Abstract This research documents a systematic bias in memory for ethical attribute information: consumers have better memory for an ethical attribute when a product performs well on the attribute versus when a product performs poorly on the attribute. Because consumers want to avoid emotionally difficult ethical information (e.g., child labor) but believe they should remember it in order to do the right thing, the presence of negative ethical information in a choice or evaluation produces conflict between the want and should selves. Consumers resolve this conflict by letting the want self prevail and forgetting or misremembering the negative ethical information. A series of studies establishes the willfully ignorant memory effect, shows that it holds only for ethical attributes and not for other attributes, and provides process evidence that it is driven by consumers allowing the want self to prevail in order to avoid negative feelings associated with the conflict. We also ameliorate the effect by reducing the amount of pressure exerted by the should self. Lastly, we demonstrate that consumers judge forgetting negative ethical information as more morally acceptable than remembering but ignoring it, suggesting that willfully ignorant memory is a more morally acceptable form of coping with want/should conflict. ethical attributes, ethical decision making, consumer memory, sustainability, want/should conflict Consumers often encounter products in the marketplace that have implications for larger ethical issues, such as environmental impact and labor practices. However, anecdotal and empirical evidence suggests that people often prefer to avoid thinking about ethical information (Lichtenstein, Gregory, and Irwin 2007; Paharia, Vohs, and Deshpandé 2013) as opposed to processing it. One way to avoid ethical information is to elect not to gather it when evaluating potential purchases. For instance, although there are numerous websites that explicate the ethicality of products (e.g., www.ethicalconsumer.org), these sources arguably have little impact on the marketplace because most consumers do not look at them. Ehrich and Irwin (2005) empirically demonstrated that sometimes consumers choose to remain willfully ignorant of ethicality during product search. However, they also showed that consumers who avoid ethical attribute information in the search phase will use it in their decision making when it is explicitly provided to them. Sociologists (Kimeldorf et al. 2006) have identified the same phenomenon; although few consumers will allow themselves to be “conscious” of ethical information, when they are made conscious of ethical issues they will often choose the more ethical item. Given that prior research suggests that even consumers who avoid ethical attribute information often use it when it is available, a market researcher or brand manager attempting to overcome willful ignorance might simply provide consumers with ethical attribute information so that they cannot avoid it. However, in actual purchasing situations, many psychological processes need to occur between attribute exposure and purchase. Like many judgments and decisions, consumer decisions can be made “online,” with all of the relevant information present in the decision environment while the decision is being made, or they can be memory-based, with some or all of the relevant information retrieved from memory at the time the decision is made (Hastie and Park 1986). The present research explores the idea that providing ethical attribute information to consumers is not enough to ensure that this information is used because any role that memory plays in subsequent processing of the information could affect the mental representation of the product. Even if consumers are presented with ethical attribute information, they still may not incorporate it into their decisions. Instead, we propose that, in line with the types of coping strategies consumers employ under conflict (Luce 1998), they may systematically forget or misremember this information to move past the affectively aversive experience of considering negative ethical information. Thus, their memories for ethical information encountered in a choice or evaluation context will be inaccurate, such that they will simply not remember how the product performs on an ethical attribute at all or will reverse the attribute sign to remember it as positive when it is actually negative. Drawing on the literature on the malleability of memory (Braun 1999; Braun, Ellis, and Loftus 2002; Cowley 2008; Cowley and Janus 2004; Kouchaki and Gino 2016; Loftus 1979; May and Irmak 2014) and the conflict that exists between consumers’ “want” and “should” selves (Bazerman, Tenbrunsel, and Wade-Benzoni 1998; Milkman, Rogers, and Bazerman 2008), we propose that consumers exhibit a particular type of willfully ignorant memory for ethical attributes. In a series of studies on memory for product attributes, we establish that there is a systematic pattern to memory for ethical product attribute information: consumers have relatively better memory for an ethical attribute when a product performs well on that attribute versus when a product performs poorly on that attribute. Because consumers find thinking about emotionally difficult ethical information (e.g., that a product is made with child labor) to be aversive but believe they should remember such information in order to do the right thing, the presence of negative ethical information produces conflict between consumers’ want and should selves. In other words, a consumer’s want self is focused on feeling good by preventing emotionally difficult experiences such as thinking about negative ethical information, whereas the consumer’s should self recognizes that s/he should do the right thing by remembering the negative ethical information. We demonstrate that consumers experience this conflict when encoding product information that varies in ethicality and then resolve it by letting the want self prevail and either forgetting the negative ethical information or misremembering the negative ethical information as being positive when retrieving it from memory. However, we show that willfully ignorant memory can be ameliorated by reducing this conflict via a reduction in the pressure exerted by the should self (e.g., by reducing the importance of the ethical issue in question so that thinking about the ethical information is ultimately not as aversive). Finally, we show that consumers judge the act of forgetting negative ethical information to be less morally wrong than remembering but ignoring it in decision making, suggesting that willfully ignorant memory appears to be a more morally acceptable form of coping with want/should conflict. Theoretically, our results add to the growing literature documenting how ethical versus nonethical attributes affect consumer decision-making processes (Ehrich and Irwin 2005; Irwin and Naylor 2009) by uncovering a specific type of willful ignorance for ethical attributes that operates through memory processes. Our work supports this account by demonstrating that it occurs specifically at the retrieval and not just the encoding phase. We also show that the negativity bias in memory (Baumeister et al. 2001) does not hold for ethical attributes, even as we show that the negativity bias does, at times, obtain for other attributes not related to ethicality. Finally, in a larger sense, we add to researchers’ understanding of how consumers process ethical product information (Haws, Winterich, and Naylor 2014; Paharia et al. 2013; Peloza, White, and Shang 2013; Reczek and Irwin 2015; Zane, Irwin, and Reczek 2016). THEORETICAL BACKGROUND AND PREDICTIONS Ethical Information and the Want/Should Conflict As Kohlberg (1984) argues, after reaching a certain stage of moral development, most people have a general desire to be good and to do the right thing by behaving in an ethical and moral manner (Bandura 1999; Bandura et al. 1996). All adults, with the exception of sociopaths, have the general desire to be prosocial (Haidt 2007) and to further the outcomes of creatures apart from themselves. Ethicality enters into consumer contexts in part through exposure to ethical product attributes, which are attributes that reflect a consumer’s conscience in some way (Haws et al. 2014; Irwin and Naylor 2009; Luchs et al. 2010; Peloza et al. 2013; Zane et al. 2016). Ethical attributes can relate to a variety of ethical issues, including the environment, labor practices, and treatment of animals; the key is that the attribute activates one’s moral values. Ethical dilemmas are commonly presented as conflicts between two different “selves.” In the formulation we will use, the self that monitors morality is the “should self” and the one that does not is the “want self” (Bazerman et al. 1998). Inherent in these conceptualizations of the two different selves is the notion that the should self focuses on larger goals (e.g., being an overall moral person in the long run), in contrast to the more affectively driven want self, which wants to feel good in the moment (e.g., forgetting or misremembering negative ethical information in any given instance in order to avoid negative emotion; Milkman et al. 2008). The idea that all individuals possess a want and a should self is consistent with various economic models that argue that people act as if their decisions are controlled by multiple, competing internal selves with different preferences (Fudenberg and Levine 2006; Thaler and Shefrin 1981). In this framework, the want self that monitors joyful, easy, everyday existence and visceral responses is often in conflict with the should self (Baumeister 2002; Hoch and Loewenstein 1991; Loewenstein 1996). Past research has shown that when the want and should selves conflict, consumers often allow the want self to triumph over the should self (Milkman 2012; Milkman, Rogers, and Bazerman 2009)—for example, when choosing an indulgent chocolate snack over an apple. We propose that an analogous want/should self conflict can also occur when one is deciding which information to incorporate into one’s decisions, and not just what options to choose, particularly when the potential to experience negative emotional consequences is high if this information is incorporated (Luce 1998). Given that past research has shown that consumers are particularly likely to experience negative emotions when decisions involve ethical issues (DePalma, Madey, and Bornschein 1995; Ehrich and Irwin 2005; Eisenberg 2000; Massi 2005; Ruedy et al. 2013, study 1; Zane et al. 2016), it is especially likely that consumers experience want/should conflict in such situations. Despite consumers’ should selves dictating that they ought to behave ethically across time, prior work has uncovered several instances when individuals stray from their long-term beliefs regarding ethical behavior (i.e., when they do not act in line with their should selves) to satisfy their want self’s desire to feel good in the short run by avoiding potentially negative ethical information. For example, Ehrich and Irwin (2005) showed that consumers engage in willful ignorance for ethical attribute information at the search stage of the buying process, choosing to avoid information about ethical attributes as a coping strategy to avoid the negative emotional consequences of incorporating this information (e.g., that products are made with child labor) into their decisions. We propose that consumers engage in another type of willful ignorance when retrieving ethical product information from memory. Just as consumers are motivated not to search for ethical attribute information, we argue that they are motivated to systematically forget or misremember negative ethical attribute information, even if they have encoded it correctly. Further, we argue that this type of motivated forgetting/misremembering is driven by the push and pull that consumers experience between their “in the moment” desire to avoid negative ethical information on the one hand and their general long-term belief that they should be morally good on the other. We propose that consumers resolve the conflict between the want/should selves by allowing the want self to prevail, just as they do in choice contexts not related to ethicality. In our context, the want self prevailing takes the form of either increased forgetting of negative ethical information or misremembering of negative ethical information as being positive. We note that in this work we do not have any particular hypotheses about the prevalence or meaning of likelihood of forgetting the information altogether versus misremembering the information and “switching” the attribute levels. We next discuss the motivated forgetting/misremembering we predict in light of the literature on the malleability of memory. The Malleability of Memory and Willfully Ignorant Memory for Ethical Attributes Memory is not always veridical; past research has demonstrated that memory can be both malleable and constructive (Anderson, Bjork, and Bjork 1994; Schacter 1999), subject to such “sins” (Schacter 1999) as blocking, misattribution, suggestibility, and bias. Human memory cannot always be counted on to provide a literal recreation of what was encountered. Instead, people construct their memories, and they are thus inclined to errors, alterations, and false impressions (Bartlett 1932; Loftus 1979). Consumer behavior researchers have shown that memory for product attributes and marketplace experiences can be malleable. For example, seeing attractive advertising after consuming a product can cause consumers to remember a more favorable product experience (Braun 1999) or a different product (Cowley and Janus 2004) than they actually consumed and even to remember having used a product with which they actually have no direct experience (Rajagopal and Montgomery 2011). There is also evidence that consumers either consciously or unconsciously exert some control over their memories. Zauberman, Ratner, and Kim (2008) demonstrate that consumers protect special memories by avoiding repeating the memorable experience. Consumers may also use hedonic editing and rosy retrospection strategies to reconstruct the past as more positive than it actually was (Cowley 2008; Mitchell et al. 1997). The memory literature most relevant to our research explores changes in memory designed to protect the self. For example, in the presence of a strong motivation to preserve self-esteem, the subset of autobiographical memories an individual retrieves from memory tends to be skewed toward memories that enhance self-image (Sanitioso et al. 1990). Even though in some contexts, such as unwanted persistence in recall of traumatic or disturbing events (Martin and Tesser 1996; Schacter 1999; Wegner 1994), memory can be more veridical than most people would like, there is also evidence that people (unconsciously or consciously) sometimes alter memories that threaten the self. For example, recent research in marketing demonstrates this type of memory bias: Dalton and Huang (2014) show that consumers employ motivated forgetting to counteract threats to the self stemming from identity-linked