Scents in The Marketplace: Explaining a Fraction of Olfaction PAULA FITZGERALD BONE, PH. D. West Virginia University
PAM SCHOLDER ELLEN, PH. D. Georgia State University
The popular and business press is enamored with the idea that the sense of smell can have strong effects on consumer responses to retail environments. The claims that odors have strong persuasive powers tantalize retailers looking for the competitive edge. Herein, we review the current paradigm of retailing-relevant olfaction research and find that “conventional wisdom” does not allow researchers or retailers to reliably predict olfaction effects. We suggest accessibility and availability theories as a way of explaining the current empirical research and as a method by which we can increase the reliability of capturing olfactory effects. We conclude by identifying fruitful areas of research in this interesting stimuli–that which we smell.
INTRODUCTION “A growing number of marketers are recognizing that the sense of smell can . . . be a powerful motivator for sales. . .” –Maxine Wilkie (1995) “Research is now making strides to discover and manufacture odors that can be used . . . to control the consumer’s emotions and thought processes.” –Dr. Alan Hirsch (1989) It’s in the news, prevalent in folklore and increasingly in the business press: “Smell sells.” It is clear that many retailers believe that the right scent can positively impact customers’ behavior. The business press reports that scent increased the desirability of Nike athletic shoes (press reports of Hirsch and Gay’s 1991 investigation), resulted in consumers spending greater amounts of time in a jewelry store (press reports of Knasko’s 1989
Paula Fitzgerald Bone, West Virginia University, PO Box 6025, Morgantown, WV 26506 – 6025; Tel: (304) 293–7959; ^
[email protected]& Journal of Retailing, Volume 75(2) pp. 243–262, ISSN: 0022-4359 Copyright © 1999 by New York University. All rights of reproduction in any form reserved.
243
244
Journal of Retailing Vol. 75, No. 2 1999
investigation), and increased bakery sales by 300% (Hirsch, 1991). Indeed, retailers have, in recent years, used olfaction to try to influence consumers in a variety of ways such as in-store displays that scent the surrounding air (Macy’s use of scent-dispensing kiosks; Power, 1998) and planned distribution systems for in-store odors (e.g., Superdrug’s use of an ambient chocolate odor during Mother’s Day promotions; Brand Strategy, 1998). Yet, while odors seemingly pervade the marketplace, there is limited academic research that captures odor effects. Indeed, controlled experimental examinations more often (63.2%) report null effects than significant effects. The problem for retailers and academic researchers, then, becomes one of separating what is really likely to happen from the hyperbole surrounding touted odor effects. Here, we examine the effects of scents in the retail environment, excluding scents that are central attributes of products (e.g., perfumes and air fresheners). We develop a paradigm representative of current olfactory research effects relevant to a retail setting and examine the evidence supporting that paradigm. We then review 206 tests of olfactory effects found in 22 studies of the delineated relationships. The details for classifying the studies and results are provided in the Appendix. We offer a theoretical rationale for the observed olfaction effects (or the lack thereof) and, finally, identify areas of promising research. CONVENTIONAL ACADEMIC WISDOM The myth associated with odor’s potential effect on consumers is strong that odors operate subliminally, that they directly affect emotion and that they strongly influence sales. Those propositions do not hold up under the scrutiny of experimental research. However, upon exploring the studies in this area, it appears that academic researchers have collectively expressed some common beliefs about how odors can operate to influence consumers. A representation of that “conventional wisdom” is presented in Figure 1. Three dimensions of an odor are felt to be important: its presence (or absence), its pleasantness and its fit or congruity with the object studied. Odors are hypothesized to affect consumers by changing approach/avoidance behaviors, altering mood state and affecting elaboration. Additionally, odors are anticipated to influence the components of the tripartite attitudinal model: the affective and cognitive responses as well as behavior toward the place or object. Finally, researchers identify many moderators. These moderators can be divided into two classes: individual characteristics (e.g., gender) and context effects (e.g., a stressful task). Below, we overview each of the three major dimensions of odor in light of the existing theories of how odors work. We then examine whether the empirical evidence supports conventional wisdom. Presence of a Scent Why might the presence of a scent affect a consumer’s response toward a shopping item, a retail environment or even impact the consumer’s overt behavior? Hvastja and
Scents in the Marketplace
245
FIGURE 1 Conventional Wisdom View of Olfactory Effects
Zanuttinit (1991) suggest that one of the primary roles of olfaction cues is to, “heighten awareness: it alerts the organism to existence of agents in the air, to check their quality for guidance of behavior on the basis of previous encounters, to avoid or approach certain substances” (p. 883). Therefore some consumer response may be an uncomplicated behavioral response (i.e., approach/avoidance (c.f., Spangenberg, Crowley, and Henderson, 1996)). Simple approach/avoidance could occur because olfactory cues are processed in a more primitive portion of the brain rather than in higher-level centers as occurs with other sensory cues (Herz and Engen, 1996). Odors require little, if any, cognitive effort to be experienced (Ehrlichman and Halpern, 1988) and basic behavioral responses can occur without conscious attention. For example, respiration deepens in the presence of a pleasant odor (Takagi, 1989). Conversely, the presence of an unpleasant odor halts breathing and even causes physical withdrawal (Levine and McBurney, 1986). In addition to a purely autonomic response, ambient odors may influence a consumer through mood state. While strong emotional responses to olfactory stimuli are relatively rare or idiosyncratic (Ehrlichman and Bastone, 1992; Rubin, Groth, and Goldsmith, 1984), many researchers suggest odors affect moods and mild affective states. Indeed, mood and affect shifts are the most frequently suggested mediators of olfaction effects (cf., Baron, 1990; Cann and Ross, 1989; DeBono, 1992; Ehrlichman and Bastone, 1992; Ellen and Bone, 1999). Such mood effects may be a pure sharing of hedonic tone. In that case, pleasant odors lead to pleasant mood states while unpleasant odors lead to unpleasant states (Ehrlichman and Bastone, 1992). Alternatively, mood shifts may result from “the complex meaning of previous social
246
Journal of Retailing Vol. 75, No. 2 1999
experience with odors” (Kirk–Smith and Booth, 1987; p. 159). These learned associations are ones common to larger groups of people, representing connections between the scent and certain foods (e.g., oranges, chocolate), products with natural or artificial scents (e.g., Playdoh® and bubblegum), places (e.g., home), and events or time periods (e.g., Christmas, birthdays, childhood). Given the opportunity and the availability of resources, odors may aid in retrieval of stored information and increase elaboration about the target object (c.f., Mitchell, Kahn, and Knasko, 1995). Assuming such elaboration is positive, one could expect a positive shift in affect. Thus, conventional wisdom offers three mechanisms by which olfactory cues may work in a retailing environment: approach/avoidance, hedonic mood shifts and cognitive elaboration. Does the empirical evidence support this? Table 1 contains the findings from those studies examining the presence of a scent on consumer responses. All the studies reviewed here included a no-odor control group. Surprisingly, there are no studies that explicitly measured scent presence effects on physical approach/avoidance tendencies.
