in the Product Development Process Augustin (2004), “A Model of Aesthetic Appreciation and Aes-
thetic Judgments,” British Journal of Psychology (95), 489–
Impacts on Company Performance,” Journal of Product Inno-
508.
vation Management, 18 (1), 28–38.
Lovett, Mitchell J., Renana Peres, and Ron Shachar (2013), “On
Gerbing, David W. and James C. Anderson (1988), “An Updated Brands and Word of Mouth,” Journal of Marketing Research,
Paradigm for Scale Development Incorporating Unidimension- 50 (November), 427–44.
ality and Its Assessment,” Journal of Marketing Research, 25 Luchs, Michael and K.S. Swan (2011), “Perspective: The Emer-
(May), 186–92. gence of Product Design as a Field of Marketing Inquiry,”
Herbst, Kenneth C., Eli J. Finkel, David Allan, and Gráinne M. Journal of Product Innovation Management, 28 (3), 327–45.
Fitzsimons (2012), “On the Dangers of Pulling a Fast One: MacInnis, Deborah J. and C.W. Park (1991), “The Differential Role
Advertisement Disclaimer Speed, Brand Trust, and Purchase of Characteristics of Music on High- and Low-Involvement
Intention,” Journal of Consumer Research, 38 (5), 909–919. Consumers’ Processing of Ads,” Journal of Consumer Research,
———, ———, and Robert W. Veryzer (2005), “The Impact of 18 (2), 161–73.
Industrial Design Effectiveness on Corporate Financial Perfor- Malhotra, Naresh K., Soumya Mukhopadhyay, Xiaoyan Liu, and
mance,” Journal of Product Innovation Management, 22 (1), Satyabhusan Dash (2012), “One, Few or Many? An Integrated
3–21. Framework for Identifying the Items in Measurement Scales,”
Hoegg, JoAndrea and Joseph W. Alba (2011), “Seeing Is Believ- International Journal of Market Research, 54 (6), 835–62.
ing (Too Much): The Influence of Product Form on Perceptions Marketing Science Institute (2010), 2010–2012 Research Priori-
of Functional Performance,” Journal of Product Innovation ties. Cambridge, MA: Marketing Science Institute.
Management, 28 (3), 346–59. ——— (2012), 2012–2014 Research Priorities. Cambridge, MA:
———, ———, and Darren W. Dahl (2010), “The Good, the Bad, Marketing Science Institute.
and the Ugly: Influence of Aesthetics on Product Feature Judg- McCracken, Grant (1986), “Culture and Consumption: A Theoreti-
ments,” Journal of Consumer Psychology, 20 (4), 419–30. cal Account of the Structure and Movement of the Cultural
Meaning of Consumer Goods,” Journal of Consumer
Hollins, Bill and Stuart Pugh (1990), Successful Product Design:
Research, 13 (1), 71–84.
What to Do and When. London: Butterworth-Heinemann.
Miller, Klaus M., Reto Hofstetter, Harley Krohmer, and Z.J.
Holt, Douglas B. (1997), “Poststructuralist Lifestyle Analysis:
Zhang (2011), “How Should Consumers’ Willingness to Pay
Conceptualizing the Social Patterning of Consumption in Post- Be Measured? An Empirical Comparison of State-of-the-Art
modernity,” Journal of Consumer Research, 23 (4), 326–50. Approaches,” Journal of Marketing Research, 48 (February),
Homburg, Christian (1991), “Cross-Validation and Information 172–84.
Criteria in Causal Modeling,” Journal of Marketing Research, Mueller, Charles W. (2004), “Conceptualization, Operationaliza-
28 (May), 137–44. tion, and Measurement,” in The Sage Encyclopedia of Social
Hu, Li-tse and Peter M. Bentler (1995), “Evaluating Model Fit,” in Science Research Methods, Michael S. Lewis-Beck, Alan Bry-
Structural Equation Modeling: Concepts, Issues, and Applica- man, and Tim F. Liao, eds. Thousand Oaks, CA: Sage Publica-
tion, Rick H. Hoyle, ed. Thousand Oaks, CA: Sage Publica- tions, 162–66.
tions, 76–99. Mugge, Ruth and Darren W. Dahl (2013), “Seeking the Ideal
Jia, Jayson Shi, Baba Shiv, and Sanjay Rao (2014), “The Product- Level of Design Newness: Consumer Response to Radical and
Agnosia Effect: How More Visual Impressions Affect Product Incremental Product Design,” Journal of Product Innovation
Distinctiveness in Comparative Choice,” Journal of Consumer Management, 30 (S1), 34–47.