promotions. May and Irmak (2014) demonstrate that, in order to justify an indulgent option in the present, individuals distort memories to make it seem as though they have previously achieved more self-regulatory goal progress. In ethical domains, individuals sometimes exhibit altered memories after engaging in ethically questionable behavior. For instance, Shu and Gino (2012) showed that after behaving dishonestly, people are less likely to correctly recall honesty pledges that they read prior to their behavior. Tenbrunsel et al. (2010) also describe memory revisionism, a process in which people employ recall strategies that protect their self-image after engaging in ethically questionable behavior. Past research thus suggests an overall theme: memories appear to be especially malleable in the presence of some sort of self-threat or potentially aversive experience. We ask a new question about the presence of ethically relevant information and how it influences memory: does evaluating products with negative ethical product information result in motivated memory errors? We propose that evaluating products that vary in ethicality produces conflict between the want and should selves, which is an aversive experience for consumers (Luce, Payne, and Bettman 1999). To resolve this conflict, consumers forget or misremember poor performance on ethical attributes to a greater degree than they forget or misremember positive performance on ethical attributes, because it is poor performance (e.g., using child labor, polluting the environment) that is unpleasant to think about and therefore creates the greatest conflict with the want self, with its desire to focus on that which is joyful, fun, and easy. The greater the conflict felt, the more likely the consumer is to incorrectly retrieve negative ethical attribute information as positive or fail to retrieve it at all once it has been encoded. To test our proposition, in our studies we first ask consumers to evaluate products that vary in ethicality that they will later be asked to retrieve from memory. Thus, in the context we explore, the conflict between the want/should self should at least occur at the encoding stage of memory. Indeed, previous research suggests that systematic biases can occur in the encoding of ethical product attributes (Ehrich and Irwin 2005). However, in our work, we establish that memory errors also occur when ethical information that has been correctly encoded is retrieved from memory. Why might retrieval also be influenced by the want/should conflict? We propose that ethical information always presents an inherent conflict because it is difficult to think about and consider. Knowing that products vary in ethicality might require uncomfortable actions, should action be required in the future (e.g., having to choose among or otherwise evaluate the products). Thus, even consumers who correctly store ethical information do not accurately retrieve the information about products that have poor performance on ethical attributes because not doing so allows them to avoid the conflict they experience at encoding. Thus, the effect we propose is similar to the “unethical amnesia” documented by Kouchaki and Gino (2016), who explore participants’ memory for their own previous unethical actions. These authors argue that people experience psychological discomfort after behaving unethically due to their general desire to maintain a positive self-view (which includes the belief that one is a moral person). As a result, when people are asked to recall past behavior, memories of unethical behavior are less clear, less detailed, and less vivid than memories of ethical actions or actions without ethical implications. These memory errors occur even when consumers do not face a current decision to act ethically. Similarly, we propose that consumers will exhibit willfully ignorant memory for ethical product information they have previously encoded even when they are not asked to evaluate or choose among these products at retrieval. The Kouchaki and Gino (2016) work establishes that people sometimes experience memory bias due to not wanting to retrieve negative ethical information, which we believe will also occur in our studies. One key difference between our work and theirs, however, is that, in our work, consumers do not have to undertake an initial unethical action to show subsequent memory errors related to ethicality. More specifically, Kouchaki and Gino (2016) show that the distress participants in their studies experience is focused on not being seen as a good person (because they have already committed a nonmoral act), whereas in our studies any psychological distress experienced is driven by the inherent tension between balancing one’s belief that they ought to care about ethicality and one’s competing desire to avoid considering ethical issues that can make decision making more emotionally difficult. In other words, experiencing want/should conflict is inherently psychologically aversive due to the negative emotions it evokes; actually taking an unethical action is unnecessary to experience this distress. We discuss this issue more in the General Discussion. Finally, we note that the directionality of the willfully ignorant memory effect is directly opposite of the negativity bias (Baumeister et al. 2001; Rozin and Royzman 2001). According to this bias, people tend to pay more attention to negative than to positive information. There has been some historic disagreement in the literature concerning whether the negativity bias extends to memory (see Rozin and Royzman 2001 for a discussion of this issue), but at least one extensive review of the literature (Baumeister et al. 2001) concludes that there is a negative memory bias. For our purposes, we argue that if differences in memory were driven solely by the negativity bias, we would expect to find that consumers would remember poor performance on an ethical attribute relatively better than good performance. Instead, we propose that consumers’ motivation to forget or misremember poor ethical attribute performance is stronger than any effects of the negativity bias for these attributes. For nonethical attributes, we would expect to observe either no bias or to see the negativity bias, given its fit with what we know about loss aversion (Kahneman and Tversky 1979) and the deeper processing of negative information due to its greater salience. Why would there not be a negativity bias for ethical attributes? We suspect that the bias may exist and, in the absence of ethical conflict, might obtain in memory tests. In our theorizing, however, the conflict between the want and should selves dominates because the motivation to avoid this conflict is so strong. Being moral is a defining feature of the self (Aquino and Reed 2002; Dunning 2007), and, in other contexts (e.g., the inconsistency-negativity neglect model: Brown and Dutton 1995; Sedikides and Strube 1997; Taylor and Brown 1988, 1994), consumers’ motivation to forget negative information that pertains to the self has been shown to outweigh the motivation to face this information and work to reconcile it (Sedikides and Green 2000). In other words, we hypothesize that whatever attentional features drive the negativity bias will be swamped by the want/should conflict. More formally, we predict the following: H1: Memory patterns will differ for ethical versus nonethical attributes such that: a: For ethical attributes, consumers will be relatively less likely to remember poor versus good performance on an ethical attribute (by forgetting the poor performance altogether or misremembering it as good performance). b: For nonethical attributes, either there will be no difference in memory for the poor versus good performance levels or there will be relatively better memory for the poor performance level. Moderating Willfully Ignorant Memory Our theory of willfully ignorant memory leads us to expect that reducing the amount of pressure exerted by the should self will have the ironic effect of decreasing willfully ignorant memory. If people feel less obligation to be ethical, they will actually remember negative ethical information better. Forgetting or misremembering negative ethical attribute information is caused by conflict between the want and should selves, and thus reducing the conflict should reduce the forgetting and misremembering. We further propose that the pressure exerted by the should self can be lessened by decreasing the perceived importance of the ethical issue associated with the ethical attribute. There is support for this prediction in the literature, as attributes with ethical implications that are not perceived as being particularly related to morality (e.g., number of vacation days for workers) do not show the same effects as ethical attributes perceived as more directly related to morality (e.g., lawsuits brought against management for poor treatment of workers; Irwin and Naylor 2009, study 1). Thus, we predict that although the want self typically triumphs, leading consumers to forget or misremember negative ethical information, diminishing the ethical weight of the issue at hand decreases the conflict between the want and should selves, which results in reduced need for willfully ignorant memory. In this case, consumers should be equally likely to remember negative and positive ethical information. More formally: H2a: Willfully ignorant memory is the result of consumers’ desire to avoid negative ethical information in order to resolve conflict between their want and should selves. b: If the ethical weight of the ethical attribute is reduced, then consumers will show relatively less willfully ignorant memory. OVERVIEW OF STUDIES We first report the results of a pilot study demonstrating that consumers experience increased conflict between their want and should selves as a result of considering a set of products that contains alternatives with poor performance on an ethical attribute. Study 1 then establishes the basic willfully ignorant memory effect using a recall task (thus testing hypotheses 1a and 1b) and rules out alternative explanations, including predecisional distortion (Russo, Meloy, and Medvec 1998), which can lead consumers to pay more attention to the attributes they like more, and choice-supportive bias (Mather, Shafir, and Johnson 2000), in which preference distorts processing in the direction of choice. We also measure both encoding and retrieval of information to show that the effect is not driven only by encoding. Although we see some evidence of avoiding ethical information even at encoding (consistent with Ehrich and Irwin 2005), participants who correctly encode the ethical information exhibit willfully ignorant memory in the predicted patterns during retrieval, affirming the role of memory in the effect. Study 2 provides a more conservative test by testing hypotheses 1a and 1b in a recognition task. Study 2 also establishes process evidence by showing that forgetting and misremembering of negative ethical information is driven by consumers’ desire to avoid the conflict that arises from thinking about this information (hypothesis 2a). Study 3 ameliorates willfully ignorant memory by priming respondents to think about the plethora of ethical issues in the world, thereby reducing the relative ethical weight of the issue relevant to the specific ethical attribute they are considering (hypothesis 2b). Finally, in study 4 we provide direct evidence that consumers judge the act of forgetting negative ethical information to be less morally wrong than remembering but ignoring it when making a purchase decision, suggesting that willfully ignorant memory appears to be a more morally acceptable form of coping with want/should conflict. PILOT STUDY The purpose of this study was to provide evidence that exposure to a set of products that contains alternatives with poor performance on an ethical attribute produces conflict between the want and should selves and a resultant desire to avoid negative ethical information to resolve this aversive experience. We asked 189 undergraduate students to evaluate brands of blue jeans that varied on three attributes. Two of the attributes, wash and style, were the same across conditions, while the third was manipulated between-subjects to be either an ethical attribute with two levels (labor practices: whether the jeans were made with adult labor only [good performance] or in a facility that uses both adult and child labor [poor performance]) or an attribute without ethical implications with two levels (material: whether the jeans were made with 100% cotton or made using a blend of cotton and Lycra). Participants read about six brands of jeans, such that they saw both levels of the respective third attribute (labor practices or material, depending on condition) and then indicated which of the brands they were most likely to buy and which they would definitely not want to buy. Thus, participants saw six pairs of jeans with labor information or six pairs of jeans with information about the jeans’ material, depending on condition (see appendix A for details). All participants then indicated their agreement with the following focal measures: “Forgetting some of the information about the jeans would help me to avoid feeling angry” [because Ehrich and Irwin (2005) found that anger was the primary affective driver of willful ignorance in search], “Thinking about which jeans I would/would not buy made me think about issues I’d rather avoid,” “Thinking about which jeans I would/would not buy made me think about issues I don’t currently want to think about,” and “It would have been more fun to evaluate the jeans with only wash and cut and no information on the third attribute,” which were combined to form a composite measure of want/should conflict (α = .82). We regressed this measure on the between-subjects variable of whether the third attribute was an ethical attribute or an attribute without clear ethical implications. As expected, participants reported significantly more conflict when the ethical attribute was part of the information they were asked to consider (M = 4.43) compared to when the third attribute was the material from which the jeans were made (M = 3.50; F(1, 187) = 28.52, p < .0001). These results support past research (Kimeldorf et al. 2006; Lichtenstein et al. 2007; Paharia et al. 2013) and our theorizing that simply evaluating products with varying ethical attribute information can produce conflict (that consumers would rather not experience) between different parts of the self. It is this desire to resolve the conflict that results in the willfully ignorant forgetting and misremembering we predict at retrieval. STUDY 1 The primary purpose of study 1 was to establish the basic willfully ignorant memory effect (thus testing hypotheses 1a and 1b). To establish that the effect occurs at retrieval, we test whether memory errors occur when the information is correctly stored in memory at the encoding stage. If respondents are simply not attending to information indicating that a product is unethical (or are actively avoiding encoding it, as Ehrich and Irwin’s 2005 research would suggest), then