Scent Present ➔ Mood Relationship Mood is typically conceived as two dimensions: (1) arousal, an activation state measured, for instance, with Mehrabian and Russell’s (1974) arousal dimension or measures of anxiety or sleepiness and (2) valence, its pleasantness or unpleasantness measured, for instance, with Mehrabian and Russell’s pleasure dimension or measures of feeling good or bad. As shown in Table 1 only a small percentage (16.1%) of the tests show any effects on either of these mood dimensions. Significant effects were found either for evaluative tasks, where the odor was ambient or one of several cues available for fulfilling the tasks (i.e., simulated shopping or other cognitive tasks), or for memories evoked by a battery of odors
TABLE 1 Effects of Scent Presence/Absence on Consumers’ Response
Approach/Avoidance Mood: Arousal Valence Elaboration Affective Response Evaluative Response Intent Behavior: Time Information Search Choice Total
Total Number of Tested Relationships
Not Statistically Significant Findings
0 12 19 36 18 26 7 6 9 10 143
10 (83.3%) 16 (84.2%) 21 (58.3%) 8 (44.4%) 16 (61.5%) 4 (57.1%) 2 (33.3%) 8 (88.9%) 8 (80.0%) 93 (65.0%)
Statistically Significant Findings Direct
Inverse
2 (16.7%) 3 (15.8%) 8 (22.2%) 6 (33.3%) 7 (26.9%) 3 (42.9%) 2 (33.3%) 1 (11.1%) 2 (20.0%) 34 (23.7%)
0 (0.0%) 0 (0.0%) 7 (19.4%) 4 (22.2%) 3 (11.5%) 0 (0.0%) 2 (33.3%) 0 (0.0%) 0 (0.0%) 16 (11.1%)
Scents in the Marketplace
247
where the effects are dependent on the specific odor’s ability to elicit associations in the consumers’ memory and the focus of the study is the odor itself. Mood arousal was affected in one of three odor-evoked memory tests and for females only in a photo rating task. Mood valence was influenced by odor presence/absence in three cognitive tasks, all moderated by either high stress, prior mood state, locus of control or the specific odors used.
Scent Present ➔ Cognitive Elaboration Many researchers examined variables indicative of elaboration, defined from both a discursive processing (using words; c.f., Kisielius and Sternthal, 1984) and imagery processing (using images; c.f., MacInnis and Price, 1987) perspective. Elaboration is a continuum ranging from simple processing (such as directing one’s mental resources to a stimulus) to more elaborate processing (such as integrating incoming information into prior knowledge structures; MacInnis and Price, 1987). Elaboration may be measured by simple recognition or recall of information, or by reports of imagining or remembering themselves (self-referencing) using the product, new, enhanced or more vivid descriptions of the stimuli, or the generation of counterarguments to the message. The reported elaboration studies examined either memories evoked by batteries of odors or olfactory effects during evaluative tasks. The majority of significant elaboration tests (75%) examined odor-evoked memories; thus, the effects are dependent on the specific odors’ ability to elicit associations in the consumer’s memory. The remaining studies involved evaluation tasks where the odor was ambient or one of several cues available. While many different indicators of elaboration were used in these tests, the significant findings were those measuring thoughts about the spokesperson or selfreferences. The null effects measured recall and inferences about the message claims. As shown in the Figure, approach/avoidance, mood and elaboration are typically viewed as mediators of olfactory effects on many traditional measures of response to retail environments (i.e., affective and evaluative response, intentions, and behavior). Since direct tests of mediation are relatively rare,1 we look at the tests of direct relationships between scent presence and these important outcome variables.
Scent Presence ➔ Affective and Evaluative Responses We defined affective responses as like/dislike responses to a stimulus whereas evaluative responses are more attribute- or quality-related. Thus, a measure asking how much the subject liked the product would be classified as an affective response and a measure which asked whether the room was comfortable or a product was of high quality would be considered an evaluative response. As shown in Table 1, the majority of tests of affective response (55.6%) are statistically significant, and all of them involved ratings of people, products or the environment. Null effects were found for ratings of odor-evoked memories. The significant results are predominantly moderated by gender (females rating more positively than males), char-
248
Journal of Retailing Vol. 75, No. 2 1999
acteristics of the rated person (more positive ratings for less formal attire) and specific scents. Spangenberg, et al., (1996) provide the most direct evidence of ambient odor effects in a retail environment with scent presence positively affecting overall ratings of a simulated store and of the store environment. Evaluation was examined in 26 tests and the majority of effects (61.5%) were null. Again, the effects that are significant are moderated by several variables including gender, attire of the rated person, motivation to process, and scent fit.
Scent Present ➔ Purchase Intentions Intentions to visit/return to the store and purchase various products were positively affected by scent in 43% of the tests. In the marketing-related tests, respondents showed greater likelihood to visit the store and purchase two out of four target products.
Scent Present ➔ Behavior Behavioral measures were divided into three categories: time spent, information search and choice. Two-thirds of the tests showed significant effects on time spent. Apparently, presence of a scent actually decreased information-search time (Mitchell, et al. 1995) but increased perceived time in the simulated store (Spangenberg, et al., 1996). Researchers used many different measures of search ranging from the number tags examined to the pattern used to acquire information (e.g., holistic). Compared to time spent, there is little evidence of odor effects on information search with only one significant effect. There was greater search of product tags for one of seven products reported in Spangenberg et al. (1996), perhaps also indicating a difference in personal relevance. Choice measures included choice among brands and measures of variety seeking and switching behavior. The two significant findings occurred in a replication study by Mitchell, et al., (1995) where the initial tests were null. Thus, using the same procedures, principal investigators, and identical dependent variables, the authors did not replicate their own findings.
Summary of Scent Presence Effects Evidence is stacked against the proposition that the simple presence of an odor affects a retail customer’s behavior. The research provides no direct evidence of approach/ avoidance effects and very limited support for mood change when an odor is present. Elaboration effects are more common but were largely dependent on the task subjects were completing (i.e., odor-evoked memories). There is more evidence of affective than evaluative odor effects, yet, many of the effects are moderated. Also there are some odor effects on intentions to visit and purchase some, but not all, products. For behavior, the stronger evidence is of effects on time spent but less actual time spent on search and
Scents in the Marketplace
249
increased perceptions of time spent. There was little effect on actual search and choice. It appears that as we move through the traditional hierarchy-of-effects, the impact of odors on consumers’ response lessens. Before proceeding, one caveat about interpreting scent-present results is necessary. In many cases, there was no information on the degree of pleasantness or congruity of the odor with the stimulus or context. In other cases, scent-present conditions collapsed across degrees of pleasantness or fit. Thus, there may be more guidance from the tests of odor pleasantness and congruity described subsequently. We now turn to other tested dimensions of odor. Scent Pleasantness Just as music and pictures can be identified by certain primary characteristics (e.g., tone, beat, shape, color), there are two primary but related characteristics of odors: quality and intensity. Quality refers to the affective tone of an odor ( i.e., its perceived pleasantness or unpleasantness) (Harper, Bate–Smith, and Land, 1968; Takagi, 1989). Intensity refers to the concentration of an odor (Takagi, 1989). Generally, as the intensity of a scent increases, its pleasantness decreases; that is, an odor in low concentrations may be perceived as pleasant while the same odor in high concentration could be considered noxious (Henion, 1971). If, as suggested before, the effects of scent are based on the sharing of hedonic tone, then it follows that the more pleasant (unpleasant) the scent, the stronger and more positive (negative) the effects on moods and evaluations. These positive or negative mood states color one’s judgments of, or can be transferred to, unrelated focal objects (c.f., Isen and Shalker, 1982; Petty, Schumann, Richman, and Strathman, 1993). We now review the empirical evidence of scent pleasantness effects on consumers. As can be seen from Table 2, there is less empirical work studying these relationships (14.8% of all tests) than studying the effects of scent presence (68.4% of all tests). Again, there were no studies of approach/avoidance effects nor were there any tests on intent or behavior.
Scent Pleasantness ➔ Mood While no tests indicate that a more pleasant scent resulted in mood arousal, we do see a direct effect of pleasantness for six of ten tests on mood valence. The significant effects occurred during specific cognitive or negotiation tasks while the null effects were found in attractiveness-rating tasks and simulated shopping experience. This, then, provides some support for researchers’ belief that scent valence positively influences mood valence.
Scent Pleasantness ➔ Cognitive Elaboration Only one test of this relationship was found and it was null. It is thus premature to draw any conclusions about this link at this time.
250
Journal of Retailing Vol. 75, No. 2 1999
TABLE 2 Effects of Scent Pleasantness/Unpleasantness on Consumers’ Response Total Number of Tested Relationships Approach/Avoidance Mood: Arousal Valence Elaboration Affective Response Evaluative Response Intent Behavior: Time Information Search Choice Total
0 6 10 1 6 8 0 0 0 0 31
Not Statistically Significant Findings 6 (100%) 4 (40.0%) 1 (100.0%) 4 (66.7%) 5 (62.5%)
20
(64.5%)
Statistically Significant Findings Direct
Inverse
0 (0.0%) 6 (60.0%) 0 (0.0%) 2 (33.3%) 3 (37.5%)
0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%)
11 (35.5%)
0 (0.0%)
Scent Pleasantness ➔ Affective Response
Only a few researchers explored this relationship. Significant results were found for an evoked-imagery study and a negotiation task. Interestingly, the study most in line with retail atmospherics (Spangenberg, et al., 1996) reported two null findings.
Scent Pleasantness ➔ Evaluative Response
Evidence of a direct relationship between scent pleasantness and evaluative response is mixed. Significant effects are reported for the evaluation of a scented room and one of several products.
Summary of Scent Pleasantness Effects
We have evidence that greater (or lesser) pleasantness leads to more (or less) positive mood valence and limited evidence that pleasantness affects evaluative responses to an environment or object. Importantly, all the findings are consistent in that the relationships are always direct–never do we find a pleasant (unpleasant) odor having a negative (positive) affect on consumers. Additionally, at lower levels of processing (i.e., mood as opposed to judgment) the evidence that scent pleasantness influences consumers is stronger.
Scents in the Marketplace
251
Scent Congruity/Fit with the Target Oject Cue congruity or fit refers to how well the cue complements the other components of a marketing stimulus (cf., MacInnis and Park, 1991). Some odors, although generally perceived as pleasant, may be viewed as inappropriate in a particular context. In other words, in the consumer’s eyes (or nose) the odor just does not “fit” with the product. Pomerantz (1981) suggests that when the cue does not fit, different components of the consumer’s experience compete for cognitive resources and, thus, may inhibit attitudinal judgments. Incongruent odors may result in the consumer retrieving irrelevant information, thus interfering with processing of relevant information. However, when the cue or odor fits, retrieval of stored information and judgments may be facilitated (Mitchell, et al., 1995). Scent congruity is a primary focus in the retailing and marketing studies of olfaction (Bone and Jantrania, 1992; Ellen and Bone, 1999; Mitchell, et al., 1995; Spangenberg, et al., 1996). Information contained in Table 3 shows the existing research on scent congruity. Once more, we found no studies of scent fit effects on approach/avoidance. Additionally, there are no tests of affective response or intention.
Congruity ➔ Mood
Only one study has examined mood effects under different congruity conditions (Ellen and Bone, 1999). In a better-fit condition, there was a positive shift in mood valence.
TABLE 3 Effects of Scent Fit/Congruity or Familiarity on Consumers’ Response Total Number of Tested Relationships Approach/Avoidance Mood: Arousal Valence Elaboration Affective Response Evaluative Response Intent Behavior: Time Information Search Choice Total
0 0 1 12 0 7 0 2 2 10 34
Not Statistically Significant Findings
Statistically Significant Findings Direct
Inverse
0 (0.0%) 9 (75.0%)
1 (100.0%) 3 (25.0%)
0 (0.0%) 0 (0.0%)
4 (57.1%)
3 (42.9%)
0 (0.0%)
1 (50.0%) 1 (50.0%) 4 (40.0%) 19 (55.8%)
1 (50.0%) 0 (0.0%) 5 (50.0%) 13 (38.2%)
0 (0.0%) 1 (50.0%) 1 (10.0%) 2 (5.8%)
252
Journal of Retailing Vol. 75, No. 2 1999
Congruity ➔ Elaboration Consistent with scent-presence effects, greater elaboration was found for three out of four imagery measures and one out of two self-referencing measures. As in the scentpresent conditions, the remaining nulls were for claim-related measures (i.e., recall).2
Congruity ➔ Evaluative Response These tests were almost equally split between direct effects and null. Tests using global product evaluations were significant while those using specific attribute ratings were not.
Congruity ➔ Behavior Mitchell et al., (1995) performed several tests of the relationship between congruity and behavior hypothesizing a distraction effect when the odor was inconsistent with the task. They found that more total time was spent in a congruent condition but that there was less variability in the information search, more brand switching and more subjects chose a poorer quality brand.
Summary of Congruity Effects There were no tests of approach-avoidance and only limited evidence of mood effects. Congruity effects on elaboration and evaluation are weak; there were more significant behavioral responses, yet these were more likely negative effects on decision-making behaviors. Careful examination of the empirical tests of congruity shows that the reported results often are not the result of a better-fitting scent boosting evaluations of the environment/object evaluations. Instead, the reported effects are more often due to the detrimental effects of an incongruent odor. Inconsistent conditions produced less imagery (Wolpin and Weinstein, 1983) and lower ad evaluations (Ellen and Bone, 1999) while there were no differences between the congruent and no-odor control groups. Only Bone and Jantrania’s (1992) investigation provides positive evidence of fit effects. Specifically, they found that the better-fitting odor for two different products resulted in more positive overall evaluations than either the no-odor control or the lower-fit odor.
INTEGRATION AND SYNTHESIS OF OLFACTORY FINDINGS: APPLYING ACCESSABILITY AND AVAILABILITY THEORIES Can we make sense of the large number of null and contradictory results, while accounting for some of the consistencies? The conventional wisdom is not significantly upheld in the
Scents in the Marketplace
253
empirical results. Instead, it seems that important variables are missing from our current model. To enhance our understanding, we suggest accessibility and availability theories, with their components, valence and diagnosticity. Accessibility (Feldman and Lynch, 1988) and availability (Kisielius and Sternthal, 1984; 1986) theories suggest that information actively accessed from the environment or available from long-term memory has the greatest influence on the consumers’ attitudes and judgments.3 These theories suggest that individuals elaborate upon incoming environmental information by relating the information to previously stored, activated information. Two characteristics of information affect a consumer’s judgment and behavior. First is the valence of that information–whether the environmental information and/or the information retrieved from memory is positive, negative or neutral. This seems to be a good fit with the olfaction research due to the fact that the primary nature of odors is pleasant (i.e., positive) or unpleasant (i.e., negative). From our review, we have evidence, albeit mixed, that exposure to odors sometimes affects retrieval and processing of information, thus having the ability to activate information stored in memory. The second characteristic of information is its diagnosticity–the degree to which the information is helpful in categorizing (i.e., high quality, low quality) or interpreting the product or service. Herr, Kardes, and Kim (1991), based on earlier work by Feldman and Lynch (1988), suggest that diagnosticity is a function of the ambiguity of the information, the valence of the information and the amount of other information available to the consumer. Olfactory information can be quite ambiguous, compared to other cues (e.g., words, pictures and sounds). Scents are difficult to recognize and label (Schab, 1991) and can produce false alarms (Engen, 1972). This lack of clarity when interpreting odors is exacerbated by the fact that surrounding cues also affect consumers’ ability to detect and recognize odors (Davis, 1981). As an example, the ability to recognize a lemon scent is greater when the scent is presented in a yellow liquid as opposed to a red liquid. Finally, olfaction researchers themselves have suggested that “response to these odours will depend on circumstances and contexts, and these cannot be easily specified” (Kirk–Smith, 1994). All of this suggests that olfactory stimuli are rather ambiguous. One will note that valence of information has now entered into our discussion twice. Not only would one expect more positive judgments and more approach behaviors when information is positive than when it is negative, but consumers tend to over-weight negatively-valenced information since they perceive it to be more diagnostic than positively-valenced information (c.f., Herr et al., 1991). The final diagnosticity criterion, the amount of information available, proposes that, when faced with a situation in which some of the available information is perceived to be diagnostic and some is felt to be ambiguous, the consumer will rely more heavily on the more diagnostic information at the expense of the ambiguous information. The next step is relating the olfaction results found to the above theories, paying particular attention to the marketing-oriented investigations (e.g., Bone and Jantrania, 1992; Ellen and Bone,1999; Mitchell et al., 1995; Spangenberg et al., 1996). The observed effect of scent pleasantness on mood valence, is clearly a case of accessible information being valenced, and in turn, affecting the consumer’s mood. Thus, if one is attempting to
254
Journal of Retailing Vol. 75, No. 2 1999
maximize the effects of odor on mood, one should use the most pleasant odor (for the particular target market) possible. However, none of the specific marketing studies have reported such an effect. Accessibility explains differences in observed elaboration and product judgment when the diagnosticity and ambiguity of the odor is taken into account. Clearly, odors can make information more available and greater congruity enhances accessibility to stored information and thus elaboration. Congruent cues that add information may enhance product judgments. Bone and Jantrania (1992) demonstrated that the addition of a consistent scent influenced product judgments. The scent of household cleansers and suntan lotions may be indicators of quality because of prior experience and beliefs about these products and because other attributes and characteristics of the products may have been difficult to evaluate in the limited experimental setting. However, when congruent odors duplicate existing information, judgments may not be enhanced. In Ellen and Bone (1999), the addition of a floral scent to an ad with a picture of wildflowers and related prose provided little new information and had no effect on product judgment beyond the control group. Incongruity interferes with elaboration reducing the amount of relevant information available to the consumer and making the task more difficult (Mitchell et al. 1995). The addition of incongruent scents to household cleaners and suntan lotions in Bone and Jantrania (1995) yielded negative effects on judgment. Similarly, Ellen and Bone’s (1999) addition of a pine scent to the wildflower ad led to a decrease in ad evaluation. Incongruent odors may cause negative responses because of the interference with elaboration or because of the greater weight attached to negatively-valenced information. In the latter case, the odor would have more influence than other positive cues (Herr et al., 1991). Using these theories also allows us to explain why Spangenberg et al. (1996) found strong support for ambient odor affecting the retail environment, but weaker effects for the same ambient odor to influence product judgments. In their study, the simple presence of a scent had consistent effects positively affecting 21 of 24 individual items regarding the store, store evaluation and general merchandise measures. The researchers couch these effects in terms of optimal arousal theory stating that “minor changes in environment, such as adding a low level of scent, increases the environment’s perceived novelty and pleasingness” (p. 70). In other words, odors in store environments may be perceived as diagnostic–they are a part of the atmospherics. Spangenberg et al.’s findings are not as strong for the three products evaluated (individual tests suggest that only one of three product evaluations appear to be affected by the scent). It is likely that the scent was not diagnostic for a particular product because it added no new relevant information to the consumer. This may have occurred because many of the product characteristics were likely to be accessible and thus competed with the effect of odor. It may also be that odor is simply not relevant for calendars and the other products examined in this study. Mitchell et al.’s (1995) investigation readily falls into the accessibility/availability framework. These researchers suggest that congruent odors lead to greater access to congruent attitudes, autobiographical memories, thoughts regarding prior experience with the product class and product-class knowledge. Incongruent odors result in accessing incongruent information and, thus create cognitive interference. Their discussion is clearly based on availability theory arguments. Mitchell et al. (1995) find such effects with respect to several measures of elaboration and product choice. At this point, however,
Scents in the Marketplace
255
retailers should be hesitant to use ambiant odor to create greater processing because, in most tests (17 out of 20), the congruent situation resulted in no difference from a no-odor control. Finally, the plethora of null olfactory effects could simply be a function of the olfactory cue’s ambiguity in a particular situation or its inability to compete (i.e., be diagnostic) in light of other, more diagnostic information. For instance, the cases in which judgments were unaffected by odor manipulations involved multidimensional stimuli (e.g., shoes, a person’s physical attractiveness) where other, more pertinent, information may have competed for resources.
EXPLORING ALTERNATIVE EXPLANATIONS
While accessibility/availability may improve prediction, greater consideration must also be given to some of the moderated effects and methodological differences found in the existing studies. Table 4 includes data for some moderators (i.e., gender and directed attention to odors), method (i.e., sample size), and measurement differences. Results are for all tests across the independent variables (i.e., combining presence, pleasantness, and congruity). T-tests of proportions and means are used to test for differences, keeping in mind that the small samples sizes led to underpowered tests.
Moderators A large percentage of odor effects were moderated by one or more variables. Accessibility theory would predict the gender differences examined in several studies. Gender is expected to moderate odor effects because women frequently have more developed schemas with regard to olfactory cues and have more heightened sensitivity to odors than do men (Koelega, 1994). While there is evidence in other olfactory literature that women can detect odors and are, perhaps more sensitive to odors, there were no statistically significant differences between tests using only men or women (See Table 4). However, gender should still be considered as we seek to refine olfaction theory. Greater accessibility and elaboration should occur when consumers are motivated to process provided information including attending to olfactory cues. In studies where attention was directed to the presence of an odor, elaboration effects were significantly more likely. Perhaps directing attention to the odor increased the salience and thought directed to the odor, and thus motivation to process, so that subjects were compelled to consider the relevance of the odor to the task at hand. On the other hand, affective responses were more likely in studies that did not direct attention to the odors. Not directing attention to the odor may have allowed subjects to simply react more holistically to the valence of the odor and transfer it to the object.
* p # .10
Mood: Arousal Valence Elaboration Affective Response Evaluative Response Intent Behavior: Time Information Search Choice Overall (% of Total)
TABLE 4
1 (100.0%) 0 (0.0%) — 1 (100.0%) 2 (66.7%) — — — — 4 (57.1%)
Women 0 (0.0%) 0 (0.0%) — 1 (100.0%) 1 (33.3%) — — — — 2 (28.6%)
Men
Significant Gender Effects
1 (14.3%) 4 (36.4%) 16 (44.4%)* 1 (16.7%) 2 (18.2%) 1 (100.0%) — — — 25 (34.7%)
Attention 1 (9.1%) 7 (35.0%) 2 (16.7%) 11 (61.1%) 14 (46.7%) 2 (33.3%) 5 (62.5%) 2 (18.2%) 14 (38.6%) 58 (39.7%)
No Attention
Drew Attention to Odor
32** 85.1 79.7 132.9 113.1 252 165.4 187.5 77.9
Significant
120.7 79.0 91.4 111.7 129.8 185 150.7 224.3 77.6
Not Significant
Mean Sample Size
Moderator, Methodological and Measurement Explanations for Findings
1 (33.3%) 3 (33.3%) — 8 (61.5%) 5 (45.5%) — — — — 17 (47.2%)*
Global
1 (6.7%) 8 (40.0%) — 0 (0.0%) 8 (36.3%) — — — — 17 (28.3%)
Specific
Specificity of Measures
256 Journal of Retailing Vol. 75, No. 2 1999
Scents in the Marketplace
257
Method Differences There was substantial variation in sample sizes across the studies with ranges from a low of 12 to a high of 308. Across all dependent variables, sample size was significant only for mood arousal effects, and then opposite that expected. In total, it appears that sample size and the resulting statistical power are poor explanations of differences between tests.
Measurement Differences Ehrlichman and Bastone (1992) suggest that different levels of measurement specificity could determine whether olfactory effects are captured. These researchers suggest that global measures, notably global mood measures such as good/bad or pleasant/unpleasant, appear to be more sensitive to changes in mood than are more specific measures (e.g., anxious or relaxing). As can be seen in Table 4, there were more significant effects when global measures were used than when specific measures were used for affective response. Across all dependent variables there were significantly more effects captured with global measures. Thus, global measures appear to be more sensitive to odor effects.
FUTURE DIRECTIONS FOR RESEARCH We offer two basic recommendations to guide researchers to more fruitful opportunities to understand and use odors in retail strategies. The first is to focus on accessibility and availability theories as a means to more reliably “capture” olfaction effects. The second recommendation is to develop methodical approaches to explore the gaps in our literature and resolve the discrepancies in the research findings. This explicitly includes the testing of mediating processes.
Reliably Capturing Olfaction Effects Reliably capturing olfaction effects depends heavily on critical tests of odors within the context of accessibility and availability hypotheses and upon using appropriate dependent variables. We identified no studies that examined diagnosticity or the effect of competing cues on olfactory effects. Varying the degree of diagnosticity could be accomplished using existing beliefs about the relevance of olfactory cues in a product category. Alternatively, consumers could be trained to use olfactory cues as an indicator of product characteristics or quality. For instance, Owen-Corning Fiberglass has successfully used the color (pink) to be used as an indicator of product quality in the fiberglass industry. This pink color is diagnostic for consumers because the product itself is difficult to evaluate and likely to be
258
Journal of Retailing Vol. 75, No. 2 1999
a credence good. The diagnosticity of odors should improve if we use the odor with “supporting” cues (e.g., add a yellow color to a lemon-scented product). Additionally, we must ensure that the environmental and product scents used are not redundant of competing information. Thus, the many dimensions of diagnosticity need to be systematically tested. Expectations about relevant outcomes also need to be realistic. For example, significant effects were more common when evoked imagery measures (e.g., self-referencing, vividness) were used rather than cognitive or learning measures such as recall. As for behavioral measures, consideration must be paid to the relevant behaviors that may be affected by mood and elaboration. Time spent is certainly one measure to be included, but product choices may be too distal for observed effects. Odors used in these studies should be perceived as congruent for the product or retail outlet since we have found that incongruent odors have a detrimental effect on the consumer’s evaluation. While Spangenberg et al. (1996) suggest the use of odors which are neither congruent nor incongruent but neutral, the real difficulty may be delineating a scent which is truly neutral and definitely not incongruent. Specifically, incongruent odors may be diagnostic, but negative, and thus work counter to the retailer’s goals. In sum, a series of systematic studies examining diagnosticity in various contexts and using measures more sensitive to odor effects would add greatly to our understanding of this area. Then academics and practitioners would be able to assess whether olfaction effects can be predicted reliably. Clearly, we cannot make such predictions at this point in time. If we fail to find reliable odor effects in contexts for which they should theoretically be diagnostic and congruent, then it may be time to write the epitaph for the pursuit of olfaction effects.
Exploration of Gaps Two clear gaps in our knowledge exist. First, we have little information about the mediating factors associated with olfaction. From a theoretical standpoint, researchers need to explicitly examine whether odor effects on affective and behavioral outcomes are direct or are partially or fully mediated by mood state, cognition or other variables. Along similar lines, one of the most cited methods by which olfaction should affect consumers’ is approach/avoidance; yet, we have no explicit tests of odors’ power to create such a response. Belizzi, Crowley, and Hasty (1983) provide an interesting test of color’s ability to create approach/avoidance. A similar type of experiment could put odors to the test. For example, subjects could be allowed to freely move about a simulated retail environment in which one area had an unpleasant odor, another a pleasant odor while a final area is left unscented. Time spent in the various areas and the subjects’ relative positioning of themselves in one area versus another are dependent variables that could directly assess approach or avoidance.
Scents in the Marketplace
259
CONCLUSION The empirical evidence reviewed suggests that conventional wisdom makes predicting specific odor effects (i.e., specific moods, thoughts, attitudes, or behaviors) a risky business. Counting on such effects is an unwise strategy at this point in time. Yet, more methodical attention to odor use may change the current situation and allow retailers in the future to use odor in a strategic manner in a cluttered competitive environment. Acknowledgement: The authors thank Terence A. Shimp for his guidance on this project and Deborah J. MacInnis for her thoughtful comments on an earlier draft of this paper. We are grateful to Randy Clark for help with the review. Both authors contributed equally to this paper.
APPENDIX: SELECTION OF STUDIES AND CLASSIFICATION OF RESULTS Studies included in the analysis of olfactory effects were those examining scent manipulations on cognitive, attitudinal or behavioral variables relevant to retailing. Specifically, these studies assessed the effects of scent presence, scent pleasantness, or scent fit on mood, elaboration, affective and evaluative response, intent and behavior (i.e., time spent, information search, and choice). Findings were classified as tests of presence, pleasance or fit using all available statistical tests in the articles. Any given test is inherently confounded with the other traits (e.g., all tests of presence involved scents which are pleasant/unpleasant to some degree). A complete list of the studies and effects is available from the authors. Several criteria were used to determine whether a particular investigation or a particular variable was included in our review. Excluded from this analysis was the substantial literature on the ability to detect, label, recognize and recall odors; such variables are of little interest when attempting to increase customers’ evaluation of a particular store and its offerings. For similar reasons, we have eliminated findings regarding task performance (such as greater ability to proofread materials or to correctly complete math problems). We have also excluded studies that examined only unpleasant odors as retailers are unlikely to intentionally introduce foul odors into their environment. When data from multiple scents and a no-odor group are available, we examined the results separately for all applicable conditions (scent presence, scent pleasantness and/or scent congruity). In all there were 206 individual test results in the final set of 22 articles. Individual effects were included and evaluated on the following criteria: 1.
2. 3.
Effects were judged significant if they reached conventional levels of statistical significance (p # 0.05). Post-hoc and univariate results were treated as statistically significant only if the relevant omnibus test is statistically significant. Results regarding unpleasant odors are included only when the research involved comparisons to pleasant odors effects. Results from summated items or scales were used rather than the results for all
260
Journal of Retailing Vol. 75, No. 2 1999
individual items composing the scale. Exceptions were made when the set of items included both cognitive and affective items.
NOTES 1. Only one identified study provided a test of mediation (Ellen and Bone, 1999). They found that scent pleasantness and mood partially accounted for the effect on attitude-toward-the-ad. 2. Herz and Cupchik (1992) looked at odor familiarity effects on elaboration. Familiarity with an odor would be a precursor to congruity judgments since familiarity would determine the existence and nature of the odor’s connections in the person’s schema. 3. Traditionally, accessibility deals with retrieving information from the environment while availability theory focuses on retrieval from memory. We believe both factors are important when considering olfaction efffects; therefore, we use both theories.
REFERENCES Baron, R.A. (1981). “Olfaction and Human Social Behavior: Effects of a Pleasant Scent on Attraction and Social Perception,” Personality and Social Psychology Bulletin, 7(4): 611– 616. . (1983). “‘Sweet Smell of Success?’ The Impact of Pleasant Artificial Scents on Evaluations of Job Applicants,” Journal of Applied Psychology, 58(4): 709 –713. . (1990). “Environmentally-Induced Positive Affect: Its Impact on Self-Efficacy, Task Performance, Negotiation, and Conflict,” Journal of Applied Social Psychology, 20(5): 368 –384. . (1992). “Exposure to Pleasant Fragrances as a Technique for Reducing the Effects of Work-Related Stress,” presented at the Centennial Convention of the American Psychological Association, Washington, DC, August. Belizzi, Joseph A., Ayn E. Crowley, and Ronald W. Hasty. (1983). “The Effects of Color in Store Design,” Journal of Retailing, 59(1): 21– 45. Bone, Paula Fitzgerald and Swati Jantrania. (1992). “Olfaction as a Cue for Product Quality,” Marketing Letters, 3(3): 289 –296. Brand Strategy. (1998). “Sensory Design” July 24. 8 –10. Cann, Arnie, and Debra A. Ross. (1989). “Olfactory Stimuli as Context Cues in Human Memory,” American Journal of Psychology, 102(1): 91–102. DeBono, Kenneth G. (1992). “Pleasant Scents and Persuasion: An Information Processing Approach,” Journal of Applied Social Psychology, 22(11):1 910 –919. Ehrlichman, Howard, and Linda Bastone. (1992). “Olfaction and Emotion.” Pp. 410 – 438 in Michael Serby and Karen Chobor (Eds.), Science of Olfaction. New York: Springer-Verlag. , and Jack N. Halpern. (1988). “Affect and Memory: Effects of Pleasant and Unpleasant Odors on Retrieval of Happy and Unhappy Memories,” Journal of Personality and Social Psychology, 55(5): 769 –779. Ellen, Pam Scholder, and Paula Fitzgerald Bone. (1999). “Olfactory Stimuli as Advertising Executional Cues,” Journal of Advertising, 27(4): 29 –39. Engen, Trygg. (1982). The Perception of Odors. New York: Academic Press. Feldman, Jack M. John G. Lynch, Jr. (1998). “Self-Generated Validity and Other Effects of
Scents in the Marketplace
261
Measurement on Belief, Attitude, Intention and Behavior,” Journal of Applied Psychology, 73(3): 421– 435. Harper, R., E. C. Bate-Smith, and D. G. Land. (1968). Odour Description and Odour Classification: A Multidisciplinary Examination. New York: American Elsevier Public Company, Inc. Henion, K. E. (1971). “Odor Pleasantness and Intensity: A Single Dimension,” Journal of Experimental Psychology, 90(2): 275–279. Herr, Paul M., Frank R. Kardes, and John Kim. (1991). “Effects of Word-of-Mouth and ProductAttribute Information on Persuasion: An Accessibility-Diagnosticity Perspective,” Journal of Consumer Research, 17(4): 454 – 462. Herz, Rachel S. and Gerald C. Cupchik. (1992). “An Experimental Characterization of Odor-Evoked Memories in Humans,” Chemical Senses, 17(5): 519 –528. , and Trygg Engen. (1996). “Odor Memory: Review and Analysis,” Psychonomic Bulletin & Review 3(3): 300 –313. Hirsch, Alan R. (1989). Expert testimony given at a December 5 public hearing of the State of New York Senate. . (1991). “Nostalgia: A Neuropsychiatric Understanding,” presented at the annual meeting of the Association for Consumer Research Conference, Chicago, Illinois. October. , and Gay, S. E. (1991). “The Effect of Ambient Olfactory Stimuli on the Evaluation of a Common Consumer Product,” presented at the Thirteenth Annual Meeting of the Association for Chemoreception Sciences. April. Hvastja, Loredana, and Luchia Zanuttini. (1991). “Recognition of Nonexplicitly Presented Odors,” Perceptual and Motor Skills, 72(1): 883– 892. Isen, Alice and Thomas Shalker. (1982). “The Effect of Feeling State on Evaluation of Positive, Neutral and Negative Stimuli: When You ‘Accentuate the Positive,’ Do You ‘Eliminate the Negative’?” Social Psychology Quarterly, 45(1): 58 – 63. Kirk-Smith, M. D. (1994). “Culture and Olfactory Communication.” Pp. 385– 406 in R. Allen Gardner et al. (Eds.), Ethological Roots of Culture. Boston: Kluwer Academic Publishers. , and D. A. Booth. (1987). “Chemoreception in Human Behavior: Experimental Analysis of the Social Effects of Fragrances,” Chemical Senses, 12(1): 159 –166. ——-, C. Van Toller, and G. H. Dodd. (1983). “Unconscious Odour Conditioning in Human Subjects,” Biological Psychology, 17(2/3): 221–231. Kisielius, Jolita and Brian Sternthal. (1984). “Detecting and Explaining Vividness Effects in Attitudinal Judgments,” Journal of Marketing Research, 21(1): 54 – 64. ——-. (1986). Examining the Vividness Controversy: An Availability-Valence Interpretation,“ Journal of Consumer Research, 12(4): 418 – 431. Knasko, Susan C. (1989). ”Ambient Odor and Shopping Behavior,“ Chemical Senses, 14(94): 718. . (1992). ”Ambient Odor’s Effect on Creativity, Mood and Perceived Health,“ Chemical Senses, 17(1): 27–35. Koelega, Harry S. (1994). ”Sex Differences in Olfactory Sensitivity and the Problem of the Generality of Smell Acuity,“ Perceptual and Motor Skills, 78(1): 203–213. Laird, Donald A. (1932). ”How the Consumers Estimate Quality by Subconscious Sensory Impressions: With Special Reference to the Role of Smell,“ Journal of Applied Psychology, 16(1): 241–246. Levine, John M. and Donald H. McBurney. (1986). ”The Role of Olfaction in Social Perception and Behavior. Pp. 179 –217 in C. Peter Herman et al. (Eds.), Physical Appearance, Stigma, and Social Behavior: The Ontario Symposium. Hillsdale, NJ: Lawrence Erlbaum. Ludvigson, H. Wayne and Theresa R. Rottman. (1989). “Effects of Ambient Odors of Lavender and Cloves on Cognition, Memory, Affect and Mood,” Chemical Senses, 14(4): 525–536. MacInnis, Deborah J. and C. Whan Park. (1991). “The Differential Role of Characteristics of Music
262
Journal of Retailing Vol. 75, No. 2 1999
on High- and Low-Involvement Consumer’s Processing of Ads,” Journal of Consumer Research, 18(2): 161–173. and Price, Linda L. (1987). “The Role of Imagery in Information Processing: Review and Extensions,” Journal of Consumer Research, 13(4): 473– 491. Mehrabian, A. and J. A. Russell. (1974). An Approach to Environmental Psychology. Cambridge, MA: MIT Press. Mitchell, Deborah J., Barbara E. Kahn, and Susan C. Knasko. (1995). “There’s Something in the Air: Effects of Congruent or Incongruent Ambient Odor on Consumer Decision-Making,” Journal of Consumer Research, 22(2): 229 –238. Petty, Richard E., David W. Schumann, Steven A. Richman, and Alan J. Strathman. (1993). “Positive Mood and Persuasion: Different Roles for Affect Under High- and Low-Elaboration Conditions,” Journal of Personality and Social Psychology, 64(1): 5–20. Pomerantz, James R. (1981). “Perceptual Organization in Information Processing.” Pp. 141–180 in Michael Kubovy and James Pomerantz (Eds.), Perceptual Organization. Hillsdale, N J: Erlbaum. Power, Denise. (1998). “Appealing to a Third Sense: Macy’s Kiosks Add Scents,” HFN The Weekly Newspaper for the Home Furnishing Network, 23 (June 8). 9. Richardson, Alan. (1991–1992). “Personal Memories: An Exploratory Investigation,” Imagination, Cognition, and Personality, 11(1): 53–74. Rubin, David C., Elisabeth Groth, and Debra J. Goldsmith. (1984). “Olfactory Cuing of Autobiographical Memory,” American Journal of Psychology, 97(4): 493–507. Spangenberg, Eric R., Ayn E. Crowley, and Pamela W. Henderson. (1996). “Improving the Store Environment: Do Olfactory Cues Affect Evaluations and Behaviors?” Journal of Marketing, 60(2): 67– 80. Takagi, Sadayuki F. (1989). Human Olfaction. Tokyo: University of Tokyo Press. Lipman, Joanne. Wall Street Journal (1990). “Advertising: Scents that Encourage Buying Couldn’t Smell sweeter,” (January 9). Sec. B, 5. Warm, Joel S., William N. Dember, and Raja Parasuraman. (1990). “Effects of Fragrances on Vigilance Performance and Stress,” Perfumer and Flavorist, 15(January/February): 15–18. . (1991). “Effects of Olfactory Stimulation on Performance and Stress in a Visual Sustained Attention Task,” Journal of the Society of Cosmetic Chemists, 42(3): 199 –210. Wilkie, Maxine. (1995). “Scent of a Market,” American Demographics, 17(8): August, 40 – 47. Wolpin, Milton and Charles Weinstein. (1983). “Visual Imagery and Olfactory Stimulation,” Journal of Mental Imagery, 7(1): 63–74.
