Abstract
Many marketing communications are carefully designed to cast a brand in its most favorable light. For example, marketers may prefer to highlight a brand's membership in the top 10 tier of a third-party list instead of disclosing the brand's exact rank. The authors propose that when marketers use these types of imprecise advertising claims, subtle differences in the selection of a tier boundary (e.g., top 9 vs. top 10) can influence consumers’ evaluations and willingness to pay. Specifically, the authors find a comfort tier effect in which a weaker claim that references a less exclusive but commonly used tier boundary can actually lead to higher brand evaluations than a stronger claim that references a more exclusive but less common tier boundary. This effect is attributed to a two-stage process by which consumers evaluate imprecise rank claims. The results demonstrate that consumers have specific expectations for how messages are constructed in marketing communications and may make negative inferences about a brand when these expectations are violated, thus attenuating the positive effect such claims might otherwise have on consumer responses.
Brands are often ranked in lists published by third parties (e.g., Fortune 500, Princeton Review, Travel & Leisure, ESPN), and these rankings can affect consumer decisions. For example, by improving their rank in U.S. News & World Report, colleges can attract better students (Griffith and Rask 2007; Meredith 2004) and hospitals can attract more patients (Pope 2009). As a result, marketers often use blogs, press releases, and other promotional materials to highlight a brand's inclusion in a third-party ranked list.
However, advertising claims do not always refer to the brand's exact rank. Instead, claims often simply communicate the brand's membership in an exclusive tier along with other top brands. To illustrate, marketers may emphasize a brand's membership in the top 10 tier rather than reporting its actual rank on the list. An analysis of master of business administration programs found that 72% of schools that display a ranking on the business school's website shroud details or coarsen the information (e.g., by citing tier membership rather than exact rank) to make it seem more favorable (Luca and Smith 2015). Such tactics raise the question of how consumers respond to imprecise rank claims, and under what conditions (if any) such claims are superior to other options. For example, when presented with an imprecise rank claim, do consumers make inferences about a brand's exact rank and, if so, what impact do these inferences have on brand evaluations?
Despite the prevalence of imprecise advertising claims that tout a particular brand's inclusion in a ranked list without specifying the brand's exact rank, the issue of how consumers respond to such claims has received little attention. Prior research on how rank information is processed has tended to assume that consumers are exposed to the entire list of ranked items and are aware of each brand's exact rank (Isaac and Schindler 2014; Leclerc, Hsee, and Nunes 2005; Luca and Smith 2013). In contrast, we examine how consumers respond to the common marketing tactic of specifying a tier in which the brand has been included (e.g., top 10) without providing consumers with the complete list or disclosing the advertised brand's exact rank.
When using an imprecise rank claim to communicate the honor of a brand's inclusion on a third party's ranked list, marketers must decide the specific tier boundary that they report. For example, a company whose mobile handset was ranked number 9 on a computer magazine's list could legitimately report that they were ranked among the top “X” brands on the list, where X is 9 or any number greater than 9. In this article, we propose that subtle differences in the selection of a tier boundary can affect both brand evaluations and consumers’ willingness to pay.
For example, which claim is likely to produce more favorable consumer evaluations—top 9 or top 10? Given the similarity of these claims, one possibility is that brand evaluations following a top 9 claim and a top 10 claim will be statistically indistinguishable. However, because a top 9 claim is more exclusive and eliminates the least favorable possibility that a brand's rank is exactly 10th, it might produce more positive brand evaluations.
In contrast to the prediction that more exclusive claims are as good as or better than less exclusive claims, we propose that claims referencing certain round numbers (e.g., top 10) can sometimes result in higher evaluations than more exclusive claims (e.g., top 9) that do not. We refer to this phenomenon as the “comfort tier effect” and define a comfort tier as a tier whose boundary is a commonly used round number. Prior research has shown that certain numbers, such as 5, 10, 20, 25, 50, and 100, are frequently used when making “approximative” rather than exact numerical expressions (Sigurd 1988). These comfort tier claims (e.g., “Top 10,” “Best 25”) can be contrasted with non–comfort tier claims (e.g., “Top 9,” “Best 27”), which use less common tier boundaries.
Our prediction that comfort tier claims may lead to more favorable brand evaluations than slightly more exclusive non–comfort tier claims is based on the notion that people typically expect comfort tiers to be used in marketing communications. Indeed, prior research has suggested that round numbers dominate verbal and written communication (Coupland 2010; Dehaene and Mehler 1992) to such an extent that the use of nonround numbers, particularly in imprecise expressions, may be considered unnatural (Sigurd 1988). In fact, the usage of numbers ending in zero and five as common cognitive reference points (Isaac, Wang, and Schindler 2016) transcends cultural boundaries (Jansen and Pollmann 2001) and may stem from an early reliance on ten fingers (five on each hand) when learning to count and perform basic arithmetic (Ho and Cheng 1997; Siegler and Robinson 1982; Yoshida and Kuriyama 1986).
We attribute the comfort tier effect to a two-stage process by which consumers exposed to imprecise rank claims evaluate a brand. Stage 1 processing focuses on the top X tier as a whole, whereas stage 2 processing focuses on the brand's likely rank within the tier and on how the advertised brand compares with other brands within the tier. To illustrate, consider a consumer who is looking for a restaurant to visit in an unfamiliar city and sees an advertisement claiming that a particular restaurant has been ranked as one of the top X in the city. His or her first reaction may be to broadly consider the exclusivity of the top X tier as a whole, which we refer to as “stage 1 processing.” During stage 1, the consumer is likely to generate a favorable impression because the restaurant has been included in an elite group. This is consistent with prior research suggesting that being recognized or honored by an independent third party will enhance a brand's evaluation (Balasubramanian, Mathur, and Thakur 2005; Dean and Biswas 2001). Consumers who focus on tier exclusivity may not even recognize that the claim was imprecise—this is consistent with prior research suggesting that consumers sometimes fail to recognize other types of strategic nondisclosure practices (Brown, Camerer, and Lovallo 2013). If no further processing occurs, seeing a top X claim should improve brand evaluations, and this positive effect is likely to be more pronounced for claims that reference a more exclusive tier (i.e., when X is a smaller number).
However, in certain cases, consumers may also engage in a second stage of processing in which they try to interpret what is implied by the way the marketer constructed the advertisement. As Grice's (1975, 1978) seminal work on conversational implicature suggests, message recipients often attempt to interpret meaning from communication beyond what is strictly denoted by the language, particularly if an expectancy violation occurs (Hilton 1995; Miller and Kahn 2005; Schwarz 1996; Sperber and Wilson 1986). In the case of imprecise rank claims, a potentially useful cue for making inferences is the tier selected by the communicator (for example, top 10 vs. top 20). We propose that, when considering what is implied by the selected tier, consumers who engage in stage 2 processing will perceive not only that an exclusive tier was selected but also that a more exclusive comfort tier was not selected. For example, a top 10 claim implies that the brand was probably not in the top 5, or else a more exclusive tier would likely have been referenced (particularly if the communicator is thought to have a persuasion motive, but this applies even in nonpersuasive communication). Thus, we anticipate that a consumer who engages in stage 2 processing will infer that the advertised brand's actual rank is likely to be less favorable than the next more-exclusive comfort tier. We propose that, as a result of this comparison to more favorably ranked brands, stage 2 processing attenuates the positive effects on brand evaluations that otherwise occur on exposure to an imprecise rank claim.
The comfort tier effect, in which a less exclusive claim (e.g., top 10) results in better evaluations than a more exclusive claim (e.g., top 9), is attributed to the notion that consumers sometimes undertake only stage 1 processing while at other times they also proceed to stage 2. We propose that, as with many other two-stage models of cognitive processes (Campbell and Kirmani 2000; Chaiken and Maheswaran 1994; Gilbert, Tafarodi, and Malone 1993; Petty, Cacioppo, and Schumann 1983), consumers must be prompted or motivated to undertake both stages. In this article, we focus primarily on how a violation of expectations can prompt consumers to undertake stage 2 processing (Ahluwalia and Burnkrant 2004; Meyer et al. 1991; Petty et al. 2001; Wilson and Gilbert 2008). More specifically, we predict that when marketing claims reference a comfort tier, consumers will normally terminate processing after stage 1. Without any additional impetus to proceed to stage 2, consumers will consider the exclusivity of the claim and evaluate the brand more favorably than if they had not been exposed to the claim.