storage differences at encoding could roll over into memory differences at retrieval. If, as we predict, there are differences in recall for poor versus good performance on ethical attributes even after successful encoding of the information, then we have uncovered a type of willfully ignorant behavior that is specific to retrieving information from memory. Pretest All participants in the main study were asked to read descriptions of six hypothetical brands of desks purportedly drawn from advertisements and then to remember this information at a later point in the study. We created the desk descriptions using stimuli adapted from Ehrich and Irwin (2005), such that the desks differed on three attributes: wood source, quality, and price. Wood source—that is, whether the wood was sourced from endangered rainforests or from sustainable tree farms—was the ethical attribute (see appendix B for details). In order to ensure that study participants recognized wood source as an ethical issue, 150 undergraduate pretest participants were shown the desk descriptions used in the main study and asked to indicate their agreement (on a seven-point scale) with eight items designed to measure whether they felt that harvesting trees from rainforests to build desks was an ethical issue (e.g., “Desks made from rainforest wood are less ethical than desks made from tree farm wood”). The eight items were averaged to create a single measure (α = .76). The mean on this index was 4.99, which was significantly greater than the scale midpoint of 4 (F(1, 147) = 218.62, p < .0001), indicating that participants do see taking trees from rainforests as an ethical issue. We also measured participants’ perceptions of how common it is to use both rainforest and tree farm wood in making desks, an issue we return to in the study discussion. See appendix C for pretest measures. Participants and Procedure A total of 236 undergraduates participated in the main study in exchange for course extra credit. Immediately after reading the desk descriptions, in order to rule out the alternate explanations that more readable desk descriptions are more memorable and/or that better-liked desks are more memorable, participants were asked to rate, on a nine-point scale: (1) their likelihood of purchasing each desk and (2) how readable they found each desk description. Likelihood of purchase measures help rule out two related alternative explanations for willfully ignorant memory: predecisional distortion (Russo et al. 1998), which can lead consumers to pay more attention to the attributes they like more, and choice-supportive bias (Mather et al. 2000), in which preference distorts processing in the direction of choice. Either of these biases could also lead consumers to perceive descriptions of desks that they liked better as more readable. If either of these biases is operating, then we should see better memory for better-liked desks and memory distortions that favor the better-liked desks. After reading all six desk descriptions and answering these two questions for each desk, participants answered additional questions about the set of desks in order to enhance processing of the information they read. These questions included: “Which desk did you like the best?” and “Which desk do you think is the best seller?” (see appendix D for a complete list). Participants were then told that the study was designed to test their memory and that they would be asked to memorize all of the information about two randomly selected desks. To select these desks, each student drew two numbers from a cup; the numbers each corresponded to one of the six desks. Participants were given 60 seconds to memorize everything they could about the desks corresponding to those numbers. The desk information was then taken away, and participants were given a blank sheet of paper on which they wrote the brand name and all the other information they could remember about the two desks they had selected. After writing down all of the information they could remember, participants then looked at the original descriptions again to see how well what they wrote matched the descriptions. They then studied the descriptions for an additional 30 seconds. Following this review of the desk information, the information was removed and participants again wrote down everything they could remember from the descriptions, as close to word-for-word as possible. The purpose of this repeated memorization task for the two selected desks was to ensure that participants correctly encoded the information. Data from this second writing task were analyzed to determine whether the information was correctly stored in memory. Upon completion of the induced memorization tasks for the two selected desks, participants then completed two unrelated distracter tasks that took approximately 15–20 minutes, after which they were given a sheet of paper listing all six desk brand names (with no additional information) and asked to write down as much as they could remember about all six desks. Memory in this final writing task served as our focal measure of the willfully ignorant memory effect at retrieval. Results In all of our studies, each participant’s responses for each attribute are coded as either incorrect (i.e., the information the participant recalled for a given brand name is not true for that brand or the information for that brand was not recalled = –1) or correct (i.e., the information recalled for a given brand name is true for that brand = 1). This coding scheme is the primary coding scheme used in our analysis, as it distinguishes between whether information was correctly remembered or not, regardless of whether the information was not recalled (i.e., left blank by participants) or recalled incorrectly (i.e., misremembered). Results of additional analyses using an alternate coding scheme distinguishing between these two types of errors are also discussed where appropriate. Storage of Ethical Attribute Information We first examined whether participants were able to accurately store the ethical attribute information at the encoding stage of memory. To check storage at this stage, we examined participants’ recall immediately after the second induced memorization of the two randomly selected desks. Because of the design of the study, we are able to compare accurate storage of the two randomly selected desks in both a within- and between-subjects analysis, because some participants randomly chose two desks made from tree farm wood, some chose two made from rainforest wood, and some chose one of each type of desk. The induced-memory task was largely successful; 94.49% of all possible wood source responses were correctly encoded. A total of 23 participants incorrectly encoded the wood source information for at least one desk. Despite this mostly accurate encoding, the pattern of incorrect responses at this stage of memory was consistent with hypothesis 1a. In the within-subjects test, the test of whether people who selected one rainforest (bad performance on the ethical attribute) and one tree farm wood (good performance) desk differed in memory for their two desks: participants were more likely to incorrectly encode rainforest wood (10.71% of all responses incorrect or missing, 89.29% of all responses correct) than they were tree farm wood (2.13% of all responses incorrect or missing, 97.87% of all responses correct; F(1, 141) = 9.54, p < .01). The results are similar for the between-subjects test, the test of whether the memory storage of people who selected two tree farm wood desks was better than that of people who selected two rainforest wood desks: participants were more likely to incorrectly encode rainforest wood (8.97% of all responses incorrect or missing, 91.03% of all responses correct) than they were tree farm wood (0.90% of all responses incorrect or missing, 99.10% of all responses correct; F(1, 93) = 5.21, p < .05). Storage of Other Attribute Information We also examined memory at encoding for price and quality for all participants for their two randomly selected desks. Although 23 participants stored wood source information incorrectly, only two stored price information incorrectly and only nine stored quality information incorrectly. There were no significant differences in memory for the high- versus low-price desks in both the within-subjects analysis (F(1, 147) = 1.00, p = .32) and the between-subjects analysis (F(1, 84) = 1.00, p = .32). Similarly, there was no difference in memory for high- versus low-quality desks in the between-subjects analysis (F(1, 88) = .73, p = .40). There was a difference in memory for quality in the within-subjects analysis: participants had better encoding of the low-quality desk than they did the high-quality desk when they drew one of each (F(1, 145) = 7.30, p < .01). This finding is consistent with a negativity bias (Baumeister et al. 2001). Retrieval of Ethical Attribute Information To test retrieval of the ethical attribute information, we dropped the 23 participants who, at the end of the induced memorization task, did not successfully encode the correct information about the wood source attribute for their two randomly selected desks. After doing this, all participants remaining in the data set had correctly stored this information, and therefore testing recall for the same information after the distracter tasks allows us to explore whether errors are made at the retrieval stage, which was our primary goal. We analyzed both the recall of (1) all six desks and (2) the two randomly selected desks (in both a within- and between-subjects analysis) for the remaining 213 participants. In order to compare whether tree farm wood was better remembered than rainforest wood across all six desks, we conducted a repeated-measures analysis with the summed memory score for rainforest wood for each participant (across desks) and the summed memory score for tree farm wood for each participant as the within-subject dependent variables. This within-subjects analysis of participants’ recall for the wood source attribute across all six desks revealed that recall for tree farm wood was significantly more likely to be correct (39.59% of all responses incorrect or missing, 60.41% of all responses correct) than recall for rainforest wood (54.77% of all responses incorrect or missing, 45.23% of all responses correct; F(1, 212) = 32.65, p < .0001). In order to rule out the alternate explanations that more readable desk descriptions are more memorable and/or that desks consumers were more likely to purchase are more memorable, we next conducted two separate mixed analyses with memory for rainforest and tree farm wood as the within-subject dependent variables and summed ratings of (1) readability and (2) purchase likelihood as between-subject predictor variables. Neither readability (p = .24) nor purchase likelihood significantly affected recall for rainforest wood alone (p = .45), tree farm wood alone (p = .79), or both types of wood together (p = .46). Additionally, neither readability (p = .24) nor purchase likelihood (p = .74) accounted for the difference in memory for rainforest compared to tree farm wood. In other words, participants did not have better memory for rainforest or tree farm wood when they considered the ethical information more readable or when it was paired with desks they liked better, and these variables did not account for the difference in memory at retrieval for rainforest compared to tree farm wood. Thus, the willfully ignorant memory effect is driven neither by predecisional distortion (Russo et al. 1998) nor by choice-supportive bias (Mather et al. 2000). We next analyzed participants’ recall for the two desks they randomly selected. We first analyzed the recall of participants who drew one desk made of each type of wood, a within-subjects test (n = 124). As in the analysis for all six desks, recall for tree farm wood (1.61% of all responses incorrect or missing, 98.39% of all responses correct) was significantly more likely to be correct than recall for rainforest wood (9.68% of all responses incorrect or missing, 90.32% of all responses correct; F(1, 123) = 7.52, p < .001). The between-subjects analysis comparing the recall of participants who drew two rainforest desks versus those who drew two tree farm desks did not have sufficient power to yield significant results. Only 34 participants drew two rainforest desks, while 55 drew two tree farm desks. Of the 110 possible tree farm responses, only two (1.82%) were incorrect or missing. Of the 68 possible rainforest responses, only four (5.88%) were incorrect or missing. Thus, although the direction of this difference is suggestive, the difference is not significant (F(1, 87) = 1.62, p = .21). Alternative Memory Coding For completeness, we conducted an analysis using coding that separates incorrect responses (i.e., misremembering) from missing responses (i.e., forgetting the information completely): correct recall of wood source = 1, incorrect recall of wood source = –1, no information about wood source recalled = 0. Replicating the results we obtained using the two-level incorrect/correct coding, in the within-subjects analysis of recall for all six desks, memory that a desk was made from tree farm wood (34.74% of all possible responses missing, 4.85% of all responses incorrect, 60.41% of all responses correct) was more likely to be present and correct than memory that a desk was made from rainforest wood (39.12% of all responses missing, 15.65% of all responses incorrect, 45.23% of all responses correct; F(1, 212) = 53.09, p < .0001). Memory for tree farm wood (1.61% of all responses missing, 0% of all responses incorrect, 98.39% of all responses correct) was also more likely to be present and correct than memory for rainforest wood in the within-subjects analysis of the two randomly selected desks (0% of all responses missing, 9.68% of all responses incorrect, 90.32% of all responses correct; F(1, 123) = 10.41, p < .01). Thus, hypothesis 1a is supported using this alternate coding. Consistent with the results using the primary coding scheme, the between-subjects test of memory for the two randomly selected desks was marginally significant in the direction supporting our hypothesis (F(1, 87) = 3.01, p = .09). Retrieval of Other Attribute Information In order to demonstrate that memory for price and quality does not follow the same pattern as memory for the ethical attribute, we next compared the recall of (1) all six desks and (2) the two randomly assigned desks (in both a within- and between-subjects analysis) for both price and quality. The two participants who stored price information incorrectly were dropped from the price retrieval analysis, and the nine participants who stored quality information incorrectly were dropped from the quality retrieval analysis. In the within-subjects analysis of recall of all six desks, there was no difference in memory for high versus low quality (F(1, 226) = 0.05, p = .82). There was, however, a difference for high versus low price (F(1, 233) = 4.92, p < .05): Participants remembered the higher price better, showing a negativity bias (Baumeister et al. 2001). In order