OUP user menu

Editors

Darren Dahl (Editor in Chief)Eileen FischerGita JoharVicki Morwitz

3.125
5.003
16 out of 115

I Imagine, I Experience, I Like: The False Experience Effect

Priyali Rajagopal , Nicole Votolato Montgomery
DOI: http://dx.doi.org/10.1086/660165 578-594 First published online: 1 October 2011

Abstract

False memories refer to the mistaken belief that an event that did not occur did occur. Much of the research on false memories has focused on the antecedents to and the characteristics of such memories, with little focus on the consequences of false memories. In this research, we show that exposure to an imagery-evoking ad can result in an erroneous belief that an individual has experienced the advertised brand. We also demonstrate that such false experiential beliefs function akin to genuine product experience beliefs with regard to their consequences on product attitude strength, a finding we call the false experience effect. We further demonstrate two moderators of this effect–plausibility of past experience and evaluation timing.

The mistaken belief that an event occurred, known as a false memory, is a topic that has attracted significant attention in recent years (Loftus 1997, 2003; Schlosser 2006). How false memories develop is important to understand because false memories not only reflect an inability to accurately remember past experiences but may also have an impact on future behaviors (Bernstein and Loftus 2009). As such, research has sought to understand how false memories are created and how they differ from true memories (see Koriat, Goldsmith, and Pansky [2000] for a review).

A consistent finding from this research is that imagery is a strong route by which incorrect beliefs about a past event may develop (Drivdahl and Zaragoza 2001; McIntyre et al. 2004). For example, Mazzoni and Memon (2003) showed that imagining an impossible event in childhood (i.e., a nurse removing a skin sample from a finger) produced an increase in the reported likelihood that the event actually occurred. Thus, imagining experiences can alter individuals’ perceptions of an experience in memory. Research contrasting the effects of false and true memories has shown that false memories may sometimes be held with the same degree of confidence and coherence as true memories (Heaps and Nash 2001). Hence, false memories can function in a manner akin to real memories. This implies that the mere act of imagining using a product or service may increase the likelihood that consumers erroneously believe that they have encountered the product in the past.

Our research extends these findings by suggesting that if consumers falsely believe that they have experience with a product, then their evaluations of that product may be similar to evaluations for products for which they have actual experience. Specifically, we demonstrate that exposure to an imagery-evoking ad can increase the likelihood that a consumer mistakenly believes s/he has experience with the advertised product when in fact s/he does not. Moreover, such a false belief produces attitudes that are as strong as attitudes based on true beliefs about previous product experience, an effect that we label the false experience effect. We demonstrate this effect with regard to several outcomes relevant to marketing (e.g., consumer attitude confidence, attitude accessibility, attitude-purchase intent correlations, and attitude valence). We also examine two moderators of the false experience effect—plausibility of past experience and evaluation timing.

Our findings contribute to the literature on false memories by focusing on the similarities of true and false beliefs about past experience in terms of their consequences as compared to most of the current literature that focuses on the antecedents to false memories (Goff and Roediger 1998; Schlosser 2006) or the qualitative differences between true and false memories (Heaps and Nash 2001). We also examine the effect of false beliefs about experience on a previously unexamined variable—attitude strength. Attitude strength is an important variable for marketers because stronger attitudes have been shown to be more predictive of behavior, more resistant to counter persuasion, and more persistent over time (Petty and Krosnick 1995). Past research has even suggested that “measures that capture aspects of attitude strength may be as, if not occasionally more, important than traditional attitude measures” (Rucker, Petty, and Priester 2007, 79). We also add to the emerging research on false memories in marketing (Braun, Ellis, and Loftus 2002; Braun-LaTour et al. 2004; Cowley and Janus 2004; Lakshmanan and Krishnan 2009; Law, Hawkins, and Fergus 1998; Schlosser 2006).

We begin by referencing key findings from the research on false memory and product experience to formulate our hypotheses. We then present three empirical studies that test our predictions and conclude with a summary of our findings and contributions.

Literature Review

False Memory

Research in the area of false memory is very broad in scope and can be characterized by whether the false memory is related to list learning (Roediger and McDermott 1995) or autobiographical events (Loftus and Pickrell 1995) and whether the research focus is the cause of the false memory (Koriat et al. 2000) or the characteristics of the false memory (Heaps and Nash 2001). Our focus in this article will be on autobiographical memories and the outcomes of erroneous beliefs about product experience compared to those produced by correct beliefs about product experience. Thus, we do not focus on why false beliefs are created but rather on the consequences of such mistaken beliefs on product attitudes. We suggest that a high-imagery ad will increase the likelihood that consumers will incorrectly believe that they have experienced a product since imagery has been shown to result in a high incidence of false memories across a variety of contexts (Thomas and Loftus 2002).

Consequences of False Beliefs about Experience

Recent research that has examined the downstream effects of false memories suggests that false beliefs may be strong enough to affect attitudes and future behavior. For example, McIntyre et al. (2004) showed that false memories of counterattitudinal events changed people’s attitudes toward gay men. Other work has found that if people were led to believe that they had an adverse experience with a food in childhood (e.g., with hard-boiled eggs), they exhibited increased avoidance of this food (Bernstein et al. 2005; Geraerts et al. 2008).

While the consequences of false beliefs about experience on future decisions have been documented, less attention has been devoted to understanding how false beliefs compare with true memories in terms of these consequences. Thus, it is unclear whether the effects of false beliefs on attitudes and intentions are different from the effects of true memories. This is an important question in a marketing context because, as the following section suggests, the creation of genuine product memories through product trial has been shown to have strong effects on consumer attitudes.

Effects of Product Experience on Attitudes

Research in marketing has found that attitudes based on product experience are stronger than attitudes based on indirect product experience (i.e., exposure to advertisement). For example, Smith and Swinyard (1983) gave respondents a snack food sample (i.e., direct experience) or an advertisement for a snack food (i.e., indirect experience). They demonstrated that attitudes formed on the basis of direct experience versus indirect experience are more favorable and more consistent with behavior. Regan and Fazio (1977) similarly found that attitudes formed on the basis of direct experience are more predictive of behavior than attitudes formed on the basis of indirect experience. Across a variety of attitude strength measures (e.g., attitude-behavior correlations, self-reported attitude confidence [Petty and Krosnick 1995]), past research has demonstrated that direct experience results in greater attitude strength than indirect experience (Marks and Kamins 1988); however, this research has not addressed the impact of the nature of the ad (e.g., vivid vs. pallid).

