Effects of Store Atmosphere On Shopping Behavior
Effects of Store Atmosphere On Shopping Behavior
Effects of Store Atmosphere On Shopping Behavior
1990
Recommended Citation
Billings '90, Wendy L., "Effects of Store Atmosphere on Shopping Behavior" (1990). Honors Projects. Paper 16. http://digitalcommons.iwu.edu/busadmin_honproj/16
This Article is brought to you for free and open access by the Business Administration at Digital Commons @ IWU. It has been accepted for inclusion in Honors Projects by an authorized administrator of Digital Commons @ IWU. For more information, please contact sdaviska@iwu.edu. Copyright is owned by the author of this document.
by
Wendy L. Billings
Department of Business Administration
..
ACKNOWLEDGMENTS
I would like to express gratitude to Illinois Wesleyan University for providing the opportunity to undertake this study. Special gratitude is expressed to Dr. Mona J. Gardner for her guidance and encouragement during the course of this work. Special thanks are expressed to Dr. Robert Leekley for his assistance with the statistical analysis of this study. Appreciation is also expressed to Dr. Fred Hoyt and Dr. Jim sikora for their suggestions and support during the
CONTENTS
Page
Introduction . . . . . . . . . . . . . . . . . . . . . .
.......... ...
12
15
.. . . . .
19
19
19
19
20
26
26
33
INTRODUCTION
It has long been recognized that consumers respond to more than just the core product or service being offered when making purchase decisions; they respond to the total product. One of the most important features of the total product can be the place where it is bought or consumed. In some
instances, the place, or to be more specific, the atmosphere of the place, is more influential than the product itself in the purchase decision (Kotler 1973). Although today there is
an increasing emphasis on store design, interior design, and overall environmental programming by retail merchandisers, many retailers still tend to underestimate the potential of using atmosphere as a marketing tool Narayana 1976). (Markin, Lillis, and
concerned with the tangible product, focusing their interest on practical and functional dimensions, while neglecting the aesthetic factor in purchase behavior. Interior designers, architects, and landscapers, however, have acknowledged the extensive influence of the environment on behavior for years. Recently, psychologists have studied
environment-behavior relationships, resulting in the swiftly growing psychological discipline known as "environmental This discipline
psychology"
attempts to predict the collective effect of stimuli in a particular environment upon different peoples'
1
feelings and
behavior
(Mehrabian
1976).
Thus,
the
main
concerns
in
environmental psychology may be summarized as "(1) the direct impact of physical stimuli on human emotions and (2) the
effect of the physical stimuli on a variety of behaviors, such as work performance or social Russell 1974, p. 4). until recently, environmental psychology has rarely been applied to the retail store environment. Previous studies interaction" (Mehrabian and
have, however, suggested using atmospherics as an important part of the overall 1976). merchandising strategy (Kotler 1973;
Markin, et ale
probability"
Markin,
Lillis,
and
acknowledge that space affects customer behavior and that design and atmosphere may be used to shape and modify the behavior of shoppers. However, these studies generally consider the atmosphere to be a component of store "image." Therefore, atmosphere is
viewed simply as being one factor influencing store patronage decisions. For instance, Kotler 1973 suggests using
atmospherics as a competitive tool in an attempt to attract and maintain a specific target market, especially where
Also, Markin,
Via design features, attitudes and images are created; that is, store personalities
are created and shaped, and these personalities friendly, upper-class, aloof, high quality, low priced, convenient, warm, inviting, cool, haughty, etc. - are in turn meant to affect customer attitudes and images and hence to shape behavior these attitudes and images affect questions of store choice and store loyalty (p. 51). There effects of is little sound documentation on shopping for the actual Some
store atmosphere
behavior.
retailers have claimed that they have influenced customers' buying behavior by manipulating store atmosphere via layout, color, lighting, and music (wysocki 1979; Stevens 1980).
Researchers have
been unable to document strong effects of store atmosphere for a variety of reasons. First, the effects evoked by store
atmosphere are primarily emotional states that are difficult to verbalize. These emotions are temporary and therefore In addition, they influence rather than more easily to
identifiable
behaviors
selecting
which
store
studies have used structured questionnaire surveys which ask respondents to rate various researcher-specified attributes according to their importance for patronage. However, this
method clearly does not capture the consumer's true emotional responses to the store's atmosphere; it simply lists
atmosphere as one component of store image. In addition, the majority of previous store-atmosphere
measurement, which was usually done in the context of store image research, has been conducted outside of the store This for
environment, long after the actual shopping experience. method is not very reliable, since it is difficult
respondents to recall accurately their emotional responses to a particular atmosphere while in a different setting. Thus, if store atmosphere can actually affect shopping behavior within the store, it is necessary to develop a
attempt to apply the Mehrabian-Russell model, an environmental psychology framework, to explore environmental variables in retail settings.
Environmental psychology focuses on two maj or topics: (1) the emotional impact of physical stimuli and of physical stimuli on a variety of (2) the
effect
behaviors
appear to have valuable applications to store environments. The theoretical model developed in work by environmental
psychologists Mehrabian and Russell 1974; Mehrabian 1980; and Russell and Pratt 1980, appears to be particularly valuable in studying the effects of store atmosphere on shopping
behavior. "centers
As defined by Mehrabian and Russell, the approach on the use of human emotional responses to
..
environments as intervening variables linking the environment to the variety of behaviors it elicits" (p. xi). The purpose
of this paper is to adapt the Mehrabian-Russell model to the retail setting and to test predictions from this model.