to examine memory across these two nonethical attributes, we also combined memory for price and quality and found that participants had marginally better memory for the negative level of the attribute (i.e., low quality and high price) versus the positive level (i.e., high quality and low price; F(1, 226) = 3.30, p = .07). These results support hypothesis 1b. In analyzing participants’ recall for the price and quality of the two randomly chosen desks, we found no differences in retrieval for the high versus low price in either the within-subjects or between-subjects analysis and no differences in retrieval for high versus low quality in the between-subjects analysis (all ps > .29), further supporting hypothesis 1b. In the within-subjects analysis for quality, however, we did find that participants had better memory for the positive level of the attribute versus the negative level (F(1, 143) = 4.09, p < .05). Relative Retrieval of Ethical versus Nonethical Attribute Information Finally, we conducted an omnibus analysis to compare relative memory across attributes and valence. Specifically, we conducted a repeated-measures analysis with two within-subjects contrasts comparing (1) the ethical attribute to the combined nonethical attributes and (2) the positive level of the attributes to the negative level of the attributes. The interaction of these two within-subjects contrasts (i.e., whether memory for positive versus negative attributes differed for ethical versus nonethical attributes) was the focal analysis, and it was significant (F(1, 212) = 28.97, p < .0001). Follow-up spotlight analyses to explicate the interaction (Irwin and McClelland 2001) revealed that participants had similar memory for tree farm wood compared to positive nonethical information (F(1, 212) = .04, p = .83). However, participants had significantly worse memory for rainforest wood compared to negative nonethical information (F(1, 212) = 65.00, p < .0001). Discussion The results of study 1 show that although some willfully ignorant memory errors (i.e., both forgetting and misremembering) occur at the storage, or encoding, phase of memory, more notably, we see evidence consistent with willfully ignorant memory at the retrieval stage even given correct storage of the information. Providing support for hypothesis 1a, retrieval of rainforest wood is more likely to be incorrect (i.e., to be forgotten or misremembered) in both the within-subjects tests and between-subjects tests of memory. Note that “rainforest” is probably a more vivid description than “tree farm.” This difference in vividness makes these stimuli a conservative test of our hypotheses because if vividness alone were driving memory, then participants would be more likely to correctly remember the rainforest level of the attribute. We also note that rainforest wood and tree farm wood vary in how typical they are considered for constructing wooden desks. In the pretest assessing the ethicality of taking wood from rainforests, we also asked participants “How atypical or unusual is it that a desk is made from…?” for both rainforest wood and tree farm wood. The results of a repeated-measures analysis revealed that rainforest wood is perceived as significantly more uncommon/atypical (M = 4.18) than tree farm wood (M = 3.32; F(1, 149) = 21.71, p < .0001). If differences in typicality at encoding were driving our results, then participants should have had better memory for rainforest than for tree farm wood, as atypical information generally receives a special tag in one’s memory and tends to be remembered better (Graesser, Gordon, and Sawyer 1979; Graesser et al. 1980; Schank and Abelson 1977; Srull 1981). The fact that rainforest wood was less likely to be recalled correctly than tree farm wood supports our contention that willfully ignorant memory occurs at retrieval and is not driven only by differences in memory at the encoding stage. We do note, however, that inferences about missing information tend to be consistent with a person’s schemata (i.e., to be the attribute level considered typical), but our argument is that there is a systematic bias at retrieval to misremember (or forget) negative ethical information regardless of its typicality. Finally, the results of study 1 also largely support hypothesis 1b by showing no difference in memory or the expected negativity bias for the nonethical attributes (price and quality) with one exception: in the within-subjects analysis of memory for the two assigned desks, participants had better memory for high quality than low quality. This finding is the only instance throughout all of our analyses across all studies showing better memory for the positive level of a nonethical attribute. Apart from this exception, the pattern of greater forgetting of the negative attribute level is unique to ethical attributes. STUDY 2 The primary purpose of study 2 is to provide process evidence that the extent to which participants experience want/should conflict drives willfully ignorant memory (thus testing hypothesis 2a). In addition, study 2 also tests whether willfully ignorant memory extends to recognition memory. People are generally more successful at recognition than recall (Loftus 1971; Shepard and Chang 1967; Singh and Rothschild 1983; Underwood, Zimmerman, and Freund 1971), and thus this test is more conservative because, overall, participants should be fairly successful at recognizing all of the attributes, reducing the probability of willfully ignorant memory. Study 2 also uses a different focal product category (jeans) and different ethical attribute (adult vs. child labor) from that used in study 1 to increase the generalizability of the findings. We also control for the fact that some of the brand names used in study 1 differed in length/complexity by anonymizing brand names in this study. Participants and Procedure A total of 402 individuals completed this study on Amazon Mechanical Turk in return for a monetary payment. All participants saw descriptions of the following clothing items: three T-shirts, two jackets, one pair of jeans (focal item), and two pairs of shoes. Participants were told, depending on condition, that their task was either to create an outfit that they would be interested in purchasing themselves or that the study was about categorization and that they were simply to select one item from each category. Using a computer interface, participants in the shopping task condition were asked to drag and drop the items to build their outfit (selecting one item from each category) into a box labeled “My Shopping Cart,” while those in the categorization condition dragged items into an unlabeled box. The purpose of this manipulation was to test whether willfully ignorant memory was greater when participants were actively considering purchase. We note, however, that although we did not explicitly encourage participants to consider purchase in the categorization task, they likely engaged in some form of evaluation of these products as part of the categorization task. Thus, the categorization task represents a conservative test of our hypothesis. Because any type of evaluation of products varying in ethicality should evoke want/should conflict, the willfully ignorant memory effect should obtain across both tasks, although we expected that the effect might be weaker in the categorization task due to the lack of a proximal purchase decision. Within participants, the descriptions of the multiple T-shirts, jackets, and shoes varied across three attributes (none of which were related to ethicality). For the focal pair of jeans, two of the attributes were not related to ethicality: the cut (skinny) and wash (dark blue). However, the third attribute related to ethicality and varied between-subjects. Participants in the adult labor condition saw a pair of jeans that was made in an overseas factory that uses adult labor. Participants in the child labor condition saw a pair of jeans that was made in an overseas factory known to use some child labor in addition to adult labor (see appendix E for complete descriptions of all clothing items). The study therefore employed a 2 (labor practices: adult or child) × 2 (task: shopping cart or categorization) between-subjects design. After dragging one of each type of clothing item into the box, all participants then answered filler questions about how readable and descriptive the item descriptions were. They then completed unrelated filler tasks that took approximately 10 minutes before returning to the second part of the study. All participants were then shown descriptions for six pairs of jeans and were asked to choose the one that matched the jeans they saw in the first task. The order in which participants saw these six jean descriptions was randomized, but we label them here as options 1–6 for ease of explication. Option 1 was the correct description that had the correct cut (skinny), wash (dark blue), and ethical information (made with adult or child labor, depending on condition). Option 2 had the correct fit and wash attributes, but incorrect ethical information (i.e., information about the labor practices did not match the condition of the participant). Options 3 and 4 had incorrect cut (boot cut) and wash (light blue) attributes, and either correct ethical information or incorrect ethical information (adult labor or child labor), depending on condition. Options 5 and 6 did not include any ethical information and either included correct cut and wash attributes or incorrect cut and wash attributes. Thus, the stimuli allowed for the possibility of both types of memory errors, misremembering and forgetting. Finally, participants reported their agreement with the same four items used in the pilot study to measure want/should conflict (α = .83). Results Data from nine participants were removed because these individuals either failed to place the pair of jeans in the box or placed more than the specified four items in the box during the first task. All results below hold if data from these participants are included. Recognition of Ethical Attribute Information First, we examined participants’ recognition of correct ethical information. We ran a logistic regression with participants’ choice of the six descriptions of jeans (1 = chose any description with correct ethical information, regardless of whether the other attributes [cut and wash] were correct, 0 = chose any description with incorrect ethical information or description with ethical information missing, regardless of whether the other attributes [cut and wash] were correct) as the dependent variable and labor practices (adult vs. child), the task (shopping cart or categorization), and their interaction as independent variables. As expected, a significant main effect of labor practices emerged (β = –.44, χ2 = 17.36, p < .0001). Participants were significantly more likely to incorrectly recognize ethical information when the jeans were made with child labor (54.12% of the 194 participants in this condition chose an option with ethical information incorrect or missing, 45.88% correct) versus adult labor (33.17% of the 199 participants in this condition chose an option with ethical information incorrect or missing, 66.83% correct), supporting hypothesis 1a. There was no main effect of the task nor an interaction between the two variables (all ps > .35), which shows that the willfully ignorant memory effect obtains even when consumers are not explicitly choosing an item for purchase.1 We expect that this effect occurred because consumers are evaluating the clothing items even when asked only to categorize them, and this evaluation is sufficient to evoke want/should conflict. Recognition of Other Attribute Information As expected, and in line with hypothesis 1b, participants’ recognition of attributes not related to ethicality (cut and wash), irrespective of whether they correctly recognized ethical information, was unaffected by either of the manipulated variables (all ps > .30). Relative Recognition of Ethical Attributes We next conducted an omnibus analysis to examine participants’ recognition of ethical information relative to their recognition of nonethical information and whether this relative difference was affected by the labor practices variable (i.e., whether participants saw jeans with positive or negative labor practices). Given the binary outcomes for recognition of both ethical and nonethical attributes, we created a generalized estimating equation (GEE) with a within-subjects contrast comparing recognition of ethical information versus recognition of nonethical information, the between-subjects labor practices variable, and their interaction as independent variables. A significant interaction between the within-subjects contrast and the between-subjects labor practices variable emerged (β = –0.56, SE = .16, Z = –3.57, p < .0005). Memory for child labor information relative to nonethical information was significantly worse (i.e., showed more evidence of both misremembering and forgetting) than memory for adult labor information relative to nonethical information. Additional chi-square analysis revealed that participants in the adult labor condition had similar levels of correct recognition for adult labor information and nonethical information (χ2 = 2.03, p > .15). However, in the child labor condition, participants had significantly worse recognition of child labor information compared to nonethical information (χ2 = 20.38, p < .0001). This difference in relative memory across the labor practices variable was driven by a decrease in memory for child labor compared to adult labor and not a change in memory across levels of the nonethical information. Mediation by Want/Should Conflict Recall that we have a measure of participants’ want/should conflict. We conducted an ANOVA with this conflict measure as the dependent variable and labor practices (adult vs. child), the task (shopping cart or categorization), and their interaction as independent variables. As expected, only a significant main effect of labor practices emerged (F(1, 389) = 6.44, p = .01). Participants in the child labor condition experienced a significantly greater desire to avoid the negative feelings associated with the want/should conflict (M = 4.60) compared to participants in the adult labor condition (M = 4.26). There was no main effect of the task nor an interaction between the two variables (all ps > .33). This lack of a significant interaction supports the notion that the categorization task elicited want/should conflict among consumers. Finally, a mediation model using PROCESS model 4 yielded the expected results (figure 2; Hayes 2013) that this conflict measure mediates the willfully ignorant memory effect. Participants in the child labor condition experienced a significantly greater desire to avoid the negative feelings associated with the want/should conflict, and this corresponded to greater willfully ignorant memory (using 10,000 bootstrapped samples: β = –.03, SE = .02, 95% CI [–.071, –.001]). FIGURE 1 View largeDownload slide STUDY 2: DIFFERENCE IN RECOGNITION MEMORY FOR CHILD LABOR VERSUS ADULT LABOR FIGURE 1 View largeDownload slide STUDY 