Attitude strength, which has been defined as “a latent psychological construct that is presumably represented in memory by various attributes of the attitude” (Petty and Krosnick 1995, 3), has increasingly become a construct of interest to consumer researchers. Research in this domain has shown that attitude strength moderates the effect of attitude valence on brand consideration such that stronger attitudes are more predictive of behavior than weaker attitudes (Krishnan and Smith 1998; Priester et al. 2004). As such, consumers with equally favorable attitudes toward a brand may nonetheless differ in the extent to which their attitudes affect their behaviors toward the brand. Since attitude strength is multifaceted, researchers have measured this construct using various outcomes, such as attitude accessibility, extremity, persistence, resistance, and certainty (Rucker et al. 2007). Overall, strong attitudes have been shown to be held with greater confidence and are more persistent over time, resistant to counterarguments, accessible in memory, and predictive of behavior.

Because marketers are interested in having an impact on consumers’ behaviors, it is important for marketers to understand how consumers develop strong attitudes. We focus on this question and predict that exposure to an imagery-evoking advertising message will increase the likelihood of consumers incorrectly believing that they have experienced the advertised product and that this false product experience belief will produce attitudes that are equivalent in strength to those produced by direct product experience, a phenomenon that we term the false experience effect. Thus, we expect that exposure to a high-imagery advertisement will elicit attitudes that are as strong as those produced by direct experience. However, if an individual is exposed to a low-imagery ad, we expect that it is unlikely that such individuals will falsely believe that they have encountered the advertised brand. Therefore, we anticipate that their attitudes will be weaker than those produced by direct experience or by exposure to the high imagery ad. Thus,

  • H1 Product experience and advertisement imagery will interact such that exposure to a high-imagery ad will elicit attitudes that are stronger than those elicited by exposure to a low-imagery ad but are equivalent in strength to attitudes formed on the basis of direct experience.

While the false experience effect suggests that consumers are likely to erroneously believe that they have encountered a brand in the past when they are exposed to a high-imagery ad, it is important to understand the conditions under which these incorrect beliefs arise versus when advertising results in correct beliefs about past product experience or the lack thereof. One of the key findings from the false memory literature is that individuals are more likely to falsely believe events have occurred if the events are consistent with their script-relevant knowledge in memory; significant overlap between an event and information stored in memory leads individuals to falsely attribute the event to past experience (Pezdek, Finger, and Hodge 1997). Put differently, consumers are more likely to form erroneous beliefs about an event if it is plausible that the event took place in the past. For example, Pezdek et al. (1997) found that Catholic (Jewish) students were more likely to incorrectly remember the false event describing performing a Catholic (Jewish) ritual than a Jewish (Catholic) ritual. This finding is important because it suggests that not all false events are equally likely to have occurred and that events with a greater likelihood of occurrence are more likely to elicit false experience beliefs.

In a marketing context, we conceptualize plausibility as the possibility that one could have had past experience with a brand and suggest that false experience beliefs are more likely to occur with familiar brands than unfamiliar brands. That is, when consumers are familiar with and have had prior experience with a brand (e.g., Hershey’s), they may perceive that it is plausible that they have had past experience with even a fictitious variant of that brand (e.g., Hershey’s Twirl). On the other hand, when consumers are unfamiliar with a brand and have not had any prior experience with the brand (e.g., Droste), then a false product experience belief about a fictitious variant of that brand (e.g., Droste’s Twirl) is less plausible, and hence the false experience effect is less likely to be elicited. Thus,

  • H2 Plausibility that a consumer has past experience with the advertised brand will moderate the false experience effect such that a high-imagery ad will elicit attitudes that are stronger than those elicited by exposure to a low-imagery ad but are equivalent in strength to attitudes formed on the basis of direct experience when plausibility of the experience is high, but not when plausibility is low.

In addition to examining plausibility as a moderator, a second issue that we address is whether the false experience effect manifests over time. Some research in the false memory literature has found that false beliefs about experience are most likely to emerge when there is a delay between the manipulation and measurement, suggesting that false experience beliefs emerge over time (e.g., Bernstein et al. 2005; Geraerts et al. 2008). Consistent with this research, we expect that the pattern of results that we hypothesized previously will occur when respondents evaluate the advertised products following a time delay between exposure to the ads and subsequent evaluations but not when respondents evaluate the products immediately after exposure to the ads. Further, if imagery affects attitudes only after a delay when memory errors are most likely to occur, it increases our confidence that false experience beliefs, as opposed to other variables (e.g., elaboration), are the underlying mechanism responsible for the false experience effect. Thus,

  • H3 Evaluation timing will moderate the false experience effect such that high-imagery ads will elicit stronger attitudes than low-imagery ads when evaluations are assessed following a delay but will elicit attitudes equivalent in strength when evaluations are assessed immediately following exposure to the ad.

Attitude Valence

While our focus in this article will be primarily on attitude strength, we expect that when product experience is positive and consistent with the expectations set up by advertising, the pattern of results for attitude valence will be similar to the pattern predicted for attitude strength. However, when product experience is negative, or does not confirm expectations set up by advertising, there are likely to be significant differences between the effects on attitude valence produced through high imagery advertising versus actual experience. Hence, we expect the false experience effect to extend to attitude valence only when product experience is positive and confirms advertising expectations.

We conducted study 1 to test hypothesis 1 on the false experience effect, study 2 to test hypothesis 2 on the moderating role of plausibility of past experience, and study 3 to test hypothesis 3 on the moderating effect of evaluation timing.

Study 1

Study 1 was conducted to examine whether exposure to an imagery-evoking advertisement can create a mistaken belief that the advertised product has been experienced that yields attitudes that are equivalent in strength to those produced by a real product experience. One hundred undergraduate students participated in the study in exchange for course credit. They were exposed to a set of six print advertisements for different products, one of which was the target ad, with the remaining five ads functioning as filler ads.