However, when marketing claims reference a non–comfort tier (e.g., top 9), we predict that the unexpected tier boundary will prompt consumers to proceed to stage 2 processing and to think about the brand's likely rank within the tier. The resulting inference that the brand's rank is likely to be outside the next more exclusive comfort tier (and therefore worse than other brands in that tier) may attenuate the otherwise positive effects of the claim. Accordingly, because we predict that non–comfort tiers are more likely than comfort tiers to trigger stage 2 processing, we expect brand evaluations to be more favorable in response to a comfort tier claim (e.g., top 10) than a proximal non–comfort tier claim (e.g., top 9), even when the comfort tier is less exclusive. For a diagram of our proposed two-stage process, see Figure 1.
TWO-STAGE PROCESS MODEL ILLUSTRATING HOW IMPRECISE RANK CLAIMS IMPACT BRAND EVALUATIONS
We test these predictions in a series of six experiments. Study 1 aims to document the predicted comfort tier effect by comparing brand evaluations following exposure to a comfort tier claim versus a non–comfort tier claim. Study 2 tests whether this effect is attenuated when consumers exposed to a comfort tier claim are explicitly prompted to engage in stage 2 processing (thus decreasing their brand evaluations). Study 3 tests whether the comfort tier effect is attenuated when consumers exposed to a non–comfort tier claim are prevented from engaging in stage 2 processing (thus increasing their brand evaluations). Study 4 tests our claim that stage 2 processing is invoked by a violation of expectations and examines whether comfort tiers can be “decomfortized” in situations in which they violate expectations (and whether non–comfort tiers can likewise be “comfortized” in situations in which they meet expectations). Study 5 tests whether contextual factors that maintain consumers’ focus on tier exclusivity can offset the negative inferences about rank brought about by stage 2 processing following exposure to a non–comfort tier. Finally, Study 6 directly tests our claim that stage 1 processing involves a focus on tier exclusivity.
Study 1
This study aims to document the predicted comfort tier effect by comparing brand evaluations following exposure to a comfort tier claim versus a non–comfort tier claim. Specifically, we predict that an advertisement containing a comfort tier claim (best 10) will result in higher brand evaluations than an advertisement containing a proximal non–comfort tier claim, even one that is more exclusive (best 9). As a control, we also expose a third group of participants to an advertisement without a rank claim. This benchmark enables us to test the basic proposition that a comfort tier claim improves evaluations, as well as to calibrate the extent to which a non–comfort tier attenuates these more positive ratings.
In this study, we also begin to test the proposed process underlying the comfort tier effect by including a fourth condition in which the advertisement reveals the brand's exact rank. If we are correct that consumers exposed to a non–comfort tier claim infer that the brand's rank is likely to be outside the next more exclusive comfort tier (and therefore worse than other brands in that tier), then a best 9 brand should be evaluated at least as favorably as a brand that is known to be ranked exactly at the tier boundary (i.e., 9th).
To further test our proposed two-stage process model, at the conclusion of the study we also asked all participants exposed to an imprecise rank claim (i.e., best 9 or best 10) to estimate the likely rank of the advertised brand. Although we expect that those exposed to a non–comfort tier (best 9) will be prompted to engage in stage 2 processing because of the unexpected nature of the claim, this question was designed to explicitly trigger stage 2 processing (if it had not already occurred). We have proposed that those who engage in stage 2 processing will infer that the brand's rank is likely to be outside the next more-exclusive tier, which for this study is best 5. If our model is correct, most participants exposed to a best 9 or best 10 claim will therefore estimate a rank that is less favorable than 5.
Method
Participants were 284 Americans (59.9% female; mean age = 38.04 years, SD = 12.72) recruited using Amazon Mechanical Turk (MTurk) who were randomly assigned to one of three treatment conditions (comfort tier, non–comfort tier, or exact rank) or a control condition. Participants in all four conditions were shown an advertising message for an actual bank (Raiffeisen Bank) that they were unlikely to be familiar with because it does not operate in the United States. In the three treatment conditions, the advertising message included a claim that the bank had been ranked by a third party (the fictional International Bank Review) as one of the 10 best banks (comfort tier), one of the 9 best banks (non–comfort tier), or the 9th best bank (exact rank) in Eastern Europe. In the control condition, no rankings were mentioned in the ad. All participants then evaluated the bank on three items using slider scales that ranged from 0 (“strongly disagree”) to 100 (“strongly agree”). The three items were (1) “If I needed to open a local bank account in Eastern Europe, Raiffeisen Bank would be an excellent choice,” (2) “I would feel confident banking at Raiffeisen Bank,” and (3) “I believe that Raiffeisen is probably one of the best places to put my money.” These three items (㬁 = .956) were subsequently averaged to form a composite brand evaluation variable. On a subsequent screen, participants were asked to provide the rationale for their bank evaluations in an open-ended text box. Participants in the two imprecise rank claim conditions (best 9 and best 10) were then prompted to estimate Raiffeisen's exact rank in the third-party list (participants could not proceed until they provided an estimate that was within the claimed tier). As an attention check, participants were shown all three claims and a “none of the above” option and asked to select the claim they had previously seen. Finally, after completing several unrelated studies, all participants indicated their gender and age. The complete stimuli for Study 1 and for all subsequent studies appear in the Web Appendix.
Results
To test for the comfort tier effect, participants’ composite brand evaluations were subjected to an analysis of variance (ANOVA) on experimental condition. The analysis was based on evaluations from the 223 participants (58.74% female; mean age = 38.08 years, SD = 12.66) who passed the attention check. This ANOVA revealed that participants’ brand evaluations differed significantly across conditions (F(3, 219) = 17.84, p < .001, ηp2 = .196). As we predicted, planned contrasts showed a comfort tier effect in which brand evaluations for the comfort tier claim (Mbest10 = 76.63, SD = 18.74, N = 68) were significantly higher than evaluations for the more exclusive non–comfort tier claim (Mbest9 = 64.76, SD = 19.85, N = 47; F(1, 219) = 10.88, p < .001). Both of these imprecise rank claims resulted in higher brand evaluations than the control; evaluations when no rank claim was given (Mcontrol = 53.39, SD = 18.14, N = 50) were significantly lower than evaluations for the comfort tier claim (F(1, 219) = 43.24, p < .001) and the non–comfort tier claim (F(1, 219) = 8.71, p < .01). In another finding consistent with our theorizing, the exact rank claim (M9th = 57.10, SD = 19.20, N = 58) led to lower evaluations than both the comfort tier claim (F(1, 219) = 33.17, p < .001) and the non–comfort tier claim (F(1, 219) = 4.24, p = .041), but not the control (F(1, 219) = 1.03, p > .31). Figure 2 illustrates these results.
COMFORT TIER CLAIMS RESULT IN HIGHER BRAND EVALUATIONS (STUDY 1)
The attention check we used was a pre-established exclusion criterion for this study. However, given that this check generated a large number of exclusions, we conducted another ANOVA on the full sample of 284 participants. Our results remained the same: using the full sample, brand evaluations again differed significantly across conditions (F(3, 280) = 10.55, p < .001, ηp2 = .102), and the comfort tier effect was still observed such that brand evaluations for the comfort tier claim (Mbest10 = 74.25, SD = 20.91, N = 76) were significantly higher than evaluations for the more exclusive non–comfort tier claim (Mbest9 = 65.30, SD = 18.32, N = 66; F(1, 280) = 7.25, p < .01). In reporting the remaining results, we follow our pre-established exclusion criterion by focusing on the sample of participants who passed the attention check. However, the pattern of results is identical irrespective of whether the full sample or the restricted sample is used.
Next, we examined participants’ estimates of actual rank to determine whether the estimates were indeed less favorable than the next more-exclusive comfort tier. As we expected, among participants exposed to the 9 best and 10 best claims, a large majority (70.2% and 63.2%, respectively) estimated the brand's rank to be outside the best 5. It is important to note that although we predict that participants who are prompted to provide rank estimates will estimate values outside the next more-exclusive comfort tier, we do not necessarily assume that participants who encounter a non–comfort tier condition will estimate the lowest possible rank. This may help explain why the non–comfort tier claim resulted in higher evaluations than the exact rank claim. These results, along with a comparison of participants’ mean rank estimates, appear in the Web Appendix for Study 1 and for all subsequent studies in which participants provided rank estimates.