2: DIFFERENCE IN RECOGNITION MEMORY FOR CHILD LABOR VERSUS ADULT LABOR FIGURE 2 View largeDownload slide STUDY 2: MEDIATION OF WANT/SHOULD CONFLICT ONTO RECOGNITION OF CHILD LABOR VERSUS ADULT LABOR FIGURE 2 View largeDownload slide STUDY 2: MEDIATION OF WANT/SHOULD CONFLICT ONTO RECOGNITION OF CHILD LABOR VERSUS ADULT LABOR Discussion Providing additional support for hypotheses 1a and 1b, the results of study 2 demonstrate that willfully ignorant memory generalizes across product categories and types of ethical attributes and extends to recognition tasks. Study 2 also provides support for hypothesis 2a, that willfully ignorant memory is driven by the desire to avoid negative feelings associated with the want/should conflict that arises when consumers are faced with negative ethical information. STUDY 3 In studies 1 and 2, we demonstrate that when consumers face want/should conflict regarding ethical attributes, the default response is to allow the want self to prevail by forgetting or misremembering negative ethical information. The primary purpose of study 3 was to test whether reducing the amount of pressure exerted by the should self reduces the amount of conflict consumers feel, thereby decreasing willfully ignorant memory. If willfully ignorant memory is driven by the want/should conflict, reducing this conflict by reducing the importance of a given ethical issue should result in relatively equal memory across positive and negative performance on an ethical attribute. We thus test hypothesis 2b in a similar recall task to that used in study 1 by manipulating the relative ethical weight of the issue at hand, child labor. Before beginning the memory task, half of the participants received a reminder of the myriad other ethical issues facing humanity (including some, like genocide, that are arguably more morally important than use of child labor), thus reducing the ethical weight of discovering that a pair of jeans has been made using some child labor. Our prediction is that providing this perspective will result in child labor seeming less important, thus reducing want/should conflict and resulting in relatively less willfully ignorant memory. Finally, we further control for the fact that some of the brand names used in study 1 differed in length/complexity by using only simple, four-letter brand names in this study (see appendix F for details). Participants and Procedure A total of 117 undergraduate students participated in this study in return for course extra credit. The study employed the same between-subjects and within-subject tests of memory for the ethical attribute (i.e., labor practices of the manufacturer) as in study 1 and also included a two-level between-subjects variable: reminder of broader ethical context (present vs. absent). In the condition in which the reminder of the broader ethical context was present, participants read the following: There are many problems in the world today. For instance, many people worry that the environment is being polluted to the point of no return. Also, animal cruelty is a problem in many areas. Poverty is increasing as well. Below is a list of current issues facing society. Please rate how helpless you feel about the chance of correcting these issues.Participants then rated how helpless each of the following issues made them feel on a seven-point scale: population growth outstripping resources worldwide, genocide, racism, natural resources running out or being degraded, pollution, and animal cruelty. These problems were selected because they are all relatively large, intractable problems with no single, easy solution. We wanted to remind participants that there are many worldwide ethical issues and that it is difficult, if not impossible, to solve all of them. The idea was that this reminder of the enormity of ethical issues in the world reduces the ethical weight of the specific issue of child labor, thus reducing the pressure exerted by one’s should self and creating less want/should conflict. The reduced want/should conflict that results should relieve the need for willfully ignorant memory. Participants in the condition in which this reminder of the broader ethical context was absent instead read a short introduction about the variety of vehicles on the road today and rated how frequently they see each of the following types of vehicles when they are out driving: truck, station wagon, convertible, sports car, minivan, and SUV. Upon completing the rating task, participants then completed what was ostensibly a separate study. All participants were told that they would be seeing information about six brands of jeans (shown in appendix F) and were asked to read the following: Each of these brands is sold in the U.S., but manufactured in Asia. The jeans differ on three main attributes, the wash (i.e., are the jeans dark or light), their cut/style (i.e., are they boot cut or straight leg), and the factory conditions in which they were produced (i.e., using child labor or using adult labor only).The procedure for this second phase of the study was similar to the memory task used in study 1, although conducted on computers instead of with paper-and-pencil questionnaires. As in study 1, participants were told that the study was designed to test their memory and that they would be asked to memorize all of the information about two of the brands assigned by the computer. After being assigned two of the brands, participants then took 30 seconds to view the information about all six of the jeans brands. The jean information was then cleared from the computer screen, and participants were asked to type the brand name and all the other information they could remember about the two assigned jeans. These two steps (30 seconds of viewing the information about all six brands of jeans and the subsequent typing task about the two selected brands) were then repeated twice. Thus, participants had three opportunities to view and memorize the information. Following this induced memorization task, as in study 1, all participants were then shown all six brand names (with no additional information) and were asked to type as much as they could remember about all six brands of jeans, and memory on this task was used as our focal measure. Finally, in order to identify what constituted good and poor performance on the nonethical attributes, participants were asked which they preferred, light or dark wash jeans and boot-cut or straight-leg jeans. Pretest In order to establish that the manipulation reminding participants about the variety of ethical issues in the world would indeed reduce the relative weight they place on child labor as an ethical attribute, we ran a separate pretest with 126 native English-speaking participants from the same population that participated in the main study in which we (1) exposed them to one of the two experimental primes from the main study and then (2) asked them to indicate their level of agreement with three statements (α = .92) assessing how important child labor is to them personally (see appendix C for measures). All responses were measured on seven-point scales. The results of the pretest indicated that participants exposed to the prime asking how helpless a host of other ethical issues made them feel reported that child labor was significantly less important to them (M = 2.88) compared to those exposed to the neutral prime (M = 3.33; F(1, 124) = 3.94, p < .05), indicating that our manipulation worked as intended. Results and Discussion As expected, the additional round of memorization in this study relative to study 1 resulted in very low levels of forgetting at the storage phase. Only eight of the adult labor responses (6.84% of all responses) and 10 of the child labor responses (8.55% of all responses) for the two assigned jeans were incorrect or missing. Due to these low levels of forgetting, there were no significant main or interactive effects of prime or level of ethical attribute at storage on memory for the ethical attribute in either a within- or between-subjects analysis. Indeed, even at retrieval, memory for the two assigned jeans remained stable. Only three of the 204 possible responses were incorrect. These results suggest that with enough active rehearsal of ethical attribute information, consumers cannot forget as they might like to. Given the high levels of storage memory (and high levels of retrieval memory for the two assigned jeans), our focal analysis is within-subjects retrieval memory for all six jeans. Memory at retrieval (using the subset of 102 participants who correctly stored the ethical labor attribute information at time 1) for all of the jeans showed the expected effects. The repeated-measures analysis revealed that there was not a significant overall difference in memory for child versus adult labor (F(1, 100) = 0.94, p = .33). However, the focal interaction between labor type and prime was significant and in the predicted direction (F(1, 100) = 5.39, p < .05). Follow-up spotlight analyses to explicate the effect of the prime on memory revealed that participants in the condition in which the reminder of the broader ethical context was absent showed the expected willfully ignorant memory effect. Memory for adult labor (24.31% of all responses incorrect or missing, 75.69% of all responses correct) was better than memory for child labor (33.33% of all responses incorrect or missing, 66.67% of all responses correct; F(1, 100) = 5.11, p < .05). In contrast, participants in the condition in which this reminder was present did not show this effect; there was no difference in memory for child versus adult labor for these participants (F(1, 100) = 0.96, p = .33; see figure 3 for details).2 FIGURE 3 View largeDownload slide STUDY 3: DIFFERENCE IN MEMORY FOR CHILD VERSUS ADULT LABOR (ACROSS ALL SIX BRANDS OF JEANS) BY PRESENCE OF REMINDER OF THE BROADER ETHICAL CONTEXT FIGURE 3 View largeDownload slide STUDY 3: DIFFERENCE IN MEMORY FOR CHILD VERSUS ADULT LABOR (ACROSS ALL SIX BRANDS OF JEANS) BY PRESENCE OF REMINDER OF THE BROADER ETHICAL CONTEXT There were no differences in memory by cut/style (straight leg or boot cut) (F(1, 92) = 0.45, p = .51) nor by wash (light or dark wash) (F(1, 97) = 0.32, p =. 57). We also combined these two nonethical attributes and found no differences in memory for the positive level (dark wash and straight cut) versus the negative level (light wash and boot cut) for this combined attribute category (F(1, 105) = 0.14, p = .70). Furthermore, the prime did not have an effect on memory for the different levels of the cut/style attribute (F(1, 92) = 0.53, p = .47) or wash attribute (F(1, 97) = 1.61, p = .21), or the variable combining these two nonethical attributes (F(1, 105) = 1.53, p = .22).3 Note that, for consistency, we removed all participants from this analysis who failed to correctly memorize the focal attribute (i.e., style/cut or wash) during the storage phase. Post-Test To further confirm that our reminder of other ethical issues reduces willfully ignorant memory by reducing want/should conflict, we conducted a post-test with 120 participants from the same population as the main study. These participants were exposed to one of the two experimental primes from the main study and then were shown the six descriptions of jeans and, as filler questions, asked to indicate which pair they would be most likely to buy and which pair they would be least likely to buy. They then indicated their agreement with the want/should conflict measures from the pilot study and study 2. Participants reminded of the myriad other ethical issues in the world reported experiencing less conflict after seeing the set of jeans containing negative ethical information (M = 4.87) than those who did not see this reminder (M = 4.31; F(1, 118) = 8.04, p < .01). Discussion The results of study 3 and the post-test therefore support hypothesis 2b, that willfully ignorant memory is attenuated when the pressure exerted by the should self is reduced. We show that the pressure of the should self can be reduced when consumers perceive the ethical issue at hand as having relatively less ethical weight. When this reduction occurs, consumers experience less want/should conflict and are more likely to remember ethical information. STUDY 4 Throughout, we have assumed that the want/should conflict presented by ethical attributes leads to willfully ignorant memory because people (perhaps unconsciously) believe that not remembering information is morally preferable to remembering it and not using it. In study 4 we aim to show that consumers indeed perceive motivated forgetting as a relatively morally acceptable coping mechanism for the resolution of conflict between the want and should self. To do so, we asked study participants to evaluate the actions of another consumer who had forgotten a negative (vs. positive) ethical attribute versus one who instead remembered this information but chose not to use it (actively choosing the want self over the should self instead of resolving the conflict via forgetting). We expected study participants to rate motivated forgetting of a negative ethical attribute as more morally acceptable compared to remembering but not using this information. Participants and Procedure A total of 341 individuals completed this study on Amazon Mechanical Turk in return for a monetary payment. All participants read a scenario about a hypothetical consumer named Chris (who we varied to be male or female) who was shopping online for a pair of jeans and found a desired pair. The description of the pair of jeans was identical to that of study 2; two of the attributes were not related to ethicality and did not differ across participants, but the third attribute did relate to ethicality and varied between-subjects (whether the jeans were made with child or adult labor). After reading the description of Chris’s desired pair of jeans, participants learned that Chris decided to wait until the following day before actually making the purchase. We then manipulated whether, while trying to remember the description of the jeans the following day, Chris forgot what type of labor practices the manufacturer used or remembered this information but chose not to consider it as part of the decision-making process. In both cases, Chris went on to purchase the jeans. Therefore, the study employed a 2 (labor practices: adult or child) × 2 (memory for ethical information: forget or remember but ignore) × 2 (Chris’s gender: male or female) between-subjects design. All participants then answered the following questions about Chris (on seven-point scales): “How ethical is Chris?” and “How good of a person is Chris?” These two measures (r = .75, p < .0001) were combined to form an overall judgment of Chris’s character. Participants were then asked to imagine themselves in Chris’s position and answered the following question, which varied slightly based on the memory for ethical information condition: “How good or bad would you feel if you forgot the information about the manufacturer’s labor practices by the next day [remembered the information about the manufacturer’s labor practices the next day, but chose to ignore it] when making