The sleeper framing effect: The influence of frame valence on immediate and retrospective judgments

Available online at www.sciencedirect.com

ScienceDirect Journal of Consumer Psychology 26, 1 (2016) 53 – 65

Research Article

The sleeper framing effect: The influence of frame valence on immediate and retrospective judgments☆ Mathew S. Isaac a,⁎,1 , Morgan Poor b,1 a b

Albers School of Business and Economics, Seattle University, 901 12th Avenue, Seattle, WA 98122, USA School of Business Administration, University of San Diego, 5998 Alcalá Park, San Diego, CA 92110, USA Accepted by Cornelia Pechmann, Editor; Associate Editor, Joel Huber Received 7 July 2013; received in revised form 5 May 2015; accepted 6 May 2015 Available online 13 May 2015

Abstract Prior research on attribute framing has documented a robust valence-consistent shift whereby positively valenced options (e.g., 75% lean beef) are preferred over equivalent negatively valenced options (e.g., 25% fat beef). However, this research has typically explored how labels influence judgments of prospective or hypothetical consumption. In contrast, we examine how frames interact with actual consumption experiences to influence both immediate and retrospective judgments. We find evidence of a sleeper framing effect wherein a valence-consistent shift emerges for retrospective judgments even when absent immediately after consumption. We attribute this effect to differences in how consumers integrate the more cognitive information of the frame with the more affective information acquired during consumption. Specifically, three experiments show that consumers attend to and rely relatively more on affective information from experience when making immediate judgments, but relatively more on cognitive information from the frame when making retrospective judgments. In addition, we identify the valence of the experience as an important boundary condition, such that the sleeper framing effect is most pronounced when the experience is relatively neutral in valence. © 2015 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved. Keywords: Attribute framing; Hedonic consumption; Affect; Cognition; Retrospection; Frame valence

Framing effects are well established in human judgment and decision-making research (for a review, see Levin, Schneider, & Gaeth, 1998). Such effects occur when logically or semantically equivalent descriptions lead to different responses. One of the most pervasive findings in the framing literature is the valence-consistent shift (Levin et al., 1998, p. 160); that is, frames with positively valenced attributes (e.g., ground beef labeled as “75% lean”) are preferred over equivalent frames with negatively valenced attributes (e.g., ground beef labeled as “25% fat”), even ☆ The authors thank the Editor, the Associate Editor, three anonymous reviewers, Nidhi Agrawal, Andrea Bonezzi, Aaron Brough, Bobby Calder, Adam Duhachek, H. Shanker Krishnan, Irwin Levin, Monika Lisjak, and Carl Obermiller for their helpful comments and suggestions on earlier versions of this manuscript. ⁎ Corresponding author. E-mail addresses: [email protected] (M.S. Isaac), [email protected] (M. Poor). 1 The authors contributed equally to this work and are listed in alphabetical order.

though the two alternatives are identical (i.e., 75% lean = 25% fat) (Levin, 1987; Levin, Johnson, Russo, & Deldin, 1985). Importantly, the valence-consistent shift is a “reliable phenomenon” (Levin, Gaeth, Schreiber, & Lauriola, 2002, p. 413) only for judgments of prospective or hypothetical consumption (e.g., LeBoeuf & Shafir, 2003; Levin, Schnittjer, & Thee, 1988; Levin et al., 1985), in which consumers do not actually consume a product. The present research investigates how consumers integrate information from actual consumption experiences with information from positively and negatively valenced frames to form not only immediate judgments (i.e., evaluations made during or immediately after a consumption episode) but also retrospective judgments (i.e., evaluations made when reflecting back on a consumption episode after a delay). In cases where frame information is disclosed just once and around the same time as the consumption episode, we show that a valence-consistent shift may emerge for retrospective judgments even when it is

http://dx.doi.org/10.1016/j.jcps.2015.05.002 1057-7408/© 2015 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved.

54

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

not evident immediately after consumption. We refer to this phenomenon as the sleeper framing effect because, similar to the well-known sleeper effect in persuasion (Kumkale & Albarracin, 2004; Mazursky & Schul, 1988), our findings suggest that the relative influence of different informational inputs on consumer judgments may vary temporally. Specifically, we argue that consumers attend to and rely relatively more on the affective information garnered from actual experience when making immediate judgments, but relatively more on cognitive information from the frame when making retrospective judgments. In addition, we show that the sleeper framing effect is moderated by the valence of the consumption experience, such that a valence-consistent shift is more likely to emerge in retrospective judgments when the experience is relatively neutral, versus extremely positive or extremely negative. Theoretical background Valence-consistent shifts in framing The valence-consistent shift has been widely documented in numerous contexts where a single attribute, object, or event is framed either positively or negatively (Banks et al., 1995; Levin, 1987; Levin et al., 1985; McNeil, Pauker, Sox, & Tversky, 1982; Quattrone & Tversky, 1988). This shift has generally been attributed to associative priming (Levin & Gaeth, 1988; Levin et al., 1988; Mittal, Ross, & Tsiros, 2002; for exceptions, see Janiszewski, Silk, & Cooke, 2003 and Freling, Vincent, & Henard, 2014), with positive or negative labels encouraging the recruitment of similarly valenced information from memory. According to Krishnamurthy, Carter, and Blair (2001), positive framing “generates more positive associations and thus seems more attractive than negatively framed options” (p. 383). Valenced frames have been shown to exert influence on judgment and decision-making in both the marketplace and the workplace. For example, consumers indicated greater willingness to spend $30 on a restaurant if they were told there was a 50% chance they would be satisfied with the restaurant as opposed to a 50% chance they would be dissatisfied with the restaurant (Peterson & Wilson, 1992). In another experiment, consumers who learned about a new stereo receiver expected it to perform better if the results of product testing were framed positively (“test market results show that 85% of the users of this product were satisfied”) rather than negatively (“test market results show that 15% of the users of this product were dissatisfied”) (Buda & Zhang, 2000). Even medical students making professional decisions related to their expertise and training were influenced by valenced frames; they were more likely to advocate a surgery described using survival rates (positive valence) as opposed to equivalent mortality rates (negative valence) (Marteau, 1989). Similarly, engineers tended to allocate more funds to projects whose prior performance was framed in terms of successes (positive valence) rather than an equivalent percentage of failures (negative valence) (Duchon, Dunegan, & Barton, 1989). Although the valence-consistent shift has been observed in a wide variety of contexts, such demonstrations have almost

exclusively been made for judgments of prospective or hypothetical consumption (e.g., LeBoeuf & Shafir, 2003; Levin et al., 1985, 1988). One important exception is a single study by Levin and Gaeth (1988) in which the authors examined how positive (i.e., 75% lean) versus negative (i.e., 25% fat) attribute frames influenced consumers' evaluations of ground beef that were made after consuming identical samples of the beef. When the frame preceded the taste experience, evaluations of the positive frame group were significantly higher than evaluations of the negative frame group for three (“fat/lean,” “greasy/greaseless,” “low quality/high quality”) out of four scales. The one scale for which the valence-consistent shift was not observed was tastiness (“bad taste/good taste”), a sensory judgment that Levin and Gaeth (1988) noted was “most related to the tasting experience (p. 376).” In the same study, if the frame was provided after the taste experience but before evaluations, the valence-consistent shift was attenuated. Although these results should be interpreted cautiously since they are from a single study, they suggest that the valence-consistent shift may be weaker in cases of actual consumption (e.g., when consumption precedes framing, when making highly sensory judgments, etc.). In the present research, we further investigate how information from valenced frames is integrated with information garnered during hedonic consumption experiences to influence judgments. However, our primary objective is to determine whether the relative influence of these two inputs differs for immediate versus retrospective judgments. We define immediate judgments as evaluations provided during or within a few seconds of consumption, when experience and frame information both reside in short-term (working) memory. Most memory models agree that information remains in short-term memory for thirty seconds or less, at which time it migrates into long-term memory (Atkinson & Shiffrin, 1968; Peterson & Peterson, 1959). In our experiments, we compare immediate judgments with retrospective judgments taken 10 to 50 min after consumption, at a point when both experience and frame information would only be available in long-term (i.e., explicit, or declarative) memory. We posit that even in instances when actual consumption attenuates the valence-consistent shift for immediate judgments, it can emerge for retrospective judgments, resulting in the sleeper framing effect. The influence of cognitive and affective inputs across time The sleeper framing effect is predicated on the assumption that frames and hedonic experiences provide consumers with qualitatively different types of information. We argue that because “a frame in communication is simply the description delivered by a speaker to a listener” (Sher & McKenzie, 2011, p. 36), it can be considered a type of cognitive input. Consistent with a cognitive conceptualization, framing effects have been discussed in terms of query theory, in which consumers seek out knowledge that is potentially relevant when encountering framed options (Hardisty, Johnson, & Weber, 2010). In contrast, hedonic consumption is characterized by an affective and sensory experience of esthetic or sensual pleasure, fantasy, and fun