The rationales that participants provided for their responses in Study 1 (and in subsequent studies) provided qualitative insight on how consumers approached imprecise rank claims, and helped us considerably in validating our proposed process model in another way. For example, comments from participants in Study 1 such as “It said it was one of the nine best banks but did not say how high on that list it is” and “I wonder which bank is number 1” were consistent with our proposal that people consider a brand's exact rank during stage 2 processing.
Discussion
The results of Study 1 identify conditions in which, counterintuitively, brand evaluations are boosted more by a weaker claim than by a stronger claim. Consistent with our proposed model, the results of Study 1 also provide initial evidence that stage 2 processing leads to negative inferences about a brand's exact rank. In particular, participants who were prompted to estimate the brand's exact rank tended to guess a number between 6 and 10, suggesting an inference that, if the actual rank had been more favorable, a best 5 claim would have been used instead. Although not central to our theorizing, the finding that a non–comfort tier claim resulted in higher evaluations than an exact rank claim suggests that any ambiguity about exact rank may help brand evaluations, even when the claim leads to an expectation that the exact rank is likely to be relatively unfavorable within the tier.
Our theory asserts that engaging in stage 2 processing attenuates the positive impact of an imprecise rank claim on brand evaluations because it induces relatively unfavorable rank estimates and comparisons to better-ranked brands within the tier. In support of the notion that believing a brand's rank to be relatively unfavorable within an exclusive tier can negate the positive effect of a rank claim, participants who were told that the brand's rank was exactly 9th provided evaluations that were no higher than those of participants in the control condition.
We argue that, taken as a whole, these study results support our assertion that only those exposed to the non–comfort tier (best 9) considered exact rank before providing their evaluations. However, one might argue that those exposed to the comfort tier (best 10) also considered exact rank before providing evaluations, and that they formed more favorable evaluations because they estimated a more favorable rank. Some evidence to counter this alternative explanation comes from the mean rank estimates provided by participants exposed to a best 10 claim (Mbest10 = 6.60, SD = 2.64, N = 68) versus a best 9 claim (Mbest9 = 6.98, SD = 2.38, N = 47), which were not significantly different from each other; t(113) = .78, p > .43. However, participants estimated these ranks after (vs. before) providing their evaluations, and the means are directionally supportive of the alternative explanation. So, in the next study, we test the viability of this alternative explanation by prompting some participants to estimate rank before providing their evaluations. If we are correct that exposure to a comfort tier claim focuses attention on tier exclusivity and does not typically lead to stage 2 processing, then prompting (vs. not prompting) consumers to estimate rank should result in lower brand evaluations. However, explicitly prompting (vs. not prompting) consumers to estimate rank following exposure to a non–comfort tier should have little effect on evaluations because these consumers would have already engaged in Stage 2 processing and estimated rank when they encountered the non–comfort tier claim. In Study 2, we test these predictions.
Study 2
Study 2 tests whether the comfort tier effect is attenuated when consumers exposed to a comfort tier claim are explicitly prompted to engage in stage 2 processing (thus decreasing their brand evaluations). We argue that stage 2 processing occurs naturally following exposure to a non–comfort tier claim but does not normally occur (unless explicitly prompted) following exposure to a comfort tier claim. According to our theorizing, those exposed to a non–comfort tier claim are naturally prompted to think about the brand's rank within the tier, which encourages them to infer that the rank is likely to be outside the next more-exclusive tier and therefore to evaluate the brand less favorably. Our theorizing also suggests that those exposed to a comfort tier claim do not engage in this processing unless explicitly prompted. To test this aspect of our theory, we expose participants to an imprecise rank claim, and then prompt some participants to engage in stage 2 processing by directly asking them to consider exact rank before they provide evaluations. For those exposed to a comfort tier claim, we predict that the explicit prompt to engage in stage 2 processing will mitigate the positive effect of exposure to a comfort tier that we demonstrated in Study 1. For those exposed to a non–comfort tier claim, we predict that an explicit prompt to engage in stage 2 processing will have little effect on evaluations. This is because we expect non–comfort tier claims to invoke stage 2 processing on their own, even without an explicit prompt. Therefore, explicitly encouraging stage 2 processing after exposure to a non–comfort tier claim should have no additional effect.
Method
Participants were 230 Americans (56.5% female; mean age = 35.48 years, SD = 12.62) recruited using MTurk who were randomly assigned to one of four conditions. This study employed a 2 (claim tier: comfort tier vs. non–comfort tier) × 2 (explicit rank estimation: prompted vs. unprompted) between-participants design. Participants were instructed to imagine that they were considering pursuing a master of business administration degree and were introduced to a fictional university, Eastmoor University. Participants encountered an advertisement that included either a comfort tier claim or a non–comfort tier claim: “It's no wonder that Eastmoor University was RANKED IN THE TOP 100 [TOP 92] by Bloomberg Businessweek.”
Following exposure to the advertisement, participants in the prompted rank estimation condition proceeded to another screen where they were asked to estimate the exact rank of Eastmoor University in the Bloomberg Businessweek list of top business schools by typing a number into a text-entry box (participants could not proceed until they provided an estimate that was within the claimed tier). Participants in the unprompted rank estimation condition were not asked this question. The time participants spent viewing the ad and estimating the rank (if applicable) was surreptitiously recorded.
Subsequently, all participants were asked to evaluate Eastmoor University by responding to two brand evaluation items. Participants provided an overall assessment of the university (0 = “extremely negative,” and 100 = “extremely positive”) and indicated their likelihood of attending Eastmoor (0 = “extremely unlikely,” and 100 = “extremely likely”) on the basis of the advertisement. As in Study 1, we averaged the scale items to create a composite brand evaluation variable (㬁 = .880). As an attention check, participants were asked to recall the tier stated in the ad (92 vs. 100) by typing a number from 1 to 100 into a text-entry box. Finally, participants indicated their gender and age. Unlike Study 1, we did not ask participants to report familiarity because we had already communicated to them that Eastmoor University was fictional.
Results
We excluded 9 participants who failed the attention check, leaving 221 participants (56.6% female; mean age = 35.57 years, SD = 12.61) for the analysis. A 2 × 2 between-participants ANOVA revealed a main effect of explicit rank estimation (F(1, 217) = 5.15, p = .024, ηp2 = .02) and a marginal effect of tier claim (F(1, 217) = 2.85, p = .093, ηp2 = .01). In a finding more relevant to our theorizing, we observed a significant interaction effect of rank estimation and tier claim on brand evaluations (F(1, 217) = 4.40, p = .037, ηp2 = .02).
As we predicted, when participants encountered a comfort tier claim, those who had not been prompted to estimate rank evaluated Eastmoor University more favorably (Munprompted100 = 61.83, SD = 19.96, N = 55) than those who had been explicitly prompted to estimate rank (Mprompted100 = 49.22, SD = 21.24, N = 59; F(1, 217) = 9.84, p < .01). However, among participants who encountered a non–comfort tier claim, evaluations of participants who were not prompted (Munprompted92 = 50.90, SD = 21.88, N = 54) did not differ from evaluations of those explicitly prompted to estimate rank (Mprompted92 = 50.40, SD = 22.67, N = 53; F(1, 217) = .01, p > .90).
Stated differently, the comfort tier effect observed in Study 1 was replicated in this study's unprompted rank estimation condition; brand evaluations for the comfort tier claim were significantly higher than evaluations for the more exclusive non–comfort tier claim (F(1, 217) = 7.08, p < .01). Furthermore, the comfort tier effect was attenuated when rank estimation was explicitly prompted; brand evaluations for the comfort tier claim did not differ from evaluations for the non–comfort tier claim (F(1, 217) = .08, p > .77). We illustrate these results in Figure 3.
RANK ESTIMATION ATTENUATES THE COMFORT TIER EFFECT (STUDY 2)
An important element of our theorizing is that when consumers consider a brand's likely rank after exposure to an imprecise rank claim, they tend to generate estimates outside the next more-exclusive comfort tier. Study 2's results replicate the results of Study 1 and are consistent with this theorizing; among participants in the top 92 and top 100 conditions, a large majority (90.6% and 81.4%, respectively) estimated the brand's rank to be outside the top 50.