your purchase?” Results Chris’s gender did not significantly impact any results, and therefore we collapse across this variable in subsequent analysis. First, we conducted an ANOVA with the labor practices and memory for ethical information variables and their interaction as independent variables and the measure of Chris’s overall character as the dependent measure. Significant main effects of both labor practices (F(1, 337) = 57.61, p < .0001) and memory for ethical information (F(1, 337) = 7.87, p < .01) emerged, but these were qualified by a significant interaction (F(1, 337) = 4.34, p < .05). As indicated by the main effect of labor practices, participants in the child labor condition viewed Chris as a worse person than those in the adult labor condition (which is unsurprising since he/she bought jeans made with child labor in this condition). The important finding for our purposes was that this difference was significantly amplified when Chris remembered and ignored the information, instead of merely forgetting the information (simple effect for remember and ignore condition: Madult labor = 4.82 vs. Mchild labor = 3.62; F(1, 337) = 46.92, p < .0001; simple effect for forget condition: Madult labor = 4.91 vs. Mchild labor = 4.23; F(1, 337) = 15.12, p = .0001). Next, we conducted the same analysis with judgments of self as the dependent measure. A significant main effect of labor practices emerged (F(1, 337) = 54.00, p < .0001), which was qualified by the same interaction (F(1, 337) = 14.51, p < .001): participants in the child labor condition judged themselves more harshly than those in the adult labor condition, but this difference was significantly amplified when participants imagined remembering and ignoring the information instead of merely forgetting the information (simple effect for remember and ignore condition: Madult labor = 4.10 vs. Mchild labor = 2.53; F(1, 337) = 62.24, p < .0001; simple effect for forget condition: Madult labor = 3.71 vs. Mchild labor = 3.21; F(1, 337) = 6.26, p < .05). Discussion The results of study 4 indicate that forgetting negative ethical information is judged less harshly than remembering this information and ignoring it. These results hold true whether participants are judging another individual or anticipating their own behavior and judging themselves. Thus, motivated forgetting seems to be an acceptable coping mechanism for dealing with want/should conflict, even when it allows the want self to triumph over the should self. GENERAL DISCUSSION Consumers often encounter product information in the marketplace that is relevant to broader ethical problems, such as discovering that a piece of furniture is made from wood from an endangered rainforest or that a pair of jeans is made using child labor in an overseas factory. Across three studies, we demonstrate that consumers exhibit willfully ignorant memory for this type of ethical product information: they have relatively better memory for good versus poor performance on ethical attributes. In our fourth study, we provide evidence that consumers find this type of forgetting more morally acceptable than remembering but failing to use ethical information in decision making. In the sections that follow, we discuss the theoretical and practical implications of these findings. Theoretical Implications Past research has documented a number of choice-related differences in how ethical versus nonethical attributes are treated in consumer behavior processes, such as the formation of consideration sets (Irwin and Naylor 2009), the formation and expression of preferences (Baron and Spranca 1997; Ritov and Baron 1999), and search for product information (Ehrich and Irwin 2005). The current research looks at another aspect of the consumer process, memory, and documents a systematic memory bias for ethical attribute information. To our knowledge, no prior research has explored whether there are systematic differences in memory for different levels of a specific type of attribute or whether there are systematic differences in memory for ethical versus nonethical product attributes. The findings in this research thus represent the first documentation of actual changes in memory for product attributes based on their ethicality. We show that the changes happen not just at the encoding stage of memory, where they might reflect a type of willful ignorance for attending to negative information, but also at retrieval. Our work is therefore consistent with Kouchaki and Gino’s (2016) work documenting “amnesia” for prior unethical actions, but makes a new contribution to the work on biased memory for unethical information by showing that such biases can occur even when a consumer has not actually engaged in any unethical actions themselves. Instead, the memory bias we observe is driven by the want/should conflict that occurs simply from evaluating products that vary in ethicality, even if one does not choose to actually purchase an unethical product. While documenting this effect, we also affirm the negativity bias for memory for some of the other attributes in our studies that do not have ethical implications (Baumeister et al. 2001), showing that the negativity bias often obtains, but that willfully ignorant memory appears to trump it when ethical product information is involved. Theoretically, these results suggest that the expectation that consumers will read and retain product attribute information, even clearly articulated information, does not apply to ethical attributes, as consumers exhibit willfully ignorant memory both during cued recall and recognition tasks. Thus, reported preference for (and likelihood of purchasing) products that have ethical attributes may not reflect interest in the underlying ethical issue. In addition, we show that willfully ignorant memory appears to be driven by the conflict between the should versus want self (Bazerman et al. 1998; Milkman et al. 2008) that is generated when one is evaluating products with negative ethical performance. The want self wants to avoid unpleasant information that can trigger negative emotions, while the should self focuses on what the person believes is right. This conflict is typically resolved in favor of the want self, by forgetting or misremembering negative ethical information at retrieval (as we show in studies 1 and 2). However, we also demonstrate that when the pressure exerted by the should self is weakened (by lessening the ethical weight of the issue at hand in study 3), conflict is reduced, thereby reducing the forgetting and misremembering of negative ethical attribute information. In study 4, we provide direct evidence that consumers consider willfully ignorant memory a relatively morally acceptable coping mechanism, as they judge others and themselves less harshly when negative ethical information is forgotten compared to when it is remembered but ignored during the decision-making process. These results suggest that motivated forgetting is a socially acceptable way to resolve want/should conflict, even though it may result in purchasing unethical products. Finally, we also demonstrate that the effect is robust across products and ethical issues. Implications for Understanding Preference for Ethical Products Many consumers report that they care about ethical issues like environmentalism and workers’ rights and that they are willing to pay more for ethically produced goods (Trudel and Cotte 2009). Social norms and a desire for status seem to enforce this behavior (Goldstein, Cialdini, and Griskevicius 2008; Griskevicius, Tybur, and van den Bergh 2010), and consumers often report that they buy ethical products. In fact, a recent Nielsen survey indicated that 50 percent of global consumers report that they are willing to pay more for products from socially responsible companies (Nielsen 2013). However, self-reports of ethical behavior in the marketplace do not always appear to match reality. Products with good performance on ethical attributes (i.e., sustainable products) are, in many product categories, not market leaders (Gleim et al. 2013), and products with poor performance on ethical attributes continue to sell. For example, despite the damage to the environment from logging rainforests, consumers continue to buy furniture and other products made from rainforest wood (Rainforest Relief 2017). Past research has addressed the causes of this gap by suggesting that consumers lie when they report their purchasing behavior, because of socially desirability (Luchs et al. 2010) or because of what Batson et al. (1999) term “moral hypocrisy,” the reporting of values that one does not actually hold. Undoubtedly, such active misrepresentation occurs. Our findings suggest, however, that even consumers who are not purposefully misrepresenting the quantity and frequency of their ethical purchases may still be mistaken in their purchasing self-reports, describing their purchases as more ethical than they actually are because of the willfully ignorant memory effect. They may simply have forgotten that the purchases they made in the past were unethical or may have actually reversed attribute values from poor performance on an ethical attribute to good performance, leading them to report having made purposefully ethical purchases. Willfully ignorant memory may therefore lead consumers to be less likely to buy ethical products, not only because they may not accurately store ethical product information in memory after they are exposed to it, but also because they are likely to retrieve an inaccurately high number of ethical purchases (and low number of unethical purchases) in their past and may feel that they therefore have a license to be less ethical in the present (Khan and Dhar 2006; May and Irmak 2014). Work on attitude-behavior consistency (Verplanken and Holland 2002) has shown that attitudes (e.g., a moral value like the desire to protect the rainforest) have to be salient at the moment of action (i.e., the point of purchase) in order for them to impact the relevant action (e.g., choosing a desk made from sustainable tree farm wood). Our research suggests not only that the value has to be activated, but also that the relationship of that value to the attributes of the product has to be successfully recalled at the point of purchase, a connection that we have shown is not consistent and foolproof. Our work also has implications for researching consumer decision making. Our results suggest that, after reading a product concept description about a product with poor performance on an ethical attribute, respondents may willfully forget the negative performance information before indicating a preference among products. There is no way to know that willfully ignorant memory has taken place unless researchers ask for recall (or recognition) of the product information (which most researchers do not do). Thus, research may conclude that consumers do not care about ethical issues because respondents report being willing to buy products with poor ethical performance. Similarly, our results suggest that surveys relying on consumer recall to measure the amount and frequency of past ethical purchases are likely biased by systematic memory errors that lead consumers to report more ethical purchases than they actually made. Consciousness and Emotion One of the questions that our research may point to is whether willfully ignorant memory is a conscious process and the role that strong emotion plays in the phenomenon. The traditional psychoanalytic treatment of motivated memory was that it was unconscious and driven by the need to avoid shameful desires and/or thinking about traumatic events (Freud 1914/2001). However, we honestly do not know whether past work on trauma and emotion applies to our findings, just that it does not seem to be completely compatible with it. Recent work has affirmed that biases in memory can occur even without traumatic emotions; Dalton and Huang (2014) showed that consumers may be (consciously or unconsciously) motivated to forget marketing promotions that threaten their identity. Future research can explore the exact nature of, and types of, willfully ignorant memory. Past cognitive research suggests that people are unable to consciously control what they do and do not remember (e.g., the persistence of traumatic memories that can result in post-traumatic stress; Schacter 1999), which is one reason why consumers employ external strategies in an attempt to protect particularly valuable memories (e.g., purchasing souvenirs on trips they want to remember; Zauberman et al. 2008) or to forget particularly painful ones (Marcoux 2017). Although consciousness is not a focus of our research, our findings also suggest that willfully ignorant memory in an ethical domain is at least partially unconscious. With six items to remember in studies 1 and 3 and no strong motivation to remember them, it is likely our participants were driven by instinct throughout. Our suspicion is that the avoidance of ethical information is learned over time, with no conscious cognitive oversight. Consumers may simply learn to discard unappealing ethical information from their minds, much as people learn to ignore all the other everyday information that can get in the way of an easy and enjoyable life. However, despite our suspicions, we cannot make any strong claims for conscious versus unconscious misremembering and can only suggest that future research might explore this issue. Exploring whether consumers still exhibit willfully ignorant memory in the face of incentives is one path that might be particularly productive for doing so. Avenues for Future Research There are a number of important ways in which future research could build on these findings. Future research could build upon study 1’s documentation of errors at the storage/encoding versus retrieval stage by incentivizing one set of processes versus the other (see Wood and Lynch 2002 for an example of this type of procedure). The study 1 results also suggest that product preference does not predict memory errors. In this way, willfully ignorant memory seems slightly different than willful ignorance in information search, which has been shown to be influenced by product attractiveness (Ehrich and Irwin 2005). Next, although we did not see any compelling patterns in failing to remember versus remembering incorrectly, future research could also explore differences between failing to remember information and remembering it incorrectly. Finally, in all of our studies we explicitly explored contexts in which consumers evaluated products that vary in ethicality, which means that want/should self conflict occurred at encoding as well as at retrieval. Future research could explore contexts in which no evaluation is required at encoding (i.e., consumers are not asked to consider the products for purchase or to categorize products but are simply shown product information that they are not asked to use in any way), but where participants do need to use the information at retrieval (e.g., they are asked to consider purchase only when retrieving the information from memory). Doing so would allow researchers to test whether willfully ignorant