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

(Hirschman & Holbrook, 1982; Holbrook & Hirschman, 1982). Thus, consumption experiences are predominantly affective in nature. A growing body of literature has examined the relative influence of affective versus cognitive information on decisions that are either temporally near or temporally distant. Converging evidence from this literature suggests that the affective system of judgment is one that is anchored in the present. For example, prior research has shown that when outcomes or events are closer to the present, affect is experienced more intensely (e.g., Ekman & Lundberg, 1971) and the influence of affect on judgments and behaviors is stronger (e.g., Huber, Van Boven, McGraw, & Johnson-Graham, 2011; Van Boven, Loewenstein, Welch, & Dunning, 2012). Recently, Chang and Pham (2013) demonstrated that even when the actual accessibility and intensity of affect is held constant, consumers nevertheless rely on affect when making decisions involving temporally proximate (vs. distal) outcomes because they consider affect to be more relevant to the decision. In line with these findings, we hypothesize that consumers will rely relatively more on the affective information garnered from actual experience when making immediate judgments and relatively more on the cognitive frame information when making retrospective judgments. Furthermore, we contribute to the literature on how affective and cognitive inputs influence judgments by proposing a more nuanced process by which a temporal shift may occur. That is, we contend that a shift in the relative reliance on affective versus cognitive information may not always be due to the perceived relevance of these inputs, for we expect that consumers will perceive the integral affect experienced during consumption to be relevant for both immediate and retrospective evaluations of the hedonic experience. Instead, we propose that the sleeper framing effect occurs because of the differential devotion of attentional resources to affective versus cognitive information during consumption and in retrospect. Previous research has shown that affective stimuli tend to capture and hold attention, leaving fewer attentional resources to fully consider other, non-affective information (Anderson, 2005; Derryberry, 1993; Vuilleumier, 2005). Kempf (1999) has also demonstrated that during a hedonic experience, consumers focus more on affective dimensions of the experience and rely more on this information (vs. cognitions) when making their evaluations. Accordingly, we expect that consumers will devote a larger share of attentional resources to the affective information garnered during actual consumption (versus frame information) and rely on this information more when making immediate judgments. Importantly, we suggest that consumers may still encode frame information when it is received at around the same time as their consumption experience, even if they do not focus on or use frame information when making immediate evaluations. This becomes relevant when consumers construct retrospective evaluations, because each and every affective experience is “subtly different” and “can neither be stored nor retrieved” (Robinson & Clore, 2002, p. 935). Instead, according to Robinson and Clore's (2002) accessibility model of emotional

55

self-report, retrospective evaluations of an affective experience will reflect other sources of non-experiential information. Thus, we propose that when retrospectively evaluating a hedonic experience, consumers will devote relatively more attentional resources to cognitively reconstructing their experience using contextual details and in doing so, are more likely to incorporate information from the frame when forming their retrospective judgment. This prediction is consistent with recent research in affective forecasting, which has shown that forecasters (who are more removed from the actual hedonic experience) tend to mispredict affective evaluations of an experience because they incorporate cognitive information that experiencers ignore. For instance, Ebert and Meyvis (2014) examined how information regarding the temporal distance of a hedonic experience (e.g., reading a sad story about an event from the recent or distant past) influenced forecasted versus actual evaluations of emotional intensity. They showed that while forecasters' evaluations of emotional intensity were influenced by the more cognitive distance information, experiencers' evaluations were not. Likewise, Buechel, Zhang, Morewedge, and Vosgerau (2014) showed that probability specifications of an outcome (e.g., winning a gamble) had more impact on forecasters' hedonic responses than experiencers' hedonic responses toward the same outcome. In both of these cases, the authors argued that whereas experiencers devoted a larger share of their limited attentional resources to the affective experience or outcome, failing to take into account the more cognitive distancing information, the forecasters who were removed from the affective experience were able to devote attentional resources to more cognitive contextual information. Similarly, we hypothesize that as a hedonic experience recedes in time, consumers will attend to and rely relatively more on the cognitive contextual details they recall from the experience, including frame information. As a result, a valence-consistent shift may emerge in retrospective evaluations even if this shift was not observed immediately after the hedonic experience. The moderating role of valence of the experience We propose that an important boundary condition of the sleeper framing effect is the valence of the consumption experience. Valence can be defined as the “level of pleasantness that is generated when a stimulus is encoded and can vary from extremely negative to neutral to extremely positive” (Gomes, Brainerd, & Stein, 2013, p. 663). We expect that the sleeper framing effect is most likely to occur when the consumption experience is relatively neutral in valence versus extremely positive or negative. Support for this assertion comes from research on remembering emotional experiences, which has found that even when holding arousal constant, positively and negatively valenced affective stimuli are better remembered than neutral ones (Bradley, Greenwald, Petry, & Lang, 1992; Gomes et al., 2013; Kensinger, Brierley, Medford, Growdon, & Corkin, 2002; Kensinger & Corkin, 2003; LaBar & Phelps, 1998; Ochsner, 2000). Unlike neutral stimuli, valenced affective stimuli may result in an enhanced elaboration process at encoding (Kensinger, 2004), which has been

56

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

shown to improve memory of the affective experience (Craik & Lockhart, 1972; Craik & Tulving, 1975). Taken together, this research suggests that when a hedonic experience is extremely positive or extremely negative in valence, consumers will be better able to remember the affect they felt during consumption and thus, may still attend to and rely on it more when making retrospective judgments. As such, we expect that the sleeper framing effect will be strongest when the hedonic experience is relatively neutral in valence. Overview of experiments Next, we describe three experiments involving different sensory domains (visual, taste, auditory) and different framing manipulations (like/dislike, artificial/natural, satisfied/dissatisfied) that were designed to demonstrate the sleeper framing effect and examine its underlying process. According to our theorizing, the sleeper framing effect is possible because frames provide consumers with more cognitive information and experience provides more affective information. As previously discussed, it is well established that hedonic consumption experiences, such as the ones we examine in this paper, are primarily affective in nature (Hirschman & Holbrook, 1982; Holbrook & Hirschman, 1982). On the other hand, even though prior research may have assumed that frames are cognitive (e.g., Sher & McKenzie, 2011), to our knowledge this claim has not been examined empirically. Therefore, as a first empirical step, we elected to test the assumption that frames are perceived as more cognitive and less affective. Pilot study In order to assess whether frames are viewed as more cognitive or more affective, we conducted a pilot study with 407 participants (242 males, 165 females; average age = 32.07 years, SD = 9.74) recruited from an online panel (Amazon Mechanical Turk), in which they encountered one of the nine different frames that we considered using in our experiments (six of which were ultimately used in our three experiments). For example, participants in one of the nine conditions received the following instruction, which maps closely onto the positive frame condition that was ultimately used in Experiment 1: “Imagine that after watching a short film clip, you read the following message: ‘50% of prior viewers liked the movie clip that you just saw.’” In each case, participants were simultaneously prompted with two questions, one related to affect and one related to cognitions. Specifically, they were asked to indicate the extent to which they had focused on “emotions and feelings” and on “thoughts and cognitions” when reading the frame description (scales adapted from Chang & Pham, 2013), each on scales ranging from 0 (not at all) to 7 (a great deal). The stimuli for each of the nine frame conditions, as well as the results of this pre-test, are shown in Table 1. We conducted a 9 (frame wording) x 2 (rating type: affect, cognition) mixed ANOVA and found a significant main effect of rating type (F(1, 398) = 113.73, p b .001, ηp2 = .222) that was consistent with our theorizing. In addition, we observed a marginal main effect of frame wording (F(8, 398) = 1.93,

p = .055, ηp2 = .037), and an interaction between frame wording and rating type that was non-significant; F(8, 398) = 1.37, p N .20, ηp2 = .027. Participants in our pilot study indicated that they had focused more on cognitions than on affect when reading the frame description that they were given; all ps b .05. Across frames, the mean cognition rating (M = 4.15, SD = 1.68) was not only significantly greater than the mean affect rating (M = 2.95, SD = 1.86; t(406) = 10.64, p b .001), but also higher than the scale midpoint of 3.5; t(406) = 7.82, p b .001. On the other hand, the mean affect rating was lower than the scale's midpoint; t(406) = − 5.95, p b .001. These data indicate that pilot study participants who encountered any of our frames focused more on their thoughts and cognitions than their emotions and feelings. Based on this pilot study, we feel confident in asserting that frames are primarily cognitive, not affective. Experiment 1 The primary objective of Experiment 1 was to provide an initial demonstration of the sleeper framing effect following a hedonic consumption experience. We purposely selected an experience that prior research has shown evokes neither extremely positive nor extremely negative affect. After participants watched a neutral film clip, they were subsequently exposed to one of two frames, either positive or negative in valence. Half of the participants immediately provided hedonic evaluations of the film clip, whereas the other half provided retrospective evaluations approximately 50 min later, following several unrelated filler tasks. Our theory predicts that participants will attend to and rely relatively more on the affective information experienced while watching the film when making immediate evaluations, but relatively more on the more cognitive frame information when making retrospective evaluations. As a result, we expect that frame valence will exert greater relative impact on retrospective judgments (as compared to immediate judgments), thereby producing the sleeper framing effect. While we argue that the relative influence of frame information (vs. affective information from the consumption experience) on evaluations increases after a delay, one could reasonably make the exact opposite prediction (i.e., that the relative influence of frame information decreases after a delay) based on the order of presentation of the two informational inputs in this experiment. Prior research has demonstrated a recency–primacy shift in memory, whereby later items in a list are better recalled immediately after exposure to the list because the later items are still present in short-term (working) memory but early items are better retained once the entire list has shifted to long-term memory (Knoedler, Hellwig, & Neath, 1999). Given that the affective experience preceded frame information in our experiment, there could be a recency effect for immediate judgments (such that frame valence influences evaluations relatively more not because it is cognitive but merely because it is presented last in this study) and a primacy effect for retrospective judgments (such that direct experience influences evaluations relatively more since it is the first informational input received in this study). Because they make opposite predictions, the design of Experiment 1 allows us to

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

57

Table 1 Means (and standard deviations) of affect and cognition ratings for each frame (pilot study). Frame

Affect rating

Cognition rating

t-Stat

Imagine that after watching a short film clip, you read the following message: “50% of prior viewers liked the movie clip that you just saw.” (Study 1) Imagine that after watching a short film clip, you read the following message: “50% of prior viewers disliked the movie clip that you just saw.” (Study 1) Imagine that while chewing a mint chocolate chip gum, you read the following message: “This gum contains more than 95% natural flavors.” (Study 2) Imagine that while chewing a mint chocolate chip gum, you read the following message: “This gum contains more than 95% artificial flavors.” (Study 2) Imagine that after listening to a ring tone, you read the following message: “50% of prior listeners were satisfied with the ring tone.” (Study 3) Imagine that after listening to a ring tone, you read the following message: “50% of prior listeners were dissatisfied with the ring tone.” (Study 3) Imagine that before listening to a ring tone, you read the following message: “Based on prior listeners, there is a 50% chance you will be satisfied with this ring tone and a 50% chance you will be dissatisfied with this ring tone.” Imagine that before listening to a ring tone, you read the following message: “Based on prior listeners, there is a 50% chance you will be dissatisfied with this ring tone.” Imagine that before listening to a ring tone, you read the following message: “Based on prior listeners, there is a 50% chance you will be satisfied with this ring tone.” Overall

3.02 (1.71)

4.20 (1.63)

3.70 ⁎⁎⁎

N 45

2.47 (1.70)

4.11 (1.57)

4.57 ⁎⁎⁎

45

2.76 (1.74)

4.48 (1.70)

5.90 ⁎⁎⁎

46

2.16 (1.88)

3.87 (1.82)

5.21 ⁎⁎⁎

45

2.93 (1.83)

3.93 (1.50)

2.66 ⁎

46

3.04 (1.78)

4.11 (1.64)

4.17 ⁎⁎⁎

46

3.36 (1.79)

4.23 (1.64)

2.03 ⁎

44

3.50 (2.02)

4.41 (1.73)

2.84 ⁎⁎

46

3.32 (2.02)

3.98 (1.88)

2.04 ⁎

44

4.15 (1.68)

2.95 (1.86)

7.82 ⁎⁎⁎

407

⁎ p b .05. ⁎⁎ p b .01. ⁎⁎⁎ p b .001.

test for both the sleeper framing effect and the recency– primacy shift. Method A total of 212 undergraduate students (97 females; average age = 21.03 years, SD = 1.69) participated in this study for course credit. The study employed a 2 (frame valence: positive, negative) × 2 (evaluation timing: immediate, retrospective) between-subjects design. Participants were run in groups of 10–16 and each experimental session was conducted entirely on personal computers. Participants were told that they would be watching a black-and-white film clip from 1919 as part of a study on visual consumption. All participants then watched a two-minute clip from the film “Don't Shove,” which is a stimulus that has been shown in prior research to evoke neither extremely positive nor extremely negative affect (Geers & Lassiter, 2002; Patrick, Macinnis, & Park, 2007).2 Next, depending on frame valence condition, participants learned that 50% of prior viewers either liked (positive frame) or disliked (negative frame) the movie clip that they had already seen. Our manipulation was similar to a standard valence manipulation that has been commonly used in the framing literature (e.g., Duchon et al., 1989; Peterson & Wilson, 1992). Timing of evaluation was manipulated by having participants judge the film clip at one of two points. Participants in the immediate evaluation condition were asked to provide judgments of the film clip right away by responding to five items using unnumbered sliders. Participants rated their overall visual 2 After viewing an extended version of the same clip, pre-test participants had rated their enjoyment just below the midpoint of a 9-point scale (M = 4.21) (Geers & Lassiter, 2002, p. 1029).

experience (“not very pleasing” to “very pleasing”), judged the film's cinematography (“very poor” to “very good”), and provided general evaluations of the visuals using three other items (“very unfavorable” to “very favorable,” “disliked it” to “liked it,” and “negative” to “positive”) on unnumbered sliding scales. Responses to these five questions corresponded to numerical values ranging from 0 to 100, with higher numbers indicating more favorable evaluations (the starting position was the far left of the scale). Subsequently, participants were asked to list up to five thoughts that came to mind while watching the film clip. Participants in the retrospective condition responded to an identical set of questions but did so approximately 50 min after the other group, following several unrelated filler tasks. Results Across experiments, we analyzed the data after all responses had been collected, and we report all data exclusions, manipulations, and measures. In addition to analyzing the full dataset, we also examined a restricted sample that excluded participants who had previously seen the film clip (and might therefore rely on their previously established preferences when making evaluations) or had failed an instructional manipulation check (adapted from Oppenheimer, Meyvis, & Davidenko (2009)). Based on these exclusion criteria, there were nine fewer participants in our restricted sample than in the full dataset (five were familiar with the film clip and four failed the instructional manipulation check). However, because the results of these two analyses were similar, we report only the analysis from the full sample of participants. We predicted that frame valence would exert relatively more influence on participants' judgments of the film clip if they were made retrospectively and less influence if they were made

58

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

immediately after viewing the clip. To test our predictions, we first combined the five evaluation items into a single measure (α = .853) after a factor analysis revealed the presence of a unidimensional scale. Next, we conducted a between-subject ANOVA and found neither a main effect of frame valence (F(1, 208) = .53, p N .46, ηp2 = .003) nor a main effect of evaluation timing (F(1, 208) = .48, p N .48, ηp2 = .002). However, we found a significant interaction between frame valence and evaluation timing; F(1, 208) = 5.32, p b .03, ηp2 = .025. Immediate evaluations made by participants in the positive frame (M = 39.81, SD = 17.14, N = 51) did not differ from evaluations made by those in the negative frame (M = 43.98, SD = 20.05, N = 55); F(1, 208) = 1.24, p N .26, ηp2 = .006. On the other hand, a valence-consistent shift was observed for retrospective judgments; participants who had encountered the positive frame (M = 47.74, SD = 19.22, N = 54) judged the film clip more favorably than those who had encountered the negative frame (M = 39.72, SD = 20.29, N = 52); F(1, 208) = 4.62, p = .033, ηp2 = .022. The results of Experiment 1 appear in Fig. 1. Thought listings Two independent coders who were blind to condition examined the thoughts that participants listed after they had provided evaluations of the film clip. Coders judged whether each thought was affective (i.e., “more about feelings and sensations”), cognitive (i.e., “more about attributes or features”), or neither. Inter-rater reliability was high (91.5% agreement), and differences were resolved through discussion. For each participant, we calculated the proportion of total thoughts that were affective. A 2 (frame valence) x 2 (timing of evaluation) between-subjects ANOVA on the proportion of affective thoughts revealed a main effect of evaluation timing; F(1, 208) = 14.11, p b .001, ηp2 = .064. Specifically, participants in the immediate evaluation conditions listed a greater proportion of affective thoughts (M = .536, SD = .302, N = 106) than participants in the retrospective evaluation conditions (M = .387, SD = .285, N = 106). This is consistent with our proposition that consumers attend to and rely more on their affective experience than the cognitive frame when making immediate evaluations. There was a non-hypothesized marginal main effect of frame valence (F(1, 208) = 2.83, p = .094, ηp2 = .013), such that participants in the positive frame Positive Frame

Negative Frame

55

Evaluation

50

*

45 40

47.67 43.19

35

40.43

40.30

30 Immediate

Retrospective

Fig. 1. Evaluation of visual experience (i.e., film clip) as a function of frame valence and evaluation timing (Experiment 1). *p b .05.

conditions listed a greater proportion of affective thoughts (M = .493, SD = .313, N = 105) than participants in the negative frame conditions (M = .430, SD = .289, N = 107). However, there was not a significant interaction between evaluation timing and frame valence (F(1, 208) = .07, p N .79, ηp2 b .001). Discussion In Experiment 1, the relative influence of frame information (vs. affective information from the experience) on evaluations increased after a delay. Therefore, these results cannot be explained by a recency–primacy shift and instead provide an initial demonstration of the sleeper framing effect. The finding that frame valence influenced retrospective evaluations is particularly compelling in view of two methodological considerations of Experiment 1, namely that frame information was presented after consumption and that participants were prompted to make sensory judgments (i.e., related to their visual experience), both of which favor experience over frames and therefore make it less likely that a valence-consistent shift would be found. Furthermore, the thoughts listed by participants in Experiment 1 support our theorizing that consumers may attend to and rely relatively more on affective information derived from experience when making immediate judgments, but relatively more on cognitive information when making retrospective judgments. Our proposal is that the affective information acquired during consumption becomes less intense as time passes, which leads consumers who make retrospective judgments to allocate more attentional resources toward cognitively reconstructing their experience. If this is the case, an intervention that increases the salience of affective information at the time of a retrospective judgment should restore some of the affective intensity and therefore its influence on evaluations. Our theorizing is supported by research examining how people recall recently experienced visceral states (Nordgren, McDonnell, & Loewenstein, 2011; Nordgren, van der Pligt, & van Harreveld, 2006). In an experiment conducted by Nordgren et al. (2006), participants who took a memory test under mildly painful conditions (i.e., keeping their non-dominant arm in a bucket of ice water for approximately five minutes) performed less well on the test than a control group whose arms were not submerged. Just ten minutes later, participants in the experimental condition were asked to indicate the extent to which they attributed their poor performance on the memory test to the cold water. When making this retrospective judgment, participants were much more likely to correctly attribute their poor performance to the cold water if their arm was again submerged in ice water at the time of attribution (the painful attribution condition) than if it was submerged in room-temperature water at the time of attribution (the pain-free attribution condition). These results suggest that just a few minutes after an (admittedly mild) experience, the intensity of the affect experienced during the hedonic episode had sufficiently faded among participants in the pain-free attribution condition (perhaps during the migration into long-term memory) to allow the misattribution to occur. However, participants who were given the opportunity to relive their affective experience made the

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

correct attribution. In Experiment 2, we adapt the approach of Nordgren et al. (2006) to provide additional evidence for the process underlying the sleeper framing effect. Experiment 2 The primary objective of Experiment 2 was to provide more direct evidence for our proposal that the sleeper framing effect arises because of differences in whether consumers attend to and rely relatively more on the affective information acquired during consumption versus cognitive information such as that provided in a valenced frame. Specifically, we attempted to show that the valence-consistent shift observed in Experiment 1 for retrospective judgments can be attenuated when affective information is again made salient at the time that participants make these retrospective judgments. In order to show the robustness of the sleeper framing effect, we also used a new framing manipulation that was presented during (vs. after) consumption, as well as a different hedonic experience emphasizing a new sensory dimension (taste). We again selected an experience that was neutral in valence, which we believe is a necessary condition for a sleeper framing effect to emerge. Method A total of 119 undergraduate students (39 females; average age = 21.07 years, SD = 2.25) participated in this study for course credit. The study employed a 2 (frame valence: positive, negative) × 2 (salience of affect at time of recall: high, low) × 2 (timing of evaluation: immediate, retrospective) mixed design, where timing of evaluation was a within-subjects factor. The study was divided into two parts, both of which occurred during a single experimental session. In Part 1, participants were invited to try a piece of chewing gum. To ensure that participants would not be familiar with the gum, we chose a flavor that was relatively new to the U.S. market at the time of the study (Extra Dessert Delights Mint Chocolate Chip gum). Results of a pre-test confirmed that the gum was relatively neutral in valence.3 While they were chewing the gum, participants were shown an advertisement for the gum that remained on their screen for exactly two minutes. For the sake of robustness, we used another framing manipulation, which was a standard product attribute framing manipulation (adapted from Levin, 1987; Levin et al., 1985), and different numbers in the frame than the ones in our first experiment. Specifically, participants encountered an advertisement in Experiment 2 that prominently noted that the mint chocolate chip gum contained either “more than 95% natural flavors” (positive frame) or “less than 5% artificial flavors” (negative frame). 3 In order to ensure that the gum tasting experience was relatively neutral in valence, we conducted a pre-test with a separate set of 28 undergraduates who had not previously tasted the gum. After tasting the gum (but without viewing any frames), pre-test participants evaluated it on three 7-point scales (1 = unpleasant/bad/negative, 7 = pleasant/good/positive). The mean rating of the gum was just above the scale midpoint (M = 4.86, SD = 1.75).

59

While chewing the gum, participants were instructed to rate it on the following four dimensions using unnumbered sliders: taste, deliciousness, flavor, and quality. Responses to these four questions corresponded to numerical values ranging from 0 to 100, with higher numbers indicating more favorable evaluations (the starting position was the midpoint of the scale). After providing their initial evaluations of the gum, participants were directed to dispose of the piece of gum that they had been chewing. Next, they proceeded to a series of unrelated filler tasks that required approximately 30 min to complete. Part 2 of the experiment followed these filler tasks. In Part 2, we manipulated the salience of the affect that participants had experienced during their gum-tasting experience by adapting the previously described manipulation used by Nordgren et al. (2006). Participants in the high affect salience condition were told to take one minute to mentally recreate their earlier experience tasting the mint chocolate chip gum, focusing on the “flavors and gustatory sensations associated with the gum.” To further focus consumers' attention on affect, we placed a mint chocolate chip gum wrapper on these participants' workstations, as previous research has shown that real presentation of a stimulus can increase the intensity of affective reactions towards it (Loewenstein, 1996; Shiv & Fedorikhin, 1999). In contrast, participants in the low affect salience condition were not exposed to the mint chocolate chip gum wrapper and instead spent one minute on an unrelated task. After the salience manipulation, all participants were again asked to evaluate the mint chocolate chip gum that they had tasted earlier in the session, using the same four scales as before. Thus, unlike Experiment 1, all participants made two judgments, one immediately after tasting the gum and one in retrospect. Once the second set of evaluations was obtained, we asked participants to list up to five thoughts that had come to mind when remembering their gum-tasting experience from earlier in the session. Participants were then instructed to self-classify each thought that they had listed as being “more about features and attributes of the gum,” “more about [their] feelings and sensations,” or “neither.”

Results None of the 119 participants in Experiment 2 were excluded. Any participants who indicated in a screening questionnaire that they had previously tried the gum flavor were not allowed to participate in this experiment (and so there was no need to exclude them subsequently), and there was no instructional manipulation check embedded in this study. We predicted that frame valence would influence participants' judgments only when the affective experience itself was relatively less salient. Thus, for immediate judgments of the gum, as well as retrospective judgments that were made when affect salience was high, we anticipated relatively less influence of frame valence (compared to experience). However, consistent with our theory and the results of Experiment 1, we expected frame valence to exert relatively more influence on retrospective judgments of participants in the low affect salience condition.

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

To test our predictions, we combined the four judgments taken in Part 1 (α = .879) and the four judgments taken in Part 2 (α = .941) into two composite measures. Next, we conducted a 2 (frame valence) × 2 (affect salience at time of recall) × 2 (timing of evaluation) mixed ANOVA on judgments of the gum. There were no significant main effects of frame valence (F(1, 115) = .88, p N .35, ηp2 = .008) or salience of affect (F(1, 115) = .34, p N .56, ηp2 = .003). However, there was a significant main effect of evaluation timing (F(1, 115) = 5.64, p b .02, ηp2 = .047). None of the 2-way interactions, between evaluation timing and frame valence (F(1, 115) = 2.69, p = .104, ηp2 = .023), between evaluation timing and affect salience (F(1, 115) = 1.41, p N .23, ηp2 = .012), or between frame valence and affect salience (F(1, 115) = 1.98, p N .16, ηp2 = .017), were statistically significant. More central to our theorizing, however, we observed the 3-way interaction that we had predicted; F(1, 115) = 5.67, p b .02, ηp2 = .047. For the high affect salience group, a 2 (frame valence) × 2 (timing of evaluation) mixed ANOVA on judgments of the gum revealed only an unanticipated main effect of timing (F(1, 60) = 5.17, p b .03, ηp2 = .079) but no main effect of frame valence (F(1, 60) = .11, p N .74, ηp2 = .002) nor a significant 2-way interaction (F(1, 60) = .22, p N .63, ηp2 = .004). For the low affect salience group, a 2 (frame valence) x 2 (timing of evaluation) mixed ANOVA on judgments of the gum revealed a marginal main effect of frame valence (F(1, 55) = 2.93, p = .093, ηp2 = .051) and no main effect of evaluation timing (F(1, 55) = .97, p N .32, ηp2 = .017). More importantly, a significant 2-way interaction between frame valence and evaluation timing emerged for the low affect salience group; F(1, 55) = 11.15, p b .01, ηp2 = .169. We first examined the immediate and retrospective evaluations of participants for whom affect salience at the time of retrospective evaluation was low. Among these participants, immediate evaluations made by those who had encountered the positive frame (M = 72.68, SD = 20.18, N = 29) did not differ from evaluations made by those who had encountered the negative frame (M = 66.83, SD = 18.73, N = 28); F(1, 115) = 1.24, p N .26, ηp2 = .011. However, the retrospective evaluations from the participants who had encountered the positive frame (M = 74.90, SD = 20.31, N = 29) were more favorable than the evaluations from the participants who had encountered the negative frame (M = 62.76, SD = 21.33, N = 28); F(1, 115) = 4.09, p b .05, ηp2 = .034. These results, which replicate the sleeper framing effect, are consistent with the findings from Experiment 1. Next, we examined the immediate and retrospective evaluations of participants for whom affect salience at the time of retrospective evaluation was high. Among these participants, immediate evaluations made by those who had encountered the positive frame (M = 67.85, SD = 18.32, N = 29) did not differ from evaluations made by those who had encountered the negative frame (M = 69.06, SD = 21.72, N = 33); F(1, 115) = .06, p N .81, ηp2 = .001. Furthermore, the retrospective judgments from participants in the positive frame (M = 64.48, SD = 20.83, N = 29) and negative frame (M = 66.86, SD = 26.85, N = 33) conditions were also not

90

Positive Frame

Negative Frame

80

Evaluation

60

*

70

60

74.90

72.68 66.83

67.85 69.06 62.76

64.48

66.86

50 Immediate

Retrospective

Low Affect Salience in Retrospect

Immediate

Retrospective

High Affect Salience in Retrospect

Fig. 2. Evaluation of taste experience (i.e., mint chocolate chip gum) as a function of frame valence, evaluation timing, and affect salience (Experiment 2). *p b .05.

significantly different; F(1, 115) = .17, p N .68, ηp2 = .001. Thus, the sleeper framing effect was observed when affect salience was low at the time of retrospective evaluation, but not when affect salience was high. The results of Experiment 2 are illustrated in Fig. 2. Thought listings We also examined how participants classified their own thoughts following their second evaluation of the gum. For each participant, we calculated the proportion of total thoughts that were affective (i.e., self-classified as being “more about feelings and sensations of the gum”) rather than cognitive (i.e., “more about attributes or features of the gum.”). A 2 (frame valence) × 2 (affect salience at the time of recall) between-subjects ANOVA on the proportion of affective thoughts revealed a main effect of affect salience, whereby participants in the high affect salience condition listed a greater proportion of affective thoughts (M = .602, SD = .255) than participants in the low affect salience condition (M = .489, SD = .247); F(1, 115) = 5.86, p b .02, ηp2 = .049. There was no main effect of frame valence (F(1, 115) = .01, p N .92, ηp2 b .001) nor a significant interaction between affect salience and frame valence (F(1, 115) = .06, p N .80, ηp2 = .001). Thus, the self-classified thought listings are consistent with our claim that when affective information (e.g., feelings and sensations) is made salient, consumers attend to it more than cognitive information (e.g., frames, attributes, features, etc.) and rely on it more when making retrospective judgments. However, when affect is not salient, consumers who make retrospective judgments are more likely to consider cognitive information. Discussion Experiment 2 demonstrates the sleeper framing effect in a second sensory domain (taste experience). Unlike Experiment 1, we utilized a repeated-measures design where all participants made judgments on two occasions, during their initial consumption experience and about 30 min afterwards. This within-participants design provided a stringent test of our effect because when making their final judgments, participants might have remembered their earlier judgments and been inclined to reproduce them for the sake of consistency. Our demonstration

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

of the sleeper framing effect in this within-participants context suggests that the effect is robust. Furthermore, we provided additional evidence for the process underlying the sleeper framing effect in Experiment 2. Specifically, we showed that the valence-consistent shift occurred for retrospective judgments only when the affective experience was relatively less salient. When participants' attention was directed toward their affective reactions, retrospective judgments and immediate judgments resembled one another and the valenceconsistent shift was absent. Indeed, self-classified thought listings confirmed that when making retrospective judgments, participants focused relatively more on their “feelings and sensations” when affect was made salient but relatively more on “attributes and features” otherwise. Taken together, these results suggest that differences in the allocation of attentional resources towards affective and cognitive information when making immediate versus retrospective judgments drive the sleeper framing effect. Our first two experiments examined hedonic experiences that are relatively neutral in valence. We propose that when experiences are extremely positive or extremely negative in valence, the valence-consistent shift will be less likely to emerge. The existence of this boundary condition, which we explicitly test in Experiment 3, is consistent with research showing that positively and negatively valenced affective stimuli are better remembered than neutral ones (Bradley et al., 1992; Gomes et al., 2013; Kensinger & Corkin, 2003; Kensinger et al., 2002; LaBar & Phelps, 1998; Ochsner, 2000). We argue that when a hedonic experience is extremely positive or extremely negative in valence, consumers will be better able to remember the affect they felt during consumption and thus, may still attend to and rely on it more (vs. cognitive information) when making retrospective judgments. This should in turn attenuate the sleeper framing effect. Experiment 3 The primary objective of Experiment 3 was to examine valence extremity of the consumption experience as a boundary condition by showing that the sleeper framing effect that occurs for experiences of relatively neutral valence is attenuated for extremely positive or extremely negative experiences. Method A total of 457 participants (217 males, 230 females, 10 gender unspecified; average age = 36.02 years, SD = 12.65) recruited from an online panel (Amazon Mechanical Turk) participated in Experiment 3 for payment. The study employed a partially crossed 3 (experience valence: positive, intermediate, negative) × 2 (frame valence: positive, negative) × 2 (timing of evaluation: immediate, retrospective) between-subjects design. Specifically, participants in the positive and negative experience conditions always encountered a frame that was opposite in valence to the valence of their experience. This was done not only to conserve subjects, but also because we thought it would be more informative and theoretically

61

interesting if the frame and the experience were potentially able to push participants' evaluations in opposite directions. However, participants in the neutral experience conditions were randomly assigned to either a positive frame valence or negative frame valence condition. Thus, our design consisted of eight total conditions. Participants were told that they would be listening to a sample of a sound that could be used as a ring tone. They were asked to confirm that their computer volume was turned up and that they were in a setting where they could listen to audio on computer speakers or headphones. Next, participants listened to one of three ten-second sound samples.4 Participants either listened to the sound of a bee buzzing (negative experience), the sound of bongos playing (positive experience), or the sound of a phone ringing (relatively neutral experience). Next, all participants were exposed to market research results related to the sound sample that they had just heard. Specifically, they were informed that prior listeners had been asked to indicate whether they were satisfied or dissatisfied with the ringtone (adapted from Buda & Zhang, 2000). As previously mentioned, participants in the positive and negative experience conditions encountered a frame whose valence was opposite to their experience. Thus, participants who had heard the sound of a bee buzzing (negative experience) learned that 50% of prior listeners were satisfied (positive frame) with the sound sample. Participants who had heard the sound of bongos playing (positive experience) instead learned that 50% of prior listeners were dissatisfied (negative frame) with the sound sample. Participants who had heard the phone ringing (relatively neutral experience) encountered either a positive (satisfaction) frame or a negative (dissatisfaction) frame. Timing of evaluation was manipulated by having participants judge the sound sample at one of two points. Participants in the immediate evaluation condition were asked to provide judgments of the sound sample right away by responding to three items using unnumbered sliders. Participants rated the sound's pleasantness (“unpleasant” to “pleasant”), the sound's quality (“bad” to “good”), and their enjoyment (“not at all enjoyable” to “very enjoyable”) on unnumbered sliding scales (the starting position was the far left of the scale). Responses to these three questions corresponded to numerical values ranging from 0 to 100, with higher numbers indicating more favorable evaluations. Participants in the retrospective condition responded to an identical set of questions but did so approximately 15 min after the other group, following several unrelated filler tasks. Results In Experiment 3, we analyzed the full dataset, as well as a restricted sample that excluded twelve participants who failed an 4 We used sounds from the IADS-2 database because it provides standardized sound stimuli (Bradley & Lang, 2007). We selected three sounds from the database that varied in terms of valence (based on pre-tests with at least 100 undergraduate students), one with an extremely positive valence (M = 7.67, SD = 1.46), one with an extremely negative valence (M = 2.16, SD = 1.33), and one with a relatively neutral valence (M = 5.49, SD = 1.98), which was just above the midpoint of the 9-point scale.

62

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

instructional manipulation check (adapted from Oppenheimer et al. (2009)). However, because the results of these two analyses were similar, we report only the analysis from the full sample of participants. We predicted that frame valence would exert relatively more influence on participants' retrospective (vs. immediate) judgments of the ring tone, but only when the experience was relatively neutral in valence (i.e., phone ringing). For positively valenced (i.e., bongos playing) or negatively valenced (i.e., bees buzzing) experiences, we predicted that frame valence would exert minimal influence on both immediate and retrospective judgments. To test these predictions, we first combined the three evaluation items into a single measure (α = .750) after a factor analysis revealed the presence of a unidimensional scale. In order to accommodate our partial factorial design, we first examined responses of participants in the neutral experience conditions. For these participants, we expected to observe a sleeper framing effect, thereby mirroring the results of our first two experiments. Among participants in the neutral experience conditions (phone ringing), we found a main effect of frame valence (F(1, 232) = 9.82, p b .01, ηp2 = .041) but no main effect of evaluation timing (F(1, 232) = .12, p N .73, ηp2 = .001). More importantly, we observed a significant interaction between frame valence and timing that is consistent with the results of our prior studies; F(1, 232) = 5.74, p = .017, ηp2 = .024. After hearing the ringing phone sound sample, immediate evaluations made by participants in the positive frame (M = 49.44, SD = 23.73, N = 63) did not differ from evaluations made by those in the negative frame (M = 47.36, SD = 19.78, N = 57); F(1, 232) = .28, p N .59, ηp2 = .001. However, in retrospect, participants who had encountered the positive frame (M = 57.18, SD = 22.45, N = 60) judged the same ring tone more favorably than those who had encountered the negative frame (M = 41.55, SD = 20.23, N = 56); F(1, 232) = 15.04, p b .001, ηp2 = .061. Next we examined the responses of participants who had a positive experience (bongos playing) but encountered a negative frame (dissatisfied). Participants in this group evaluated their ring tone similarly, irrespective of whether their evaluation occurred immediately (M = 57.95, SD = 23.06, N = 58) or retrospectively (M = 59.12, SD = 21.69); t(109) = .28, p N .78, d = .05.

70

Positive Frame

Negative Frame

***

Evaluation

60 50 40

57.95

57.18

30 20 10

49.44 47.36 34.08

59.12

41.55

31.33

Immediate Retrospective Immediate Retrospective Immediate Retrospective Negative Experience Neutral Experience Positive Experience (Bees Buzzing) (Phone Ringing) (Bongos Playing)

Fig. 3. Evaluation of auditory experience (i.e., ring tone sample) as a function of frame valence, evaluation timing, and experience valence (Experiment 3). ***p b .001.

Likewise, participants who had a negative experience (bees buzzing) but encountered a positive frame (satisfied) provided lower overall evaluations but evaluated their ring tone similarly, irrespective of whether their evaluation occurred immediately (M = 34.08, SD = 20.75, N = 56) or retrospectively (M = 31.33, SD = 18.11, N = 54); t(108) = .74, p N .46, d = .14. The results of Experiment 3 appear in Fig. 3.

Discussion The results of Experiment 3 support our contention that an important boundary condition of the sleeper framing effect is the valence of the hedonic consumption experience. When the listening experience was relatively neutral in valence, we again observed a valence-consistent shift, as we had previously found in Experiments 1 and 2. However, when the listening experience was either extremely positive or extremely negative, immediate and retrospective judgments did not differ from one another.

General discussion The three experiments described in this article provide evidence of a sleeper framing effect across three sensory domains (i.e., visual, gustatory, auditory) using three different framing manipulations. We demonstrate that while exposure to a positive or negative frame has relatively less effect on judgments made immediately after an identical hedonic consumption experience, frame valence has greater influence on retrospective judgments that are made following a short delay. As a result, it is possible to observe a valence-consistent shift (whereby positive frames lead to more favorable evaluations than equivalent negative frames) for retrospective judgments even if it is absent for immediate judgments. We further show that this sleeper framing effect is more likely to emerge when the experience itself is of relatively neutral valence. We also find evidence that is supportive of the proposed underlying process driving the sleeper framing effect. Specifically, we show that the effect arises because consumers attend to and rely relatively more on affective information acquired during consumption when making immediate judgments, but relatively more on cognitive frame information when making retrospective judgments. In support of our theorizing, we demonstrate that post-evaluation thoughts tend to be more cognitive (vs. affective) following a retrospective judgment, as compared to an immediate judgment (Experiment 1) or a retrospective judgment in which affective information is made salient (Experiment 2). And finally, we find that the influence of frames on retrospective judgments is attenuated if the hedonic consumption experience is itself extremely positive or negative in valence (Experiment 3) or if consumers are prompted to focus on affective information during retrospection (Experiment 2). Taken together, this evidence supports our conclusion that the affective information from experience is attended to and relied on more [less] than the cognitive frame information when consumers make immediate [retrospective] judgments.

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

Theoretical contributions The present research contributes to the extensive literature on framing effects in two important ways. First, we provide a more thorough investigation of how consumers integrate information from frames with information from actual experience. Although the valence-consistent shift in framing has been documented consistently for judgments of prospective or hypothetical consumption (e.g., LeBoeuf & Shafir, 2003; Levin et al., 1985, 1988), there is limited evidence that this effect persists in contexts where consumers have their own direct experience on which to base their evaluations. We are aware of only the single study by Levin and Gaeth (1988) showing that actual consumption might mitigate the effects of valenced frames, especially for sensory judgments that are most related to the consumption experience. In the present research, we provide robust evidence that when consumers are exposed to both a frame and a hedonic consumption experience at about the same time, the valence-consistent shift is attenuated for immediate judgments because the more cognitive frame information is overshadowed by the primarily affective information garnered during the consumption experience. Furthermore, this research is the first to our knowledge that examines the influence of valenced frames on both immediate and retrospective judgments. Similar to other documented sleeper effects in persuasion (Kumkale & Albarracin, 2004; Mazursky & Schul, 1988) and priming (Srull & Wyer, 1980), our findings demonstrate that information can under certain circumstances become paradoxically more influential after a delay versus immediately. The present research also contributes to the growing body of literature examining the differential effects of affective and cognitive information on evaluations across time. Previous research has provided cogent evidence suggesting that the affective system is inherently anchored in the present (e.g., Chang & Pham, 2013; Huber et al., 2011; Pham, 2004, 2007; Van Boven et al., 2012). Chang and Pham (2013) recently demonstrated this phenomenon, showing that incidental affect influences judgments more if outcomes and targets are closer to the present than when outcomes and targets are temporally more distant (past or future). The authors argued that these effects were driven by the greater perceived information value of affective information in judgments whose outcomes and targets were closer to the present. Our findings may be considered an instantiation of this phenomenon in a novel context, wherein we show that the integral affect acquired during hedonic experiences is relied on more than the more cognitive frame information when making immediate hedonic judgments, but less so when making retrospective judgments. Furthermore, we show that an important and unexplored boundary condition of this phenomenon is the valence of the experience, such that in the case of extremely positive or extremely negative experiences, individuals may continue to rely on affect over cognitive information much longer than in the case of relatively neutral experiences. In addition, we extend previous research by suggesting a more nuanced process through which the sleeper framing effect occurs. That is, our findings suggest that consumers do not simply rely relatively more on affective information (versus cognitive

63

information) when making immediate judgments because affect has greater perceived information value. Instead, drawing on research showing systematic mispredictions in affective forecasting (e.g., Buechel et al., 2014; Ebert & Meyvis, 2014), our findings suggest that affective information captures a larger share of consumers' limited attentional resources when making immediate judgments, but less attentional resources when consumers are removed in time from the experience. Indeed, we show that when attention is once again drawn to affective information when consumers are making retrospective evaluations, they once again rely relatively more on this information. Comparison to prior research Although our research is one of the first to explicitly examine the interaction between a valenced frame and an actual hedonic consumption experience, and the first to consider both immediate and retrospective judgments, related studies have explored how advertising claims (e.g., Hoch & Ha, 1986), which typically provide positively valenced information about the advertised product, or product labels (e.g., Allison & Uhl, 1964; Lee, Frederick, & Ariely, 2006) influence post-trial product judgments. With some exceptions, this literature has documented that secondhand information often influences immediate judgments. In contrast, we found that valenced frames consistently failed to influence immediate judgments, even in cases where they influenced retrospective judgments. While our results may seem at odds with previous research, the apparent discrepancy might be explained by any number of methodological differences. For example, participants in our experiments were exposed to relatively neutral hedonic experiences and primarily made judgments related to these sensory experiences (e.g., “tastiness”). Further, the cognitive information provided by the frames in our experiments differs on multiple dimensions from the cognitive information provided in most advertisements and product descriptions (e.g., perceptions of the source's credibility, use of numbers/percentages, etc.), all of which may contribute to different effects. We certainly acknowledge that there may be circumstances where frames do in fact have an effect on immediate judgments. However, this does not invalidate our theorizing, as our main proposition is that the relative influence that the more cognitive frame information versus the more affective information acquired through consumption has on evaluative judgments changes over time. We suspect that even when a valence-consistent shift is observed immediately, it may become more pronounced in retrospective judgments. In our view, findings such as these would be wholly consistent with our explanation for the sleeper framing effect. Finally, the sleeper framing effect relates to research on the reconstructive nature of memory, which has similarly found that non-experiential information can impact judgments that follow a consumption experience (Braun, 1999; Braun-LaTour & LaTour, 2005). Whereas Braun (1999) showed that advertising information acquired after a consumption experience (and after the information acquired during that consumption experience has shifted to long-term memory) can transform memory of that experience, our findings indicate that frame information acquired

64

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65

at around the same time as an experience can also influence retrospective judgments, even without influencing immediate judgments of the experience. Relatedly, while research on reconstructive memory might have predicted differences in immediate and retrospective judgments only if new information were acquired in the interim between these two judgments, we find that differences are possible even if all informational inputs precede the immediate judgment. Ultimately, however, our findings are aligned in that they challenge the prevailing view that post-consumption advertising will have relatively little effect on product evaluations and show that even postconsumption information can sometimes overshadow respondents' memory of their actual consumption experience. Managerial implications This research also has important practical implications for managers. Marketers may be under the mistaken impression that advertising primarily influences product trial, but that subsequent product purchase (and repurchase) decisions will be based on consumers' first-hand experience with the product. Despite the importance of direct experience in influencing consumer judgments immediately following consumption, our research shows that framed information provided in advertising and other marketing materials may influence retrospective consumer judgments within minutes of a direct consumption experience. Our findings suggest that secondhand information (e.g., frames, advertisements) encountered just after or even during an experience may quickly overshadow memories of the experience. Thus, our research implies that frames and other cognitive inputs can be important influencers of retrospective evaluations and downstream repurchase decisions. As such, marketers would be wise to always ensure that non-experiential information is framed as positively as possible and to bear in mind that the failure of an advertisement to influence immediate judgments does not necessarily imply that the advertisement has failed, because it may still influence how the experience is reconstructed. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jcps.2015.05.002. References Allison, R. I., & Uhl, K. P. (1964). Influence of beer brand identification on taste perception. Journal of Marketing Research, 1, 36–39. Anderson, A. K. (2005). Affective influences on the attentional dynamics supporting awareness. Journal of Experimental Psychology: General, 134, 258. Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. Psychology of Learning and Motivation, 2, 89–195. Banks, S. M., Salovey, P., Greener, S., Rothman, A. J., Moyer, A., Beauvais, J., et al. (1995). The effects of message framing on mammography utilization. Health Psychology, 14, 178–184.

Bradley, M. M., Greenwald, M. K., Petry, M. C., & Lang, P. J. (1992). Remembering pictures: Pleasure and arousal in memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 379. Bradley, M. M., & Lang, P. J. (2007). The international affective digitized sounds: Affective ratings of sounds and instruction manual. Gainesville, FL: University of Florida. Braun, K. A. (1999). Postexperience advertising effects on consumer memory. Journal of Consumer Research, 25, 319–334. Braun-LaTour, K. A., & LaTour, M. S. (2005). Transforming consumer experience: When timing matters. Journal of Advertising, 34, 19–30. Buda, R., & Zhang, Y. (2000). Consumer product evaluation: The interactive effect of message framing, presentation order, and source credibility. Journal of Product & Brand Management, 9, 229–242. Buechel, E. C., Zhang, J., Morewedge, C. K., & Vosgerau, J. (2014). More intense experiences, less intense forecasts: Why people overweight probability specifications in affective forecasts. Journal of Personality and Social Psychology, 106, 20. Chang, H. H., & Pham, M. T. (2013). Affect as a decision-making system of the present. Journal of Consumer Research, 40, 42–63. Craik, F. I., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11, 671–684. Craik, F. I., & Tulving, E. (1975). Depth of processing and the retention of words in episodic memory. Journal of Experimental Psychology: General, 104, 268. Derryberry, D. (1993). Attentional consequences of outcome-related motivational states: Congruent, incongruent, and focusing effects. Motivation and Emotion, 17, 65–89. Duchon, D., Dunegan, K. J., & Barton, S. L. (1989). Framing the problem and making decisions: The facts are not enough. IEEE Transactions on Engineering Management, 36, 25–27. Ebert, J. E., & Meyvis, T. (2014). Reading fictional stories and winning delayed prizes: The surprising emotional impact of distant events. Journal of Consumer Research, 41, 794–809. Ekman, G., & Lundberg, U. (1971). Emotional reaction to past and future events as a function of temporal distance. Acta Psychologica, 35, 430–441. Freling, T. H., Vincent, L. H., & Henard, D. H. (2014). When not to accentuate the positive: Re-examining valence effects in attribute framing. Organizational Behavior and Human Decision Processes, 124, 95–109. Geers, A. L., & Lassiter, G. D. (2002). Effects of affective expectations on affective experience: The moderating role of optimism–pessimism. Personality and Social Psychology Bulletin, 28, 1026–1039. Gomes, C. F., Brainerd, C. J., & Stein, L. M. (2013). Effects of emotional valence and arousal on recollective and nonrecollective recall. Journal of Experimental Psychology: Learning, Memory, and Cognition, 39, 663. Hardisty, D. J., Johnson, E. J., & Weber, E. U. (2010). A dirty word or a dirty world? Attribute framing, political affiliation, and query theory. Psychological Science, 21, 86–92. Hirschman, E. C., & Holbrook, M. B. (1982). Hedonic consumption: Emerging concepts, methods and propositions. Journal of Marketing, 92–101. Hoch, S. J., & Ha, Y. -W. (1986). Consumer learning: Advertising and the ambiguity of product experience. Journal of Consumer Research, 221–233. Holbrook, M. B., & Hirschman, E. C. (1982). The experiential aspects of consumption: Consumer fantasies, feelings, and fun. Journal of Consumer Research, 9, 132–140. Huber, M., Van Boven, L., McGraw, A. P., & Johnson-Graham, L. (2011). Whom to help? Immediacy bias in judgments and decisions about humanitarian aid. Organizational Behavior and Human Decision Processes, 115, 283–293. Janiszewski, C., Silk, T., & Cooke, A. D. (2003). Different scales for different frames: The role of subjective scales and experience in explaining attributeframing effects. Journal of Consumer Research, 30, 311–325. Kempf, D. S. (1999). Attitude formation from product trial: Distinct roles of cognition and affect for hedonic and functional products. Psychology & Marketing, 16, 35–50. Kensinger, E. A. (2004). Remembering emotional experiences: The contribution of valence and arousal. Reviews in the Neurosciences, 15, 241–252.

M.S. Isaac, M. Poor / Journal of Consumer Psychology 26, 1 (2016) 53–65 Kensinger, E. A., Brierley, B., Medford, N., Growdon, J. H., & Corkin, S. (2002). Effects of normal aging and Alzheimer's disease on emotional memory. Emotion, 2, 118. Kensinger, E. A., & Corkin, S. (2003). Memory enhancement for emotional words: Are emotional words more vividly remembered than neutral words? Memory & Cognition, 31, 1169–1180. Knoedler, A. J., Hellwig, K. A., & Neath, I. (1999). The shift from recency to primacy with increasing delay. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 474–487. Krishnamurthy, P., Carter, P., & Blair, E. (2001). Attribute framing and goal framing effects in health decisions. Organizational Behavior and Human Decision Processes, 85, 382–399. Kumkale, G. T., & Albarracin, D. (2004). The sleeper effect in persuasion: A meta-analytic review. Psychological Bulletin, 130, 143–172. LaBar, K. S., & Phelps, E. A. (1998). Arousal-mediated memory consolidation: Role of the medial temporal lobe in humans. Psychological Science, 9, 490–493. LeBoeuf, R. A., & Shafir, E. (2003). Deep thoughts and shallow frames: On the susceptibility to framing effects. Journal of Behavioral Decision Making, 16, 77–92. Lee, L., Frederick, S., & Ariely, D. (2006). Try it, you'll like it: The influence of expectation, consumption, and revelation on preferences for beer. Psychological Science, 17, 1054–1058. Levin, I. P. (1987). Associative effects of information framing. Bulletin of the Psychonomic Society, 25, 85–86. Levin, I. P., & Gaeth, G. J. (1988). How consumers are affected by the framing of attribute information before and after consuming the product. Journal of Consumer Research, 15, 374–378. Levin, I. P., Gaeth, G. J., Schreiber, J., & Lauriola, M. (2002). A new look at framing effects: Distribution of effect sizes, individual differences, and independence of types of effects. Organizational Behavior and Human Decision Processes, 88, 411–429. Levin, I. P., Johnson, R. D., Russo, C. P., & Deldin, P. J. (1985). Framing effects in judgment tasks with varying amounts of information. Organizational Behavior and Human Decision Processes, 36, 362–377. Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Processes, 76, 149–188. Levin, I. P., Schnittjer, S. K., & Thee, S. L. (1988). Information framing effects in social and personal decisions. Journal of Experimental Social Psychology, 24, 520–529. Loewenstein, G. (1996). Out of control: Visceral influences on behavior. Organizational Behavior and Human Decision Processes, 65, 272–292. Marteau, T. M. (1989). Framing of information: Its influence upon decisions of doctors and patients. British Journal of Social Psychology, 28, 89–94. Mazursky, D., & Schul, Y. (1988). The effects of advertisement encoding on the failure to discount information: Implications for the sleeper effect. Journal of Consumer Research, 15, 24–36.

65

McNeil, B. J., Pauker, S. G., Sox, H. C., & Tversky, A. (1982). On the elicitation of preferences for alternative therapies. New England Journal of Medicine, 306, 1259–1262. Mittal, V., Ross, W. T., & Tsiros, M. (2002). The role of issue valence and issue capability in determining effort investment. Journal of Marketing Research, 39, 455–468. Nordgren, L. F., McDonnell, M. -H. M., & Loewenstein, G. (2011). What constitutes torture? Psychological impediments to an objective evaluation of enhanced interrogation tactics. Psychological Science, 22, 6. Nordgren, L., van der Pligt, J., & van Harreveld, F. (2006). Visceral drives in retrospect: Explanations about the inaccessible past. Psychological Science, 17, 635–640. Ochsner, K. N. (2000). Are affective events richly recollected or simply familiar? The experience and process of recognizing feelings past. Journal of Experimental Psychology: General, 129, 242. Oppenheimer, D. M., Meyvis, T., & Davidenko, N. (2009). Instructional manipulation checks: Detecting satisficing to increase statistical power. Journal of Experimental Social Psychology, 45, 867–872. Patrick, V. M., Macinnis, D. J., & Park, C. W. (2007). Not as happy as I thought I'd be? Affective misforecasting and product evaluations. Journal of Consumer Research, 33, 11. Peterson, L., & Peterson, M. J. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology, 58, 193. Peterson, R., & Wilson, W. (1992). Measuring customer satisfaction: Fact and artifact. Journal of the Academy of Marketing Science, 20, 61–71. Pham, M. T. (2004). The logic of feeling. Journal of Consumer Psychology, 14. Pham, M. T. (2007). Emotion and rationality: A critical review and interpretation of empirical evidence. Review of General Psychology, 11, 155. Quattrone, G. A., & Tversky, A. (1988). Contrasting rational and psychological analyses of political choice. The American Political Science Review, 82, 719–736. Robinson, M. D., & Clore, G. L. (2002). Belief and feeling: Evidence for an accessibility model of emotional self-report. Psychological Bulletin, 128, 934–960. Sher, S., & McKenzie, C. R. (2011). Levels of information: A framing hierarchy. Perspectives on Framing, 35–63. Shiv, B., & Fedorikhin, A. (1999). Heart and mind in conflict: The interplay of affect and cognition in consumer decision making. Journal of Consumer Research, 26, 278–292. Srull, T. K., & Wyer, R. S. (1980). Category accessibility and social perception: Some implications for the study of person memory and interpersonal judgments. Journal of Personality and Social Psychology, 38, 841. Van Boven, L., Loewenstein, G., Welch, E., & Dunning, D. (2012). The illusion of courage in self‐predictions: Mispredicting one's own behavior in embarrassing situations. Journal of Behavioral Decision Making, 25, 1–12. Vuilleumier, P. (2005). How brains beware: Neural mechanisms of emotional attention. Trends in Cognitive Sciences, 9, 585–594.