I. Introduction
Expectations have been shown in many contexts to exert a strong influence on the interpretation of subjective experiences, affecting not only our perceptions of such experiences but also our memories of them (e.g., Carlsmith and Aronson, Reference Carlsmith and Aronson1963; Darley and Gross, Reference Darley and Gross1983; Klaaren et al., Reference Klaaren, Hodges and Wilson1994; Wilson et al., Reference Wilson, Lisle, Kraft and Wetzel1989). The source of such expectations could be one's own experiences, the reports of others, or the presence of certain cues in a particular setting. In the specific domain of foods and beverages, it is well known that expectations formed by exposure to a variety of external cues can enhance or degrade the pre-consumption perception of a product and that such enhancement or degradation can influence subsequent sensory perception during consumption (Cardello, Reference Cardello, MacFie and Thomson1994; Deliza and MacFie, Reference Deliza and MacFie1996). The influence of expectations on sensory perception has been demonstrated with a wide range of foods and beverages, including natural yogurt, soy, liver paté, and beer (e.g., Bowen et al., Reference Bowen, Tomoyasu, Anderson, Carney and Kristal1992; Hurling and Shepherd, Reference Hurling and Shepherd2003; Lee et al., Reference Lee, Frederick and Ariely2006; Schifferstein et al., Reference Schifferstein, Kole and Mojet1999; Tuorila et al., Reference Tuorila, Meiselman, Cardello and Lesher1998; Wansink et al., Reference Wansink, Park, Sonka and Morganosky2000).
The sensory perception of wines is a case in point. Studies in which wines are tasted blind, except for nonsensory cues that are provided to consumers prior to tasting, find that such cues have a substantial impact on tasters' sensory perceptions. Nonsensory cues that have been studied include the wine's price (Plassmann et al., Reference Plassmann, O'Doherty, Shiv and Rangel2008; Veale and Quester, Reference Veale and Quester2008), its country of origin (Veale and Quester, Reference Veale and Quester2008), and the type of label that appears on the bottle (Brochet, Reference Brochet2001). In these studies, consumers perceive that wines they believe are more expensive, to come from France, instead of, for example, Chile, and that wines poured from bottles with Grand Cru labels taste better, even though they are unknowingly tasting the same wine.
In contrast, blind-tasting studies in which no external cue is provided find that consumers are generally unable to distinguish among wines that differ on nonsensory attributes often thought to be important. Examples are Ashton (Reference Ashton2014) and Goldstein et al. (Reference Goldstein, Almenberg, Dreber, Emerson, Herschkowitsch and Katz2008), who studied wines that varied widely in price; Weil (Reference Weil2001), who studied wines from “excellent” vs. “average” vintages; and Weil (Reference Weil2005), who studied wines designated as “reserve” vs. “regular” bottlings, but who did not disclose any of these distinctions prior to tasting.
The present paper reports three studies that, together, exploit what is known about the sensory perceptions of wine consumers when they do and do not have their expectations set by the provision of nonsensory cues prior to tasting. In these studies, three different groups of participants blind-taste the same six red wines, four from California and two from New Jersey, and rate the extent to which they enjoy each. In Study 1 (wine club members) and Study 2 (wine professionals), participants know only the color of the wines prior to tasting—similar to the Ashton (Reference Ashton2014), Goldstein et al. (Reference Goldstein, Almenberg, Dreber, Emerson, Herschkowitsch and Katz2008), and Weil (Reference Weil2001, Reference Weil2005) studies. The results of Studies 1 and 2 provide a “baseline” of ratings for Study 3, in which participants are given a nonsensory cue prior to tasting—similar to the Brochet (Reference Brochet2001), Plassmann et al. (Reference Plassmann, O'Doherty, Shiv and Rangel2008), and Veale and Quester (Reference Veale and Quester2008) studies. Specifically, participants in Study 3 are informed that two of the six wines (but not which two) are from New Jersey. Participants first identify the wines they believe to be from New Jersey and then rate all six wines. If their expectations about the wines, rather than the wines themselves, dominate their sensory perceptions, participants will “downgrade” the wines that they believe to be from New Jersey relative to participants who believe that the same wines are from California—regardless of whether the wines identified as from New Jersey are actually from New Jersey or California.
