Вы находитесь на странице: 1из 12

Beyond First Impressions:

The Effects of Repeated Exposure


on Consumer Liking of Visually
Complex and Simple Product Designs
Dena Cox
Anthony D. Cox
Indiana University

This article presents an experiment examining the effects


of stimulus complexity on consumers' aesthetic preferences. The results suggest that preferences for visually
complex product designs tend to increase with repeated exposure, while preferences for visually simple product designs tend to decrease with repeated exposure. In addition,
the results suggest that perceived complexity partially mediates the exposure-preference relationship. The authors
discuss implications of these findings for market researchers conducting aesthetic product design concept tests, as
well as more basic research on the affective impact of repeated exposure.

Traditionally, aesthetics dominated consumer product


choices in only a few product categories, most notably
women's and men's apparel. In recent decades, however,
the competitive importance of product aesthetics has
spread into a growing number of industries (see, e.g.,
Schmitt and Simonson 1997). Lorenz (1986) documented
the increasing emphasis on aesthetics in the competitive
strategies of Sony, Ford Motor Company, and a variety of
other firms not normally associated with fashion. Management consultant Tom Peters devoted an entire chapter of
his 1992 book Liberation Management to the increasing
importance of design (both aesthetic and ergonomic) in
global competition, describing how even high-technology
Journal of the Academy of Marketing Science.
Volume 30, No. 2, page 119-130.
Copyright 9 2002 by Academy of Marketing Science.

products such as electronic personal planners are being


marketed as "fashionable accessories." Bloch (1995)
noted that "not since the 1930's has product design been
more creatively and strategically employed to gain advantage in the global marketplace" Aesthetic product design
has become so important that even discount retailers like
K-Mart and Target are rushing to deliver aesthetic
designer touches on everyday objects, like teapots and
kitchen utensils. K-Mart has used the aesthetic touches of
Martha Stewart in their housewares division, while Target
has hired the famed architect, Michael Graves, to use his
aesthetic powers on 200 everyday items in their store. This
venture has been so successful to Target that they are entering into a multimillion-dollar partnership with furniture
designer Philippe Starck to add flair to their branded furniture (Thau 2000).
On the positive side, the increasing importance of product aesthetics provides marketers with another tool for
competitive differentiation. However, this trend also
forces an increasing number of companies into the difficult and risky business of forecasting consumers' tastes.
One way in which companies attempt to reduce this
fashion risk is through visual concept tests, in which consumers react to drawings or prototypes of potential aesthetic designs before a commitment is made to bring them
to market (see, e.g., Kotler and Rath 1984; Lorenz 1986;
Page and Rosenbaum 1987). In theory, this approach
seems very reasonable, but in practice it often has problems. Lorenz (1986) documented several instances in
which companies have found wide discrepancies between
the aesthetic designs for which consumers express preference in early concept tests and what they ultimately buy.

120

JOURNALOF THE ACADEMYOF MARKETINGSCIENCE

SPRING2002

Similar observations have been made by Robinson (1975)


and P. Morgan (personal communication, June 1992).
What accounts for this apparent discrepancy? One
explanation may lie in the way premarket design preferences are measured. When design concept tests are conducted in industry, respondents' evaluations are almost
always collected after a single exposure to the product (R.
Barber, personal communication, June 1992; P. Morgan,
personal communication, June 1992; Schwartz 1987).
However, once a newly designed product (e.g., a new garment or automobile model) is actually brought to market,
consumers typically have multiple opportunities to
observe it before making a final evaluation. This raises
some important questions: Do consumer preferences for
product designs change with repeated exposure? Does the
nature of this change depend on the nature of the design?
The answers to these questions could have important
implications for the conduct and interpretation of
premarket design concept tests; however, the literature on
concept testing neither discusses nor empirically examines them. Furthermore, as Bloch (1995) noted, "Despite
the centrality of product design to marketing practice and
society as a whole, empirical studies on design issues are
rare in marketing journals" (p. 17). What design factors
will positively affect consumer's affective reaction to
products? In this study, we will seek to examine these
questions.
This article examines the effect of repeated exposure on
consumers' aesthetic preferences and how this effect may
vary depending on the nature of a product's visual design.
In the following sections, we review past research from
psychology on how repetition affects aesthetic preferences
for various types of stimuli and draw some implications
for understanding consumer reactions to new product
designs. We formulate hypotheses and then test them in a
controlled experiment. Finally, we discuss the implications of our study, both for product concept testing and for
future research on consumers' aesthetic preferences.

liking. What has not been discussed widely in the marketing literature is the fact that several studies subsequent to
Zajonc's did not confirm his hypothesis. For example,
Cantor (1968) found that children's liking of geometric
shapes decreased with repeated exposure.
This apparent contradiction led Berlyne (1970) to seek
a contingency variable that might moderate the effects of
exposure on affect. Berlyne suggested that the inconsistent
findings of Zajonc and Cantor could be explained by the
relative complexity of their stimuli. Zajonc had used relatively complex stimuli (e.g., Chinese characters), while
Cantor had used simple geometric shapes. Berlyne
hypothesized that exposure effects were due to the interplay of two factors, "positive habituation" and "tedium,"
which occur at different exposure levels for complex and
simple stimuli. He suggested that the initial presentation of
a complex stimulus (such as those used by Zajonc) creates
an uncomfortably high level of uncertainty and that
repeated exposures would decrease this uncertainty and
enhance individuals' comfort or liking for the stimulus.
However, individuals could familiarize themselves with
simple stimuli on the first exposure, and repeated exposures might decrease liking due to boredom. Stimulus
uncertainty eventually drops to a point where little more is
learned from additional exposure, tedium sets in, and further exposures decrease liking of the stimulus. Berlyne
hypothesized that complex stimuli would induce tedium
only after a large number of exposures. However, individuals could familiarize themselves with simple stimuli on
the first exposure, and repeated exposures would more
quickly lead to tedium and a drop in liking. Stang and
O' Connell (1974) also theorized that the process of learning about a complex stimulus across numerous exposures
is itself satisfying, and this also increases liking of the
stimulus.
Bornstein (1989:270-271) reviewed nine experiments
that examined Berlyne's hypothesis. Six of these nine
experiments found that repeated exposure had a more positive effect on the liking of visually complex stimuli. In
addition, two studies not cited by Bornstein (Cox and Cox
1988; Smith and Dorfman 1975) found more positive
effects of exposure on complex stimuli. However, none of
these studies have examined how design complexity might
moderate the effect of exposure on consumer product preferences. Furthermore, Bornstein noted that all six confirming studies employed a within-subjects design, which
allowed individual participants to compare the complex
and simple stimuli. To ensure that Berlyne's results were
not simply due to demand artifacts, Bornstein recommended that future studies include a between-subjects
manipulation of stimulus complexity.
None of the studies reviewed above has examined consumer response to product design complexity and how this
response changes with repeated exposure. Since Bornstein
(1989) stressed that exposure effects can vary markedly

LITERATURE REVIEW
Repeated Exposure and Aesthetic Preferences
In 1968, social psychologist Robert Zajonc reviewed a
diverse set of evidence that seemed to show that the mere
repeated exposure to unfamiliar or novel stimuli (e.g.,
visual patterns and Chinese characters) increased individuals' liking of these stimuli. After the publication of
Zajonc's article, many studies reported similarly positive
effects of exposure on affect (for reviews, see Bornstein
1989; Bornstein and D'Agostino 1992).
In addition, several marketing studies (e.g., Cox and
Cox 1988; Janiszewski 1988, 1993; Obermiller 1985)
have documented positive effects of repetition on stimulus

Cox,Cox/ CONSUMERS'PREFERENCES I21


from one class of stimulus to another, one cannot merely
assume that effects observed for nonsense syllables and
Chinese characters can be generalized to new product
designs. Instead, it is important to conduct empirical
research to examine these phenomena. In so doing, we
seek to form an empirical basis for extending previous theory into a new stimulus domain, which is highly relevant to
marketing. In addition, our study seeks to add to prior theory by examining how changes in perceived complexity
may mediate the effects of repeated exposure on stimulus
liking.

Relevant Marketing Literature


There have been very few studies in marketing that have
explored product design issues. Specifically, researchers
have given little attention to the effects of product design
complexity on liking or how this relationship may change
with repeated exposure. Janiszewski (1988) found that a
single prior exposure to a picture of a woman wearing a fur
coat seemed to enhance consumers' liking of this product.
However, Janiszewski only examined the impact of one
prior exposure and did not vary any product design characteristics; he only examined a single product design.
Cox and Cox (1988) examined the effects of repetition
on liking for advertisements. They found that consumer
evaluations of ads with complex layouts increased
between first and second exposures, while evaluations of
simple layout ads did not. This study used only one prior
exposure to the stimulus, had no mediation analysis, and
(most important) did not manipulate characteristics of
product design. In addition, this study used a within-subjects design, which Bornstein (1989) indicated may be
vulnerable to demand effects.
Finally, Veryzer and Hutchinson (1998) empirically
examined the effects of two aesthetic design characteristics, prototypicality and unity, on consumer liking for the
product. The authors concluded that product designs that
were highly unified and prototypical were most liked by
consumers. However, consumer responses were measured
after only one exposure to the product designs. The authors
did not examine how consumers' preferences for different
product designs might change with repeated exposure.
In sum, previous studies have not examined the effects
of exposure on consumer liking of alternative product
designs. In general, exposure research in both psychology
and marketing (e.g., Obermiller 1985) has tended to
employ abstract or meaningless stimuli (e.g., nonsense
syllables, series of musical tones) rather than recognizable
objects such as products. Furthermore, Bornstein
(1989:269-270) noted that exposure effects vary among
categories of stimuli. Specifically, he noted that mere
exposure experiments employing drawings and paintings
(a common presentation mode in new product concept
testing) have been sparse and have produced inconsistent

findings. For these reasons, it is important to conduct


research examining how repeated exposure affects consumer evaluations of differing product design concepts.

HYPOTHESES
According to research in experimental aesthetics
(Berlyne 1970; Berlyne and Lawrence 1964; see also
Hekkert and Von Wieringen 1990), aesthetic preference is
related to a stimulus's arousal potential in an inverted-U
shaped pattern, in which the most preferred stimuli are
those that are moderately complex. Berlyne's research
suggests that individuals tend to dislike stimuli that they
perceive to be either too simple or too complex. Thus, we
expect the following:

Hypothesis 1: There will be an inverted-U relationship


between individuals' perceived product design complexity and product liking: products perceived to be
moderately complex will be most liked.
Berlyne (1970) suggested that when presented with
complex stimuli, respondents feel uncertainty and arousal.
From Gestalt psychology, we know that there is a human
tendency to find patterns in stimuli, even those that are randomly generated (Hochberg 1998). Each repetition of the
stimulus presents more opportunities for the randomness
to be simplified or encoded by the respondent. As this encoding process takes place, the perceived complexity of
the stimulus should decrease (Falk and Konold 1997).
Therefore, we expect respondents' judgments of products'
perceived complexity to decrease across repeated exposures to those products. Thus, we hypothesize the following:

Hypothesis 2: The perceived complexity of a product's


design will decrease with repeated exposures to the
stimulus.
Repeated exposure is expected to move respondents'
perceptions of high-complexity stimuli closer to the optimal (moderate) complexity level. In contrast, repeated exposure is expected to cause perceptions of low-complexity
stimuli to move even further away from the optimal (moderate) perceived complexity level. Although Berlyne
(1970) suggested that tedium may eventually set in for
complex stimuli, this is only expected to occur after a very
large number of repetitions. Within the range of repetitions
(one to four) examined in this study, we hypothesize the
following:

Hypothesis 3: Repeated exposure will move objectively


complex stimuli closer to the optimal perceived
complexity level and objectively simple stimuli
away from the optimal perceived complexity level.

122

JOURNALOF THE ACADEMYOF MARKETINGSCIENCE

SPRING2002

Consistent with these hypotheses, we predict that the


effects of repeated exposure on consumers' liking of a
product design will vary depending on the nature of that
product's visual design. Specifically, past research suggests that after several exposures to complex stimuli, liking
tends to increase at a fairly consistent level.
The findings concerning the effects of exposure on simple stimuli are somewhat less clear-cut than those involving complex stimuli. Most studies suggest that liking
drops (presumably as a result of tedium) at a much lower
number of exposures for simple stimuli than for complex
stimuli. However, the evidence is less consistent regarding
the exact point at which such negative effects are likely to
occur. Some studies suggest that visually simple stimuli
increase in liking with initial exposure (albeit at a slower
rate than complex stimuli) and that a drop occurs only after
several exposures (e.g., Saegert and Jellison 1970; see also
Cox and Cox 1988). Others show an immediate decline in
the liking of visually simple stimuli, even after one exposure (e.g., Berlyne 1970; Cantor 1968; Smith and Dorfman
1975). A common thread among these studies is that exposure tends to have a less positive effect on the liking of simple versus complex stimuli. Thus, within the range of
repetitions examined in this study, we hypothesize the following:

exposures in with other stimuli), we hoped to make it less


obvious to the participants that we were focusing on
repeated exposure.

Hypothesis 4a: Repeated exposure to visually complex


product designs will have a positive effect on consumers' liking of these products.
Hypothesis 4b: Repeated exposure will have a less positive effect on consumers' liking of visually simple
product designs than on liking of visually complex
product designs.

METHOD
Overview and Experimental Design
To test our hypotheses, we designed a 2 x 3 full factorial
between-subjects experiment with two levels of product
design complexity (simple and complex) and three levels
of preevaluation exposure (0, 1, 3 times). We also
employed multiple operationalizations (3) of each level of
product design complexity to increase the construct validity of this manipulation (Cook and Campbell 1979), using
women's fashion products as the stimuli.
Two aspects of this experimental design were intended
to reduce the chance of participants' hypothesis guessing
(a potential problem in mere exposure studies; see, e.g.,
Bornstein 1989). First, by employing a between-subjects
experimental design (in which each participant was
exposed to only one level of both repetition and complexity), we hoped to disguise the fact that these factors were
being varied. Second, by limiting the maximum
preevaluation exposures to three (and mixing these

Participants
Three hundred ninety-four volunteer participants were
recruited from undergraduate business classes at a large
midwestern university. As an incentive to participate, participants' names were placed in a lottery to win a Sony
compact disc player. Thirteen cases were excluded
because of excessive missing data, leaving a usable sample
of 381 participants. Participants' ages ranged from 21 to
41, with a median age of 22. The sample included both
female (36%) and male (64%) participants. While men do
not generally purchase female fashions themselves
(except perhaps as gifts), male opinions of fashion attractiveness are rated as extremely important by many women
in this age-group (e.g., Reynolds and Wells 1977) and thus
are likely to exert an indirect influence on purchase
behavior.

Stimuli and Pretests


The stimuli were professionally prepared drawings of
fashion apparel, designed specifically for this study (see
appendix). Fashion is a product for which consumer aesthetic response is very important (Schmitt and Simonson
1997) and that can believably be presented in a wide variety of designs.
Initially, 42 dress designs were developed for pretesting. Both the shapes and fabrics were designed to vary on
visual complexity (having many heterogeneous elements,
irregular in arrangement) as defined by Berlyne (1970; see
also Albert and Witryol 1990). Thus, fashion shapes varied in visual complexity (some had many "turns" and
embellishments, while others were very simple) as did the
fabrics (some fabrics were very simple, such as a solid
shade; others had complex prints such as an irregular
arrangement of circles and rectangles). The average number of turns for the three simple fashion shapes was 10,
while the average number of turns for the three complex
fashion shapes was 60. The average number of design elements within the fabric of the three simple shapes was 0,
while the average number of design elements within the
fabric of the three complex shapes was 263. Thus, objectively, these stimuli fit the operationalization set forth by
Berlyne (1970).
Next, these 42 product designs were pretested to gauge
participants' perceptions of their complexity and to ensure
that the complexity manipulation was not confounded
with other constructs that might moderate the exposureliking relationship. For example, Zajonc (1968) had
reported that exposure effects may be moderated by stimulus novelty or unfamiliarity; thus, our participants were

Cox, Cox / CONSUMERS'PREFERENCES 123


asked to rate the novelty of the product designs. In addition, pretest participants reported their liking of the product designs.
The pretests were conducted among 400 undergraduate
students, using measures modeled after Cox and Cox
(1988). Initial perceived complexity was measured on two
7-point semantic differential scales anchored by complicated-simple and not complex-complex (1 = simple, 7 =
complex). The coefficient alpha for this scale was .85. Perceived novelty was measured on five 7-point semantic differential scales anchored by new-old, original-unoriginal,
unusual-common, familiar-novel, and typical-atypical.
The coefficient alpha for this scale was .81. Stimulus
likability was measured by three 7-point semantic differential scales anchored by bad-good, pleasant-unpleasant,
and likable-not likable. The coefficient alpha for this scale
was .89.
From these pretests involving 42 product designs, 6
product designs were found to meet our criteria. The mean
perceived complexity of the three simple designs (.~" =
2.60) was significantly lower than that of the three complex designs (.~"= 5.95, t = 10.2, p < .001), with all three
complex designs (~'4 = 6.18, ~'5 = 5.2, ~'6 ---- 5.6) being
rated substantially higher than all three simple designs
(~'1 = 1.78, ~'2 = 2.83, -~3 = 2.2). The mean perceived novelty of the simple designs (~" = 5.01) was not significantly
different from the mean of the complex designs (,~ = 4.87).
All designs were perceived as at least moderately novel,
with means ranging from 4.47 to 5.25 on the 7-point scale.
Finally, the mean initial liking of the simple designs (~" =
4.47) did not differ significantly from that of the complex
designs (.~ = 4.27).
Additional confounding checks were conducted with
47 undergraduate students. Each participant saw one of the
six product designs and assessed them on the following
characteristics:

1. Schema incongruity (see Mandler 1982) was


measured using scales adapted from Saliagas
(1984) and Meyers-Levy and Tybout (1989).
Participants were asked to indicate how "different" or how "typical" each dress design was
compared with the dress design most representative of that product category, on 9-point scales
ranging from 1 (not at all different) to 9 (very different) and 1 (very typical) to 9 (not at all typical). The coefficient alpha for this two-item averaged scale was .865. The results indicated that
both the simple (.~ = 6.85) and complex (~" =
6.95) product designs were perceived as moderately incongruous but were not significantly different on incongruity.
2. Meaningfulness was measured using 7-point
semantic differential scales anchored by not
meaningful-meaningful, not important-important,
and not significant-significant. The coefficient

alpha for this summed scale was .79. The perceived meaningfulness for the simple designs
(~" = 10.68) was not significantly different from
that of the complex designs (X = 10.95). Both
the simple and complex designs were perceived
to be moderately meaningful.
. Usefulness was measured using three semantic
differential scales anchored by not useful-useful,
not functional-functional, and not practicalpractical. This summed scale had a coefficient
alpha of .78. The perceived usefulness of the
simple designs (X = 12.04) did not differ significantly from that of the complex designs (X =
12.59).
After pretesting, the six stimulus product designs were
placed in the context of advertisements for a fictitious retail store. The finished ads, like many fashion ads, were
very simple. They contained minimal copy (the headline,
"Spring at Goldstucker's," and store location information)
and no product-related copy. Aside from the variations in
dress design, all of the experimental ads were identical on
all other factors: store name; type; letter size; as well as the
model, her accessories, her facial expression, and so on.
Each participant received a booklet containing 15
pages of ads, 1 ad per page. There were 18 versions of this
booklet, each containing either zero, one, or three exposures to the target ad, depicting one of the six product
designs. The exposures to the target ads were interspersed
between "filler" black-and-white advertisements, half of
which were for well-known brands, half for unknown
brands. To avoid drawing undue attention to the repetition
of the target product, the booklet also contained multiple
ads for three other brands.

PROCEDURE
The booklets were randomly distributed to volunteer
participants within three large classes. Participants were
given the following instructions:
Our study concerns thoughts and feelings that people have about advertisements. In a moment, you
will be given about 20 seconds to view each of 15
proposed advertisements. The examiner will let you
know when the 20-second time limit is over. When
he or she says "Next," please turn the page to view
the next ad. Do not look back at any ad. When this
task is completed, we will ask you some questions.
We will be asking different people different questions due to the limited time period. Please answer
each question as honestly as you can. You will not
place your name on the questionnaire, so no one will
know who you are.
Participants were given 20 seconds to view each ad in
the booklet. (A pretest had indicated that most partici-

124

JOURNALOF THE ACADEMYOF MARKETINGSCIENCE

pants' attention seemed to wander after about a 20-second


exposure to these ads.) Participants were instructed not to
look back at any previous advertisement. After this task,
they were asked to write all of the brand, store, or manufacturer names that they could recall. Next, they were presented with a list of product categories and asked to place
checks by those depicted in the preceding ads. Both of
these tasks were designed to disguise the true purpose of
the study. Next, participants viewed a picture of the target
product design (to which they had received zero, one, or
three prior exposures) and were asked to rate its visual appeal (their liking of the design and their rating of the design
on measures of novelty and complexity). They viewed another product and were asked similar questions about this
product. Next, participants completed several demographic items.
Finally, participants were asked what they thought the
purpose of the study was. The responses to this item suggested that our efforts to disguise the purpose of the study
were successful. Most participants guessed that we were
interested in how different kinds of people respond to different kinds of advertisements; none came remotely close
to guessing the true hypotheses of the study.

Measures
Consumer evaluations of the product designs' visual
appeal (which for brevity, we will hereafter refer to as
product liking) were measured by averaging six 7-point
scales, anchored by bad-good, pleasant-unpleasant, likablenot likable, flattering-unflattering, unattractive-attractive,
and stylish-not stylish. The first three adjective pairs were
taken from Cox and Cox (1988), and the last three were
designed to be specifically relevant to fashion. The polarity of some adjectives was reversed to avoid acquiescence
bias and then recoded before being combined into a mean
score. This scale had a coefficient alpha of.93. To confirm
the scale's unidimensionality, we performed a principal
components analysis. This analysis extracted a single
overall factor, which explained 72 percent of the total variance of the scale items.
Perceived complexity was measured on two 7-point semantic differential scales anchored by complicated-simple
and not complex-complex (1 -- simple, 7 = complex). The
coefficient alpha for this scale was .85.

RESULTS
Hypothesis 1 predicted an inverted-U relationship
between participants' perceived complexity ratings of the
product designs and their liking of them. To test this
hypothesis, nonlinear effects of perceived complexity on
product liking were examined using polynomial regression analysis. Each potential predictor was entered in a

SPRING2002
hierarchical fashion (the linear function of perceived complexity, then the quadratic, then the cubic) and then
retained only if it made a statistically significant marginal
contribution to the explained variance of the regression
equation. In this analysis, the linear regression coefficient
was statistically significant and positive (B = .9058, t =
4.58, p < .001), and the quadratic coefficient was statistically significant and negative (B = -. 1176, t -- -4.794, p <
.001). Taking the first derivative of this quadratic function
and setting it to zero, we can find the peak of the invertedU curve, or the most preferred level of perceived complexity. Peak product liking occurs at a perceived complexity
level of 3.85, very near to the midpoint (4) of the 7-point
complexity scale. Thus, Hypothesis 1 is supported by the
data.
Hypotheses 2 predicted that participants' perceived
level of complexity for the product designs would decrease
over repetitions for both the simple and complex designs.
Hypothesis 3 predicted that perceptions of high-complexity designs would move toward the optimal level, while
perceptions of low-complexity designs would move away
from the optimal level.
Participants' ratings of the fashion stimuli were analyzed in a 2 x 3 ANOVA, with product design complexity
(2 levels) and exposure (3 levels) as manipulated factors.
The error term in this ANOVA represents the variance
across participants' scores within cells of the experiment.
Since (as noted earlier) we employed multiple stimuli in
each cell, this error term can be viewed as pooling two
sources of variance: variance across stimuli within a cell
and variance across participants exposed to a given stimulus. To examine the appropriateness of this pooling of variance, we conducted the pooling test recommended by
Keppel (1982:264). This test revealed that there was no
significant variance across stimulus means within each
cell. Thus, according to Keppel, the pooled analysis is
appropriate.
To ensure that effects of the experimental variables
were consistent across age, gender, and student classification groups, each of these background variables was
entered as a blocking factor in initial ANOVAs, along with
product complexity and repetition. None of these variables
significantly interacted with either of the manipulated factors of complexity or exposure, indicating that the impact
of the experimental variables appeared to be consistent
across demographic groups. Therefore, these demographic variables were dropped from all subsequent
analyses.
As predicted in Hypothesis 2, ANOVA revealed a significant main effect of repetition on perceived complexity
(F2,374 = 3.324,p < .05,112 = .017). There was no significant
interaction between repetition and stimulus type (complex
vs. simple). Figure 1 shows the mean levels of perceived
complexity at each level of repetition for both the simple
and complex designs. A post hoc Scheff6 test revealed that

Cox, Cox / CONSUMERS'PREFERENCES 125

FIGURE 1
The Effect of Repetition on
Perceived Complexity of Simple
and Complex Product Designs

FIGURE 2
The Effect of Repeated Exposure on Consumer
Liking of Complex and Simple Designs
50

5.5

4.5

Complex
Design

Opttmal level 0 85)

Perceived
Complexity

Complex
Design

4.0
CoI1sLlmcr

Liking
Simple
Design

3,0

3.5
Simple
Design
25
2

Total Exposures to Product Design

Total Exposures to Product Design

the overall mean for perceived complexity at one repetition


(X = 4.42) was significantly different from the means at
two and three repetitions (~" = 3.97, p < .05 and ,g = 4.01,
p < .05, respectively). However, the mean levels of perceived complexity at two and three repetitions were not
significantly different from each other.
The data in Figure 1 indicate not only that there is a consistent decrease in perceived complexity for both the simple and complex designs with repeated measures but that
the perceived levels of complexity for the complex designs
tend to move toward the optimal level of perceived complexity (3.85), while the perceived complexity ratings of
the simple designs tend to move a w a y from the optimal
level, as predicted in Hypothesis 3. Thus, both Hypotheses
2 and 3 were supported by the data.
Finally, liking of the fashion stimuli was analyzed in a 2
x 3 ANOVA, with product design complexity (two levels)
and repetition (three levels) as manipulated factors. Again,
a pooling test revealed no significant variance across stimulus means within each cell; therefore, the error term in
this ANOVA represents the variance across participants'
scores within cells of the experiment (Keppel 1982).
The findings regarding changes in consumers' liking of
complex and simple product designs are shown in Figure
2. Repeated exposure did not have a main effect on participants' product liking (Fz, 376= .69, p = .50), while product
complexity did have a main effect on participants' product
liking (F2. 376~'~ 3.96, p < .05, ,q2 = .010). As suggested in
Hypotheses 4a and 4b, there was a significant interaction
between exposure level and complexity (F z 376= 4.40, p <
.05, q2 = .023), with the complex stimuli showing an overall increase in liking across three exposures and the simple
stimuli exhibiting a slight drop.
After conducting the ANOVA, we examined possible
nonlinear effects of exposure on product liking, using
polynomial regression analysis (similar to that used by

Berlyne 1970). Each potential predictor was entered in a


hierarchical fashion (first, the linear function of repetition, then the quadratic, then the cubic) and then retained
only if it made a statistically significant marginal contribution to the explained variance of the regression equation. For both the simple and complex designs, only the
linear function was significant (simple, F1, ~85= 3.98, p <
.05; complex, Fl, 194= 4.78, p < .05), which indicates only a
linear trend.

Tests of Mediation
To help interpret our findings, we examined whether
the effects of repetition were mediated by changes in the
participants' perception of stimulus complexity. The analysis followed James and Brett's (1984) logic for tests of
mediation. If the effects of an antecedent (A) on an outcome (O) are completely mediated by a third variable (M),
this should produce a certain pattern of correlations. Specifically, the three simple correlations (rao, rAM, and rMo)
should be statistically significant, but the correlation
between the antecedent and the outcome should become
nonsignificant when the mediator is controlled (i.e., rao M=
ns). Since the relationship between perceived complexity
and liking is curvilinear, each participant's perceived complexity score was converted into a distance score by taking
the absolute value of the difference between each participant's perceived complexity score and the optimal level of
perceived complexity (3.85)9 This distance score for perceived complexity was used in the test of mediation.
The mediation analysis suggests that the impact of repetition on liking for complex designs was completely
mediated by perceptions of stimulus complexity. Consistent with James and Brett's criteria, all three simple
correlations were statistically significant (rrepe m . . . . l,kmg =
9155, p <. 05; rrepetmon ' p. . . . . . plex,ty = --" 177, p <.05; rp~..... ple~,ty, hkmg = - - . 2 0 8 , p < .01), while the impact of repetition on

126

JOURNALOF THE ACADEMYOF MARKETINGSCIENCE

product liking became nonsignificant when perceived


complexity was controlled (rrep~t,uo,.h k i n g p . . . . . . . plexity= ' 123,
ns). In contrast, the relationship between perceived complexity and product liking remained significant even after
controlling for repetition (rr~. . . . plex,ty, hklng.repetltlon = - - " 186, p
< .01). These findings suggest that repetition affects the
liking of complex designs by changing participants' perceptions of their complexity.
The mediation analysis for the simple product designs
is not as straightforward. For the simple product designs,
the correlation between repetition and liking (r = -. 145, p
< .05) and perceived complexity and liking (r = -.209, p <
.01) was significant, but the correlation between repetition
and perceived complexity was not significant (r = .09).
However, the impact of repetition on liking of the simple
product designs became nonsignificant when perceived
complexity was controlled (rrepetttton ' hk,ng p. . . . . . plextty= -" 130,
ns). The relationship between perceived complexity and
product liking remained significant even after controlling
for repetition (r o...... p~x,ty,hkmgr~pet,t,on= --"199, p < .01). The
interpretation of these results for the simple product
designs is less conclusive, but there is still some evidence
that perceived complexity plays a role in mediating the
exposure-liking relationship.

DISCUSSION
As noted at the outset, consumers' aesthetic reactions
are an increasingly important factor in competition among
products. However, marketers who attempt to predict
these reactions through premarket concept tests may
obtain misleading results, suggesting a divergence
between consumers' initial and subsequent reactions to
new aesthetic designs. This article has proposed and tested
a partial explanation for this discrepancy: consumers' liking of new product designs changes with repeated exposure, and these changes vary systematically according to
the visual complexity of the design. Our findings suggest
that consumers' preferences for visually complex product
designs are more prone to increase with repeated exposures than preferences for visually simple product designs.
In the remaining paragraphs, we will discuss some limitations and potential implications of our findings both for
future research on consumers' aesthetic product judgments and for manufacturers and retailers of aesthetically
oriented consumer goods.

Limitations
Our study has several limitations that both suggest caution in generalizing our results and suggest potential avenues for future research. First, we examine only one product class, fashion apparel. Apparel was used because of the
predominant importance of aesthetic design in consumer

SPRING2002
reactions to this product class and because it allows a wide
range of believable aesthetic changes without creating
obvious functional problems. However, future research
should examine the extent to which our findings can be
replicated for other product classes, particularly those
(e.g., automobiles) in which aesthetic design is more constrained by functional considerations.
While this study attempts to inject some realism into
visual preference testing (by incorporating the repeated
product exposure that typically occurs in the marketplace),
there are several additional elements of marketplace reality that future research should attempt to introduce into
such tests. It should be noted that in our experiment, both
the duration of each product exposure and the interval
between exposures were considerably shorter than is often
the case in the marketplace. Bornstein (1989) found that
naturalistic studies (which tend to employ longer intervals
between exposures) tend to produce stronger exposure
effects than laboratory studies. His review also indicated
that increasing the duration of each exposure (beyond the
level employed in our study) is likely to have little impact
on the observed effects. Again, however, the impact of
both of these factors needs to be examined directly in a
new product context. In this study, we also used student
participants. While they are very familiar with clothes and
fashion, the robustness of these findings should be examined with consumers of all age ranges. Finally, our
research examined the effect of only a limited number of
exposures; future research should examine the effects of
higher levels of repetition on consumer liking.
While acknowledging these limits in our scope, we feel
that our findings, if confirmed in subsequent studies, have
several potential implications.

Implications for Research on


Preference and Product Exposure
Most discussions of exposure effects in the marketing
literature (e.g., Janiszewski 1988, 1993; Obermiller 1985)
tend to view any affective impact of initial stimulus exposure to be positive and have focused on the level of consumer awareness required to achieve these positive effects.
Our findings suggest that exposure effects can vary markedly depending on a product's visual design and that for
some designs the effects of initial exposure can actually be
negative. This suggests that future research should pay
greater attention to the moderating role of a product's
design in the exposure-preference relationship, as well as
other contingencies that may alter this relationship.
In general, our findings are consistent with the "uncertainty reduction" theory of exposure effects (see, e.g.,
Bornstein 1989): liking of complex products tends to
increase with repeated exposure, while liking of the simple
stimuli declined with repeated exposure.

Cox,Cox / CONSUMERS'PREFERENCES I27


Our research also provides some support for Berlyne's
(1970) hypotheses concerning the relationship between
perceived complexity and aesthetic preference. Berlyne
posited that human aesthetic preferences are a function of
a stimulus's "arousal potential" (which was largely determined by its visual complexity) and that aesthetic preference is related to a stimulus's arousal potential in an
inverted-U fashion. Our data confirm this relationship:
moderate levels of product design complexity were most
well liked by consumers. However, perceived complexity
changes with repeated exposure. Upon repeated exposure,
the perceived complexity ratings of product designs
decrease. Our data show that this phenomenon occurs with
both complex and simple designs. With repeated exposure,
the perceived complexity of the objectively complex
designs decreases toward the optimal point, while the perceived complexity of the simple designs decreases away
from the optimal point. The result is an increase in liking
across repeated exposures of the complex stimuli and a
decrease in liking across repeated exposures for the simple
designs. Furthermore, a test of mediation showed that for
complex stimuli, the effects of repetition on liking were
totally mediated by participants' perception of stimulus
complexity. There also was some evidence that perceived
complexity played a similar mediating role for simple
designs; however, the evidence here was less clear-cut.
Future research should examine other potential mediators
of the exposure-liking relationship in consumer evaluations of products' aesthetic designs. For example, future
research might examine whether the relationship between
exposure and preference for some types of product designs
is mediated by changes in consumers' feelings of perceptual fluency while evaluating those designs (see, e.g.,
Reber, Winkielman, and Schwartz 1998).

Implications for Consumer Product Marketers


The differential impact of repeated exposure on the liking of complex and simple designs may have implications
for the way in which manufacturers gather premarket consumer reactions to new designs. As noted earlier, visual
concept evaluations are almost always collected after a
single exposure to the proposed design (R. Barber, personal communication, June 1992; P. Morgan, personal
communication, June 1992; Schwartz 1987). In the marketplace, however, consumers typically have multiple
opportunities to observe a newly designed product before
making a final evaluation. Our results suggest that as consumers receive these repeated exposures, they will come to
like some designs more, and others less, than their singleexposure evaluations would have predicted. Specifically,
single-exposure consumer pretests appear to systematically underestimate eventual consumer preferences for
visually complex product designs. If this finding is confirmed in future studies, it would seem to hold important

implications for managers who conduct and/or interpret


premarket tests of potential product designs. First, it
would suggest that managers interpreting conventionally
conducted (i.e., single-exposure) concept tests should
view low initial preference scores for complex designs
with some skepticism, realizing that these scores are likely
to improve with repeated exposure. Second, our findings
suggest that it may be possible to improve the predictive
validity of concept testing by incorporating repeated exposures into concept testing designs. This possibility seems
worthy of future empirical investigation.
Since consumer evaluations of visually complex
designs seem to improve with exposure, it would be interesting to see if this is generally true of designs that are high
on what Berlyne (1970) called "arousal potential" (i.e.,
complexity, novelty, incongruity, etc.). Berlyne argued
that upon initial exposure, most people resist stimuli that
are highly novel, incongruous, or complex, but with
repeated exposure their evaluations of such stimuli
increases. Consistent with this notion, P. Morgan (personal
communication, June 1992) reported that many fashion
retailers feel that consumers' premarket evaluations
invariably favor familiar designs and (if relied on too
heavily) can lead to overly conservative merchandising
decisions. Similarly, Moulson and Sproles (2000)
reported that "the reason that clinical research can lead to
seriously flawed decisions is that the average consumer's
opinions are inherently conservative. Really innovative
themes generally earn poor likeability ratings . . . . The
more innovative a style is, the worse it fares" (p. 48).
Finally, Lorenz (1986) reported that until recently, Ford
Motor Company's (single-exposure) consumer tests of
proposed designs tended to "support only slight design
improvements to the previous model, and resistance to any
dramatic change" (p. 103). The result was a general conservatism in design and a somewhat stodgy image. Ford's
remedy to this bias was to ask consumers to first imagine
what cars should look like in 4 to 5 years and then evaluate
proposed new designs. Again, an alternative approach
might be employing multiple exposures in consumer pretests of new designs. This methodology might help separate consumers' initial bias against designs that are novel
or complex from resistance having a more enduring
source.
Repeated exposure should be incorporated not only
into commercial design testing but into academic product
design research as well. In a recent study, Veryzer and
Hutchinson (1998) concluded that consumers tend to prefer designs that are highly "unified" and highly
"prototypical" of the designs they are accustomed to seeing in a given product category. At first glance, these findings seem to suggest that product designers should avoid
novelty and adhere closely to traditional or familiar product designs. However, Veryzer and Hutchinson's study,
examining several product categories, incorporated only a

128

JOURNALOF THE ACADEMYOF MARKETINGSCIENCE

single consumer exposure to these product designs. Might


their participants have responded differently to "typical"
and "atypical" product designs had they experienced the
repeated exposures that t y p i c a l l y occur in the

SPRING2002
marketplace? Future research needs to examine how consumer responses to a variety of product design elements
may change with repeated exposure. In the absence of such

APPENDIX
Pictures of Stimuli
The following pictures show all six styles that were presented in the target advertisements/conceptpresentation.

FIGURE A1
Simple Stimuli

\
i

//

q
Style 1

Style 2

SIMPLE STIMULI

Style 3

Cox, Cox / CONSUMERS' PREFERENCES

129

FIGURE A2
Complex Stimuli

Style 5

Style 4

Style 6

COMPLEX STIMULI

research, companies may continue to be biased toward


overly conservative approaches to product design.

ACKNOWLEDGMENTS
The authors thank John O. Summers and Thomas E
Hustad for their helpful comments on an earlier draft of
this manuscript and Jackie Thompson for her help in the
manuscript preparation.

REFERENCES
Albert, Eric and Sam Witryol. 1990. "Children's Preference for Complexity as a Function of Perceived Units in Collative Motivation"
Journal of Genetic Psychology 151 (1): 91-101.
Berlyne, D. E. 1970. "Novelty, Complexity and Hedonic Value." Perception and Psychophysics 8 (November): 279-286.
--and
G. H. Lawrence. 1964. "Effects of Complexity and Incongruity Variables on GSR, Investigatory Behavior and Verbally Expressed
Preference." Journal of General Psychology 71:21-45.
Bloch, E 1995. "Seeking the Ideal Form: Product Design and Consumer
Response." Journal of Marketing 59 (3): 16-30.

Bornstein, R. E 1989. "Exposure and Affect: Overview and Meta-Analysis of Research, 1968-198%" Psychological Bulletin 106 (2): 265289.
--and
E R. D'Agostino. 1992. "Stimulus Recogmtion and the Mere
Exposure Effect." Journal of Personality and Social Psychology 63
(4): 545-552.
Cantor, G. N. 1968. "Like-Dislike Ratings of Familiarized and
Nonfamiliarized Visual Stimuli." Journal of Experimental Child
Psychology 6 (December): 651-657.
Cook, T. and D. Campbell. 1979. Quasi-Experimentation: Design and
Analysis Issuesfor Field Settings. Chicago: Rand McNally.
Cox, D. and A. Cox. 1988. "What Does Familiarity Breed? Complexity
as a Moderator of Repetition Effects in Advertisement Evaluation."
Journal of Consumer Research 15:111-116.
Falk, Ruma and Clifford Konold. 1997. "Making Sense of Randomness:
Implicit Encoding as a Basis for Judgment." Psychological Review
104 (2): 301-319.
Hekkert, P. and E Von Wieringen. 1990. "Complexity and Prototypicality
as Determinants of Appraisal of Cubist Paintings." BritishJournal of
Psychology 8 l:483-495.
Hochberg, J. 1998. Perception and Cognition at Centurv's End. San
Diego, CA: Academic Press.
James, Lawrence and Jeanne Brett. 1984. "Mediators, Moderators, and
Tests of Mediation,' JournalofApplied Psychology69 (2): 307-321.
Janiszewski, Chris. 1988. "Preconscious Processing Effects: The Independence of Attitude Formation and Conscious Thought." Journalof
Consumer Research 15 (September): 199-209.

130

JOURNAL OF THE ACADEMY OF MARKETING SCIENCE

SPRING 2002

. 1993. "Preattentive Mere Exposure Effects." Journal of Consumer Research 20 (3): 376-393.
Keppel, Geoffrey. 1982. Design and Analysis: A Researcher's Handbook. Englewood Cliffs, NJ: Prentice Hall.
Kotler, P. and G. A. Rath. 1984. "Design: A Powerful But Neglected Strategic Tool." Journal of Business Strategy 5 (Fall): 16-21.
Lorenz, C. 1986. The Design Dimension: Product Strategy and the Challenge of Global Marketing. Oxford, UK: Basil Blackwell.
Mandler, George. 1982. "The Structure of Value: Accounting for Taste"
In Affect and Cognition: The Seventeenth Annual Carnegie Symposium. Eds. Margaret Clark and Susan Fiske. Hlllsdale, NJ: Lawrence
Erlbaum, 3-36.
Meyers-Levy, Joan and Alice Tybout. 1989. "Schema Congruity as a Basis for Product Evaluation." Journal of Consumer Research 16
(June): 39-54.
Moulson, T. and G. Sproles. 2000. "Styling Strategy." Business Horizons
43 (5): 45-52.
Obermiller, C. 1985. "Varieties of Mere Exposure: The Effects of Processing Style and Repetition on Affective Response." Journal of
Consumer Research 12 (June): 17-30.
Page, A. and H. Rosenbaum. 1987. "Redesigning Product Lines With
Conjoint Analysis: How Sunbeam Does It." Journal of Product Innovation Management 4:120-137.
Peters, Tom. 1992. Liberation Management. New York: Knopf.
Reber, Rolf, Piotr Winkielman, and Norbert Schwartz. 1998. "Effects of
Perceptual Fluency on Affectwe Judgments." Psychological Science
45 (January): 45-48.
Reynolds, Fred and William Wells. 1977. Consumer Behavior. New
York: McGraw-Hill.
Robinson, D. 1975. "Style Changes: Cyclical, Inexorable and Foreseeable." Harvard Business Review 53 (November-December): 121131.
Saegert, S. and J. Jelhson. 1970. "Effects of Initial Level of Response
Competition and Frequency of Exposure on Liking and Exploratory
Behavior." Journal of Personality and Social Psychology 16 (November): 553-558.
Sahagas, Dena. 1984. "An Investigation of the Affective Response: Applications to Advertising Communications?' Ph.D. Dissertation. University of Houston, TX.
Schmin, B. and A. Simonson. 1997. Marketing Aesthetics: The Strategic
Management of Brands, ldenti~ and Image. New York: Free Press.

Schwartz, David. 1987. Concept Testing: How to Test New Product Ideas
Before You Go to Market. New York: American Management Association.
Smith, G. and D. Dorfman. 1975. "The Effect of Stimulus Uncertainty on
the Relationship Between Frequency of Exposure and Liking." Journal of Personality and Social Psychology 31 (January): 150-155.
Stang, D. and E. O'Connell. 1974. "The Computer as Experimenter in
Social Psychology Research." Behavior Research Methods and Instrumentation 6:223-231.
Thau, Barbara. 2000. "Target's Hip Mix Hits Bull's Eye, Revolutionizes
Mass Retailing?" HFN The Weekly Newspaper for the Home Furnishing Network, May 29, p. 40s.
Veryzer, R. W. and J. W. Hutchinson. 1998. "The Influence of Unity and
Prototypicality on Aesthetic Responses to New Product Designs."
Journal of Consumer Research 24 (4): 374-395.
Zajonc, R. B. 1968. "Attitudinal Effects of Mere Exposure." Journal of
Personality and Social Psychology Monograph Supplement 9 (2, Pt.
2): 1-28.

ABOUT THE AUTHORS


D e n a Cox is an associate professor of marketing at the Kelley
School of Business at Indiana University, Indianapolis, Indiana.
She received her Ph.D. from the University of Houston. She publishes research primarily on aspects of consumer behavior and
promotion effects and marketing research. She has published her
research in the Journal of Marketing, the Journal of Consumer
Research, and the Journal of Retailing.

Anthony D. Cox is an associate professor of marketing at the


Kelley School of Business at Indiana University, Indianapolis,
Indiana. He received his Ph.D. from Indiana University. His research focus is on forecasting demand and consumer behavior
and advertising effects. He has published in the Journal of Marketing, the Journal of Marketing Research, and the Journal of
Consumer Research.

Вам также может понравиться