Stimulus Materials

Each of the advertised products consisted of a real brand name with a fictitious product variant name (e.g., Dial Naturals soap). A high-imagery and a low-imagery print ad were developed for the target brand, Orville Redenbacher’s Gourmet Fresh popcorn, while the five filler ads were all of low imagery and consistent across conditions. The verbal description of the product contained in the target ad was varied such that the description in the high-imagery ad was more image-provoking than its low-imagery counterpart; however, the same information about the advertised brand was presented in both versions of the ad (see fig. A1 in app. A). The target ad was pretested according to the procedure outlined in Unnava and Burnkrant (1991) to ensure that the two versions of the ad differed only with respect to their imagery-evoking ability (not image-provoking/image-provoking, dull/vivid, boring/interesting), and not on their favorableness, believability, understandability, meaningfulness, distinctiveness, self-referencing, informativeness, or perceived argument strength. A pretest with 56 undergraduate students revealed that the high- and low-imagery versions of the target ad differed on imagery (MHigh = 5.95, MLow = 4.18; F (1, 54) = 23.91, p <.05) but not on any of the other measures (p’s >.05).

An additional pretest was conducted to increase our confidence that any differences in the target measures in study 1 were not attributable to differences in the amount of elaboration elicited by the high-imagery ad as compared to the low-imagery ad. Eighty-nine undergraduate students viewed one of the two versions of the ad and reported their thoughts about the ad. Thoughts were coded into three categories: positive thoughts (e.g., “It looks good”), negative thoughts (e.g., “The ad is not as good as others”), and neutral thoughts (e.g., “It made me hungry”). An analysis of variance indicated that respondents who read the high-imagery version of the ad listed similar numbers of positive (0.76 vs. 0.90), negative (1.02 vs. 1.08), and total thoughts (3.46 vs. 3.49) as participants who read the low-imagery version of the ad (all p’s >.05). Thus, differences in elaboration are unlikely to explain our results.

Design and Procedure

A 2 (advertisement imagery: high vs. low) × 2 (product experience: yes vs. no) between-subjects design was used with imagery and experience as the factors of interest. When respondents entered the room, they were informed that they would be participating in an advertising evaluation study, after which they were given the six advertisements to read. Each ad was presented to them for 35 seconds, consistent with the time limits used in both pretests. This time limit was selected as one that provided adequate time for respondents to read through the ad at a comfortable pace and was set forth to ensure that any differences obtained could not be attributed to participants processing the high-imagery ad for a longer period of time. After viewing all six ads, participants completed a short survey that measured their involvement during presentation of the ads on three 7-point scales anchored by paying little attention to paying a lot of attention, concentrating very hard to concentrating very little (reverse-coded), and very uninvolved to very involved (α =.68). This measure was included to ensure that any differences in the target measures were not attributable to differences in task involvement. There were no differences in involvement between any of the experimental conditions (p’s >.05).

Immediately following presentation of the ads, respondents in the no-experience conditions completed an unrelated survey, while respondents in the experience conditions were given half of a cup of the target product to taste (Orville Redenbacher’s Gourmet Fresh popcorn). Since the target product was fictional, we used Orville Redenbacher’s Movie Theater Butter popcorn for the taste experience, and care was taken to ensure that the brand name was not visible. To this end, the experimenter placed the popcorn kernels in identical individual plastic cups before the respondents entered the room. After the respondents finished tasting the product, they moved to another unrelated study. One week after the original study, all respondents were asked to complete a follow-up online survey that contained the dependent measures. Ninety respondents completed the follow-up survey (90% response rate). An analysis of the respondents revealed that there were no significant differences in the number of drop-outs across the experimental conditions (p’s >.05).

Dependent Measures

Our primary measures of interest were attitudes toward the target brand and product experience beliefs (true vs. false depending on experience condition; table 1). We chose attitude confidence as our measure of attitude strength because this measure has been shown to be predictive of behavior and indicative of an attitude’s likely persistence over time (Rucker et al. 2007). Product experience beliefs were assessed using five different measures presented to respondents in the following order: perceived usage, belief statements, belief confidence, memory quality, and subjective memory accuracy.

View this table:
Table 1.

Study 1 Dependent Measures

graphic

Perceived usage was a dichotomous measure wherein respondents were provided with a list of 36 products and asked to select all of the brands they believed that they had tried at least once. The list contained the six advertised products (one target and five fillers) and 30 other products from the same product categories as the advertised products (e.g., popcorn, soap, chocolate, etc.). For each product category, two different product variants for each of three brands (advertised brand and two additional real brands) were used. For each advertised brand, the two products consisted of the advertised product variant (e.g., Orville Redenbacher’s Gourmet Fresh Popcorn) and another fictitious product variant (e.g., Orville Redenbacher’s Blue Corn popcorn). The fictitious variant was included to ensure that the belief about product experience with the target brand was specific to the advertised product variant rather than generally for the overall brand name.

Subsequently, respondents were presented with the remaining product experience belief measures. The belief statements were a set of three statements (e.g., “I do not know this product”) for each of the advertised brands from which respondents were asked to select all that applied. This measure attempted to capture lack of knowledge about the brand in addition to beliefs about having previously encountered the brand. Belief confidence was a continuous measure adapted from Hyman and Pentland (1996) that incorporated the false belief itself (e.g., “I have/have not tried the brand”) and the strength with which the belief was held (e.g., “I am very confident in my memory”). Finally, consistent with past research on false memory that has included qualitative measures to evaluate the quality of false memories relative to true memories (Hyman and Pentland 1996), respondents who reported that they had used the target product at least once (i.e., chose the target product on the perceived usage measure) were asked to indicate the quality of their memory of trying the product and whether they had any doubts in the subjective accuracy of their memory for trying the product.

In the present study, attitudes were measured prior to product experience beliefs to ensure that the effects on attitudes could not be attributed to the measurement of product experience beliefs prior to measuring attitudes. In subsequent studies, we measure product experience beliefs before attitudes and obtain consistent results, suggesting that the ordering of the dependent variables does not have an impact on our results.

Results

Attitude Confidence

An analysis of variance revealed the expected significant interaction between imagery and experience (F(1, 86) = 7.58, p <.05) on attitude confidence. Planned contrasts showed that respondents without product experience exhibited more confident attitudes toward the target brand when they were exposed to the high-imagery ad (M = 5.61) than the low-imagery ad (M = 4.74; t(86) = −2.77, p <.05). However, respondents in the two experience conditions did not significantly differ in terms of their attitude confidence (MHigh = 5.31, MLow = 5.79; p >.05), nor did they differ from the high-imagery–no-experience respondents (p >.05; table 2). Finally, respondents did not differ in their attitude confidence for the filler brands (p’s >.05). Hence hypothesis 1 was supported.

View this table:
Table 2.

Study 1 Results

graphic

Attitude Valence

An analysis of variance revealed a significant interaction between imagery and experience (F(1, 86) = 5.18, p <.05). Planned contrasts indicated that respondents who were exposed to the high-imagery ad without product experience exhibited more favorable attitudes toward the target brand (M = 5.87) than those who were exposed to the low-imagery ad without product experience (M = 5.11; t(86) = 2.97, p <.05). However, respondents in the two experience conditions reported attitudes that were equivalent on favorability (MHigh = 5.51, MLow = 5.82; p >.05) and equivalent to those reported by the high-imagery–no-experience respondents (p’s >.05). Finally, respondents did not differ in their attitudes toward the filler brands (p’s >.05).

Product Experience Beliefs

An analysis of the perceived usage measure indicated that respondents across conditions did not differ in the total number of brands they selected from the list when asked to indicate which products from the list they had used at least once (p >.05). However, respondents differed on whether or not they selected the target product (χ2(3, N = 90) = 13.10, p <.05). A larger percentage of respondents without product experience selected the target product if they were exposed to the high-imagery ad versus the low-imagery ad (χ2(1, N = 53) = 7.67, p <.05). Respondents in the experience conditions, however, did not differ from each other in their selection of the target product (χ2(1, N = 37) = 0.25, p >.05) nor did they differ from the high-imagery–no-experience respondents (p’s >.05). Finally, respondents did not differ in their selection of the nonadvertised product variant with the same brand name as the target (Orville Redenbacher’s Blue Corn popcorn; p >.05), suggesting that the effect is specific to the advertised variant and not to the overall brand.

In addition to examining the target brand, we examined the perceived usage measure for the other advertised brands to ensure that imagery, not simply advertising exposure, resulted in the obtained results. An analysis of variance revealed that contrary to the target brand, the effects of imagery and experience were not significant (p’s >.05). Finally, we also analyzed the perceived usage measure for the other filler brands that were from the same category as the target brand (popcorn). These brands were included to ensure that the effects that were found for the target brands were not generalizable to other members of the target brand’s product category, that is, to rule out a “category” effect. We found no significant effects of imagery and experience on the memory for these brands (p’s >.05).

For the belief statements, our expectation was that respondents in the low-imagery–no-experience condition would select the “purchase” and “trial” statements less frequently and the “do not know” statement more frequently than the other three conditions. As such, we combined the first two statements for analysis. An analysis of variance revealed a significant interaction of imagery and experience on the selection of the “trial” and “purchase” statements (F(1, 86) = 5.14, p <.05), with respondents in the low-imagery–no-experience condition selecting fewer of these statements for Orville Redenbacher’s Gourmet popcorn than respondents in the other three conditions (t(86) = −2.58, p <.05). Finally, imagery and experience significantly predicted the selection of the “do not know” statement (χ2(3, N = 90) = 10.71, p <.05), with respondents in the low-imagery–no-experience condition selecting this statement more frequently for the target than respondents in the other three conditions (χ2(3, N = 41) = 5.11, p <.05). Selection of the three statements did not differ for respondents in the experience conditions or in the high-imagery–no-experience condition (p’s >.05). Finally, respondents across all conditions did not differ on their statement selection for the advertised filler brands (p’s >.05).

An analysis of variance on the belief confidence measure revealed a significant interaction between imagery and experience (F(1, 86) = 3.93, p <.05). Planned contrasts revealed that without product experience respondents who were exposed to the high-imagery target ad were more confident that they had tried the product than respondents who were exposed to the low-imagery ad (t(86) = −4.57, p <.05). Also as predicted, respondents in the two experience conditions and the high-imagery–no-experience condition did not differ in their reported confidence (p’s >.05). Finally, confidence did not differ for any of the filler products (p’s >.05).

The memory quality and subjective memory accuracy measures sought to show that individuals who falsely believed that they had tried the target product had memories of trying the product that did not differ qualitatively from those of individuals who had genuine memories of trying the target product. To that end, only those individuals who indicated that they had tried Orville Redenbacher’s Gourmet Fresh popcorn in the perceived usage measure completed these measures. Because we are interested in the qualitative nature of false versus true product experience beliefs, we examined all individuals who reported having a false (ad-only conditions) versus true product experience (experience conditions) regardless of the version of the ad that they saw. An analysis of the memory quality measure with experience as the factor of interest indicated that individuals who reported false product experience did not differ from individuals who reported true product experience (MFalse = 4.83, MTrue = 5.34; F(1, 58) = 3.67, p >.05). Further, respondents reported equivalent subjective accuracy in their memory of trying the popcorn regardless of whether or not they had experienced the product (MFalse = 5.81, MTrue = 5.97; F(1, 58) <. 1, p >.05).

Mediation Analysis

To support our contention that false product experience beliefs resulting from exposure to vivid ads result in stronger attitudes for respondents without product experience, we conducted a mediation analysis in line with the procedure outlined in Baron and Kenny (1986). We tested the following three relationships: (i) imagery and experience interact to predict attitude confidence, (ii) imagery and experience interact to predict product experience belief confidence, and (iii) when imagery, experience, and product experience belief confidence are included as predictor variables, only product experience belief confidence significantly predicts attitude confidence. The first two relationships have already been documented in the analyses reported above. To test for the third relationship, when imagery, experience, and product experience belief confidence were included in the regression equation, only product experience belief confidence significantly predicted attitude confidence (βFalse Beliefs = 0.40, t(86) = 3.53, p <.05; βImagery × Experience = −0.31, t(86) = −1.91, p >.05, Sobel test = −2.04, p <.05), indicating mediation of product experience beliefs on attitude confidence. A mediation analysis using the dichotomous perceived usage measure of false memory yielded similar results.

Discussion

The results of study 1 support our theorizing in that the false product experience beliefs created through high-imagery advertising appear to be similar to the genuine beliefs created through direct product experience in terms of their effects on attitude confidence. Hence, respondents who were exposed to the high-imagery target ad without experiencing the products reported attitudes that were as confident and favorable as respondents who actually tried the popcorn. This is an important finding since it implies that at least some of the effects of product experience may be elicited without actual product experience. The finding that the effects on product experience beliefs only hold for the advertised variant of the target brand and not an unadvertised variant lend confidence to our contention that the false experience effect is not an artifact of previous memories of the brand in general but is specific to the advertised variant. Similarly, the lack of effects for the advertised filler brands and unadvertised brands from the same product category as the target brand lend further support to our results.

Notably, the results also demonstrate that our conceptual framework can predict the conditions under which we can replicate findings from past research, as well as eliminate these findings. Specifically, we demonstrate that when participants were exposed to low-imagery ads without direct experience, they exhibited less confident and less favorable attitudes than participants who had direct experience with the target brand, consistent with the findings from Smith and Swinyard (1983). However, contrary to past research, we find that imagery-evoking advertisements and direct experience have equivalent effects on attitude confidence and favorability, a finding that was predicted by our conceptual framework.

Study 2

The objective of study 2 was to examine the moderating effect of plausibility on the creation of false product experience beliefs. To this end, we utilized a 2 (advertisement imagery: high vs. low) × 2 (plausibility: high vs. low) between-subjects design. In addition, we included two product experience control conditions. As compared to study 1, in which the experience condition participants were exposed to the high-imagery or low-imagery ad for the target brand and then experienced the product, we used pure experience conditions in this study. That is, we did not expose the experience conditions participants to the ads for the target brand but instead gave them the brand to try without any advertising information. This was done to rule out the possibility that in the first study, perceptions of product experience were influenced by advertising information presented prior to the product experience (Hoch and Ha 1986) and, hence, resulted in similarities in attitudinal response between the experience and high-imagery-ad–no-experience conditions. Thus, a total of six experimental conditions were used. Participants in the experience control conditions only viewed advertisements for some of the products. For the remainder of the products, including the target product, they did not see an advertisement for the product; they only received the product name and a sample to try.

Manipulations

Plausibility was manipulated using a product consisting of a real brand name and fictitious product variant name (high-plausibility condition) versus a product consisting of a fictitious brand name and fictitious product variant name (low-plausibility condition). The target brand was either Orville Redenbacher’s Gourmet Fresh popcorn or Pop Joy’s Gourmet Fresh popcorn. In addition to the target brand, seven filler brands were also used and were either all real brands (e.g., Dial) or fictitious brands (e.g., Margo; see app. C). The same high-imagery and low-imagery versions of the ad were used as in study 1, with the exception of an additional version of each that consisted of the fictitious brand name.

Procedure

One hundred and fifty-five undergraduate students participated in the study in return for course credit. For respondents in the ad conditions, the same procedure was utilized as in study 1, with eight ads presented for 35 seconds each. Respondents in the experience conditions were informed that they would be presented with some information about different brands and would be given a sample of some of the brands to try. They were exposed to four of the filler ads that were used in the main study, but for the other four products, including the target, they were shown a page that contained only the brand name for the product (e.g., Orville Redenbacher Gourmet Fresh popcorn or Pop Joy Gourmet Fresh popcorn). Subsequently, they were given a sample of each product to try. Participants experienced four different products in order to reduce focus on the target brand alone. For the target product, they were given the same amount and type of popcorn as in study 1. As with study 1, all respondents were asked to complete a self-reported task involvement measure (α =.92) before being dismissed, which revealed no differences in involvement between any of the experimental conditions (p’s >.05). One week later, all respondents were contacted via e-mail and asked to complete an online survey that contained the dependent measures. One hundred and forty-four respondents completed the follow up survey (92% response rate), and an analysis revealed no significant differences in the dropout rate by experimental condition (p >.05).

Because of the consistency in results across all the product experience belief measures in study 1, we utilized only the continuous belief confidence measure in study 2. The measures for attitude valence and attitude confidence were identical to those of study 1 (αValence =.98 and αConfidence =.96). In addition, we examined a second measure of attitude strength—attitude accessibility. Accessibility was measured by response time (in milliseconds) to the attitude scale (Fazio, Powell, and Williams 1989). We also included two questions ostensibly as practice questions (e.g., “The undergraduate program at [name of university] is … ”), which were similar in format to the attitude valence scales to use as covariates while analyzing the response time measure in order to rule out individual differences in reading and typing.

Results

We present the analysis for the ad-only conditions (i.e., no experience) first and then present the relevant planned contrasts for the experience conditions (table 3). We expected to replicate the findings from study 1 when respondents viewed ads for fictitious product variants with real brand names (i.e., high-plausibility conditions) but not when they viewed ads for products with fictitious brand names (i.e., low-plausibility conditions). In line with our expectations, an analysis of variance revealed the expected significant interaction between imagery and plausibility on product experience belief confidence (F(1, 92) = 8.86, p <.05), attitude confidence (F(1, 94) = 4.81, p <.05), attitude accessibility (F(1, 91) = 4.33, p <.05), and attitude valence (F(1, 94) = 7.99, p <.05). Planned contrasts supported our theorizing that the interactions were driven by differences between the high- and low-imagery conditions when the brand was Orville Redenbacher but not when the brand was Pop Joy.

View this table:
Table 3.

Study 2 Results

graphic

Attitude Strength

As expected, respondents who viewed the high-imagery ad for the real brand reported greater attitude confidence (t(94) = 3.98, p <.05) and exhibited greater attitude accessibility (t(94) = −2.81, p <.05) than respondents who viewed the low-imagery ad. There were no such significant differences between the high-imagery and low-imagery conditions when the brand was fictitious (p’s >.05), offering support for hypothesis 2. Analyses of the filler brands revealed no significant effects (p’s >.05).

Attitude Valence

A similar pattern of results was found for attitude valence with respondents who viewed the high-imagery ad for the real brand reporting more favorable attitudes toward the target brand than respondents who viewed the low-imagery ad (t(94) = 3.98, p <.05). Respondents who viewed the two versions of the ad for the fictitious brand did not differ in their reported attitude favorability (p >.05).

Product Experience Belief Confidence

As expected, when the brand was real, high-imagery respondents were more likely to confidently believe that they had tried the product than low-imagery respondents (t(94) = 3.96, p <.05). However, when the brand was fictitious, there was no difference in the reported confidence between the two imagery conditions (p >.05).

Mediation Analysis

We conducted a mediation analysis to further examine whether the effects of imagery and plausibility on attitudes were mediated by the false beliefs about previously encountering the popcorn brand. We earlier reported the interactive effects of imagery and plausibility on attitude confidence, attitude valence, and belief confidence. In line with our expectations, when imagery, plausibility, and belief confidence were all included as predictors, only belief confidence significantly predicted attitude confidence (βImagery × Plausibilty = −.09, t(94) = 1.72, p >.05, βFalse Beliefs =.50, t(94) = 5.02, p <.05, Sobel statistic = −3.6, p <.05) and attitude valence (βImagery × Plausibilty = −.14, t(94) = −1.69, p >.05; βFalse Beliefs =.62, t(94) = 7.26, p <.05, Sobel statistic = −4.3, p <.05).

True Brand Experience versus False Brand Experience

Our expectation (hypothesis 2) was that when the plausibility of having encountered the target product in the past was high (i.e., the target brand was Orville Redenbacher), the high-imagery ad condition and the experience conditions would produce equivalent results for attitudes because both would lead to an increased likelihood that individuals would believe that they had experienced the product. On the other hand, when the target brand was fictitious (i.e., Pop Joy), we expected participants who viewed the high-imagery ad would mimic participants who viewed the low-imagery ad in that they would exhibit weaker and less favorable attitudes toward the target brand than participants who tasted the product (i.e., experience conditions) because the implausibility of having tried the product previously would decrease the likelihood of the false experience effect.

As expected, when the target brand was real (Orville Redenbacher), we found no significant differences between the high-imagery respondents and the experience respondents on their reported attitude confidence, attitude accessibility, attitude valence, or belief confidence (all p’s >.05). However, when the target brand was fictitious (Pop Joy), we found significant differences between the high-imagery ad and experience conditions on their reported attitude confidence (t(94) = −4.31, p <.05) and product experience belief confidence (t(94) = −2.75, p <.05), but contrary to our expectations, we did not find any differences between the two groups of respondents in terms of their attitude valence or attitude accessibility (p’s >.05).

These results suggest that the experience of the fictitious brand may not have been as powerful as the experience of the real brand, perhaps due to the limited information respondents received about the target brand in the experience conditions (i.e., brand name only). Given the paucity of information about the target product, respondents for whom the brand was fictitious may have had difficulty encoding the experience compared to the respondents in the real brand experience condition who likely had more knowledge and information about the brand. In line with this explanation, we found significant differences between the real and fictitious brands in terms of belief confidence (MReal = 6.48, MFalse = 4.37; t(94) = 3.31, p <.05). While testing this line of reasoning is outside the scope of the current research, future research may further examine this finding.

Discussion

The results of study 2 support hypothesis 2 and our contention that the false experience effect is moderated by the plausibility of the product experience event. When provided with vivid information about a fictitious brand, respondents were less likely to generate false beliefs about their past experience with the product and therefore did not differ significantly in terms of attitudinal outcomes from respondents who were provided with pallid information about the fictitious brand. Further, respondents in both the vivid and pallid ad fictitious brand conditions differed significantly from the respondents in the fictitious brand experience conditions in terms of attitude confidence and beliefs about past product usage. However, when respondents read vivid information about a real brand, rendering the imaginary experience plausible in real life, the high-imagery ad respondents did not differ from the respondents who experienced the brand in terms of their attitude valence or strength. Similarly, respondents who viewed the low-imagery ad for the real target brand reported significantly weaker and less favorable attitudes than respondents who viewed the high-imagery ad or who experienced the product.

Thus, when the plausibility of having encountered the target product in the past was high, we replicated the false experience effect found in study 1. Moreover, we extended the findings from study 1 to include an additional measure for attitude strength (accessibility). However, when the plausibility of having encountered the target product in the past was low, we did not find evidence of the false experience effect, supporting our claim that plausibility of the false memory is a moderator of this phenomenon. Further, the results of study 2 mirror the mediation results found in study 1, with false product experience beliefs mediating the effect of imagery and plausibility on attitude confidence.

Study 3

The objective of study 3 was to examine evaluation timing as a boundary condition of the false experience effect to determine if the effects found in studies 1 and 2 could be attributed to factors other than false memory. We hypothesized that if an erroneous belief about experiencing a product in the past underlies the effects demonstrated in studies 1 and 2, then instructing participants to complete the target measures immediately following exposure to the ads, when memory accuracy errors are usually small, versus following a delay, when memory errors are expected to occur, should attenuate the false experience effect. To accomplish this, respondents were exposed to a set of five print advertisements for different products, and they were asked to evaluate the experience either immediately or 1 week following exposure to the ads. As in our previous studies, each of the five products contained in the ads was a fictitious product variant with a real brand name (e.g., Dial Naturals soap). Respondents were exposed to a low-imagery advertisement for each of four filler brands. For the target brand (Colgate All-Day), which was placed in the second position in the series, participants saw either a high-imagery or a low-imagery ad (see app. B). The target ad was pretested with 56 undergraduates according to the procedure utilized in studies 1 and 2 to ensure that the two versions differed on their image-provoking ability (MHigh = 4.50, MLow = 3.40; F(1, 54) = 6.01, p <.01) and did not differ on other dimensions (p’s >.05).

Design and Procedure

A 2 (type of advertisement: high-imagery vs. low-imagery) × 2 (evaluation timing: immediate vs. 1 week delay) between-subjects design was utilized. One hundred and three undergraduate students received course credit in exchange for their participation. Respondents were given 25 seconds to view each of the five ads, consistent with the use of time limits in studies 1 and 2 and with the time limit used in the pretest. Immediately following the presentation of the advertisements, participants completed a short survey in which they reported their involvement with the ads in the same manner as in studies 1 and 2 (α =.93). There were no differences in involvement across any of the conditions (p’s >.05). Respondents in the immediate condition subsequently completed the target measures, while respondents in the delay condition completed a set of filler questions. One week after the original study, respondents in the delay condition were asked to complete a follow-up survey that contained the dependent measures.

All of the dependent measures from study 2 (product experience belief confidence, attitude valence [α =.95], attitude confidence [α =.96], and attitude accessibility) were included in study 3. In addition, we included purchase intentions to examine a third measure of attitude strength: attitude–behavioral intention correlations. Purchase intent was measured by asking respondents to indicate how likely it was that they would purchase Colgate All-Day toothpaste the next time they had a need to purchase toothpaste (1 = not at all likely, 7 = very likely). Ninety-six respondents completed the follow-up survey (93% response rate), and an analysis revealed no significant differences in the dropout rate by experimental condition (p >.05).

Results

Attitude Confidence

An analysis of variance revealed the expected interaction between imagery and evaluation timing (F(1, 92) = 4.33, p <.05). Planned contrasts revealed that when respondents completed the dependent measures immediately, they exhibited no differences in attitude confidence regardless of their exposure to the high-imagery ad or the low-imagery ad. However, when respondents completed the dependent measures following a delay, those who were exposed to the high-imagery ad exhibited more confident attitudes toward the target brand than those who were exposed to the low-imagery ad (t(92) = −3.03, p <.05; table 4). No other effects were significant (p’s >.05). An analysis of the filler brand attitudes revealed no significant interactive effect of imagery and timing (p >.05).

View this table:
Table 4.

Study 3 Results

graphic

Attitude Accessibility

Attitude accessibility exhibited the same pattern of results with a significant interaction between imagery and evaluation timing (F(1, 92) = 4.05, p <.05). Following a delay, respondents who viewed the high-imagery ad exhibited greater attitude accessibility than respondents who viewed the low-imagery ad (t(92) = 2.17, p <.05). However, immediately following ad exposure, respondents in the high-imagery and low-imagery ad conditions did not exhibit a significant difference in attitude accessibility (p >.05).

Attitude-Purchase Intention Correlation

A regression analysis revealed that attitudes significantly predicted purchase intentions for the target product (β =.62; t(94) = 7.68, p <.05), and respondents in all conditions exhibited a significant attitude-intention correlation (p’s <.05). However, following a delay, respondents exposed to the low-imagery ad reported a lower correlation between target brand attitudes and purchase intentions than respondents exposed to the high-imagery ad (z = −1.97, p <.05), while immediately after exposure to the ads, respondents in the two ad conditions did not differ in their correlations (z = −1.90, p >.05). Thus, hypothesis 3 was supported.

Attitude Valence

An analysis of variance with imagery and evaluation timing as the factors of interest showed only a significant main effect of imagery such that respondents who were exposed to the high-imagery ad reported more favorable attitudes (M = 5.35) than respondents who were exposed to the low-imagery ad (M = 4.86; F(1, 92) = 5.75, p <.05). Planned contrasts revealed that when respondents completed the dependent measures immediately, they exhibited no differences in their reported attitudes regardless of their exposure to the high-imagery versus low-imagery ad (p >.05). However, following a delay, respondents who viewed the high-imagery ad reported more favorable evaluations than respondents who viewed the low-imagery ad (t(92) = −2.91, p <.05). An analysis of the filler brand attitudes revealed no significant effects (p’s >.05).

Product Experience Belief Confidence

An analysis of variance revealed the expected significant interaction between imagery and evaluation timing (F(1, 92) = 14.08, p <.05). Respondents who completed the target measures immediately following exposure to the ads reported equivalent confidence in their belief about trying or not trying the target product if they viewed the high-imagery versus the low-imagery target ad (p >.05). However, when respondents completed the target measures following a delay, respondents who viewed the high-imagery versus low-imagery ad reported greater confidence in having tried Colgate All-Day (t(92) = −3.30, p <.05). No other effects were significant (p’s >.05). The analysis of the filler brands revealed no significant main effects or interaction effects (p’s >.05).

Mediation Analysis

Consistent with studies 1 and 2, we expected to find that false beliefs about previously encountering a product mediated the effect of imagery and evaluation timing on attitude confidence and attitude valence. We earlier reported the interactive effects of imagery and evaluation timing on attitude confidence, attitude valence, and belief confidence. As expected, with evaluation timing, imagery, and belief confidence in the regression equation, only belief confidence significantly predicted attitude confidence (βFalse Beliefs = 0.30, t(92) = 2.84, p <.05; βTiming × Imagery = 0.16, t(92) = 0.98, p >.1) and attitude valence (βFalse Beliefs = 0.25, t(92) = 2.35, p <.05; βTiming × Imagery = 0.19, t(92) = 1.17, p >.1), indicating full mediation by false product beliefs on attitude confidence (Sobel test = 2.26, p <.05) and attitude valence (Sobel test = 1.99, p <.05). These results provide additional evidence that false product experience memory is the underlying mechanism responsible for the effects of imagery on attitude confidence and valence.

Discussion

Consistent with hypothesis 3, we find that the timing of the evaluations moderates the false experience effect. Participants exposed to a high-imagery ad versus a low-imagery ad exhibited stronger and more favorable product attitudes when their evaluations were assessed following a delay, replicating the results from studies 1 and 2. However, this difference dissipated when evaluations were assessed immediately following exposure to the ads. Thus, we demonstrate another condition under which the false experience effect can be reduced. Further, the effect on attitude strength and favorability was noticed only after a delay of 1 week, not immediately, a finding consistent with a memory explanation for the results. That is, respondents must have relied on their beliefs about past experience with the advertised brands while reporting their judgments, increasing our confidence that false memory is driving the attitude effects rather than the reverse.

General Discussion

Past research on false memory has recently begun documenting its downstream effects (i.e., attitudes and behavioral intentions); however, research has not been clear with respect to its comparative effects with genuine experience (Bernstein and Loftus 2009; Heaps and Nash 2001). Our research makes a significant contribution to this literature by documenting that a mistaken belief that an event occurred (i.e., false memory) and genuine memory can produce similar effects on attitudes (i.e., false experience effect). Specifically, we find that a single exposure to a high-imagery ad elicits attitudes that are similar to real product experience in terms of their confidence (studies 1–3), accessibility (studies 2 and 3), predictive ability (study 3), and favorability (studies 1–3) as a result of the increased likelihood that individuals erroneously believe that they tried an advertised product. The effects on attitudes are more pronounced for plausible brands (study 2) and after a delay between exposure and measurement (study 3), which is consistent with a false memory explanation.

Our research contributes to the false memory literature by considering additional consequences of false memories and, specifically, by showing equivalent attitudinal outcomes of true and false beliefs about an event. Previous research in false memory has typically focused on the characteristics of false memories (Smith, Gleaves, and Pierce 2003) and the variables that have an impact on the creation of such memories (Loftus and Pickrell 1995). In addition, false memory research in marketing has shown that advertising can create false memories (Braun-LaTour et al. 2004; Lakshmanan and Krishnan 2009). Our findings add to this emerging stream of research by providing evidence that false beliefs about past events that arise through exposure to advertising may produce attitudes that are as strong as actual product experience. This finding is important because stronger attitudes have been shown to be more accessible in memory, more predictive of behavior, more resistant to counter persuasion, and more persistent over time (Petty and Krosnick 1995). Further, research has demonstrated that strong positive attitudes increase the likelihood that consumers will include the brand in their consideration sets and subsequently purchase the brand (Priester et al. 2004).

While our findings are consistent with past research on false memory that demonstrates such effects through advertising exposure, our finding that evaluation timing enhances the false experience effect is discrepant with some research that has found the effects of false memory immediately after exposure to an ad (Braun-LaTour et al. 2004; Schlosser 2006). A comparison of our work with this past research shows two key differences. First, while some of this research examines false memory in a context akin to a learning context (e.g., recognition of product attributes shown in the ad), our research focuses on autobiographical memories of experiencing the product shown in the ad. This difference suggests that future research should continue to address the theoretical similarities and differences between false memory in various contexts. Second, our research focuses on establishing false beliefs about relatively recent product experiences, unlike some past research, which has focused on childhood memories (e.g., a visit to Disneyland). As such, an important area for future research is an investigation of how the distance of the past experience for which one is attempting to create false beliefs has an impact on the emergence of false memory.

We also contribute to the literatures in imagery and product experience. The imagery literature has typically focused on elaboration as the process underlying imagery effects on attitudes and behavior (Bone and Ellen 1992; MacInnis and Price 1987). However, Petrova and Cialdini (2005) and Schlosser (2006) found differences in the effects of imagery on consumers’ judgments without differences in respondents’ elaboration. Their findings suggest that additional underlying mechanisms may be responsible for the differential effects of imagery. Consistent with their theorizing, our studies suggest that an additional process—false product experience beliefs—may also explain imagery effects on attitudes. While we exposed respondents to ads for a predetermined amount of time rather than allowing them to view ads at their own pace, in the actual marketplace consumers are not subject to such time limitations. Thus, we do not rule out the possibility that elaboration may operate in conjunction with other processes to affect attitudes, but we posit that processes other than elaboration may also underlie the effects of imagery on attitudes. Future research should examine the relationship between elaboration and false beliefs as the underlying mechanisms responsible for the effects of high-imagery advertising.

Contrary to previous research on product experience (Smith and Swinyard 1983), our findings indicate that experience may not always lead to stronger and more favorable attitudes than advertisements. Instead, we find that experience only results in stronger attitudes than does an an ad if the ad is pallid. An extremely vivid ad leads to equivalently strong and polarized attitudes as experience. Thus, in examining false memory and, in particular the false experience effect, we are able to replicate the conditions under which the past experience findings hold and to predict the conditions under which differences between experience and ad exposure are eliminated.

As with any research, there are potential caveats to our findings. First, while our results document that false beliefs about having experienced a brand arise on exposure to high-imagery advertising, it may be debatable as to whether these beliefs are best termed false memories or should be termed judgment errors. The possibility that respondents are referring to actual experiences with the genuine brands cannot be ruled out. While our first study included qualitative measures about the characteristics of these beliefs to show that they were similar to beliefs generated through actual product experience, we acknowledge that there can nonetheless be differences between false and true beliefs that our measures do not capture. We leave this issue on the extent to which the false beliefs elicited through advertising are similar to genuine experience beliefs for future research.

Second, our research examines some measures of attitude strength (attitude accessibility, attitude-purchase relationship, and attitude confidence), but it does not examine other strength measures, such as attitudinal resistance and persistence. While past research suggests that increasing attitude confidence tends to enhance the attitude-behavior relationship, as well as lead to attitudes that persist over time (Rucker et al. 2007), other research has suggested that the different dimensions of attitude strength may operate independently (Petty and Krosnick 1995) and may not always follow consistent patterns. Hence, it is unclear whether false beliefs about past product experience can protect the advertised brand from counterattitudinal information or whether the attitudes elicited via false product experience beliefs will last over time.

Third, our findings suggest that false and true product experience memories both result in more favorable attitudes toward a product; however, our investigation is limited to positive, confirmatory experiences when assessing the impact on attitude valence. If the product experience exceeds a consumer’s expectations, it is likely that consumer will exhibit favorable attitudes toward the product, but if expectations of the same experience are not met, a consumer will likely exhibit less favorable attitudes toward the product. Similarly, the attitudes elicited by ads and experience may not always be positive or consistent. For example, an ad aimed at reducing smoking may generate negative attitudes toward smoking, while the act of smoking may elicit positive attitudes toward smoking. Past research in the false memory domain has demonstrated the creation of false beliefs about a negative experience (Bernstein et al. 2005). However, an interesting area for future research is an examination of the downstream effects of positive versus negative false beliefs that confirm or disconfirm expectations about an experience. In addition, an investigation of how such outcomes affect individuals’ relationships with the brands via other dimensions, such as brand commitment, would be important to marketers.

Enhancement

  • Figure A1

  • Figure B1

Appendix A

Figure A1.

Studies 1 and 2 Target Print Ads

Appendix B

Figure B1.

Study 3 Target Print Ads

Appendix C Target and Filler Brands in Studies 1–3

Study 1
  • Target brand: Orville Redenbacher’s Gourmet Fresh popcorn

  • Advertised filler brands: Hershey’s Twirl chocolate, Dial Naturals soap, Kleenex Fresh tissues, Levi’s 702 jeans, Scope All-Day mouthwash

Study 2
  • Target brand (fictitious brand in parentheses): Orville Redenbacher’s Gourmet Fresh popcorn (Pop Joy’s Gourmet Fresh popcorn)

  • Advertised filler brands (fictitious brands in parentheses): Tropicana (Rasna) with Vitamin A orange juice, Dial (Dabur) Naturals shower gel, Tylenol (Panadol) with Vitamin A pain reliever, Kleenex (Softel) Fresh tissues, Scope (Margo) All-Day mouthwash, Levi’s (Woodland’s) 702 jeans, Gatorade (Limca) Zoom sports drink

Study 3
  • Target brand: Colgate All-Day toothpaste

  • Filler brands: Kleenex Fresh tissues, Orville Redenbacher’s Gourmet Fresh popcorn, Dial Naturals soap, Levi’s 702 jeans

References

View Abstract