Because stage 2 processing is an additional step beyond stage 1, it should require more cognitive resources. Reviewing a comfort tier claim might therefore be expected to take less time than reviewing a non–comfort tier claim and subsequently result in slower rank estimates (because stage 2 processing had not previously occurred). Due to skewness in the distributions of reading time (skewness = 1.94, SD = 2.43, p < .001) and rank estimation time (skewness = 2.51, SD = 2.41, p < .001), we log-transformed both measures prior to analyses and report means and standard deviations after reversed transformation. Consistent with this expectation, participants in the comfort tier condition took less time to review the ad (Mtop100 = 19.56 seconds, SD = 13.79, N = 111) than participants in the non–comfort tier condition (Mtop92 = 22.05 seconds, SD = 16.26, N = 106), though this effect was not significant (F(1, 215) = 1.43, p > .23). The time required to estimate rank (among participants who were prompted to do so) also did not differ significantly by condition (Mtop100 = 14.79 seconds, SD = 10.80, N = 58 vs. Mtop92 = 15.73 seconds, SD = 11.13, N = 53; F(1, 109) = .31, p > .58). Although significant differences in both response times would have provided further support for our two-stage model, response latency measures can be noisy and “extraordinarily messy” (Fazio 1990, p. 75). In Study 3, we more clearly demonstrate that stage 2 processing requires more cognitive resources.
Discussion
In this study, we explicitly prompted some participants to estimate rank following exposure to a comfort tier claim or a non–comfort tier claim. Consistent with our theorizing and the results of Study 1, the effect of a comfort tier claim on brand evaluations was more positive than the effect of a more exclusive non–comfort tier claim. Study 2 also shows that directly prompting stage 2 processing can attenuate this positive effect, making responses to a comfort tier claim statistically indistinguishable from responses to a non–comfort tier claim. We argue that these findings are due to the fact that unprompted participants exposed to the comfort tier claim thought primarily about the exclusivity of the tier as a whole, whereas participants exposed to a non–comfort tier claim or prompted to engage in stage 2 processing made negative inferences about the brand's exact rank relative to other brands in the tier.
In contrast to Study 1, which manipulated the claim tier using adjacent numbers (9 vs. 10), Study 2 manipulated the claim tier using nonadjacent numbers (92 vs. 100). This highlights the robustness of the comfort tier effect. Compared with a top 100 claim, a top 92 claim excludes eight of the least favorable possibilities for a brand's exact rank and is therefore more exclusive. Despite this difference in exclusivity, Study 2 replicated the comfort tier effect (for unprompted participants), showing that it holds for both adjacent and nonadjacent tier boundaries. Moreover, although non–comfort tiers are sometimes odd numbers, Study 2 shows that a non–comfort tier can also be an even number.
According to our theory, the comfort tier effect is driven by the idea that some consumers terminate processing after stage 1, whereas other consumers engage in stage 2 processing. Study 2 provides evidence that the comfort tier effect can be attenuated when consumers exposed to a comfort tier claim are explicitly prompted to engage in stage 2 processing (thus lowering their brand evaluations relative to unprompted consumers exposed to the same claim). With the next study, we examine whether the comfort tier effect can be attenuated when consumers exposed to a non–comfort tier are prevented by a time constraint from engaging in stage 2 processing (thus producing higher brand evaluations relative to unhindered consumers exposed to the same non–comfort tier claim).
Study 3
In Study 3, we use a time pressure manipulation to test our claim that those exposed to a non–comfort tier claim are naturally prompted to undertake stage 2 processing and consider rank. We anticipate that the time constraint will decrease consumers’ ability to progress to stage 2 after exposure to a non–comfort tier claim, despite the notion that the claim naturally prompts them to do so. We therefore aim to show that although the comfort tier effect can be replicated among consumers who have cognitive resources available, it is attenuated by time pressure because those exposed to a non–comfort tier do not have the cognitive resources to proceed to stage 2.
A data pattern showing this effect would also address an alternative explanation for our results thus far. It could be argued that the negative effect we have documented in response to non–comfort tiers is not the result of systematic consideration of rank but is rather a heuristic or automatic negative reaction to a nonnormative tier claim (e.g., unusual = bad). If that explanation is valid, the comfort tier effect should not be affected by restraining cognitive resources because heuristic processing involves minimal cognitive effort and is therefore not sensitive to cognitive constraint (Chaiken 1980). However, if time pressure has an effect on the comfort tier effect, this would provide evidence consistent with our argument that stage 2 processing involves more systematic processing of imprecise rank claims and a consideration of exact rank.
Method
Participants were 450 Americans (59.6% female; mean age = 30.08 years, SD = 9.12) recruited using MTurk who were randomly assigned to one of four conditions. This study employed a 2 (claim tier: comfort tier vs. non–comfort tier) × 2 (time pressure: high vs. low) between-participants design. All participants read an introductory paragraph explaining that approximately 40 million Americans go camping each year, and that to find an ideal campsite, many campers consult lists published by travel experts (e.g., the Travel Channel) that rank different campgrounds. Participants were told that they would see a promotional advertisement produced by a particular campground. The advertisement depicted a campground with either a comfort tier (top 100) or a non–comfort tier (top 92) claim that said, “Cades Cove ranked in the Top 100 [92] Travel Channel Family Campgrounds.”
We manipulated time pressure by instructing participants in the high time pressure condition to review the advertisement as quickly as possible because the screen would advance automatically after only two seconds. In contrast, participants in the low time pressure condition were instructed to take as long as they would like to review the advertisement and reflect on it. This manipulation, which was modeled after prior research (e.g., Chu and Spires 2001), has several advantages in our research context over another commonly used type of cognitive load manipulation in which participants try to perform a task while keeping a long versus short string of numbers in mind (e.g., Shiv and Fedorikhin 1999). First, given that our primary manipulation of comfort versus non–comfort tier claims involves numeric processing, it enabled us to constrain cognitive resources without introducing other numbers that could interfere with the comfort tier effect. Second, it enhances the study's ecological validity, because consumers often view ads under time pressure (e.g., while driving past a billboard) but may less frequently view ads while attempting to recall a string of numbers.
Following exposure to the advertisement, participants were asked to evaluate the brand using two scale items (“Cades Cove Campground is a truly exceptional family campground,” and “If planning a camping trip, I would definitely choose Cades Cove Campground over other family campgrounds”), both measured on unnumbered sliders ranging from 0 (“strongly disagree”) to 100 (“strongly agree”). We averaged these two scale items to create a composite brand evaluation variable (㬁 = .874). As an attention check, participants were asked to recall the tier stated in the ad (92 vs. 100) by typing a number from 1 to 100 into a text-entry box. Participants were also asked to estimate the exact rank of the campground in the Travel Channel's list of Top Family Campgrounds by typing a number from 1 to 100 into a text-entry box. Finally, participants indicated whether they had heard of Cades Cove Campground prior to the study, as well as their gender and age.
Results
We excluded 17 participants who were familiar with the campground, as well as an additional 90 participants who failed the attention check, leaving 343 participants (47.0% female; mean age = 30.38 years, SD = 9.02) for the analysis. A 2 × 2 between-participants ANOVA revealed a marginal main effect of tier claim (F(1, 339) = 3.29, p = .070, ηp2 = .01) but no main effect of time pressure (F(1, 339) = 2.19, p = .140, ηp2 = .006). More relevant to our theorizing, we observed a significant interaction effect of tier claim and time pressure on brand evaluations (F(1, 339) = 3.91, p < .05, ηp2 = .011).
The attention check we used was a pre-established exclusion criterion for this study. However, given that our attention check generated a large number of exclusions, we conducted another 2 × 2 between-participants ANOVA on the full sample of 450 participants. The full-sample analysis yielded no main effects of either tier claim (F(1, 446) = 1.91, p = .168, ηp2 = .004) or time pressure (F(1, 446) = 1.84, p = .176, ηp2 = .004), but more importantly, we again found a significant interaction effect of tier claim and time pressure on brand evaluations (F(1, 446) = 4.06, p < .05, ηp2 = .01). In the more detailed results that we present in the next paragraph, we follow our pre-established exclusion criterion by focusing on the sample of participants that passed the attention check. However, the pattern of results is identical irrespective of whether the full sample or the restricted sample is used.
As we predicted, the comfort tier effect was replicated in the low time pressure condition; brand evaluations for the comfort tier claim (Mtop100 = 53.33, SD = 21.30, N = 101) were significantly higher than evaluations for the more exclusive non–comfort tier claim (Mtop92 = 44.21, SD = 24.79, N = 90; F(1, 339) = 8.10, p < .01). Also consistent with our predictions, this comfort tier effect was attenuated under high time pressure; brand evaluations for the comfort tier claim (Mtop100 = 52.14, SD = 21.11, N = 79) did not differ significantly from evaluations for the non–comfort tier claim (Mtop92 = 52.53, SD = 20.76, N = 73; F(1, 339) = .01, p > .91). Figure 4 illustrates these results.
TIME PRESSURE ATTENUATES THE COMFORT TIER EFFECT (STUDY 3)
The results of this study again support our theory that when consumers consider exact rank after exposure to an imprecise rank claim, they tend to generate estimates outside the next more-exclusive comfort tier (which leads to more negative evaluations). Consistent with the results of the previous studies, when participants were explicitly prompted to estimate rank, they were likely to generate estimates outside the next more-exclusive comfort tier; among participants in the top 92 and top 100 conditions, a large majority (78.5% and 71.7%, respectively) estimated the brand's rank to be outside the top 50.
Discussion
In Study 3, we provide evidence consistent with our two-stage process model by showing that when cognitive resources were constrained by time pressure, participants did not engage in stage 2 processing and their evaluations were therefore not sensitive to whether the claim referenced a comfort tier or a non–comfort tier. However, when cognitive resources were unconstrained, we replicated our earlier results and showed that a comfort tier claim leads to higher brand evaluations than a non–comfort tier claim. The finding that brand evaluations following a top 100 claim were unchanged by time pressure suggests that stage 1 processing may be more heuristic in nature and based on automatic, unconscious, or intuitive responding. In contrast, the finding that brand evaluations following a top 92 claim were influenced by time pressure suggests that stage 2 processing may be based more on conscious deliberation and “system 2” thinking (Kahneman 2003). These results are consistent with prior research on two-stage models of cognitive processes that have found similar differences when contrasting evaluative strategies that involve heuristic versus systematic evaluative strategies (Bettman, Luce, and Payne 1998; Dijksterhuis et al. 2005; Goldsmith and Amir 2010; Kahneman 2003; Lee, Amir, and Ariely 2009; Wilson and Schooler 1991).
An important factor in our model is whether the tier boundary reported by the marketer meets or violates expectations. We have argued that when a rank claim meets expectations, consumers are likely to terminate processing after stage 1. However, when a claim violates expectations, consumers proceed to stage 2. In the first three studies, we manipulated expectancy violation by exposing participants to a comfort tier claim (which met expectations) or a non–comfort tier claim (which violated expectations). However, an imprecise rank claim can meet or violate expectations in other ways. For example, consumers may not expect marketers to communicate a tier boundary that is close to (but different from) the length of the third-party list. For example, if the third-party has ranked 101 brands, a top 100 claim might seem unexpected even though it references a comfort tier. As a result of the stage 2 processing induced by this violation of expectations, brand evaluations should be attenuated in this situation despite the use of a comfort tier claim. In contrast, if the third-party has ranked 101 brands, a top 101 claim might seem expected and may improve evaluations despite its use of a non–comfort tier claim.
Study 4
Study 4 tests our claim that stage 2 processing is invoked by a violation of expectations and examines whether comfort tiers can be “decomfortized” in situations in which they violate expectations (and whether non–comfort tiers can likewise be “comfortized” in situations in which they meet expectations). To accomplish this, we manipulate expectation violation through two sources. As our previous studies have shown, one source of meeting or violating expectations is whether the claim references a comfort tier (100) or a non–comfort tier (99 or 101). Another source of meeting or violating expectations is whether the claim matches or mismatches the length of the third-party list. Thus, for this study, we additionally manipulate the length of the third-party list to be either 100 or 101. We predict that brand evaluations will be highest when both expectations are met—in other words, when a comfort tier claim matches the length of the third-party list (e.g., a claim to be in the top 100 of 100 brands ranked by the third party). When one expectation is met but the other is violated, the positive impact of the claim on brand evaluations should be attenuated. We expect this to occur when a comfort tier claim is “decomfortized” by making it mismatch the third-party list length and when a non–comfort tier claim is “comfortized” by making it match the third-party list length. We further predict that brand evaluations will be lowest when both expectations are violated—that is, following exposure to a non–comfort tier claim that mismatches the third-party list length.
Investigating the moderating role of third-party list lengths that are non–comfort tiers has considerable theoretical value because it enables us to test the critical role that expectancy plays in the comfort tier effect. However, this investigation also has practical value because although the typical length of a third-party list is a comfort tier, certain third-party lists reference non–comfort tiers, including Tabelog's list of America's top 11 burgers (http://www.tabelog.us/summary_articles/top-11-burgers-in-america), BarkPost's list of the top 23 cities to own a dog (http://barkpost.com/best-dog-cities/), Golf Digest's list of the top 36 “buddy” destinations (http://www.golfdigest.com/gallery/top-buddies-destinations-photos), and Docurated's list of the top 47 sales force knowledge management tools (http://www.docurated.com/all-things-productivity/top-47-salesforce-knowledge-management-tools-software) (all websites accessed July 22, 2015).
Method
Participants were 403 Americans (38.5% female; mean age = 33.71 years, SD = 11.01) recruited using MTurk who were randomly assigned to one of four conditions. Study 4 employed a 2 (claim tier: comfort tier vs. non–comfort tier) × 2 (third-party list length: 100 vs. 101) between-participants design. We manipulated the third-party list by informing participants that the fictional ONEUniverse Travel Guide publishes an annual list of either the best 100 or best 101 hotels in the region. Subsequently, participants were exposed to an advertisement for a specific hotel with an imprecise rank claim that referenced a comfort tier (best 100) or a non–comfort tier (best 99 or best 101). To ensure that participants had attended to this information, participants were required to correctly identify the length of the third-party list as well as the tier boundary referenced in the hotel's ad before they could proceed.
After viewing the ad, all participants rated the hotel on three brand evaluation scales (㬁 = .984) using sliders ranging from 0 to 100 (0 = “very negative,” and 100 = “very positive”; 0 = “unfavorable,” and 100 = “favorable”; 0 = “very bad,” and 100 = “very good”). On a separate screen, they then estimated the exact rank of the hotel on the ONEUniverse list and provided their rationale for their response. As in the previous studies, participants’ rank estimates were constrained to the range that was implied by the claim. Finally, all participants indicated their gender, age, and whether English was their native language.
Results
A 2 × 2 between-participants ANOVA revealed a significant effect of claim tier on brand evaluations (F(1, 399) = 19.90, p < .001, ηp2 = .05) but no significant main effect of third-party list length (F(1, 399) = 1.34, p > .24, ηp2 < .01). More central to our theorizing is the significant interaction observed between claim tier and third-party list length (F(1, 399) = 35.42, p < .001, ηp2 = .08), which indicates that the (mis)match between the claim and the third-party list length influenced brand evaluations. We first compared brand evaluations when the claim met expectations in two ways (i.e., by using a comfort tier and matching third-party list length) versus when the claim violated expectations in two ways (i.e., by using a non–comfort tier and mismatching third-party list length). This analysis revealed the comfort tier effect documented in our previous studies, in which participants who learned that the third-party list consisted of 100 hotels evaluated the advertised hotel as significantly better when it used a top 100 claim (M100–100 = 77.96, SD = 14.76, N = 101) versus a slightly more exclusive top 99 claim (M99–100 = 57.49, SD = 27.38, N = 106; F(1, 399) = 55.74, p < .001, ηp2 = .12). When one expectation was met and the other was violated, participant responses fell between these two extremes; there was no difference in brand evaluations between a comfort tier claim that did not match the third-party list length (M100–101 = 68.54, SD = 18.33, N = 101) and a non–comfort tier claim that matched the third-party list length (M101–101 = 71.47, SD = 14.93, N = 95; F(1, 399) = 1.08, p > .29, ηp2 < .01).
To look at the data another way, consumers exposed to a comfort tier claim that matched the third-party list length (100–100) evaluated the claim more favorably than consumers exposed to the same comfort tier claim that mismatched the third-party list length (100–101) (F(1, 399) = 11.53, p < .001, ηp2 < .03). Likewise, consumers exposed to a non–comfort tier claim that matched the third-party list length (101–101) evaluated the brand more favorably than consumers exposed to a non–comfort tier claim that did not match the third-party list length (99–100), despite the latter claim being more exclusive (F(1, 399) = 25.18, p < .001, ηp2 < .06). We illustrate these results in Figure 5.
(MIS)MATCH BETWEEN THE RANK CLAIM AND THE THIRD-PARTY LIST LENGTH ATTENUATES THE COMFORT TIER EFFECT (STUDY 4)
Next, we examined the rank estimates provided by participants in each of the four conditions. In our previous studies, in which participants did not receive any information about third-party list length, we found that when prompted to estimate rank, participants generally estimated ranks that were less favorable than the next more-exclusive comfort tier. Rank estimates in Study 4 exhibited a similar pattern but, notably, only when there was a mismatch between the rank claim and the third-party list length. Specifically, when a top 99 claim mismatched the third-party list length of 100, 76.4% of participants estimated the brand's rank to be less favorable than 50, suggesting that most participants would have expected the marketer to use at least a top 50 claim instead had the actual rank been 50 or better. Similarly, when a top 100 claim mismatched the third-party list length of 101, 59.4% of participants estimated the brand's rank to be less favorable than 50. These results support our claim that consumers often assume that the marketer would have strategically selected a more favorable boundary if the brand's actual rank had been more favorable.
However, in the conditions in which the rank claim matched the third-party list length (e.g., top 100 in a list with 100 hotels), participants may have assumed that rather than strategically choosing the most exclusive tier possible, the marketer instead just reported the boundary defined by the third party. If participants made such an assumption, they would be less likely to make negative inferences when prompted to estimate rank. Consistent with this explanation, when there was a match between the rank claim and the third-party list length, most participants did not assume that the brand's rank would be less favorable than the next more-exclusive comfort tier. Specifically, we found that only 6.3% of participants who encountered a top 101 claim that matched the third-party list length of 101 estimated that the brand was ranked exactly 101st (i.e., outside the top 100). And fewer than half (45.6%) of participants who encountered a top 100 claim that matched the list length of 100 estimated that the brand would be outside the next more-exclusive comfort tier (i.e., the top 50).
Discussion
By informing participants about the length of the third-party list, and by manipulating the length of the list to either match or mismatch the tier boundary referenced in an advertising claim, Study 4 provides further evidence that the comfort tier effect is driven by meeting or violating expectations. In our previous studies, we showed that stage 2 processing can be prompted by the violation of expectations induced by a non–comfort tier claim or by explicitly asking consumers to consider a brand's exact rank. In Study 4, we showed that stage 2 processing can also be caused by referencing a tier boundary that is close to, but does not match, the length of the third-party list (even if the claim happens to reference a comfort tier). The results suggest that the effect of expectancy violations is compounded when a non–comfort tier mismatches the third-party list length; brand evaluations in such a case are significantly lower than if the expectancy violation occurs in only one of these two ways.
Comparing the results of Study 4 with the results of Study 1 is informative because both studies included conditions in which the X in a top X claim differed by just one rank. In Study 1, we demonstrated the comfort tier effect by comparing responses to a best 9 claim versus a best 10 claim. Although the marginal effect of changing a tier boundary by a single rank might be expected to decrease as the tier becomes larger and more inclusive, Study 4 nevertheless showed a substantial decrease in brand evaluations when the top X claim referenced a tier boundary of 99 versus 100. These findings again highlight the robustness of the comfort tier effect.
Results from the first four studies are consistent with our theory that when consumers exposed to an imprecise rank claim undertake stage 2 processing, they not only think about the exclusivity of the tier (stage 1) but also deduce that the exact rank of the advertised brand is likely to be relatively unfavorable. In the next study, we further test our assumptions about stage 2 processing by introducing a manipulation designed to maintain a consumer's focus on the exclusivity of a claim even when the claim references a non–comfort tier, thus attenuating the otherwise negative effect of stage 2 processing. Whereas Study 3 showed that the negative inferences about rank that normally accompany stage 2 processing can be tempered through high time pressure, Study 5 is designed to show that even when cognitive resources are available, the advantage of using a comfort tier over a slightly more exclusive non–comfort tier can be mitigated by maintaining consumers’ focus on tier exclusivity.
Study 5
Study 5 tests whether contextual factors that maintain consumers’ focus on tier exclusivity can offset the negative inferences about rank brought about by Stage 2 processing following exposure to a non–comfort tier. We have proposed that consumers who engage in stage 1 processing focus primarily on the exclusivity of a top X claim, whereas those who subsequently engage in stage 2 processing make inferences about the brand's relatively unfavorable rank on the list of X brands. Study 5 tests this theorizing by introducing a condition designed to maintain consumers’ focus on tier exclusivity, even when they encounter a non–comfort tier claim. Specifically, we propose that joint (vs. separate) evaluation of a comfort tier claim and a non–comfort tier claim can attenuate the comfort tier effect. We anticipated that evaluating a brand in a non–comfort tier (e.g., top 19) at the same time as a brand in a slightly less exclusive tier (e.g., top 20) would emphasize that the first tier is more exclusive than the second. According to our model, focusing attention on tier exclusivity should offset the negative effect that (as we have demonstrated in previous studies) a non–comfort tier (vs. comfort tier) claim tends to have on brand evaluations. In contrast, we expected that when evaluated separately (by different participants), the comfort tier effect will again be replicated, such that a comfort tier claim of top 20 will generate more positive brand evaluations than a non–comfort tier claim of top 19. In addition to testing this aspect of our theorizing regarding Stage 2 processing, Study 5 examines a downstream consequence of brand evaluations by measuring consumers’ willingness to pay as the primary dependent variable.
Method
Participants were 225 undergraduates who were randomly assigned to one of three conditions: top 19 separate evaluation, top 20 separate evaluation, or joint evaluation (top 19 and top 20 together). This resulted in 299 total observations, because each participant in the joint evaluation condition provided two willingness-to-pay measures.
All participants were asked to suppose that they were planning a skiing trip and were exposed to either one or two imprecise rank claims, depending on condition. Participants in the joint evaluation condition saw an ad for a ski resort claiming that it had been ranked by SKI magazine as one of the top 19 resorts in North America, as well as an ad for a different ski resort claiming that it had been ranked by SKI magazine as one of the top 20 resorts in North America. In the separate conditions, participants saw only one of the two ads. The identity of the advertised ski resort, Alta or Canyons (both real resorts), was systematically varied such that each was advertised as one of the top 19 resorts for half of the participants and advertised as one of the top 20 resorts for the remaining participants.
After seeing these advertisements, participants in the joint evaluation condition were asked how much they would be willing to pay for a lift ticket to each resort. As a reference point, participants were informed that the average price of a one-day lift ticket at North American ski resorts was $50. Subsequently, they indicated their willingness to pay for a one-day lift ticket at Alta and Canyons (order was counterbalanced across participants) on sliding scales anchored at $0 and $100. Participants in the separate evaluation condition similarly expressed their willingness to pay, but only for the resort for which they had seen an advertisement (i.e., either Alta or Canyons). Finally, participants were asked to indicate their attitude toward skiing on a five-point scale (1 = “Hate it,” and 5 = “Love it”), whether they had heard of Alta or Canyons ski resort prior to the study (yes/no), as well as their gender, age, and whether English was their native language.
Results
Because real ski resorts were referenced in the advertisements, we were concerned that familiarity with the resorts could affect willingness to pay in ways that are independent of the comfort tier effect (in the “General Discussion” section, we further discuss the potential role of familiarity on our effects). Indeed, we found that the 38 participants who had heard of either resort before the study were willing to pay marginally more for a one-day lift ticket than participants who were unfamiliar with the resorts (Mfamiliar = $68.59, SD = $17.15, N = 46 observations vs. Munfamiliar = $63.90, SD = $14.79, N = 253 observations; t(167) = 1.67, p = .095). Therefore, we excluded these 38 participants, leaving us with a sample of 187 participants (60.43% female; mean age = 20.61 years, SD = 1.64) and 253 total observations.
Participants’ willingness to pay measures were analyzed first for participants in the separate evaluation conditions (for whom willingness to pay was a between-participants measure) and then for participants in the joint evaluation condition (for whom willingness to pay was a within-participant measure). Replicating the comfort tier effect we demonstrated in previous studies, willingness to pay in separate evaluation mode was significantly lower following exposure to a top 19 claim than a top 20 claim (Mtop19 = $60.28, SD = $16.96, N = 60 vs. Mtop20 = $66.16, SD = $11.66, N = 61; t(119) = 2.23, p = .028, d = .404). Although participants’ attitude toward skiing significantly predicted their willingness to pay in separate evaluation mode (F(1, 118) = 5.93, p = .016, ηp2 = .048), we still observed a comfort tier effect even when this covariate was included in the analysis (F(1,118) = 4.61, p = .034, ηp2 = .038). However, as we expected, in joint evaluation mode, willingness to pay for a lift ticket at each ski resort did not differ on the basis of claim tier (Mtop19 = $64.56, SD = $14.98 vs. Mtop20 = $64.44, SD = $14.81, N = 66; t(65) = .12, p > .90, d < .01). Although attitude toward skiing significantly predicted willingness to pay in joint evaluation mode (F(1, 64) = 7.52, p < .01, ηp2 = .105), controlling for this covariate yielded the same pattern of results, such that there was still no difference between claim tiers when it was included in the analysis (F(1, 64) = .24, p > .62, ηp2 = .004). Figure 6 illustrates these results.
EVALUATION MODE ATTENUATES THE COMFORT TIER EFFECT ON CONSUMERS’ WILLINGNESS TO PAY (STUDY 5)
Discussion
The results of Study 5 show that the comfort tier effect on consumers’ willingness to pay is observed when each claim is evaluated separately but is attenuated when both claims are evaluated jointly. The separate evaluation conditions were procedurally akin to the conditions in our previous studies that also demonstrated the comfort tier effect. The separate evaluation conditions again support the notion that exposure to a comfort tier (top 20) tends to encourage only stage 1 processing (and a focus on tier exclusivity), whereas a non–comfort tier (top 19) tends to additionally encourage stage 2 processing (and negative inferences about how the brand is likely to be worse than others on the list). The finding that this effect was attenuated in the joint evaluation condition is consistent with our view that simultaneously seeing the two claims encouraged participants to balance the negative inference that the exact rank of the top 19 brand is likely worse than others on the list with the positive fact that top 19 is a more exclusive tier than top 20. This is consistent with our theorizing that a focus on tier exclusivity (which is the natural result of stage 1 processing and in this study was highlighted by joint evaluation) has a favorable impact on brand evaluations, while a consideration of exact rank in stage 2 processing has a negative impact on brand evaluations.
Study 6
To further support our proposed two-stage model of how consumers process information when exposed to imprecise rank claims, in Study 6 we aim to provide additional evidence that stage 1 processing involves a focus on tier exclusivity. Specifically, if we are correct in theorizing that consumers focus during stage 1 on a brand's membership in an elite group rather than on its exact rank within a list, then a comfort tier claim is theoretically analogous to an unbounded claim that simply states that a brand is “one of the best” (without referencing a specific tier). Indeed, an unbounded claim inherently restricts consumers’ ability to engage in stage 2 processing because there is no numerical boundary specified that might prompt consideration of exact rank. The resulting focus on being “one of the best” is akin to what we argue happens during stage 1 processing of an imprecise rank claim and we therefore expect it to have a similar effect on brand evaluations as a comfort tier claim. To further situate the effects of an unbounded claim versus a comfort tier claim, we also included in this study other potentially important reference points used in our previous studies, including a non–comfort tier claim, a specific rank claim, and a no-claim control.
Method
Participants were 236 participants recruited on MTurk. Participants were randomly assigned to one of five conditions. In each condition, participants saw an advertisement from the fictional AmCorp Bank. Depending on condition, the ad contained one of the following: an unbounded claim in which the ad simply stated that AmCorp Bank had been ranked by National Bank Review as one of the best banks in the nation (without referencing a specific tier), a comfort tier claim stating that AmCorp had been ranked as one of the 50 best banks, a non–comfort tier claim stating that AmCorp had been ranked as one of the 47 best banks, or an exact rank claim stating that AmCorp had been ranked as the 47th best bank. In addition, we included a no-claim control condition, in which the ad included an identical description of AmCorp's value proposition, but no ranked list was mentioned.
Following exposure to the ad, all participants provided three brand evaluations (㬁 = .933) using slider scales that ranged from 0 (“strongly disagree”) to 100 (“strongly agree”). The items were (1) “If I needed to open a bank account, AmCorp Bank would be an excellent choice,” (2) “I would feel confident banking at AmCorp Bank,” and (3) “I believe that AmCorp is probably one of the best places to put my money.” Participants were then asked to estimate AmCorp Bank's exact rank in the third-party list and provide a rationale for their evaluation of AmCorp. Finally, all participants indicated their age and gender.
Results
We excluded 11 participants in the imprecise rank claim conditions from the analysis for estimating an exact rank that was outside the feasible range described in the ad. The analysis was based on the responses of the remaining 225 participants (61.3% female; mean age = 35.07 years, SD = 11.85).
A between-participants ANOVA on the composite brand evaluation measure revealed a significant effect of condition (F(4, 220) = 7.82, p < .0001, ηp2 = .124). We observed a comfort tier effect in which brand evaluations were higher among participants exposed to a comfort tier claim (Mbest50 = 62.33, SD = 19.42, N = 45) than among participants exposed to a more exclusive non–comfort tier claim (Mbest47 = 52.00, SD = 22.16, N = 39; F(1, 220) = 5.36, p = .022), consistent with prior studies.
A key prediction in this study was that an unbounded claim would have the same effect on brand evaluations as a comfort tier claim. As we expected, brand evaluations in the unbounded claim condition (Mbest = 64.07, SD = 17.74, N = 49) significantly outperformed every other condition (all ps < .04), with the predicted exception of the comfort tier claim condition (F(1, 220) = .17, p > .68). Consistent with the notion that stage 2 processing can offset the positive effect of stage 1 processing, brand evaluations among participants exposed to the non–comfort tier claim were no better than those among participants exposed to the no-claim control (Mcontrol = 55.18, SD = 18.94, N = 43; F(1, 220) = .50, p > .48), and evaluations were lowest among participants exposed to the exact rank claim (M47th = 43.69, SD = 23.38, N = 49). These results appear in Figure 7.
THE COMFORT TIER EFFECT ON BRAND EVALUATIONS IS REPLICATED USING AN UNBOUNDED CLAIM (STUDY 6)
Next, we examined participants’ estimates of actual rank in the two imprecise rank claim conditions (i.e., best 47 and best 50) to determine whether the estimates were indeed less favorable than the next comfort tier. Among participants exposed to the best 47 and best 50 claims, a large majority (82.1% and 64.4%, respectively) estimated the brand's rank to be below the next comfort tier, the best 25.
Discussion
The results of this study provide additional support for our theorizing that comfort tier claims encourage consumers to focus on the general exclusivity of the top X tier without considering exact rank. The finding that an unbounded claim has a similar effect on brand evaluations as a comfort tier claim suggests that, in both cases, the improvement in brand evaluations is driven by the consideration of a brand's inclusion in an elite group, not the particular tier that is referenced. This is consistent with our theory that when exposed to a comfort tier claim, consumers typically terminate processing after stage 1 and do not consider a brand's likely rank or how it compares with other brands within the tier.
The evaluations for this study again replicate the comfort tier effect demonstrated in previous studies, whereby a less exclusive claim (one of the 50 best) produces higher evaluations than a more exclusive tier claim (in this case, one of the 47 best). We have demonstrated that those exposed to a non–comfort tier tend to think about the brand's exact rank and to infer that the rank is likely to be lower than the next more-favorable comfort tier. We therefore anticipated that evaluations in response to the non–comfort tier claim (one of the 47 best) would be either slightly better than or statistically indistinguishable from the specific rank claim at the tier boundary (47th). Although not critical to our theorizing, the significant difference between evaluations in response to these two claims (which replicates our finding in Study 1, in which a non–comfort tier claim led to more favorable evaluations than an exact rank claim) might be explained by a plausible inference participants may have made in the absence of more information. Specifically, the phrase “top X” or “best X” in an imprecise rank claim may be perceived to imply that the universe of brands is significantly larger than X. If this is true, any brand in the top X may be perceived as relatively favorable compared with the many brands that were not included in the third-party list. However, this favorable inference may not be made when “top” or “best” is excluded from the imprecise rank claim. Although the comparisons most central to our theorizing in this research are between imprecise claims that reference comfort versus non–comfort tiers, the result that imprecise claims seem to boost evaluations compared with certain exact rank claims is a finding that has managerial relevance; we discuss this further in the following section.
GENERAL DISCUSSION
In this research, we examine how consumers’ expectations affect their interpretation of imprecise marketing claims. Unlike prior research on how consumers process information about ranked lists, which typically assumes that consumers are exposed to the entire list of ranked items and are aware of each brand's exact rank (Isaac and Schindler 2014; Leclerc, Hsee, and Nunes 2005), our research examines how consumers respond, without seeing the complete list, to marketing claims in which the exact rank of an advertised brand is unspecified. This context is important to study because consequential consumer decisions, such as which college to attend or which hospital to select, may be influenced by imprecise advertising claims about a brand's rank in a third-party list. We provide the first experimental evidence of how consumers interpret these imprecise rank claims, and we specify a conceptual model regarding the psychological mechanisms these claims elicit.
Results from a series of six experiments show that consumers’ response to imprecise rank claims depends not only on the exclusivity of the claim but also on the extent to which the claim is consistent with consumers’ expectations. Specifically, we document a comfort tier effect in which brand evaluations are more favorable when a marketing communication meets expectations by referencing a comfort tier, even when a more exclusive rank claim could be made by referencing a proximal non–comfort tier. Theoretically, our work contributes to a greater understanding of how consumers respond to imprecision (vs. transparency) in marketing communications. In particular, prior research has suggested that consumers prefer transparency in marketing communications and penalize nondisclosure (Darke and Ritchie 2007; John, Barasz, and Norton 2016). Yet marketers often shroud information when communicating rank to make it seem more favorable (Luca and Smith 2013), Thus, in line with prior research, one might expect consumers to penalize marketers who use imprecise rank claims. However, our findings show the opposite—brand evaluations are often higher when marketers use imprecise rank claims than when marketers provide full transparency by disclosing a brand's exact rank. This new evidence opens the door for future studies to further examine when and why transparency can help versus hurt firm performance.
In addition to demonstrating that, counterintuitively, brand evaluations can be improved by weakening an advertising claim, our research contributes to a greater understanding of the processes by which imprecise rank claims influence consumer reactions to marketing communications. First, our identification of the comfort tier effect complements prior research that has shown how brand evaluations can be improved by weakening a claim through other methods. In particular, prior research has shown that evaluations can be improved by weakening a claim through the inclusion of mildly negative information in a product description (Ein-Gar, Shiv, and Tormala 2012) or acknowledgment of a negative quality (Ward and Brenner 2006). Similarly, we show that evaluations can be improved by weakening a claim through the use of slightly less exclusive comfort tiers rather than more exclusive non–comfort tiers. Whereas prior research has attributed this improvement in evaluations to an accentuation of positive information caused by the presence of negative information (Ein-Gar, Shiv, and Tormala 2012), our theoretical model suggests that in the case of consumers’ reaction to imprecise advertising claims, the positive effect of a weaker claim on brand evaluations can be attributed to the idea that it meets consumer expectations instead of violating them as a slightly stronger claim might.
Second, by examining how consumers respond to ambiguous advertising claims in which marketers convey numeric information using a bounded range rather than a precise numeric value, our research complements prior research that has examined a related question in the context of price-related communications (Mobley, Bearden, and Teel 1988). Indeed, prior research has suggested that when exposed to a tensile pricing claim such as “save 30% or more,” in which marketers do not specify the exact discount, savvy consumers may assume that the savings will not be substantially higher than 30% (Hardesty, Bearden, and Carlson 2007). Our identification of a similar effect in the context of imprecise rank claims (when consumers engage in stage 2 processing) not only provides additional evidence that tensile claims are sometimes interpreted pessimistically by consumers but also offers a novel explanation. Specifically, whereas related findings in a pricing context have been attributed primarily to anchoring or resistance to persuasive attempts (Biswas and Burton 1993; Gupta and Cooper 1992), our research suggests that consumer expectations may help explain the finding in tensile pricing research that more extreme (and therefore, unexpected) tensile claims tend to be discounted more than less extreme claims (Licata, Biswas, and Krishnan 1998; Mobley, Bearden, and Teel 1988). Moreover, our proposed two-stage process by which consumers process imprecise advertising claims suggests that these negative inferences about price savings could potentially be avoided if firms are able to prevent consumers from engaging in stage 2 processing and instead maintain their focus on the tier as a whole.
Our findings suggest that consumers who engage in stage 2 processing consistently infer that marketers would have made a more favorable claim had the exact rank warranted it. This result has relevance to the literature on persuasion knowledge, which suggests that consumers make inferences about the “possible background conditions” that encourage marketers to construct claims in particular ways (Friestad and Wright 1994, p.30). Recognizing marketers’ motivation to cast a brand in its most favorable light (Artz and Tybout 1999; Kirmani and Zhu 2007), knowledgeable consumers may be sensitive to the implicatures of imprecise rank claims, particularly if they are likely to access and use their persuasion knowledge about the marketing tactic (either because of dispositional tendencies or contextual factors). Specifically, the use of imprecise rank claims may be viewed as an attempt by marketers to mask a brand's unfavorable rank by including it in the same category as better-ranking brands. Such an assumption logically leads to an expectation that the brand's exact rank will be unfavorable relative to other brands in the tier. Although we agree that persuasion knowledge may play a role in the comfort tier effect, there is evidence to suggest that it cannot fully explain our results. For example, we collected data not reported in this article that shows that consumers make similar inferences even in nonpersuasive communication, when the top X claim is communicated by a neutral third-party rather than by the advertised brand itself. Nevertheless, our findings that stage 2 processing involves deliberation about what may have led marketers to select a particular tier boundary are in line with Campbell and Kirmani's (2000) influential finding that persuasion knowledge use requires cognitive resources and that the application of persuasion knowledge often has a negative effect on evaluations.
Our assertion that consumers’ decisions can be affected not only by changes in a brand's actual rank (Pope 2009) but also by the way that rank is communicated to consumers by marketers has important implications for marketing practice. In particular, our research provides guidance to marketers on how best to promote a brand's inclusion in a ranked list. First, results from Studies 1 and 6 suggest that at least in certain conditions, imprecise advertising claims can result in more positive brand evaluations than precise advertising claims that communicate a brand's exact rank. Second, our finding across all six studies that consumers tend to react more favorably to imprecise advertising claims that meet rather than violate their expectations suggests that marketers may be better off referencing a comfort tier rather than a more exclusive non–comfort tier in their advertisement. Third, our two-stage process model suggests that even when a non–comfort tier claim is used, marketers may be able to maximally improve brand evaluations by maintaining consumers’ focus on tier exclusivity and taking actions that are likely to keep consumers from advancing to stage 2 processing (by matching the claim to third-party list length, providing two rank claims simultaneously, etc.).
From a practical perspective, marketers must also consider the effect of brand familiarity on consumers’ interpretation of rank claims. Our analysis of the 38 participants who were previously familiar with either of the ski resorts used in Study 5 showed that these participants were willing to pay marginally more for a one-day lift ticket than those participants who were unfamiliar with the resorts. This could be due to many factors, such as prior positive associations with the resort or prior knowledge of the resort's exact rank. For example, if a familiar brand claims to be ranked more or less favorably than expected, consumers’ evaluations or willingness to pay may be influenced in ways that are independent of the comfort tier effect. Moreover, consumers exposed to an imprecise rank claim of a brand they like (dislike) may be motivated to formulate reasons why its exact rank should be more (less) favorable, which may in turn affect their brand evaluations. To avoid these confounds and improve experimental control, our studies focused primarily on the evaluation of unfamiliar or fictional brands.
In summary, the present research provides robust evidence that consumers’ brand evaluations and willingness to pay can be influenced by small variations in rank claims, including cases in which tier boundaries differ by just a single integer. Further research might explore how consumers integrate rank claims with other information frequently provided in an advertisement (e.g., product specifications, consumer reviews) or with their own pre-existing knowledge about the brand or the third-party list. Such investigations may be able to shed additional light on how and when consumers are influenced by rank claims.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