The contention that wine consumers will downgrade wines they believe to be from New Jersey relative to wines they believe to be from more familiar wine-producing locations such as California takes as given that consumers expect New Jersey wines will not be enjoyable, a notion that seems uncontroversial. Davidson (Reference Davidson2013), for example, discussing the success of some New Jersey winemakers in securing the Outer Coastal Plain (OCP) designation for their wines, says: “The OCP has only one real challenge: It's in southern New Jersey, a state associated with many things—Springsteen, Snooki, industrial pollution, the mob—but not great wine.” A similar point is made in a Morning Edition segment on National Public Radio, which also focused in part on the OCP designation, which is now prominently displayed on many New Jersey wine labels. The segment was given the following title (Smith, Reference Smith2013): “The Trick to Selling Fancy Wine from New Jersey: Don't Say It's from New Jersey.” A third anecdote concerning consumers' expectations about New Jersey wines is provided by one of the participants in Study 1. After blind-tasting the four California and two New Jersey wines (knowing only their color) and rating the extent to which they enjoyed each wine, participants were informed that two of the wines were from New Jersey. One of the participants, without waiting to hear which two wines were from New Jersey, exclaimed: “Nothing good ever came from New Jersey!”—hence the title of this paper.
II. The Judgments of Princeton and Paris
The decision to focus on New Jersey and California wines was inspired by the “Judgment of Princeton,” a blind tasting of New Jersey and French wines held in Princeton, NJ, on June 8, 2012, at the annual meeting of the American Association of Wine Economists (AAWE). The tasting involved nine wine industry professionals (producers, columnists, restaurateurs, etc.) from the United States, France, and Belgium who tasted six New Jersey and four French reds (and six New Jersey and four French whites), rating each on a 20-point scale.Footnote 1 The Princeton tasting was patterned closely after the “Judgment of Paris,” the May 24, 1976, blind tasting of French and California wines that revolutionized the wine world (Taber, Reference Taber2006): In that tasting, nine renowned members of the French wine establishment, to whom California wines were largely unfamiliar, tasted six California and four French reds (and six California and four French whites), rating each on a 20-point scale.
In the Paris tasting, both the top-scoring red wine and the top-scoring white wine were from California. In the Princeton tasting, both the top-scoring red and white wines were from France, but the largely unknown New Jersey wines fared remarkably well against the much more famous (and expensive) French wines.Footnote 2 For example, one of the New Jersey reds, from Heritage Vineyards, placed third in terms of average scores, and one of the New Jersey whites, from Unionville Vineyards, placed second. Therefore, just as the Paris tasting had done 36 years earlier, the Princeton tasting demonstrated that, in blind tastings, wines that are largely unknown can elicit similar overall sensory perceptions as their more familiar counterparts, even among experienced wine tasters. The present studies both extend and deepen research on this issue by pitting New Jersey wines against California wines and by highlighting the influence of pre-tasting expectations.
Two of the four California wines chosen for the present studies are marketed as cabernet sauvignons, which, under California law, requires them to be at least 75 percent from the cabernet sauvignon grape: Heitz Cellar Cabernet Sauvignon Napa Valley 2007 ($40) and Worthy Sophia's Cuvée Cabernet Sauvignon Napa Valley 2007 ($30). The other two California wines are marketed as blends: Orin Swift Papillon Napa Valley 2009 ($70) and Artesa Elements Red Wine Sonoma County 2009 ($20). The two New Jersey wines, both marketed as “Bordeaux Blends,” are Heritage Vineyards Estate Reserve BDX 2010 ($35/$70) and Unionville Vineyards Red Montage 2010 ($30).Footnote 3 The goal in choosing these six wines was to include representative wines that are widely available to consumers.
III. Study 1
A. Method
The procedures involved in Study 1 were very similar to those of the blind tastings reported by Ashton (Reference Ashton2014), and the participants were drawn from the same pool of wine consumers. The tasters were 15 members of the Wine Appreciation Club in the full-time MBA program at the Fuqua School of Business, Duke University. Before they began tasting, the only thing that they knew about each wine was its color. The tasting, conducted in April 2013, lasted about one hour, divided about equally between the tasting itself and a follow-up debriefing.
Participants were instructed to rate the extent to which they enjoyed each wine on a scale from 1 (“I did not enjoy this wine at all”) to 10 (“I enjoyed this wine as much as or more than any red wine I have ever tasted”). Participants were also instructed to indicate what they believed to be the most likely place of origin of each wine by choosing one of the following responses: Argentina, Australia, California, France, Italy, and Other (in which case, a blank was provided for anyone who wished to be more specific).
B. Results
The mean (standard deviation) of enjoyment ratings for each wine is shown in the top half of Table 1, which lists the wines as they were numbered in the tasting (Artesa was labeled number 1, Heitz number 2, etc.). While participants use the entire 10-point response scale (with the exception of 10) to rate the wines, the mean ratings differ relatively little across wines. None of the differences in means even approaches conventional levels of statistical significance. The overall mean is 5.19. More importantly, the mean ratings of the California wines (5.22) and the New Jersey wines (5.14) are almost identical. Thus, in this blind tasting in which nonsensory cues are not provided, participants do not distinguish between the New Jersey and California wines in terms of personal enjoyment.
Table 1 Means (Standard Deviations) of Enjoyment Ratings and Place-of-Origin Judgments of California and New Jersey Wines: Studies 1 and 2
Participants also do not distinguish between the New Jersey and California wines in terms of most likely place of origin, as shown in the bottom half of Table 1. The modal response for the California wines is Argentina, followed by California and France. The modal response for the New Jersey wines is California, followed by Argentina and Australia/Italy. Overall, the results suggest that this group of wine consumers has neither a clear taste preference between these California and New Jersey wines nor a clear sense of the wines' geographic origins.
IV. Study 2
A. Method
A potential concern with Study 1 is that participants are novices with relatively little experience or knowledge in wine tasting. Clearly, both sensory and cognitive mechanisms are involved in wine tasting, both of which likely can be honed with experience (Hughson and Boakes, Reference Hughson and Boakes2001, Reference Hughson and Boakes2002; Morrot, Reference Morrot1999). It is an open question, however, whether greater experience or knowledge will translate into preferences different from those observed in Study 1.
To address this concern, Study 1 was repeated with four experienced wine professionals—the wine buyer for a large gourmet specialty store (with 25 years of experience in the wine business), a retail wine store owner (17 years), a sales representative for a wine distributor who previously had been an assistant sommelier (10 years), and a wine importer/supplier who sells to distributors (2 years). Obviously, caution must be exercised in interpreting findings from such a small number of participants, but their variety and length of experience as wine professionals may nevertheless allow Study 1′s results to be supplemented in a useful way.
Study 2, conducted in July 2013, involved procedures similar to those in Study 1. In addition to rating each wine and indicating what they believed to be its most likely place of origin, the wine professionals were told to write a few words or phrases that they thought best described each of the wines. Participants in Study 1 were not asked to do this because research finds that substantial experience in tasting and communicating about wine is a necessary (though not sufficient) condition for meaningful “wine talk” (e.g., Brochet and Dubourdieu, Reference Brochet and Dubourdieu2001; Gawel, Reference Gawel1997; Lawless, Reference Lawless1984; Lehrer, Reference Lehrer1975, Reference Lehrer1983, Reference Lehrer2009; Solomon, Reference Solomon1990, Reference Solomon1997; Weil, Reference Weil2007).
B. Results
The mean ratings and place-of-origin results are presented in Table 1, alongside the corresponding results from Study 1. The wine professionals rate all six wines more highly than the novices do, by an average of about 1.25 on the 10-point scale.Footnote 4 However, their mean ratings of the California wines (6.50) and the New Jersey wines (6.38) are almost identical, and, as in Study 1, none of the differences in means approaches statistical significance. Moreover, the wine professionals’ place-of-origin beliefs are no more accurate than those of the novices.Footnote 5 Thus, the results in this particular setting do not support the idea that professionals and novices differ in their appreciation for New Jersey vs. California wines.
Descriptors provided by the wine professionals are shown separately for each wine in Table 2. A total of 125 descriptors were provided or an average of about five descriptors per wine per person. I have grouped them into four categories: (1) Fruit, which includes descriptors related to detection of fruits, flowers, and spices in the wine's aroma or taste (52% of all descriptors); (2) Balance, which includes terms related to tannins, acidity, integration, balance, and finish (22%); (3) Oak, which is self-explanatory (10%); and (4) Other, which includes descriptors related to grape variety, Old World vs. New World, and a few terms that are difficult to classify elsewhere (16%).
Table 2 Descriptors Provided by Wine Professionals: Study 2
It is important not to “overinterpret” the data in Table 2, so perhaps the main observation should be that it is difficult to discern major differences across wines in the descriptors provided. However, one thing stands out in Table 2 (at least for me): Substantially more “fruit”-related descriptors are provided for Heritage (15) than for any of the other wines (10, on average); moreover, the variety of such descriptors is greater for Heritage (coffee, vanilla, sage, tea leaf, toast, and moss, in addition to the usual red- and black-fruit descriptors). Thus, these wine professionals perceive a broader range of fruit-related aromas/tastes in this New Jersey wine than in any of the others.
V. Study 3
A. Method
Study 3, conducted in October 2013, involved the same six wines as the earlier studies. Participants were 18 members of the Fuqua Wine Appreciation Club who had not participated in Study 1. For this study, the wines were grouped into two subsets, with two California and one New Jersey wine in each. Thus, Study 3 was designed as two “triangle tests” (Amerine and Roessler, Reference Amerine and Roessler1983; Peynaud, Reference Peynaud and Schuster1987), and the tasters' first task was to identify which wine in each triangle was from New Jersey. Participants were then asked to explain why they believed that the wine they chose in each triangle was the New Jersey wine and finally to rate all six wines on the same 10-point scale used in the earlier studies. The Worthy (CA), Unionville (NJ), and Heitz (CA) wines comprised one triangle, while the Orin Swift (CA), Heritage (NJ), and Artesa (CA) wines comprised the other. Aside from this design change, the tasting procedures were the same as those in Study 1.
The results of Study 1, in which the mean ratings of the California and New Jersey wines are essentially identical in the absence of the “New Jersey cue,” suggest that this new sample of participants from the same population of tasters will have considerable difficulty in correctly identifying the New Jersey wines based solely on sensory perception. Other research, showing that expectations in general strongly influence the interpretation of subjective experiences, as well as the findings of Brochet (Reference Brochet2001), Plassmann et al. (Reference Plassmann, O'Doherty, Shiv and Rangel2008), and Veale and Quester (Reference Veale and Quester2008) in the specific context of wine, suggests that these tasters will “downgrade” their ratings for the wines that they believe (correctly or incorrectly) to be from New Jersey.
B. Results
I first consider the extent to which the tasters correctly identify the New Jersey wine in each of the two triangles. With 18 tasters and three choices, there should be six correct identifications in each triangle by chance alone. This is precisely what occurs. Moreover, simply by chance there should be two tasters who correctly identify both New Jersey wines. However, there is only one taster who does so.
The means and standard deviations of participants' ratings of each wine are presented in Table 3. Overall, the mean ratings in Study 3 are very similar to those in Study 1 (see Table 1). The grand mean across all wines is 5.24 in Study 3, compared to 5.19 in Study 1. The mean ratings for California (New Jersey) wines in Study 3 are 5.25 (5.22), compared to 5.22 (5.14) in Study 1. Again, none of the mean differences approaches statistical significance. Thus, these two samples of tasters from the same population express very similar preferences, on average, across the six wines.
Table 3 Means (Standard Deviations) of Enjoyment Ratings of California and New Jersey Wines: Study 3
The key comparison in Study 3 concerns the ratings of participants who believe that a particular wine is from California vis-à-vis those who believe that the identical wine is from New Jersey. If the participants' expectations about the sensory properties of California and New Jersey wines dominate their sensory experience of the wine itself, wines that they believe to be from California will receive higher ratings than wines that they believe to be from New Jersey, regardless of the wines' actual geographic origins.
This result does indeed obtain. Across all six wines, the mean rating when the wines are believed to be from California (New Jersey) is 5.75 (4.22), a significant difference (t = 4.44; p = .000, one-tailed). Moreover, this result holds without regard to whether the wine is actually from California or New Jersey. California wines correctly (incorrectly) believed to be from California (New Jersey) are rated 5.73 (4.29), a significant difference (t = 3.47; p = .001). Similarly, New Jersey wines incorrectly (correctly) believed to be from California (New Jersey) are rated 5.79 (4.08), again a significant difference (t = 2.70; p = .005).Footnote 6
The reasons that participants provide for identifying a particular wine in each triangle as being from New Jersey are shown in Table 4. With 18 tasters and two New Jersey wines, there are 36 entries in Table 4. Twelve of the 36 entries correspond to the wines that actually are from New Jersey, while the other 24 correspond to the wines that are from California.
Table 4 Reasons Provided for Identifying Each Wine as Being from New Jersey: Study 3
Many participants say their reason for identifying a particular wine as being from New Jersey is that it is their least favorite or that they simply do not like it, with several stating explicitly that they do not believe good wine is produced in New Jersey. The following examples (from Table 4) are typical:
It is stronger and more acid. I believe good wines are not from New Jersey.
I don't like it very much and I have an image of New Jersey as not very strong.
It seems “weaker” for me, less dominant flavor, missing scent/smell, the alcohol a little bit odd. I assume the weather in NJ is not so ideal, the landscape neither, and maybe the soil is weak, too.
Clearly, these are statements about the participants' expectations concerning New Jersey wines. However, these types of reasons are attached to wines that are actually from California as often as to wines from New Jersey.
Other participants provide specific reasons for not liking the wines that they identify as being from New Jersey, often describing various faults they perceive (e.g., off taste, diluted taste, unnatural aftertaste, smelly, immature grapes, tastes like something other than grapes). Still others, while not mentioning faults per se, describe ways in which they perceive their chosen wines to differ from the others (e.g., sweet, bitter, sour, stronger, weaker, duller, earthiest, least earthy). Again, however, such descriptors are attached to wines that are actually from California as often as to wines from New Jersey.
VI. Discussion and Conclusion
The studies reported here demonstrate the influence of expectations on the sensory perception of wines. Study 1, involving wine club members, finds no difference in the mean enjoyment ratings of a sample of New Jersey and California wines when tasters know only that they are tasting red wines. Study 2 replicates this result with experienced wine professionals. In contrast, Study 3, in which tasters from the same wine club are informed that some of the wines are from New Jersey and some are from California, finds that when wines are believed to be from New Jersey they receive significantly lower mean ratings than when the same wines are believed to be from California, regardless of the wines' actual place of origin.
In Study 3, the average magnitude of the ratings difference when wines are believed to be from California vs. New Jersey is about 1.5 points (5.75 vs. 4.22) on the 10-point response scale employed. I am not aware of direct evidence that bears on the “practical significance” of a difference of this magnitude. At an intuitive level, however, it seems reasonable to maintain that a difference equal to 15 percent of the response scale is likely to be important for both wine producers and wine consumers. If this is correct, then the question for producers is how to achieve greater awareness and acceptance of their wines vis-à-vis those of better-known competitors, and the question for consumers is whether (and how) they wish to experiment with wines for which their expectations may be low.
From the producers' standpoint, the research literature on consumer learning may be helpful, especially the literature on learning from first-hand or direct experience (as opposed to learning from education, i.e., second-hand, indirect learning based on verbal or written descriptions). Hoch and Deighton (Reference Hoch and Deighton1989) provide a comprehensive account of the process of consumer learning from experience, as well as tactics that producers can use to influence various points in the process. Effective tactics depend on whether the product is a “top dog” brand (well established with substantial market share) or an “underdog” brand (relatively unknown with low market share); in general, underdogs have much to gain and little to lose by encouraging and facilitating consumer learning from direct experience, whereas top dogs benefit by impeding such learning.
The problem for underdogs is that consumer learning from experience can be slow. Hoch and Deighton (Reference Hoch and Deighton1989) identify three determinants of the effectiveness of learning from experience: familiarity, motivation, and ambiguity. Consumers who are unfamiliar with a particular domain may be more easily persuaded to try something new, but learning from experience can be difficult because they do not have well-developed knowledge structures for assimilating what they are learning. Motivation to learn is a necessary condition for learning but entails the possibility that novice consumers may not know what they don't know, while more experienced consumers may think that they already know all they need to know. Finally, learning from experience can be problematic in ambiguous information environments, for example, settings in which quality has a large subjective component (as in wines). In spite of these difficulties, certain tactics have been found effective for underdog brands, including comparative advertising, side-by-side displays or sales promotions, blind taste tests, personal selling, targeting small groups of customers who are considered opinion leaders, and developing alternative distribution channels such as “home parties,” where customers are exposed only to the underdog brand.
As mentioned earlier, some New Jersey winemakers have already pursued tactics aimed at changing the underdog status of their wines, such as securing the OCP designation to differentiate their more “serious” wines from traditional New Jersey wines and publicizing the results of blind tastings of their offerings vis-à-vis more established wines. Overcoming the negative New Jersey wine stereotype will undoubtedly be extremely challenging given the wide availability of high-quality, moderately priced wines with which consumers are already familiar. However, a similar statement regarding California wines could have been made some years ago—before the Judgment of Paris forever changed consumers' expectations.
To the extent that underdog wines are successful at overcoming consumers' “lethargy and lack of curiosity” (Hoch and Deighton, Reference Hoch and Deighton1989, 11), consumers face a dilemma: how to allocate their resources between repeatedly consuming wines that are familiar and for which expectations are high and experimenting with wines that are unfamiliar and for which expectations may be low. Those who try to do both may find the right balance difficult to achieve.
At a broader level, this same dilemma is analyzed by James March (Reference March1991) in a classic paper on organizational learning. March describes two sets of adaptive processes that are pursued by organizations—“the exploration of new possibilities and the exploitation of old certainties” (March, Reference March1991, 71). March notes that exploration involves search, variation, experimentation, discovery, and risk taking, while exploitation involves efficiency, implementation, execution, and refinement. Both exploitation and exploration are essential for organizations, but they compete for limited resources—the more an organization engages in one, the less it engages in the other. Achieving an optimal balance between the two is challenging, with the result that exploitation often drives out exploration. In a similar way, wine consumers may favor re-experiencing the familiar over exploring the unfamiliar unless something disrupts their expectations.