memory occurs when the moral conflict is experienced specifically at retrieval and not an encoding (as it is in all of our studies). We note, however, that any time multiple stimuli are presented together, participants will naturally be likely to compare across them (e.g., participants are likely to compare any products shown together because they are used to shopping). Such comparisons, even when unintended, are likely to lead to evaluation and thus conflict. DATA COLLECTION INFORMATION The pilot study was collected during spring 2017 at the Fisher Behavioral Lab at the Ohio State University by Daniel M. Zane and Rebecca Walker Reczek, and by a lab manager and research assistants working under their supervision. The study 1 and study 3 pretests were collected at the Fisher Behavioral Lab at the Ohio State University by Rebecca Walker Reczek and by a lab manager and research assistants working under her supervision during fall 2014. Study 1 data was collected during spring 2004 in the McCombs Behavioral Lab at the University of Texas at Austin. This study was conducted by Kristine Ehrich and research assistants working under her supervision. Study 2 data was collected by Daniel M. Zane and Rebecca Walker Reczek during March 2017 using Amazon Mechanical Turk. Study 3 data was collected during summer 2011 in the McCombs Behavioral Lab at the University of Texas at Austin by Julie R. Irwin and by research assistants working under her supervision, and at the Fisher Behavioral Lab at the Ohio State University, by Rebecca Walker Reczek and by a lab manager and research assistants working under her supervision. The study 3 post-test was collected during spring 2017 at the Fisher Behavioral Lab at the Ohio State University by Daniel M. Zane and Rebecca Walker Reczek, and by a lab manager and research assistants working under their supervision. Study 4 data was collected by Daniel M. Zane and Rebecca Walker Reczek in April 2017 using Amazon Mechanical Turk. Study 1 was designed by the second and fourth authors. Study 2 was designed by the first and third authors. Study 3 was designed by the first, second, and fourth authors. Study 4 was designed by the first, second, and third authors. The data analysis for the pilot study, study 2, study 3 post-test, and study 4 was conducted jointly by the first and third authors. Data analysis for the study 1 pretest, study 1, study 3 pretest, and study 3 was conducted jointly by the first and second authors. The third author also conducted additional analysis for study 1 and study 3 at a later time. The authors thank Aradhna Krishna, Chris Janiszewski, and Cait Lamberton for their comments on earlier versions of this research, and Amy Dutton, Jordan Fulk, Stephanie Marshall, Christopher Summers, and Sarah Zimmerman for their assistance in data collection and coding. APPENDIX A JEANS BRAND DESCRIPTIONS USED IN PILOT STUDY Cole Jeans: These jeans feature a light blue wash and fashionable straight-leg cut. This brand of jeans is made in a facility that uses adult labor [is made using a blend of cotton and Lycra for added stretch]. West Jeans: These jeans are made with a classic boot-cut leg and dark blue wash. This brand of jeans is made in a factory that uses child labor [using 100% cotton]. Park Jeans: These jeans are known for their very dark wash and stylish boot cut. This brand of jeans is made using only adult labor [using a cotton/Lycra blend for added stretch]. Nash Jeans: These jeans have a deep, dark wash and straight leg. This brand of jeans is made by a company known to use all adult labor [is made using a cotton/Lycra blend fabric for added stretch]. Rowe Jeans: These jeans have a very light wash and hip straight leg. This brand of jeans is made in a factory with child labor conditions [is made using 100% cotton fabric]. Tate Jeans: These jeans are straight-leg jeans with an attractive dark wash. This brand of jeans is produced in factories that use child labor [is produced using 100% cotton fabric]. APPENDIX B DESK DESCRIPTIONS USED IN STUDY 1 International Furniture Company Desk: This desk is a designer desk with excellent workmanship and style. The desk is made out of wood from tree farms, where the trees and forest are not endangered. This desk costs $750. Washington Industries Desk: This desk is a bargain at only $350. The desk is made from wood taken from rainforests in Thailand. The desk is well-made, although not a designer desk and should last a reasonably long time. Picklebury Desk: This desk is made from tree farm wood, where the trees and forests are not threatened by endangerment. The desk is only $350. It is well-made, although not designer, and should prove to be a sturdy companion for the future. Stevensville Desk: Stevensville is a brand with good, although not designer, quality workmanship. This desk costs $750. The wood is taken from tree farms, where the wood and forests are not endangered. Eubank Designs Desk: This desk is made out of wood taken from rainforests in the tropics. It is very beautifully crafted with designer style and workmanship. It costs $750. Hawthorne Woods Desk: This desk is made from trees taken from a rainforest in Asia. It is a beautifully designed desk, with fabulous workmanship. It only costs $350. APPENDIX C QUESTIONS USED IN PRETESTS Study 1 Pretest: Some of the desk descriptions you read described the wood being taken from sustainable tree farms and some described the wood being taken from rainforests. How much do you agree with the following statement about these different wood sources? Wood taken from tree farms is not endangered. Wood taken from rainforests is endangered. Whether wood comes from rainforests or tree farms is an ethical issue that companies and consumers must decide. Many people likely consider it unethical for a company to take wood from a rainforest. Protection of rainforests (and the animal and plant life in them) is an ethical issue. Desks made from rainforest wood are made from endangered types of wood. Desks make from rainforest wood are less ethical than desks made from tree farm wood. An ethical consumer would be more likely to buy one of the desks made from tree farm wood than one made from rainforest wood. Wood can come from many different sources. How atypical or unusual is it that a desk is made from rainforest wood? How atypical or unusual is it that a desk is made from tree farm wood? Study 3 Pretest: There are many things that consumers take into account when buying products. How much do you agree with each of the following statements? I try to pay careful attention to whether a company uses child labor when deciding what brands to buy. Child labor is an issue I care deeply about when making purchases. It is important to me, personally, to “vote with my wallet” to stop child labor by not buying products from companies that use it. APPENDIX D QUESTIONS USED TO ENSURE PROCESSING OF DESK (STUDY 1) AND JEAN INFORMATION (STUDY 3) AT STORAGE Which desk [brand of jeans] do you like best? Which desk [brand of jeans] do you like the least? Which description do you think is the most readable? Which description do you think is the least readable? Which description has the best adjectives (descriptive words)? Which desk [brand of jeans] seems most like a desk [brand] you might find in a store? Which desk [brand of jeans] seems least like a desk [brand] you might find in a store? Which desk [brand of jeans] do you think is the best seller among these desks [brands]? APPENDIX E DESCRIPTIONS OF CLOTHING ITEMS USED IN STUDY 2 A T-shirt is a must-have piece of clothing to own! Shirt This T-shirt is a black V-neck and is made with cotton. It can be worn on its own or underneath sweaters. Shirt B This crewneck T-shirt is charcoal in color. It is made from polyester and makes for a very versatile piece. Shirt C This multi-occasion T-shirt has navy blue and white stripes. It is a crewneck and made out of cotton. A light jacket is great to have around! Jacket X This lightweight jacket is solid black in color and waterproof. It is machine washable. Jacket Y This blue and gray jacket is made with a medium weight material and is wind resistant. It is machine washable. Everyone needs that go-to pair of jeans! Jeans J This pair of jeans features a fashionable skinny cut and a dark blue wash. They were made in an overseas factory that is known to use some child labor in addition to adult labor [only adult labor]. Shoes always round out an outfit! Shoes M This pair of shoes can be worn with most casual outfits. They are black and white low-tops that are made from suede material and have a nonslip sole. Shoes N This pair of shoes has a sporty look, but can be worn in many settings. They are brown and made from leather, with increased arch support. APPENDIX F DESCRIPTIONS OF JEANS BRANDS USED IN STUDY 3 Cole Jeans: These jeans feature a dark blue wash and fashionable boot cut. Cole jeans are made in a facility that uses adult labor. Nash Jeans: Nash jeans are made with a classic straight leg and light blue wash. These jeans are made in a factory that uses child labor. Park Jeans: These jeans are known for their very light wash and slim-fit straight legs. Park jeans are made using only adult labor. Rowe Jeans: Rowe jeans have a deep, dark wash and straight leg. These jeans are made by a company known to use all adult labor. Tate Jeans: These jeans have a very dark wash and hip boot-cut leg. Tate jeans are made in a factory with child labor conditions. West Jeans: West jeans are stylish boot-cut jeans with an attractive light wash. They are produced in factories that use child labor. Footnotes 1 Note that the above results replicate when we use a three-way coding scheme for the dependent variable in a linear regression: 1 = chose any description with correct ethical information, regardless of whether the other attributes (cut and wash) were correct, 0 = chose any description with ethical information missing, regardless of whether the other attributes (cut and wash) were correct, –1 = chose any description with incorrect ethical information, regardless of whether the other attributes (cut and wash) were correct. The same pattern of results also replicates when we use the following alternate coding scheme for the dependent variable of the logistic regression: 1 = chose the single overall correct description, 0 = chose any of the other five overall incorrect descriptions. 2 This pattern of results replicated with the alternate coding scheme. When memory was coded as correct, incorrect, or missing, there was, again, no significant overall difference in memory for child versus adult labor (p = .51), but there was a marginally significant interaction between labor type and prime (F(1, 100) = 3.48, p = .07). When the reminder of the broader ethical context was absent, memory for adult labor (11.11% of all responses missing, 13.19% of all responses incorrect, 75.69% of all responses correct) was marginally better than memory for child labor (18.75% of all responses missing, 14.58% of all responses incorrect, 66.67% of all responses correct; F(1, 100) = 3.03, p = .09). In contrast, when the reminder was present, there was no difference in memory for child versus adult labor (F(1, 100) = 0.76, p = .38. 3 Further analysis comparing the relative memory for the positive versus negative performance level for each attribute suggests that the prime did not affect relative memory for all three attributes the same way. Specifically, we conducted a mixed analysis with two within-subjects contrasts and the between-subjects prime. The two within-subjects contrasts compared (1) the ethical attribute to the nonethical attributes and (2) the positive level of the attributes to the negative level of the attributes. These within-subjects contrasts ultimately allowed us to examine relative memory for child labor compared to the negative level of nonethical attributes and relative memory for adult labor compared to the positive level of the nonethical attributes. We then examined whether the difference in these relative memories was affected by the between-subjects prime. This analysis revealed a trending three-way interaction (F(1, 100) = 1.77, p = .19). We first examined the two-way interaction between the two within-subjects contrasts in the between-subjects condition in which the reminder of the broader ethical context was absent. This interaction was trending (F(1, 47) = 1.77, p = .19), indicating that memory for child labor relative to the negative level of the nonethical attributes was worse than memory for adult labor relative to the positive level of the nonethical attributes. In the between-subjects condition in which the reminder of the broader ethical context was present, the two-way interaction between the two within-subjects contrasts was nonsignificant (F(1, 53) = 0.25, p = .62). Though relative memories for adult and child labor in this condition were not significantly different from one another, they did directionally flip, indicating greater relative memory for child labor compared to relative memory for adult labor. REFERENCES Anderson Michael C., Bjork Robert A., Bjork Elizabeth L. ( 1994), “Remembering Can Cause Forgetting: Retrieval Dynamics in Long-Term Memory,” Journal of Experimental Psychology: Learning, Memory and Cognition , 20 5, 1063– 87. Google Scholar CrossRef Search ADS Aquino Karl, Reed AmericusII ( 2002), “The Self-Importance of Moral Identity,” Journal of Personality and Social Psychology , 83 6, 1423– 40. Google Scholar CrossRef Search ADS Bandura Albert ( 1999), “Moral Disengagement in the Perpetration of Inhumanities,” Personality and Social Psychology Review , 3 3, 1193– 209. Google Scholar CrossRef Search ADS Bandura Albert, Barbaranelli Claudio, Caprara Gian Vittorio, Pastorelli Concetta ( 1996), “Mechanisms of Moral Disengagement in the Exercise of Moral Agency,” Journal of Personality and Social Psychology , 71 2, 364– 74. Google Scholar CrossRef Search ADS Baron Jonathan, Spranca Mark ( 1997), “Protected Values,” Organizational Behavior and Human Decision Processes , 70 1, 1– 16. Google Scholar CrossRef Search ADS Bartlett Frederic C. ( 1932), Remembering: A Study in Experimental and Social Psychology , Cambridge, UK: Cambridge University Press. Batson Daniel C., Thompson Elizabeth R., Seuferling Greg, Whitney Heather, Strongman Jon A. ( 1999), “Moral Hypocrisy: Appearing Moral to Oneself without Being So,” Journal of Personality and Social Psychology , 77 3, 525– 37. Google Scholar CrossRef Search ADS Baumeister Roy F. ( 2002), “Yielding to Temptation: Self-Control Failure, Impulsive Purchasing, and Consumer Behavior,” Journal of Consumer Research , 28 4, 670– 6. Google Scholar CrossRef Search ADS Baumeister Roy F., Bratslavsky Ellen, Finkenauer Catrin, Vohs Kathleen ( 2001), “Bad Is Stronger than Good,” Review of General Psychology , 5 4, 323– 70. Google Scholar CrossRef Search ADS Bazerman Max H., Tenbrunsel Ann E., Wade-Benzoni Kimberly ( 1998), “Negotiating with Yourself and Losing: Making Decisions with Competing Internal Preferences,” Academy of Management Review , 23 2, 225– 41. Google Scholar CrossRef Search ADS Braun Kathryn A. ( 1999), “Postexperience Advertising Effects on Consumer