Figure 1
The Mehrabian-Russell Model
Environmental stimuli
Emotional states
Behavioral Responses
The Mehrabian-Russell approach uses a Stimulus-Organism Response model (See Figure 1). Thus, it requires a stimulus
taxonomy, a set of intervening variables, and a taxonomy of responses. There should be a clear relationship between the
stimuli and responses by way of the intervening variables. Environmental psychologists assume that individuals'
feelings and emotions ultimately determine their behavior. They also assume that environments can evoke various feelings which cause certain behaviors. Mehrabian (1976) states that
--.
"human
emotions
are
amenable and
to
precise statistical
description, analysis.
quantitative
measurement,
Environmental psychologists working under this assumption have provided a sound descriptive framework for emotions .
(which) forms one of the crucial elements of the system that has been developed in order to evaluate whole environments and people's reactions to them" (p.9). that a particular in an environment This framework suggests certain turn, emotional cause the
causes in
responses
individual,
which,
individual to approach or avoid the environment to a greater or lesser degree (Mehrabian 1976). The Mehrabian-Russell model clearly defines the mediating variables and the response taxonomy. However, the selection
of useful environmental descriptors, or stimulus variables, is very difficult. Previous environmental studies use a
number of alternative sets of stimulus categories (Craik 1970; Ittelson, Rivlin, and Proshansky 1970). The most common
procedure is to describe an environment in terms of various obj ects in it and the relations among these obj ects. example, a park may be described as a lake with For
trees,
descriptors could continue forever; therefore, one list does not form a complete description of the setting. the items are too vaguely defined. analyze behavioral changes In addition,
resulting
1974). Another possible set of environment descriptors is the emotional reactions to variables which stimulate the senses, such as those for color, sound, temperature, and texture
(Crane and Levy 1962; Schaie 1961; Bedford 1961; Middleton, Fay, Kerr, and Amft 1944). is also quite long and However, this list of descriptors awkward since most environments
simultaneously include stimulation in all the sense modes, as well as along numerous stimulus dimensions within each
modality (for instance, a color may be a certain hue and of a certain brightness). These descriptors also vary in time
a park may be bright and sunny one moment and cloudy the next. To account for the overall effect of the various stimuli in a given environment, theory. Mehrabian and Russell apply the concept (1974) of use
information
They
average
information rate to characterize complex spatial and temporal arrangements of stimuli within and across settings. concept will be more fUlly discussed later. This
of establishing the validity of the link between the set of mediating addressed. variables and the response taxonomy will be
Response Taxonomy
Mehrabian and Russell propose that individuals' reactions to all environments may be categorized as either approach or
which
include
the environment as opposed to a tendency to remain inanimate in (avoid) the environment; (3) a desire to communicate with (approach) others in the environment versus a tendency to
avoid interacting with others; (4) enhancement (approach) of performance and satisfaction of task performances or hindrance (avoidance) of task performances. Donovan and Ross iter (1982)
propose that these aspects may easily be applied to shopping behaviors in a retail environment, as shown in Figure 2. Figure 2
Approach and Avoidance Responses in a Retail Environment
Approach Behavior Patronize store Browse through merchandise Interact with sales personnel Repeat shopping in store frequently
Avoidance
Behavior
Avoid store Look at minimum number of items Avoid inter action with personnel Do not return to store
communication
Intervening Variables
An adequate model requires intervening variables that are basic, immediate, and measurable reactions to all types of In addition, these variables must
environmental stimulation.
relate directly to the stimulus variables and also account for variations in other behaviors of concern (Mehrabian and
Mehrabian and Russell assert that three basic act as and mediating variables between
states
environmental
stimuli
approach-avoidance and
behaviors: dominance
arousal-disarousal,
individual may be described as a combination of these three dimensions. which an Pleasure-displeasure refers to the degree to individual feels happy, joyful, to contented, one's level or of
satisfied.
Arousal-nonarousal
refers
unrestricted and in control of the situation. Mehrabian and Russell present ample supporting evidence for the selection of three emotional states as the three intervening variables. individuals' They state that in order to understand it is
essential to identify those responses that are the immediate result of stimulation and that occur in varying degrees in all
10
environments.
an exact
do not yield a complete list since it is necessary to consider numerous dimensions of response within each sense modality. Thus, in an attempt to identify responses common to all types of stimuli, regardless of the sense modality stimulated,
Mehrabian and Russell turn to the study of intermodality. Resul ts of intermodality studies show that emotional reactions represent the common core of human response to all types of environments (Mehrabian and Russell 1974). Mehrabian and Russell propose that pleasure, arousal, and dominance are the three basic emotional environments. demonstrated reactions to all
They point to physiological studies that have that there is a well-defined physiological
mechanism associated with the experience of pleasure-pain. Electrical stimulation of areas of the hypothalamus sensation, and and
causes a
pleasant
stimulation of lower parts of the midline system causes pain (Heath 1954; 1963; 1964a; 1964b; Olds 1956). is common to the all exact the sensory of modalities. the This mechanism Furthermore, arousal
although
nature
physiological
response is unknown, support is given to the notion of basic cross-modal i ty responding (Lacey 1967) . Thus, the
physiological mechanisms support the idea that pleasure and arousal are two dimensions which cut across sense modality distinctions. Therefore, they are responses common to all
11 types of stimuli. Further support is given by semantic differential studies (Osgood, Suci, and Tannenbaum 1957). These studies have shown
that human jUdgments of complex samples of stimuli can be characterized in terms of three dimensions: evaluation,
corresponds to the emotional response of pleasure; activity corresponds to arousal; and potency corresponds to an
and Russell, pleasure, arousal, and dominance constitute the common core of human emotional responses to all environmental stimuli. However, Russell and Pratt 1980 suggest that the
dominance dimension should be deleted from the Mehrabian Russell model. dominance Russell argues in his later work that since a knowledgeable interpretation by the
requires
individual, it is not purely applicable in situations calling for emotional responses. Russell and Pratt (1980) claim that
the two dimensions of pleasure and arousal are sufficient to represent individuals' situations. affective responses to all types of
They point out that evidence for the suitability on the other hand, is quite
and Russell's original tridimensional model and test to find out if, in fact, the dominance dimension is significant or
insignificant.
12
Although the three basic emotional states are orthogonal (that is, none causes the other and they are completely
independent),
that there is a conditional interaction between pleasure and arousal in determining approach-avoidance behaviors. In a
neutral environment (i.e., one that is neither pleasing nor displeasing), mild arousal tends to enhance approach
behaviors, while extremely high or extremely low arousal tends to cause avoidance behaviors. higher the level of In a pleasant environment, the the greater the approach
arousal,
behavior;
in an unpleasant environment,
and Russell argue that these emotional dimensions do interact under these conditions, as shown in Figure 3 on the following page.
stimulus factors is extremely difficult because of the complex and changing combinations of stimuli encountered in any
environmental setting.
using in-store experimentation to learn which particular types of in-store stimulus variables (such as store layouts, color arrangements, 1 ighting, noise levels, and in-store promotions) cause which types of emotional responses, thereby resulting in approach or avoidance behaviors (Donovan and Rossiter
13
PLEASANT
APPROACH
NEUTRAL
AVOIDANCE
Low
Moderate AROUSAL
High
this descriptive system is the concept of "information rate," or the amount of information contained or perceived in the environment per unit of time (Mehrabian 1976). be described as the "load" This rate may The more
of an environment.
information in the form of stimuli that an observer must process, the higher the load of the environment. Mehrabian
and Russell (1974) refer to the load of any environment as a combination of its novelty and complexity. The novelty of an
environment has to do with the degree of unfamiliarity and uncertainty. Complexity refers to the number of elements, When the a reI iable
features, or changes in an environmental setting. factors of novelty and complexity are summed,
its load.
Mehrabian and Russell assume that environmental load is a direct correlate of the emotional response of arousal. higher the load, the higher a person's arousal The
level.
Therefore,
an environment that
is unfamiliar,
surprising,
crowded, and complex will cause a person to become stimUlated, excited, jittery, and alert. Conversely, an environment that
is common, probable, usual, and expected will cause feelings of relaxation, calmness, and sluggishness. However, a person's individual method of responding to external information must be taken into consideration when
15
measuring one's arousal response to the environmental load. Mehrabian (1976) suggests that one's degree of arousal may be related to the extent to which that person screens or filters incoming stimuli, referred only out to to less as "stimulus screening." They of the
selective important
effectively
reducing
load.
"Nonscreeners," on the other hand, are less selective in what they respond to. They tend to sense more stimuli in all
environments and are more sensitive to stimulus changes than are screeners. Therefore, nonscreeners experience settings
as being more complex and more loaded, resulting in higher arousal levels. stimulus screening can be used to draw
implications for the environmental preferences of individuals. Since nonscreeners become more aroused in high-load
environments,
screeners, they are more likely to approach high-load and pleasant places and to avoid high-load and unpleasant ones (Mehrabian 1976). Thus, Mehrabian and Russell's model
specifies how individual differences are expected to relate to the other variables in their system.
An
individual's
reaction
to
any
environment
may
be
16
categorized Approach
as
either
approach
or
avoidance moving
behavior. toward,
behaviors
include
physically
exploring, communicating, and performing in an environment, as well as returning to that environment. Avoidance behaviors
include a desire to leave, disinterest, lack of interaction, and poor performance in an environment, returning to that environment. as well as never
by an individual's emotional states evoked by the environment. Mehrabian and Russell's model posits that three basic
emotional states - pleasure, arousal, and dominance - form the palette from which all feelings are created. dimension is independent of the other two. and arousal are hypothesized to interact: Each emotional
arousal level, the greater the approach behavior in pleasant environments, and the greater the avoidance behavior iR
unpleasant settings. The concept of information rate, or load, is used as the stimulus taxonomy of an environment that arouses the various emotional states. The degree of arousal caused by an
environment may be directly correlated with the information load of the setting. An individual's characteristic way of
responding to external stimulation also modifies his or her arousal response to the environmental load. A given load
evokes less arousal in those who tend to screen out irrelevant stimulation than versa. in those who are nonscreeners, and vice
17
is summarized in
Russell model to the retail setting, it may be predicted that customers will spend more time and perhaps make more purchases in those retail atmospheres which evoke feelings of pleasure and a moderate to high degree of arousal. In this study, unlike previous studies, I will also
compare and contrast the intended shopping behavior of college students with various educational backgrounds. Results may
help predict whether majors have an effect on the degree of approach-avoidance setting. behavior students exhibit in a retail
more knowledgeable of marketing tactics cause them to behave any differently in a retail store than individuals with other maj ors might? The answer to such questions may be of
importance to retailers who wish to use atmospherics as a marketing tool when their products are aimed at distinct psychographic segments. I will also compare and contrast the intended shopping behavior of males and females. Results may suggest that one
gender is more highly affected by the store environment than the other. For example, it may be that retail atmospheres This would
suggest that females may exhibit more approach behavior in pleasant environments than would males. Therefore, it may be
18
Figure 4
Approach-avoidance (Which includes phys ical approach, exploration, affiliation, per formance, or other verbal and non verbal communica tions of preference)
Characteristic
emotions asso
ciated with
PERSONALITY
19 variables in departments frequented by women to enhance approach behavior of female shoppers.
METHOD
overview
The empirical portion of this study uses the Mehrabian Russell model states evoked to establish a in a retail relationship between emotional environment and statements of
Sample
Fifty-five Illinois Wesleyan University students served as SUbjects, including 28 with business majors and 27 with social science maj ors. The sample included 35 males and 20 femal,es. Each person in the
All subjects were between the ages of 19 and 22. was randomly assigned to two retail
environments
J.e.
To ensure various shopping times, respondents were instructed to visit each retail store on a different day and at a different time of day.
Procedure
Respondents
entered
each
store
and moved
to
central
20
location within the store; in multilevel stores, this was on the first floor. Then, while in the store, they completed the
Measures
The first
section evaluated sUbjects' emotional states while in the store environment, differential using Mehrabian of and Russell's state (1974) semantic
measures
emotional
(See
Appendix A).
Several of the original dominance scales (in control-cared for, autonomous-guided, important-awed) were replaced by more
context-appropriate items (restricted-free, crowded-overcrowded, important-insignificant) 1982. Mehrabian and Russell devised this self-report measure of the three emotional dimensions over the course of three separate studies. In the first study, based on intuitive grounds, they as suggested by Donovan and Rossiter
devised a tentative set of descriptors for the three emotional factors. most They proceeded directly to construct scales that would and uniquely as noted measure earlier, pleasure, their arousal, review of and the
directly
dominance
because,
applicable literature had shown that various combinations of these three factors may to and sUfficiently environments Russell also represent (Mehrabian wrote the and diverse Russell verbally
emotional 1974) .
reactions Mehrabian
forty
21
described
situations to provide a
wide variety
of physical
environments which would elicit many diverse emotional states. One hundred thirty-four sUbjects were given a random selection of eight situations and asked to describe how they would feel in each one by using the twenty-eight adjective pairs (which were the descriptors for the three emotional dimensions). The resulting correlation matrix was factor analyzed, These
factors were labeled pleasure, arousal, and dominance; the six highest loading items in each factor were then viewed as
adequate measures for that factor (Mehrabian and Russell 1974). A second study was performed again, there to improve three the emotional with and
Once
were
factors arousal,
exceeding one,
measuring pleasure,
A third study was executed to cross-validate the findings from the second study and determine the six best items for each emotional dimension. Thus, the semantic differential measures in this study all the three six of
of emotional state used (but slightly modified) (Appendix Mehrabian A) and are based on the results and from
Russell's
studies
include
best
descriptive adjective pairs for each of the three emotional factors. To compute factor scores for a respondent who rates
his or her emotions in a store environment, his or her responses to items loading highest on each factor are simply added. The second section of the questionnaire used in this study
Information Rate, which measured the environmental load factor (See Appendix B). The first step Mehrabian and Russell took in
developing this measure of information rate was to devise a set of adjective pairs which may be used to characterize
environments.
Examples of these adjectives are simple-complex, familiar-novel, and sparse-dense. with the
patterned-random,
use of these adjective pairs, Mehrabian and Russell (1974) were attempting (1) to provide preliminary validation and, (2) for their
to develop a
more adequate verbal scale of information rate by eliminating those items in the original set that were related more to
objective judgment than to emotions. In their study, to read Mehrabian a set of and six Russell situations asked from 214 those
undergraduates
Each sUbject
then characterized his emotional reaction to each situation, after which he or she rated the same situation on the adjective pairs used to measure information rate. were interested in determining were how Thus, the researchers emotional in states and
environmental situations.
descriptors
correlated
identical
items was factor analyzed, and a principal component solution was obtained. than one. There were five factors with eigenvalues greater These factors accounted for 60% of the total
variance.
23
evaluative jUdgments.
Thus,
bias, the respondent's emotional reactions of pleasure, arousal, and dominance to each situation were computed. Next, regression
equations were written to express each adjective pair in terms of these emotional reactions. By using the information rate-arousal hypothesis, this study validated Mehrabian and Russell's assertion that their set of adjective pairs formed measures of information rate. The
coefficients in the regression equations showed that arousal was a significant component in all but two cases. In addition, the
regression equations provided useful information for selecting a subset of the adjective pairs to be used as a verbal measure of information rate. The two factors that appeared to be
evaluatively biased in the beginning were eliminated, because the regression equations of showed that these that two factors affected
characterized
aspects
environments
mainly
pleasure rather than arousal. Thus, the final scale in Appendix B includes the remaining fourteen adjective pairs which may be used to describe an
are computed by reversing the signs of sUbjects' responses to the negatively signed items (which are those adjective pairs in which the adjective rate, referring to an environment novel, of higher was
information
e. g.,
complex,
random,
dense,
placed on the left) and then by using an algebraic sum over all items. Thus, the higher the sum, the higher the sUbject rated
24 the information load of the environment. The third section of the questionnaire used in this study measured subjects' Append ix C). intentions of behavior in the store (See
Mehrabian and Russell conducted three experiments to test their hypothesis that preference, exploration, work performance, and affiliation are intercorrelated aspects of response to a situation and can all be subsumed under the generic concept of approach-avoidance (These four dimensions were described and These studies showed that
all the behaviors that were assumed to be part of an approachavoidance reaction to As a situations result, were indeed significantly (1974)
intercorrelated.
concluded that the following verbal attempts to measure approach (+) - avoidance (-) were accurate (p. 221):
(+) 1. (-) 2.
How much time would you like to spend in this situation? How much would you try to leave or get out of this situation?
(+) 3.
Once in this situation, how much would you enjoy exploring around? How much would you try to avoid
(-) 4.
25
5.
To what extent is this situation a good opportunity to think out some difficult task you have been working on? How much would you dislike
having to work in this
situation? (0 = no dislike)
(-) 6.
7.
To what extent is this a situation in which you would feel friendly and talkative to a stranger who happens to be near you? Is this a situation in which you might try to avoid other people, avoid having to talk to them? (0 = no avoidance)
(-) 8.
In their study,
questions in this list in a random order, without the category name (e.g., "Desire to stay in the Situation ll ) . The of the
respondent answered each question by circl ing one al ternatives similar to those found in Appendix C. scores on each of the four dimensions,
To compute
responses to the negatively signed items are assigned minus signs. Next, the scores for each pair are summed. the resulting score, the greater the sUbj ect I s The higher approach
26
desire for affiliation was identified as a separate factor. In addition, there was evidence suggesting that desire to work should be treated as a separate dependent measure. Therefore,
Mehrabian and Russell suggest that the four approach-avoidance factors be analyzed separately when detailed information is needed on how the particular environment influences each of these factors.
The primary objective of this analysis is to determine the extent to which respondents' approach-avoidance responses may be predicted from their reported emotional states while in the various retail environments.
Factor Analysis Results Separate factor analyses (principal components, varimax rotation of factors with eigenvalues greater than or equal to 1.0) were carried out on the 18 emotional measures, responses, and the 14 the 8
approach-avoidance items.
information-rate
Results of these respective factor analyses appear in Reliability estimates (coefficient alpha) The
Tables 1, 2, and 3.
27 coefficient alpha is based on the internal consistency of a test. Thus, it is based on the average correlation of item It ranges from 0 to 1. The higher
alpha is, the more reliable the scale. Pleasure-Arousal-Dominance. The three factors resulting
from the 18 emotional measures are clearly identifiable as pleasure, arousal, and dominance (See Table 1 on the following page). Pleasure is by far the strongest factor. The relative
insignificance of the dominance dimension is consistent with Russell's Overall, more recent the findings (Russell and Pratt and 1980).
however,
pleasure,
arousal,
dominance
dimensions that were developed by Mehrabian and Russell in the laboratory situation retained their nature and factorial
independence in actual retail environments. For later analysis, I computed pleasure, arousal, and
dominance scores for each sUbject by computing the average of the six highest loading items on factor 1, the five highest loading items on factor 2, and the three highest loading items on factor 3 (Table 1). These scores were then designated as Reliability Pleasure
a respondent's Pleasure, Arousal, and Dominance. coefficients (alpha) are high for all measures:
28 Table 1
Factor Analysis Results for the 18 P1easure-Arousa1-Dominance Measures!
FACTOR LOADINGS 2 MEASURE Factor 1 (Pleasure) .74164 .86851 .87761 .85958 .72792 .67311 .64739 .44487 Factor 2 (Arousal) .32290 Factor 3 (Dominance)
Happy-Unhappy Pleased-Annoyed Satisfied-Unsatisfied Contented-Depressed Hopeful-Despairing Relaxed-Bored Important-Insignificant Free-Restricted stimulated-Relaxed Excited-Calm Jittery-Dull Interested-Disinterested Frenzied-Sluggish Overcrowded-Uncrowded Wideawake-Sleepy Controlling-Controlled Dominant-Submissive Influential-Influenced Variance
.36020 .69880 .81573 .75177 .41538 .70380 .34820 .77889 .81405 .84010 .68499
.75211 .32249
39.4%
>
10.0% 61. 7
Approach-Avoidance Responses.
eight dependent measures originally yielded two factors with eigenvalues greater than 1.0, which is inconsistent with the findings of Mehrabian and Russell (1974). Mehrabian and
29 ("friendly to strangers" and "avoid talking to people") were generally quite independent of the other approach-avoidance responses. However, in this study, one of these items loaded Thus, they
on factor 1, while the other loaded on factor 2. were not independent of the other responses.
In addition, the
second factor in this study had an eigenvalue of only 1.05. Therefore, I am using a unidimensional scale to represent (See Table 2 on the following
computed an approach-avoidance
score for each sUbject by computing the average of the five highest loading items on factor 1. The reliability
is acceptably high
an environment should be a unidimensional variable. Mehrabian and Russell their 14 measures. (common-rare, "complexity" (1974)
(continuous-intermittent,
heterogeneous, sYmmetrical-asYmmetrical, similar-contrasting, patterned-random, redundant-varied) ; and "spaciousness" (small scale-large scale, simple-complex, sparse-dense, uncrowded crowded, distant-immediate). Donovan and Rossiter (1982)
found five factors with eigenvalues greater than 1.0, which they labeled "novelty," "variety," "irregularity," "density,"
30
Table 2
MEASURE Do you like the environment? Would you enjoy shopping in this store? Would you avoid returning?
(reverse scoring) Would you feel friendly to a stranger? Would you avoid other people?
(reverse scoring) Would you spend more than you set out to? How much time would you spend browsing? Would you avoid exploring?
(reverse scoring) Variance
FACTOR LOADINGS
Factor 1
.90066
.92711
.81671
.38294
.55997
.68744
.82860
.86822
58.9%
31 and "size." The present study also yielded five factors with However, since the fifth factor 1.00033 and represented I no
forced my
labeled identically to those of Donovan and Rossiter with the omission of the "size" factor (See Table 3 on the following page) . For measures: later analysis, I defined of four information-rate ordinary
novelty
(average
novel-familiar,
(average of sparse-dense, intermittent-continuous, immediate distant) and irregularity (average of uncrowded-crowded, The reliability high. (-.12), that The and their as
unreliability irregularity
variety
(.63),
(.38)
measures,
however,
should be
regarded
32
Table 3
Factor Analysis Results for the Information-Rate Measures l
FACTOR LOADINGS 2 MEASURE Factor 1 (Novelty) .85649 .89695 .84178 .44341 .39069 .38088 Factor 2 (Variety) Factor 3 (Density) Factor 4 (Irreg. )
Usual-surprising Common-Rare Familiar-Novel Redundant-varied similar-Contrasting simple-Complex HomogeneousHeterogeneous Sparse-Dense continuous Intermittent Distant-Immediate Small scaleLarge scale Patterned-Random Uncrowded-Crowded SYmmetricalASYmmetrical Variance
.50302 .71123 .65508 .71026 .65427 .43682 -.69958 .69745 .44366 .75508 .54596 .54812
.31828
25.7%
13.9%
>
11.6% accounted
1.0
33
Table 4
Reliability Estimates (Coefficient Alpha) for Index Measures
Information-rate Measures Novelty .88 Variety .63 Density -.12 Irregularity .38
.91
Once the large set of candidate measures for Pleasure Arousal-Dominance and Approach-Avoidance had been reduced to single scores for each dimension for each respondent, the
is Approach-Avoidance behavior, P is Pleasure, A is Arousal, D is Dominance, Sex is the respondent's Sex, Major is the and E is a random error term.
Since the model deals with nonstandard units of measure (what is an increase of one unit of pleasure?), beta coefficients are used. Therefore, the constant drops out of the equation.
Pleasure-Arousal-Dominance and Approach-Avoidance. Table 5 (on the following page) shows the regression of the independent variables, including the pleasure, arousal, and dominance dimensions and the respondent's major and sex,
34 against behavior. the dependent variable, approach and avoidance
study, the respondent's sex is also a significant predictor. The subject's academic background or major, however, does not appear to have any affect on shopping behavior intentions. Likewise, dominance and arousal are insignificant predictors of behavior. The failure of the dominance dimension to
predict approach-avoidance intentions is consistent with the later findings of Russell. Thus, the sex of the sUbject and the perceived
pleasantness of the within-store environment are the major predictors in the equation, with the equation itself
accounting for a substantial 65 percent of the variation in intended approach-avoidance behaviors in the retail stores., Table 5
Beta Coefficients for the Dimensions Used to Predict the Approach-Avoidance Behavior
Dimension Pleasure
Arousal
Dominance
Sex
Major
Multiple R
R2 (adjusted)
35
These results suggest that arousal is not significantly related to approach-avoidance behavior. However, as stated
earlier, Mehrabian and Russell hypothesized that there is a conditional interaction between pleasure and avoidance. They
proposed that in a pleasant environment, the higher the level of arousal, the greater the approach behavior; in an
unpleasant environment, the greater the arousal, the greater the avoidance behavior. provided support for Donovan and Rossiter's study in 1982 this pleasure-arousal interaction
hypothesis.
I tested Mehrabian and Russell's hypothesis by computing two regressions: one for pleasant environments (pleasure scores
above zero) and another for unpleasant environments (pleasure scores below zero). According to the hypothesis, arousal
should have a high, positive coefficient (and therefore emerge as a significant predictor of approach intentions) in pleasant retail environments, and a negative coefficient in unpleasant environments. cases. In this study, arousal is insignificant in both
avoidance behavior of the respondents in the present sample. Of particular interest in this study is the finding that sex is a significant predictor of shopping behavior. To
determine which sex's approach-avoidance behaviors are most affected by variables: store environment, I created two interaction
Sex) + BdA
Sex)
+ E.
sex,
Thus, I computed a regression of the pleasure, arousal, and two interaction variables against the dependent
Since the interaction variable between insignificant, variable, I computed in a a second better
regression
without
this
resulting
Table 6
Beta Coefficients for the Pleasure, Arousal, Sex, and Sex-Pleasure Interaction Dimensions Used to Predict the Approach-Avoidance Behavior
The effect of the pleasure evoked by the retail environment on male and female shopping behavior is shown in Figure 5 (on the following page). Surprisingly, male college students
store environment increases, the rate of approach-avoidance behavior of the males increases at a faster rate than does that of the females. This finding may present interesting Perhaps they should place more
implications to retailers.
emphasis on certain environmental variables in the departments frequented by men to evoke a more pleasurable feeling,
Figure 5
Effect of Pleasure Evoked by the Retail Environment
on the Approach-Avoidance Behavior of Kales and Females
AA
Approach Behavior
Pleasure
In sum, the pleasure-arousal-dominance emotional states do correlate with approach-avoidance intentions in retail
38
avoidance
nor
suggested
Russell's more recent studies. In addition, this study found the sex of the respondent to be a major predictor of shopping behavior intentions. Information Rate, Pleasure-Arousal-Dominance, and
Approach-Avoidance.
regression of the dependent approach-avoidance behavior on the pleasure-arousal-dominance dimensions (and sex and maj or) with the addition of the information-rate measures. R and adjusted R2 are increased only The multiple from the
slightly
original regression which did not contain the information-rate measures. Thus, this study, like that of Donovan and Rossiter
(1982) suggests that the information rate is not particularly useful in helping to predict approach-avoidance behavior.
According to the Mehrabian-Russell model, arousal is a direct response to the information rate in the environment. However,
in this study, perceived information rate is not an accurate predictor of an individual's level of arousal. the following page) Table 8 (on
the regression of the information-rate measures on arousal. Only one information-rate measure (novelty) increases arousal, while the others (variety, density, and irregularity) have However, as noted
39 Table 7
Beta Coefficients for pleasure-Arousal-Dominance, Sex, Kajor, and Information-Rate Dimensions Used to Predict Approach-Avoidance Behavior
Dimension Pleasure
Arousal
Dominance
Novelty
Variety
Irregularity
Density
Sex
Major
MUltiple R
R2 (adjusted)
Estimated Coefficients .67228 .04180 .00893 .01883 .04834 .00021 .01148 .21263 .00021 .83114 .66297
statistics 7.3100& .5800 .1460 .2290 .7260 .0030 .1830 3.5920& .0041
Table 8
Beta Coefficients for Information-Rate Dimensions Used to Predict the Arousal Dimension
40 earlier, all of the information-rate measures but novelty are of questionable reliability. predictive potential. This fact may have reduced their
extent a function of the information-rate in the environment, it is not a direct response to the information rate as was implied by the Mehrabian-Russell model. Perceived information rate is helpful in predicting an individual's level of pleasure, however. The regression of
the information-rate measures on pleasure in Table 9 shows that both the novelty and irregularity of an environment are significant results predictors that of a one's novel level of pleasure. (one that The is
suggest
environment
surprising,
rare, varied,
contrasting,
and complex)
would
increase pleasure, while an irregular atmosphere (one that is random, crowded, and asymmetrical) would appear to decrease pleasure. Table 9
Beta Coefficients for Information-Rate Dimensions
Used to Predict the Pleasure Dimension
41
Overall, this study suggests that the Mehrabian-Russell model is useful for studying approach-avoidance behaviors However, the study is just a
starting point since it was strictly correlational rather than experimental, behavioral and it tested rather the than model with only actual with stated
intentions
behaviors. the
These points
should be kept
in mind while
analyzing
following implications. This study implies that store-induced pleasure, the first variable in the Mehrabian-Russell model, is an extremely
powerful determinant of approach-avoidance behaviors within the store. If an environment is perceived as pleasurable, and
hence rewarding, it seems logical that the activities which transpire in that environment would also be pleasurable and rewarding. Thus, consumers would tend to exhibit more
through the merchandise, interact with the sales personnel, and perhaps spend more money than they had planned. In contrast to the Mehrabian-Russell model and the work of Donovan and Rossiter (1982), results of this study suggest that arousal is not a particularly significant predictor of approach-avoidance behaviors. the students in my sample It is possible that a few of did not fully understand the
42
meanings or the applicability of some of the terms used in the questionnaire to rate this measure. Dominance, the third emotional measure in the model, also appears to be unrelated to in-store behaviors. This finding
correlates with that of Russell and Pratt (1980), which stated that since dominance requires a knowledgeable interpretation by the individual, it is not purely applicable in situations calling for emotional responses. An interesting finding of this study is the significance of one's sex in predicting approach-avoidance behavior within a retail environment. sex and pleasure, Results suggest an interaction between
affected by higher levels of store-induced pleasure than are female students. Thus, it may by predicted that male
consumers would exhibit more approach behavior in a pleasing environment than would female consumers. In summary, the pleasure-arousal-dominance (especially
the pleasure) part of the Mehrabian-Russell model, with the addition of a gender variable, is very useful in predicting in-store behavioral intentions: 65.326 percent. accounted-for variance was is needed to develop a
Further research
stimulus taxonomy for retail environments that relates to the emotional dimensions, but this study, like that of Donovan and Rossiter, confirms that these emotional states are apparently valid mediating variables with considerable predictive power (Donovan and Rossiter 1982).
43
This
finding for
presents retail
some
particularly Because
useful consumer
implications
merchandisers.
behavior is so complex and so dynamic, retailers must become more behaviorally sensitive by shifting their analysis from a simple economics and engineering emphasis to include
variety, and quality of merchandise), rather than focusing on the influence of emotional affect on consumers. Donovan and
Rossiter (1982) propose that, .. whereas cognitive factors may largely account for store selection and for most of the
planned purchases within the store, the emotional responses induced by the environment within the store are primary
determinants of the extent to which the individual spends beyond his or her original expectations." Retailers may want
to pay particular attention to departments frequented by men, since this study suggests that males are more affected by store-induced pleasure than are females. In addition,
merchandisers may benefit by creating an atmosphere which is more surprising, rare, and novel, since novelty seems to Over a period of of pleasure may
increase sales dramatically! Because this study is just a starting point for analyzing approach-avoidance behaviors within retail environments, there are many suggestions for future research. First, as noted,
44 future researchers must devise a stimulus taxonomy for retail environments. stimulation specifically Second, The measure of information rate was not a good measure in this to study. in-store A taxonomy factors be is that is
applicable
needed. in
individual
differences
should
taken
consideration, since individual reactions to environments may vary considerably. In addition, respondents of various age
groups should be used - both this study and that of Donavan and Rossiter use sUbjects category. Third, in the 20 to 24 year old age
and lastly,
should be tested with actual purchasing behavior rather than just with consumer-stated intentions.
45
Instructions to Subjects:
Take a few moments to get into the mood of the situation. Then, using the adjective pairs below, rate your feelings in this setting. Though some of the pairs may seem unusual, you'll probably feel more of one emotion than the other. So, for each pair, place a check mark (Example: { ) closest to the adjective which you believe describes your feelings the best. The more appropriate the adjective, the closer the check mark should be placed to that adjective. Happy Pleased satisfied Contented Hopeful Relaxed Important Free stimulated Excited Jittery Interested Frenzied Overcrowded Wideawake Controlling Dominant Influential Unhappy Annoyed Unsatisfied Depressed Despairing Bored Insignificant Restricted Relaxed Calm Dull Disinterested Sluggish Uncrowded Sleepy Controlled Submissive Influenced
46
Instructions to Subjects:
Please use the following adjective pairs to describe the environment surrounding you. Each of the following adjective pairs helps define the setting or the relations among the various aspects of the setting. Place a check mark closest to that adjective which best describes the environment. Please consider the environment as a whole - do not focus on specific design features or products. varied Simple Novel Small scale Similar Dense Intermittent ordinary Heterogeneous Uncrowded Asymmetrical Immediate Common Patterned Redundant Complex Familiar Large scale Contrasting Sparse continuous Surprising Homogeneous Crowded symmetrical Distant Rare Random
47
APPENDIX C
VERBAL MEASURES OF APPROACH-AVOIDANCE
Would you enjoy shopping in this store? Not at all Very slightly Slightly Slightly to moderately
(4) (5) (6) (7)
2)
(0)
How much time would you like to spend browsing in this store? None A few minutes Half an hour One hour
(4) (5) (6) (7)
(1)
(2) (3)
Would you avoid ever having to return to this store? Not at all Very slightly Slightly Slightly to moderately
(4) (5) (6) (7)
4)
Is this a place in which you would feel friendly and talkative to a stranger who happens to be near you? Not at all Very slightly Slightly Slightly to moderately
(4) (5) (6) (7)
Would you want to avoid looking around or exploring this environment? Not at all Very slightly Slightly Slightly to moderately (4) (5) (6)
(7)
48
Do you like this store environment? Not at all Very slightly Slightly Slightly to moderately (4) (5) (6) (7) Moderately Much Very much Extremely
7) Is this a place where you might try to avoid other people, and avoid having to talk to them? (0) (1) (2) (3) Not at all Very slightly Slightly Slightly to moderately (4) (5) (6) (7) Moderately Much Very much Extremely
8) Is this the sort of place where you might end up spending more money than you originally set out to spend? (0) (1) (2) (3) Not at all Very slightly more Slightly more Slightly to moderately more (4) (5 ) (6) (7) Moderately more Much more Very much more Very, very much more
49
REFERENCES
on
Thermal
Comfort,"
craik, K.H. (1970), "Environmental Psychology," New Directions in Psychology, 1-121. Crane, R.R. and B.I. Levy (1962), "Color Scales in Responses to Emotionally Laden Situations," Journal of Consulting Psychology, 26, 515-519. Donovan, Robert and John Rossiter (1982), "Store Atmosphere: An Environmental Psychology Approach," Journal of Retailing, 58 (Spring), 34-57. Heath, R.G. Mass. : (1954), Studies in Schizophrenia, Harvard University Press. Cambridge,
Heath, R.G. (1963), "Electrical Self-Stimulation of the Brain in Man," American Journal of Psychiatry, 120, 571-577. Heath, R.G. (1964a), "Pleasure Response of Human Subjects to Direct stimulation of the Brain: Physiologic and Psychodynamic considerations," The Role of Pleasure in Behavior, New York: Harper & Row. Heath, R.G. (1964b), The Role of Pleasure in Behavior, New York: Harper & Row. Ittelson, W.H., L.G. Rivlin and H.M. Proshansky (1970), "The Use of Behavioral Maps in Environmental Psychology," Environmental Psychology: Man and His Physical Setting, New York: Holt, Rinehart & Winston, Inc., 658-668. Kotler, Philip (1973), "Atmosphere as a Marketing Tool," Journal of Retailing, 49 (Winter), 48-64. Lacey, J.I. (1967), "Somatic Response Patterning and Stress: Some Revisions of Activation Theory," Psychological Stress: Issues in Research, New York: Appleton-Century Crofts, 14-37. Markin, Rom, Charles Lillis and Chem Narayana (1976), "Social Psychological Significance of Store Space," Journal of Retailing, 52 (Spring), 43-54. Mehrabian, Albert (1976), Public Spaces and Private Spaces: The Psychology of Work, Play and Living Environments, New York: Basic Books, Inc. Mehrabian, Albert (1980), Basic Dimensions for a General
Mehrabian, Albert and James Russell (1974), An Approach to Environmental Psychology, Cambridge, Mass.: MIT Press. Middleton, W.C., P.J. Fay, W.A. Kerr and F. Amft (1944), "The Effect of Music on Feelings of Restfulness-Tiredness and Pleasantness-Unpleasantness," Journal of Psychology, 17, 299-318. Olds, J. (1956), "Pleasure Centers in the Brain," Scientific American, 195, 105-116. Osgood, C E., G. J . Suci and Measurement of Meaning, Illinois Press. P.H. Tannenbaum (1957), The Urbana: The University of
Russell, James and G. Pratt (1980), "A Description of the Affective Quality Attributed to Environments," Journal of Personality and Social Psychology, 38 (August), 311 322. Schaie, K. W. (1961) , "A Q-Sort Study of Color-Mood Association," Journal of Projective Techniques, 25, 341 346. Stevens, C.W. (1980), "K-Mart Stores Try New Look to Invite More Spending," The Wall Street Journal, Nov. 26, 29-35. Wilson, G. D. (1966), "Arousal Properties of Red Versus Green," Perceptual and Motor Skills, 23, 947-949. Wysocki, B. (1979), "Sight, Smell, Sound: They're All Arms in Retailer's Arsenal," The Wall Street Journal, Nov. 17, 1979, 1,35.