Research, 41 (2), 342–60. Noble, Charles H. and Minu Kumar (2010), “Exploring the Appeal
Jöreskog, Karl G. and Dag Sörbom (1982), “Recent Developments of Product Design: A Grounded, Value-Based Model of Key
in Structural Equation Modeling,” Journal of Marketing Design Elements and Relationships,” Journal of Product Inno-
Research, 19 (November), 404–416. vation Management, 27 (5), 640–57.
Karjalainen, Toni-Matti and Dirk Snelders (2010), “Designing Okada, Erica M. (2005), “Justification Effects on Consumer
Visual Recognition for the Brand,” Journal of Product Innova- Choice of Hedonic and Utilitarian Goods,” Journal of Market-
tion Management, 27 (1), 6–22. ing Research, 42 (February), 43–53.
Orth, Ulrich R. and Keven Malkewitz (2008), “Holistic Package
Keller, Kevin L. (2003), “Brand Synthesis: The Multidimensional-
Design and Consumer Brand Impressions,” Journal of Market-
ity of Brand Knowledge,” Journal of Consumer Research, 29
ing, 72 (May), 64–81.
(4), 595–600. Park, C. Whan, Deborah J. MacInnis, Joseph Priester, Andreas B.
——— and Donald R. Lehmann (2006), “Brands and Branding: Eisingerich, and Dawn Iacobucci (2010), “Brand Attachment
Research Findings and Future Priorities,” Marketing Science, and Brand Attitude Strength: Conceptual and Empirical Differ-
25 (6), 740–59. entiation of Two Critical Brand Equity Drivers,” Journal of
Kleine, Robert E., Susan Schultz Kleine, and Jerome B. Kernan Marketing, 74 (November), 1–17.
(1993), “Mundane Consumption and the Self: A Social-Identity Peck, Joann and Terry L. Childers (2003), “Individual Differences
Perspective,” Journal of Consumer Psychology, 2 (3), 209– in Haptic Information Processing: The ‘Need for Touch’
235. Scale,” Journal of Consumer Research, 30 (3), 430–42.
Koffka, Kurt (1962), Principles of Gestalt Psychology. London: Peters, Thomas J. (2005), Design: Tom Peters Essentials. London:
Lund Humpheries. Dorling Kindersley.

New Product Design / 55


Preacher, Kristopher J. and Andrew F. Hayes (2008), “Asymptotic Sprott, David, Sandor Czellar, and Eric Spangenberg (2009), “The
and Resampling Strategies for Assessing and Comparing Indi- Importance of a General Measure of Brand Engagement on
rect Effects in Multiple Mediator Models,” Behavior Research Market Behavior: Development and Validation of a Scale,”
Methods, 40 (3), 879–91. Journal of Marketing Research, 46 (February), 92–104.
Radford, Scott K. and Peter H. Bloch (2011), “Linking Innovation Srinivasan, Raji, Gary L. Lilien, Arvind Rangaswamy, Gina M.
to Design: Consumer Responses to Visual Product Newness,” Pingitore, and Daniel Seldin (2012), “The Total Product
Journal of Product Innovation Management, 28 (S1), 208–220. Design Concept and an Application to the Auto Market,” Jour-
Raghubir, Priya and Eric A. Greenleaf (2006), “Ratios in Propor- nal of Product Innovation Management, 29 (S1), 3–20.
tion: What Should the Shape of the Package Be?” Journal of Stompff, Guido (2003), “The Forgotten Bond: Brand Identity and
Marketing, 70 (April), 95–107. Product Design,” Design Management Journal, 14 (1), 26–32.
Ramani, Girish and V. Kumar (2008), “Interaction Orientation and Strauss, Anselm L. and Juliet M. Corbin (1998), Basics of Qualita-
Firm Performance,” Journal of Marketing, 72 (January), 27– tive Research: Grounded Theory Procedures and Techniques.
45. Newbury Park, CA: Sage Publications.
Reber, Rolf, Norbert Schwarz, and Piotr Winkielman (2004), Tian, Kelly T., William O. Bearden, and Gary L. Hunter (2001),
“Processing Fluency and Aesthetic Pleasure: Is Beauty in the “Consumers’ Need for Uniqueness: Scale Development and
Perceiver’s Processing Experience?” Personality & Social Psy- Validation,” Journal of Consumer Research, 28 (1), 50–66.
chology Review, 8 (4), 364–82. Verganti, Roberto (2008), “Design, Meanings, and Radical Inno-
Reimann, Martin, Judith Zaichkowsky, Carolin Neuhaus, Thomas vation: A Metamodel and a Research Agenda,” Journal of
Bender, and Bernd Weber (2010), “Aesthetic Package Design: Product Innovation Management, 25 (5), 436–56.
A Behavioral, Neural, and Psychological Investigation,” Jour- Veryzer, Robert W. (1995), “The Place of Product Design and Aes-
nal of Consumer Psychology, 20 (4), 431–41. thetics in Consumer Research,” in Advances in Consumer
Richardson, Paul S., Alan S. Dick, and Arun K. Jain (1994), Research, Vol. 22, Frank R. Kardes and Mita Sujan, eds. Provo,
“Extrinsic and Intrinsic Cue Effects on Perceptions of Store UT: Association for Consumer Research, 641–45.
Brand Quality,” Journal of Marketing, 58 (October), 28–36. Voss, Kevin E., Eric R. Spangenberg, and Bianca Grohmann
Rindova, Violina P. and Antoaneta P. Petkova (2007), “When Is a (2003), “Measuring the Hedonic and Utilitarian Dimensions of
New Thing a Good Thing? Technological Change, Product Consumer Attitude,” Journal of Marketing Research, 40
Form Design, and Perceptions of Value for Product Innova- (August), 310–20.
tions,” Organization Science, 18 (2), 217–32. Wanous, John P., Arnon E. Reichers, and Michael J. Hudy (1997),
Rossiter, John R. (2002), “The C-OAR-SE Procedure for Scale “Overall Job Satisfaction: How Good Are Single-Item Mea-
Development in Marketing,” International Journal of Research sures?” Journal of Applied Psychology, 82 (2), 247–52.
in Marketing, 19 (4), 305–335. Wertheimer, Max (1938), “Über Gestalttheorie,” in A Source Book
Schmitt, Bernd (2012), “The Consumer Psychology of Brands,” of Gestalt Psychology, Willis D. Ellis, ed. New York: Harcourt,
Brand Insights from Psychological and Neurophysiological Brace and Co.
Perspectives, 22 (1), 7–17. Yeung, Catherine W. and Robert S. Wyer (2005), “Does Loving a
——— and Alex Simonson (2009), Marketing Aesthetics. New Brand Mean Loving Its Products? The Role of Brand-Elicited
York: The Free Press. Affect in Brand Extension Evaluations,” Journal of Marketing
Sirgy, M. Joseph (1982), “Self-Concept in Consumer Behavior: A Research, 42 (November), 495–506.
Critical Review,” Journal of Consumer Research, 9 (3), 287– Zhang, Jie, Michel Wedel, and Rik Pieters (2009), “Sales Effects
300. of Attention to Feature Advertisements: A Bayesian Mediation
Spears, Nancy and Atefeh Yazdanparast (2014), “Revealing Analysis,” Journal of Marketing Research, 46 (October), 669–
Obstacles to the Consumer Imagination,” Journal of Consumer 81.
Psychology, 24 (3), 363–72.

56 / Journal of Marketing, May 2015

You might also like