Memory,” Journal of Consumer Research , 25 4, 319– 34. Google Scholar CrossRef Search ADS Braun Kathryn A., Ellis Rhiannon, Loftus Elizabeth F. ( 2002), “Make My Memory: How Advertising Can Change Our Memories of the Past,” Psychology and Marketing , 19 1, 1– 23. Google Scholar CrossRef Search ADS Brown Jonathon D., Dutton Keith A. ( 1995), “Truth and Consequences: The Costs and Benefits of Accurate Self-knowledge,” Personality and Social Psychology Bulletin , 21 12, 1288– 96. Google Scholar CrossRef Search ADS Cowley Elizabeth ( 2008), “The Perils of Hedonic Editing,” Journal of Consumer Research, 35 1, 71– 84. Google Scholar CrossRef Search ADS Cowley Elizabeth, Janus Eunika ( 2004), “Not Necessarily Better, but Certainly Different: A Limit to the Advertising Misinformation Effect on Memory,” Journal of Consumer Research , 31 1, 229– 35. Google Scholar CrossRef Search ADS Dalton Amy N., Huang Li ( 2014), “Motivated Forgetting in Response to Social Identity Threat,” Journal of Consumer Research , 40 6, 1017– 38. Google Scholar CrossRef Search ADS DePalma Mary Turner, Madey Scott F., Bornschein Susan ( 1995), “Individual Differences and Cheating Behavior: Guilt and Cheating in Competitive Situations,” Personality and Individual Differences , 18 6, 761– 9. Google Scholar CrossRef Search ADS Dunning David ( 2007), “Self-Image Motives and Consumer Behavior: How Sacrosanct Self-beliefs Sway Preferences in the Marketplace,” Journal of Consumer Psychology , 17 4, 237– 49. Google Scholar CrossRef Search ADS Ehrich Kristine R., Irwin Julie R., ( 2005) “Willful Ignorance in the Request for Product Attribute Information,” Journal of Marketing Research , 42 3, 266– 77. Google Scholar CrossRef Search ADS Eisenberg Nancy ( 2000), “Emotion, Regulation, and Moral Development,” Annual Review of Psychology , 51 ( February), 665– 97. Google Scholar CrossRef Search ADS Freud Sigmund ( 1914/2001), On Dreams , trans. Eder M.D., Mineola, NY: Dover. Fudenberg Drew, Levine David K. ( 2006), “A Dual-Self Model of Impulse Control,” American Economic Review , 96 5, 1449– 76. Google Scholar CrossRef Search ADS Gleim Mark R., Smith Jeffrey S., Andrews Demetra, Cronin J. JosephJr. ( 2013), “Against the Green: A Multi-Method Examination of the Barriers to Green Consumption,” Journal of Retailing , 89 1, 44– 61. Google Scholar CrossRef Search ADS Goldstein Noah, Cialdini Robert B., Griskevicius Vladas ( 2008), “A Room with a Viewpoint: Using Social Norms to Motivate Environmental Conservation in Hotels,” Journal of Consumer Research , 35 3, 472– 82. Google Scholar CrossRef Search ADS Graesser Arthur C., Gordon Sallie E., Sawyer John D. ( 1979), “Recognition Memory for Typical and Atypical Actions in Scripted Activities: Tests of a Script Pointer + Tag Hypothesis,” Journal of Verbal Learning and Verbal Behavior , 18 3, 319– 32. Google Scholar CrossRef Search ADS Graesser Arthur C., Woll Stanley B., Kowalski Daniel J., Smith Donald A. ( 1980), “Memory for Typical and Atypical Actions in Scripted Activities,” Journal of Experimental Psychology: Human Learning and Memory , 6 5, 503– 15. Google Scholar CrossRef Search ADS Griskevicius Vladas, Tybur Joshua M., Van den Bergh Bram ( 2010), “Going Green to Be Seen: Status, Reputation, and Conspicuous Conservation,” Journal of Personality and Social Psychology , 98 3, 392– 404. Google Scholar CrossRef Search ADS Haidt Jonathan ( 2007), “The New Synthesis in Moral Psychology,” Science , 316 5827, 998– 1002. Google Scholar CrossRef Search ADS Hastie Reid, Park Bernadette ( 1986), “The Relationship between Memory and Judgment Depends on Whether the Judgment Task Is Memory-Based or Online,” Psychological Review , 93 3, 258– 68. Google Scholar CrossRef Search ADS Haws Kelly L., Winterich Karen Page, Naylor Rebecca Walker ( 2014), “Seeing the World through GREEN-tinted Glasses: Green Consumption Values and Responses to Environmentally Friendly Products,” Journal of Consumer Psychology , 24 3, 336– 54. Google Scholar CrossRef Search ADS Hayes Andrew F. ( 2013), Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach , New York: Guilford. Hoch Stephen J., Loewenstein George F. ( 1991), “Time-Inconsistent Preferences and Consumer Self-Control,” Journal of Consumer Research , 17 4, 492– 507. Google Scholar CrossRef Search ADS Irwin Julie R., McClelland Gary H. ( 2001), “Misleading Heuristics and Moderated Multiple Regression Models,” Journal of Marketing Research , 38 ( February), 100– 9. Google Scholar CrossRef Search ADS Irwin Julie R., Naylor Rebecca Walker ( 2009), “Ethical Decisions and Response Mode Compatibility: Weighting of Ethical Attributes in Consideration Sets Formed by Excluding versus Including Product Alternatives,” Journal of Marketing Research , 46 2, 234– 46. Google Scholar CrossRef Search ADS Kahneman Daniel, Tversky Amos ( 1979), “Prospect Theory: An Analysis of Decision under Risk,” Econometrica , 47 2, 263– 91. Google Scholar CrossRef Search ADS Khan Uzma, Dhar Ravi ( 2006), “Licensing Effect in Consumer Choice,” Journal of Marketing Research , 43 2, 259– 66. Google Scholar CrossRef Search ADS Kimeldorf Howard, Meyer Rachel, Prasad Monica, Robinson Ian ( 2006), “Consumers with a Conscience: Will They Pay More?” Contexts , 5 1, 24– 9. Google Scholar CrossRef Search ADS Kohlberg Lawrence ( 1984), The Psychology of Moral Development: The Nature and Validity of Moral Stages , New York: Harper and Row. Kouchaki Maryam, Gino Francesca ( 2016), “Memories of Unethical Actions Become Obfuscated Over Time,” Proceedings of the National Academy of Sciences , 113 22, 6166– 171. Google Scholar CrossRef Search ADS Lichtenstein Sarah, Gregory Robin, Irwin Julie R. ( 2007), “What’s Bad Is Easy: Taboo Values, Affect, and Cognition,” Judgment and Decision Making , 2 3, 169– 88. Loewenstein George ( 1996), “Out of Control: Visceral Influences on Behavior,” Organizational Behavior and Human Decision Processes , 65 3, 272– 92. Google Scholar CrossRef Search ADS Loftus Elizabeth F. ( 1979), “ The Malleability of Human Memory”, American Scientist , 67 3, 312– 20. Loftus Geoffrey ( 1971), “Comparison of Recognition and Recall in a Continuous Memory Task,” Journal of Experimental Psychology , 91 2, 220– 6. Google Scholar CrossRef Search ADS Luce Mary Frances ( 1998), “Choosing to Avoid: Coping with Negatively Emotion-Laden Consumer Decisions,” Journal of Consumer Research , 24 4, 409– 33. Google Scholar CrossRef Search ADS Luce Mary Frances, Payne John W., Bettman James R. ( 1999), “Emotional Trade-off Difficulty and Choice,” Journal of Marketing Research , 36 2, 143– 59. Google Scholar CrossRef Search ADS Luchs Michael, Naylor Rebecca Walker, Irwin Julie R., Raghunathan Rajagopal ( 2010), “The Sustainability Liability: Potential Negative Effects of Ethicality on Product Preference,” Journal of Marketing , 74 5, 18– 31. Google Scholar CrossRef Search ADS Marcoux Jean-Sébastien ( 2017), “Souvenirs to Forget,” Journal of Consumer Research , 43 6, 950– 69. Martin Leonard L., Tesser Abraham ( 1996), “Some Ruminative Thoughts,” in Advances in Social Cognition , Vol. 9, ed. Wyer Robert S., Mahwah, NJ: Erlbaum, 1– 47. Massi Lisa L. ( 2005), “Anticipated Guilt as Behavioral Motivation,” Human Communication Research , 31 4, 453– 81. Google Scholar CrossRef Search ADS Mather Mara, Shafir Eldar, Johnson Marcia K. ( 2000), “Misremembrance of Options Past: Source Monitoring and Choice,” Psychological Science , 11 2, 132– 8. Google Scholar CrossRef Search ADS May Frank, Irmak Caglar ( 2014), “Licensing Indulgence in the Present by Distorting Memories of Past Behavior,” Journal of Consumer Research , 41 3, 624– 41. Google Scholar CrossRef Search ADS Milkman Katherine L. ( 2012), “Unsure What the Future Will Bring? You May Overindulge: Uncertainty Increases the Appeal of Wants Over Shoulds,” Organizational Behavior and Human Decision Processes , 119 2, 163– 76. Google Scholar CrossRef Search ADS Milkman Katherine L., Rogers Todd, Bazerman Max H. ( 2008), “Harnessing Our Inner Angels and Demons: What We Have Learned about Want/Should Conflicts and How That Knowledge Can Help Us Reduce Short-Sighted Decision Making,” Perspectives on Psychological Science , 3 4, 324– 38. Google Scholar CrossRef Search ADS Milkman Katherine L., Rogers Todd, Bazerman Max H. ( 2009), “Highbrow Films Gather Dust: Time-Inconsistent Preferences and Online DVD Rentals,” Management Science , 55 6, 1047– 59. Google Scholar CrossRef Search ADS Mitchell Terence R., Thompson Leigh, Peterson Erika, Cronk Randy ( 1997), “Temporal Adjustments in the Evaluation of Events: The Rosy View,” Journal of Experimental Social Psychology , 33 4, 421– 48. Google Scholar CrossRef Search ADS Nielsen ( 2013), “50% of Global Consumers Surveyed Willing to Pay More for Goods, Services From Socially Responsible Companies, Up From 2011,” http://www.nielsen.com/us/en/press-room/2013/nielsen-50-percent-of-global-consumers-surveyed-willing-to-pay-more-fo.html. Paharia Neeru, Vohs Kathleen D., Deshpandé Rohit ( 2013), “Sweatshop Labor Is Wrong Unless the Shoes Are Cute: Cognition Can Both Help and Hurt Moral Motivated Reasoning,” Organizational Behavior and Human Decision Processes , 121 1, 81– 8. Google Scholar CrossRef Search ADS Peloza John, White Katherine, Shang Jingzhi ( 2013), “Good and Guilt Free: The Role of Self-Accountability in Influencing Preferences for Products with Ethical Attributes,” Journal of Marketing , 77 1, 104– 19. Google Scholar CrossRef Search ADS Rainforest Relief ( 2017), “Avoiding Unsustainable Rainforest Wood,” http://www.rainforestrelief.org/What_to_Avoid_and_Alternatives/Rainforest_Wood.html. Rajagopal Priyali, Montgomery Nicole Votolato ( 2011), “I Imagine, I Experience, I Like: The False Experience Effect,” Journal of Consumer Research , 38 3, 578– 94. Google Scholar CrossRef Search ADS Reczek Rebecca Walker, Irwin Julie R. ( 2015), “Ethical Consumption,” in The Cambridge Handbook of Consumer Psychology , ed. Norton Michael, Rucker Derek, Lamberton Cait, New York: Cambridge University Press, 507– 29. Google Scholar CrossRef Search ADS Ritov Ilana, Baron Jonathan ( 1999), “Protected Values and Omission Bias,” Organizational Behavior and Human Decision Processes , 79 2, 79– 94. Google Scholar CrossRef Search ADS Rozin Paul, Royzman Edward B. ( 2001), “Negativity Bias, Negativity Dominance, and Contagion,” Personality and Social Psychology Review , 5 4, 296– 320. Google Scholar CrossRef Search ADS Ruedy Nicole E., Moore Celia, Gino Francesca, Schweitzer Maurice E. ( 2013), “The Cheater’s High: The Unexpected Affective Benefits of Unethical Behavior,” Journal of Personality and Social Psychology , 105 4, 531– 48. Google Scholar CrossRef Search ADS Russo J. Edward, Meloy Margaret G., Medvec Victoria Husted ( 1998), “Predecisional Distortion of Product Information,” Journal of Marketing Research , 35 4, 438– 52. http://dx.doi.org/10.2307/3857327 Google Scholar CrossRef Search ADS Sanitioso Rasyid, Kunda Ziva, Fong Geoffrey T. ( 1990), “Motivated Recruitment of Autobiographical Memories,” Journal of Personality and Social Psychology , 59 2, 229– 41. Google Scholar CrossRef Search ADS Schacter Daniel L. ( 1999), “The Seven Sins of Memory: Insights from Psychology and Cognitive Neuroscience,” American Psychologist , 54 3, 182– 203. Google Scholar CrossRef Search ADS Schank Roger C., Abelson Robert P. ( 1977), Scripts, Plans, Goals, and Understanding: An Inquiry into Human Knowledge Structures , Hillsdale, NJ: Erlbaum. Sedikides Constantine, Green Jeffrey D. ( 2000), “On the Self-Protective Nature of Inconsistency-Negativity Management: Using the Person Memory Paradigm to Examine Self-Referent Memory,” Journal of Personality and Social Psychology , 79 6, 906– 22. Google Scholar CrossRef Search ADS Sedikides Constantine, Strube Michael J. ( 1997), “Self-Evaluation: To Thine Own Self Be Good, to Thine Own Self Be Sure, to Thine Own Self Be True, and to Thine Own Self Be Better,” in Advances in Experimental Social Psychology , Vol. 29, ed. Zanna Mark P., New York: Academic Press, 209– 69. Shepard Roger N., Chang Jih-Jie ( 1967), “Recognition Memory for Words, Sentences and Pictures,” Journal of Verbal Learning and Verbal Behavior , 6 1, 156– 63. Google Scholar CrossRef Search ADS Shu Lisa L., Gino Francesca ( 2012), “Sweeping Dishonesty under the Rug: How Unethical Actions Lead to Moral Forgetting,” Journal of Personality and Social Psychology , 102 6, 1164– 77. Google Scholar CrossRef Search ADS Singh Surendra N., Rothschild Michael L. ( 1983), “Recognition as a Measure of Learning from Television Commercials,” Journal of Marketing Research , 20 3, 235– 48. Google Scholar CrossRef Search ADS Srull Thomas K. ( 1981), “Person Memory: Some Tests of Associative Storage and Retrieval Models,” Journal of Experimental Psychology: Human Learning and Memory , 7 6, 440– 63. Google Scholar CrossRef Search ADS Taylor Shelley E., Brown Jonathon D. ( 1988), “Illusion and Well-being: A Social Psychological Perspective on Mental Health,” Psychological Bulletin , 103 2, 193– 210. Google Scholar CrossRef Search ADS Taylor Shelley E., Brown Jonathon D. ( 1994), “Positive Illusions and Well-being Revisited: Separating Fact from Fiction,” Psychological Bulletin , 116 1, 21– 7. Google Scholar CrossRef Search ADS Tenbrunsel Ann, Diekman Kristina A., Wade-Benzoni Kimberly A., Bazerman Max H. ( 2010), “The Ethical Mirage: A Temporal Explanation as to Why We Aren’t as Ethical as We Think We Are,” Research in Organizational Behavior , 30, 153– 73. Google Scholar CrossRef Search ADS Thaler Richard H., Shefrin Hersh M. ( 1981), “An Economic Theory of Self-Control,” Journal of Political Economy , 89 2, 392– 406. Google Scholar CrossRef Search ADS Trudel Remi, Cotte June ( 2009), “Does It Pay to Be Good?” MIT Sloan Management Review , 50 2, 61– 8. Underwood Benton J., Zimmerman Joel, Freund Joel S. ( 1971), “Retention of Frequency Information with Observations on Recognition and Recall,” Journal of Experimental Psychology , 87 2, 149– 62. Google Scholar CrossRef Search ADS Verplanken Bas, Holland Rob W. ( 2002), “Motivated Decision Making: Effects of Activation and Self-Centrality of Values on Choices and Behavior,” Journal of Personality and Social Psychology , 82 3, 434– 47. Google Scholar CrossRef Search ADS Wegner Daniel M. ( 1994), “Ironic Processes of Mental Control,” Psychological Review , 101 1, 34– 52. Google Scholar CrossRef Search ADS Wood Stacy L., Lynch John G.Jr. ( 2002), “Prior Knowledge and Complacency in New Product Learning,” Journal of Consumer Research , 29 3, 416– 26. Google Scholar CrossRef Search ADS Zane Daniel M., Irwin Julie R., Reczek Rebecca Walker ( 2016), “Do Less Ethical Consumers Denigrate More Ethical Consumers? The Effect of Willful Ignorance on Judgments of Others,” Journal of Consumer Psychology , 26 3, 337– 49. Google Scholar CrossRef Search ADS Zauberman Gal, Ratner Rebecca K., Kim B. Kyu ( 2008), “Memories as Assets: Strategic Memory Protection in Choice over Time,” Journal of Consumer Research , 35 5, 715– 28. Google Scholar CrossRef Search ADS © The Author 2017. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Journal of Consumer Research – Oxford University Press
Published: Dec 8, 2017
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera