Exploring Consumers' Interpretations of a Product Related Illness

Zeynep Gnrhan, New York University
Elizabeth H. Creyer, University of Iowa
ABSTRACT - Two studies explored the role of counterfactual processing on consumers' assignments of blame when evaluating a product related illness. The first study underscored the importance of counterfactual processing in consumers' willingness to award money to the injured party. We found that the willingness to award money is partly determined by the blame assigned to the manufacturer, the manufacturer's market share, event foreseeability, and the number of counterfactual alternatives generated related to the manufacturer and to the store where the product was purchased. In the second study, we determined that whether or not the product was being promoted also influenced the assignment of blame.
[ to cite ]:
Zeynep Gnrhan and Elizabeth H. Creyer (1995) ,"Exploring Consumers' Interpretations of a Product Related Illness", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 526-531.

Advances in Consumer Research Volume 22, 1995      Pages 526-531

EXPLORING CONSUMERS' INTERPRETATIONS OF A PRODUCT RELATED ILLNESS

Zeynep Gnrhan, New York University

Elizabeth H. Creyer, University of Iowa

ABSTRACT -

Two studies explored the role of counterfactual processing on consumers' assignments of blame when evaluating a product related illness. The first study underscored the importance of counterfactual processing in consumers' willingness to award money to the injured party. We found that the willingness to award money is partly determined by the blame assigned to the manufacturer, the manufacturer's market share, event foreseeability, and the number of counterfactual alternatives generated related to the manufacturer and to the store where the product was purchased. In the second study, we determined that whether or not the product was being promoted also influenced the assignment of blame.

After an event or incident has occurred, people often evaluate its outcome by constructing alternative outcomes (Kahneman and Miller 1986). Consider the following scenario about a hypothetical consumer.

Mrs. Watson had a severe headache when she was working at the office. Because an important meeting was scheduled for the next day, she was not able to go home. She went to the nearest store but she could not find her regular brand of pain reliever. Since she did not have time to visit another store, she decided to buy a new brand that was currently being promoted. Unfortunately, after taking the medicine, she became ill and was rushed to the hospital. Doctors determined that she had an allergic reaction to one of the ingredients.

After reading such a story it is possible to imagine a number of "what ifs..." "If Mrs. Watson had found her regular brand of pain reliever..." "If the new brand was not on promotion..." "If she had time to look for another store..." Thus, our interpretation of the event is guided by not only what happened, but also by what might have happened. That is, behaviors, thoughts, and feelings are determined by both reality and imagined alternatives to reality, which are known as counterfactuals (Wells and Gavanski 1989). Counterfactual processing has been identified as playing a possible role in the assessment of causality (Wells and Gavanski 1989), victim compensation (Miller and McFarland 1986), attribution of responsibility (Miller and Gunasegaram 1990), and feelings of happiness and regret (Creyer and Gnrhan 1994; Kahneman and Tversky 1982).

Therefore a better understanding of counterfactual processes should improve our understanding of consumer behavior. Specifically, the purpose of this research is to explore the role counterfactual processing plays in consumers' assignments of blame when evaluating a product related illness. Blame assignment for product related injuries and illnesses seems to be an important concern to marketers, consumers, and policy makers (Griffin et al. 1992; McGill 1990) and offers an interesting context in which to analyze counterfactual processes.

THEORETICAL BACKGROUND

How people create imagined alternatives (that is, counterfactuals) has long been a topic of interest among psychologists and philosophers alike (Johnson 1986; Kahneman and Varey 1990). Extensive prior research has demonstrated that the extent to which an event is judged to play a causal role in an outcome depends on the extent to which a change, or mutation, to that event would undo the outcome (Kahneman and Tversky 1982; Wells and Gavanski 1989). As Wells and Gavanski point out, events are neither mutable nor immutable, but instead vary in terms of their mutability. Not surprisingly, substantial research has explored those factors influencing the mutability of events.

Kahneman and Miller (1986) hypothesize that easy to visualize alternatives are more mutable than alternatives that are more difficult to imagine. For example, exceptional events are clearly more mutable than unexceptional events (Kahneman and Miller 1986; Kahneman and Tversky 1982). Kahneman and Tversky (1982) asked people to consider the fate of a passenger killed in a crash of a commercial airliner; one of the passengers of the ill-fated flight switched planes just moments before the flight took off. The results indicate that people perceive the fate of this passenger to be more tragic than the fate of other passengers who were booked on the flight weeks in advance.

Theorists (Kahneman and Miller 1986; Kahneman and Tversky 1982; Miller, Turnbull, and McFarland 1990; Wells, Taylor, and Turtle 1987) suggest that this is because the counterfactual alternative of "not dying in a plane crash" is somehow closer to reality in the case of the passenger who switched flights than in the case of passengers who were booked weeks in advance. The mutability of an event also seems to be influenced by the number of alternative behaviors which are available to the main, or focal, actor in the event. Wells and Gavanski (1989) demonstrated that the greater the number of options which were available to the actor, the more likely observers are to imagine an alternative to the outcome. Counterfactual processing has also been found to influence the victim who had experienced a more common fate. Observers offered greater compensation for the same injury which occurred in the exceptional context. Greater compensation for the victim injured in the exceptional case may imply that the victim injured in the exceptional case is blamed less for his injuries than the victim in the common case.

We draw upon prior research to explore how consumers' interpretations of a product related illness are influenced by counterfactual processes. Two studies in which a short vignette describing the circumstances surrounding the illness of a hypothetical consumer were presented to subjects. The first study examined whether the extent to which the consumer, the manufacturer, and the store are blamed for the consumer's illness is influenced by time pressure, the consumer's level of experience with other brands, the product's market share, the number of counterfactual alternatives generated, and the foreseeability of the event. In the second study, we explored whether assignments of blame were also influenced by whether or not the product causing the illness was on promotion.

STUDY 1

Prior research has found that as the exceptionality of the event increases, people are more likely to engage in counterfactual processing (Kahneman and Miller 1986; Wells and Gavanski 1989). The amount of counterfactual processing that occurs, in turn, influences observers' reactions towards the victim; subjects offered greater compensation for a victim injured in an exceptional circumstance compared to a victim injured in an unexceptional circumstance (Miller and McFarland 1989).

We therefore suggest that when consumers evaluate a product related illness, their assignments of blame and their willingness to award money to the injured party will be influenced by the exceptionality of the circumstance. Usually there are a number of factors involved in a product related illness. One factor likely to influence the exceptionality of the circumstance surrounding a product related illness is the market share of the product, which in this case, is an over-the-counter pain reliever. It is proposed that when a high-share brand fails, it is considered to be more exceptional as compared to the case when a low-share brand fails. This is because high-share brands presumably enjoy higher reputations with respect to their quality. Hence, subjects are more likely to perceive a high-share brand's failure as an exceptional event. Therefore, we hypothesize that they will be more likely to blame the manufacturer and award money to Mrs. Watson when she buys a high-share brand.

Another factor likely to influence the exceptionality of the event is whether or not the victim had previous experience with other brands of pain reliever. If the victim had only used one brand of pain reliever, then becoming ill after trying a new brand is not likely to be perceived as being particularly surprising C the victim may have had similar reactions to the other brands. However, if the victim had used different brands of pain reliever in the past, then becoming ill after using the new brand would seem to be more unlikely, and thus, more exceptional.

Furthermore, we suggest that whether or not the choice of pain reliever was made under time pressure will influence subjects' perceptions of the exceptionality of the circumstance. When the consumer only had a short time to purchase a pain reliever, the choices available were limited. However, when a purchase was not made under time pressure, then the choices available to the consumer were not as limited (e.g., the consumer could have visited another store). Consequently, subjects are expected to generate a greater number of alternative courses of action that could have been taken by the consumer when there was a lack of time pressure. This in turn, implies that subjects will be more likely to place greater blame on the manufacturer, and less blame on the victim, when the purchase was made under time pressure.

Previous research also suggests that the order of the mutation has an impact on causal attributions (Wells and Gavanski 1989). Wells and Gavanski define a mutation as "a deletion, substitution, or other distortion of an event" (p. 161). In their experiments, subjects were asked to list four things that could have been different in the story to prevent the freak events. They found that the influence of counterfactual processing was greater when subjects engaged in the mutation task prior to causal attributions; they awarded more money to the focal individual as a result of his/her injuries. Therefore, we also expect that subjects will blame Mrs. Watson less when they perform the mutation task first.

Event foreseeability is another variable which is likely to influence the blame assignment and willingness to award money (Miller et al. 1990). Foreseeable fates are more easily imagined than unforeseeable fates. Subjects should be less likely to award money to Mrs. Watson when they believe that she could have easily foreseen this event. The above discussion leads to the following hypothesis.

H1: Subjects are more likely to blame the manufacturer and award money to Mrs. Watson when the product's market share is high, when she has previous experience with the other brands, when she is under time pressure, when subjects perform the mutation task prior to the blame assignment and when they think that Mrs. Watson could not have foreseen this freak event.

Previous research also indicates that mutability mediates perceived causality (Wells and Gavanski 1989). Thus, it may be possible to argue that as the number of counterfactuals related to the accused party (manufacturer) increases, perceivers are more likely to assign a causal role to the accused party. Therefore, the greater the number of counterfactuals generated related to the manufacturer, the greater the blame assigned to the manufacturer.

H2: Subjects are more likely to blame the manufacturer and award money to Mrs. Watson when the number of counterfactuals related to the manufacturer is high.

Whereas prior research (Wells and Gavanski 1989) has suggested that the greater the mutability of an event, the greater its causal role, it has not examined whether the causal role of one contributing factor increases as the causal role of another factor decreases. We examine whether the blame assigned to the manufacturer varies with the blame assigned to Mrs. Watson, to the store, and to fate. It is intuitive to expect that the blame assigned to the manufacturer is inversely related to the blame assigned to the focal actor, Mrs. Watson. However, it is more interesting to investigate whether the blame assigned to the manufacturer will vary with the blame assigned to fate and to the store. Kahneman and Miller (1986) have suggested that options which are easier to visualize are more mutable than options which are harder to imagine. This may indicate that if people are more likely to assign blame to the tangible features of an event (e.g., manufacturer), they may be less likely to assign blame to intangible features such as fate or bad luck. Thus, we expect an inverse relationship between the blame assigned to the manufacturer and to fate. We expect to observe a positive relationship between the blame assigned to the manufacturer and to the store because as less blame is assigned to the victim, more blame should be assigned to the other, tangible aspects of the event.

H3: Blame assigned to the manufacturer will be negatively related to the blame assigned to Mrs. Watson and fate and positively related to the blame assigned to the store.

Method

Subjects. One hundred thirty-nine undergraduates enrolled in introductory marketing and economics classes at two large northeastern universities served as subjects. In the vignette, we did not specifically mention that the ingredients were listed on the package; a pretest indicated that the students were generally aware that manufacturers are required to list the ingredients on the package. However, an analysis of the subjects' written responses revealed that 11 subjects incorrectly assumed that the ingredients were not listed on the package; data from these subjects were therefore not included in the analysis. Additionally, data from five other subjects were deleted because they did not complete the tasks.

Stimuli and Design. This study is a 2 (high or low market share) X 2 (absence or presence of previous experience) X 2 (absence or presence of time pressure) X 2 (order of the mutation task: before or after the blame assignment) between subjects factorial design. Subjects were presented a short vignette describing a product related illness. A hypothetical consumer identified as Mrs. Watson, was at work when she suddenly got a headache. Unfortunately, she was not able to go home to rest because she needed to prepare for an important meeting the next day. She went to a store to buy a pain reliever and decided to try a new brand, called Bomex, which was on promotion. Shortly after taking Bomex, she became very ill and was rushed to the hospital where doctors determined that she was allergic to one of the ingredients. Mrs. Watson is now asking for monetary compensation from the drug manufacturer. Subjects' tasks were to assign blame to the different parties involved in this situation, generate counterfactual thoughts identifying how the illness might not have occurred, and determine whether Mrs. Watson should receive money from the drug manufacturer.

Several different factors in the situation were varied. Market share of the hypothetical drug was either high or low. Information presented about the drug stated the following: "According to the results of a recent market survey, 70% (30%) of consumers use Bomex when they have a headache." Time pressure was also manipulated. In the high time pressure condition, subjects were told that the first store Mrs. Watson visited did not have her regular brand of pain reliever; since she didn't have time to go to another store she purchased Bomex, a brand heavily promoted. In the low time pressure condition, Mrs. Watson had the time to look for her regular brand in another store but still chose to buy Bomex, which was on promotion. The prior experience of Mrs. Watson also varied. In the high experience condition, subjects were informed that Mrs. Watson had tried other brands of pain reliever in the past. In the low experience condition, subjects were told that Mrs. Watson had only used her favorite brand of pain reliever.

Finally, the order of the mutation task differentiated the conditions. Half the subjects read the vignette and then listed ways that the outcome could have been avoided. The other subjects completed the other tasks first.

Procedure

Subjects, randomly assigned to one of the conditions, were run in groups of approximately 25 during a normal class period. At the beginning of class booklets containing the vignette and dependent measures were handed out to subjects. Subjects completed the task at their own pace. After completing the tasks, which took approximately 10 to 15 minutes, subjects were thanked and debriefed.

Dependent Measures

Blame assigned to the manufacturer was assessed by two different measures. The first measure asked whether they would be likely to award Mrs. Watson monetary compensation (yes or no) and its amount. Blame assigned to the manufacturer was also measured on two 7 point items ranging from 1 (very little) to 7 (a great deal). The first item asked subjects to indicate "To what extent is the drug manufacturer to blame for this situation?" The second inquired "To what extent can Mrs. Watson's sickness be attributed to the drug manufacturer?" The higher the value, the greater the blame assigned to the manufacturer. These items were highly correlated (r=.64) and averaged to form an overall index for blame assigned to the drug manufacturer. Blame assigned to fate, the store, and to Mrs. Watson were each measured by an item which assessed the blame assignment on a 7 point scale anchored by "very little" and "a great deal".

Additionally, subjects indicated their agreement with two items which stated "Mrs. Watson could not have done anything to avoid this event" and "Mrs. Watson had no way of knowing that she could have an allergic reaction." These items were averaged (r=.56) to form an index for event foreseeability; lower values indicate that Mrs. Watson could not have foreseen this freak event.

Results and Discussion

First, we examined the effect of counterfactual processing and experimental manipulations on the willingness to award money to Mrs. Watson. Next, we explored how the blame assigned to the manufacturer varied with the blame assigned to the other factors.

Subjects' counterfactual thoughts were coded into four categories: counterfactuals related to Mrs. Watson (e.g., "Mrs. Watson could have gone to another store"; "Mrs. Watson could have asked her doctor"), counterfactuals related to the drug manufacturer (e.g., "If the manufacturer did not promote the product..."; "If the manufacturer used another ingredient..."), counterfactuals related to the store (e.g., "If Mrs. Watson's regular brand was in stock..."; "If the store was closed..."), and counterfactuals related to fate (e.g., "If Mrs. Watson never got a headache..."; "If there was no meeting the next day...").

A logistic regression was conducted to explore the impact of counterfactual processing and experimental manipulations on the willingness to award money to Mrs. Watson. The willingness to award money (binary outcome) served as the dependent variable and the experimental manipulations, assignments of blame, event foreseeability, and the number of counterfactuals generated served as independent variables. The four experimental manipulations have been recorded as dummy variables.

The chi square value is 119.44, the value of -2 Log Likelihood for this model. The observed significance level (p=.18) indicates that this model does not differ significantly from the perfect model. The goodness of fit statistic (chi-square=107.44, p=.43) also leads to a similar conclusion. The logistic regression coefficients and associated statistics are presented in Table 1.

As Table 1 reveals, market share of the manufacturer, blame assigned to the manufacturer, event foreseeability, and the number of manufacturer and store related counterfactual thoughts significantly influenced the willingness to award money. Specifically, subjects were more likely to award money to Mrs. Watson when (1) the manufacturer's market share was high, (2) greater blame was assigned to the manufacturer, (3) they believed that Mrs. Watson could not have foreseen this freak event, (4) a high number of counterfactual alternatives involving the manufacturer were generated, and (5) a high number of counterfactual alternatives involving the store were generated.

The finding that subjects were more likely to award money when they believed the consumer could not have foreseen the event and when the manufacturer was assigned a large portion of the blame for the illness are intuitive. More interesting are the findings regarding the market share and the number of counterfactuals generated related to both the manufacturer and the store. The manufacturer of the low-share brand was blamed less for the consumer's illness compared to the manufacturer of the high-share brand. This may be explained by the fact that an illness as a result of a product failure is considered to be more exceptional when the product is used by many consumers and enjoys a higher-share. Therefore, subjects were more sympathetic towards Mrs. Watson when she met with an exceptional fate (injury as a result of using a high-share brand), as compared to the situation when she used the low-share brand (Miller and McFarland 1986). These findings give partial support to the first and second hypotheses. Our expectations regarding the impact of market share, event foreseeability, and the number of manufacturer related counterfactuals were confirmed. Although it was not hypothesized, the results also suggest that the number of store related counterfactuals has an impact on subjects' willingness to award money. However, our data did not provide enough evidence to suggest that the order of the mutation task, time pressure, and previous experience with the product category influence the subjects' willingness to award money.

A regression analysis which included the blame assigned to the manufacturer as the dependent variable was used to analyze how the blame assigned to the manufacturer varies with the blame assigned to the other factors. Dummy coded experimental manipulations and the blame assigned to Mrs. Watson, to fate, and to the store served as independent variables. The overall model was significant (F(7,114)=5.45, p<.001) and accounted for 25% of the variation. The analysis suggests that the blame assigned to the manufacturer is significantly related to the blame assigned to Mrs. Watson (beta=-.41, t=-4.91, p<.001), the blame assigned to the store (beta=.24, t=2.88, p<.005), and the blame assigned to fate (beta=-.26, t=-3.21, p<.001).

TABLE 1

LOGISTIC REGRESSION RESULTS FOR THE WILLINGNESS TO AWARD MONEY

The results of the first study underscored the importance of counterfactual processing and market share of the manufacturer in subjects' willingness to award money to the victim. Additionally, we showed that the blame assigned to the manufacturer was positively related to the blame assigned to the store and negatively related to the blame assigned to Mrs. Watson and fate. Furthermore, an informal analysis of the mutation task revealed that some of the subjects generated counterfactuals which referred to the fact that the brand was on promotion. However, since the presence of promotion was not manipulated in the first study, we could not directly test its impact. We address this issue in the second experiment.

STUDY 2

Hypotheses

In the second study, we explored the impact of market share and promotion on the assignment of blame within the same context of product related illness. Based on the previous research and the results of the first study, we expect to replicate the findings regarding the impact of market share on the assignment of blame to the manufacturer. It has been suggested that when the market share is high, subjects are more likely to blame the manufacturer because they have more sympathy towards the victim in an exceptional circumstance. Additionally, we propose that buying the product on promotion increases the perceived mutability of the outcome. Thus, subjects are more likely to engage in counterfactual processing when the product is bought on promotion. Furthermore, when the product is bought on a promotion, subjects will be more likely to attribute the purchase to the promotion as opposed to internal factors (Scott 1976). Therefore, they will blame Mrs. Watson less when the product is on promotion. We again propose that blame assignments will be affected by the order of the mutation task. Greater blame will be assigned to the manufacturer when subjects perform the mutation task prior to the blame assignment.

H4a: Blame assigned to the manufacturer (Mrs. Watson) will increase when the market share of the product is high (low).

H4b: Blame assigned to the manufacturer (Mrs. Watson) will increase (decrease) when the product is bought on a promotion.

H4c: Blame assigned to the manufacturer (Mrs. Watson) will increase (decrease) when the subjects perform the mutation task prior to the blame assignment.

Method

Subjects. One-hundred two undergraduates in a large northeastern university volunteered to participate in the study.

Stimuli and Design. The same vignette used in study 1 was again used in study 2 with the following exceptions. The levels of market share used to define the manufacturer differed slightly and whether the product was or was not promoted was varied. In the high market share condition, subjects were told the following: "According to the results of a recent market survey, Bomex is the best selling brand of all pain relievers and more than half of the consumers use Bomex when they have a headache." In the low market share condition, subjects were presented the following sentence: "The results of a recent market survey revealed that Bomex has a low market share. Only 10% of the consumers use Bomex when they have a headache." To manipulate promotion, half of the subjects were told that she saw Bomex on promotion and its price was reduced by 25%. The other half were not given any information about the promotion or the price.

Subjects filled out a questionnaire containing the market share information, the hypothetical scenario, and the dependent measures which took approximately 10 minutes.

Dependent Measures

Blame assigned to the manufacturer was measured on two 7 point items ranging from 1 (very little) to 7 (a great deal). The higher the value, the greater the blame assigned to the manufacturer. As these items were highly correlated (r=.71) they were averaged to form an overall index for blame assigned to the drug manufacturer. Blame assigned to fate, to the store, and to Mrs. Watson were each measured by an item which assessed the blame assignment on a 7 point scale anchored by "very little" and "a great deal."

TABLE 2

BLAME ASSIGNED TO MRS. WATSON AS A FUNCTION OF MARKET SHARE, PROMOTION, AND ORDER OF THE MUTATION TASK

Results and Discussion

The data were analyzed as a 2 (high or low market share) X 2 (presence or absence of promotion) x 2 (order of the mutation task: before or after the blame assignment) between-subjects factorial design.

An ANOVA model which included the blame assigned to the manufacturer was significant (F(7,74)=2.53, p<.02). As expected, greater blame was reported when the product was on promotion (F(1,94)=2.79, p<.05 (one-tailed), Ms=2.86 vs. 3.37). Moreover, subjects assigned greater blame when they performed the mutation task prior to the blame assignments (F(1,94)=8.84, p<.004, Ms=3.61 vs. 2.63). However, the effect of the market share was insignificant (p=.28).

Another model which included the blame assigned to Mrs. Watson was marginally significant (F(7,93)=1.98, p<.07). As expected, blame assigned to Mrs. Watson was greater when the market share is low (F(1,93)=3.93, p<.05, Ms=4.22 vs. 3.67). Additionally, the results suggest a significant two-way interaction between promotion and the order of the mutation task (F(1,93)=3.37, p<.07) and a three-way interaction among market share, promotion, and the order of the mutation task (F(1,93)=4.16, p<.04). The means of the model are presented in Table 2.

In order to explore the three-way interaction, we analyzed the data separately for the promotion absent and promotion present conditions. In the no promotion condition, Mrs. Watson was blamed more when she bought the low-share brand and when the mutation task was performed before the blame assignment as compared to when the mutation task was performed after the blame assignment. A series of planned contrasts revealed a significant difference only for these two cells (t=2.19, df=93, p<.03). In this case, blame assigned to Mrs. Watson is significantly correlated with the number of counterfactual thoughts related to Mrs. Watson (r=.70) and fate (r=-.58). However, when Mrs. Watson bought Bomex on promotion, we observed a reverse pattern. She was blamed more when she bought the low-share brand on promotion and when the mutation task was performed after the blame assignment. In this case, the order of the mutation task influenced the blame assigned to Mrs. Watson in the expected way; subjects blamed Mrs. Watson less when they performed the mutation task prior to the blame assignment.

An ANOVA model which included the blame assigned to the store was also significant (F(7,94)=2.41, p<.03). There are three significant main effects. The store is blamed more for selling a low-share brand as opposed to a high-share brand (F(1,94)=4.96, p<.03, Ms=1.75 vs. 1.28). Additionally, subjects blamed the store more when the product was on promotion (F(1,94)=3.05, p<.05, Ms=1.34 vs. 1.69) and when the mutation task was performed before the blame assignment (F(1,94)=3.11, p<.05 (one-tailed), Ms=1.69 vs. 1.34).

GENERAL DISCUSSION

These two experiments suggest that counterfactual processing can influence how blame is assigned when evaluating a product related illness or injury. The first study underscored the importance of counterfactual processing in consumers' willingness to award money to the injured party. We found that willingness to award money is partly determined by the blame assigned to the manufacturer, the manufacturer's market share, event foreseeability, and the number of counterfactuals generated related to the manufacturer and to the store. Subjects are more likely to award money to the injured party when the manufacturer's market share is high, when greater blame is assigned to the manufacturer, when they think that Mrs. Watson could not have foreseen this freak event, and when the number of counterfactual alternatives related to the manufacturer and to the store is higher. Additionally, we showed that blame assigned to the manufacturer is negatively related to the blame assigned to Mrs. Watson and to "fate" and positively related to the blame assigned to the store.

In the second study, we determined that in addition to market share and counterfactual processing, whether or not the product was being promoted also influenced the assignment of blame. The manufacturer was blamed more when the product was on promotion.

Our data did not support the hypotheses about the impact of previous experience with the product category and time pressure on counterfactual processing and assignment of blame. One explanation for these insignificant results might be that our experimental manipulations were not strong enough to reveal the hypothesized effects, therefore, a larger sample size might be required to observe any effect. Still another explanation might be that previous experience and time pressure do not influence the exceptionality of the situation and do not lead to counterfactual processing. We tend to favor the former explanation because there are theoretical reasons to expect these hypothesized effects. An important task for future studies can be exploring the robustness of these effects.

The present research suffers from the usual shortcomings of laboratory research. However, these shortcomings provide us with an opportunity for future research. More realistic stimuli, such as videotapes of actual product related accidents or injuries and more detailed descriptions of the accidents could improve the external validity of the research. Additionally, we only collected outcome measures which made it neccessary for us to conjecture how people assessed the product related illness. Future studies should employ cognitive process measures to inquire mediating variables that might affect blame assignments.

The current findings suggest that counterfactual processing seem to influence consumers' blame assignments in a product failure. It is likely that counterfactual processing might play a significant role in explaining consumers' reactions to various other stimuli in different contexts. For example, a consumer's attitude toward a product might be more favorable if the product performs very well in a situation where the counterfactual alternative related to its malfunctioning is highly available. Future research should also explore the impact of counterfactual processing in these different contexts.

REFERENCES

Creyer, Elizabeth H. and Zeynep Gnrhan (1994), "Who's to Blame?: Counterfactual Reasoning and the Development of Regret," unpublished manuscript.

Griffin, Mitch, Barry J. Babin, and William R. Darden (1992), "Consumer Assessments of Responsibility for Product Related Injuries: The Impact of Regulations, Warnings, and Promotional Policies," in Advances in Consumer Research, John F. Sherry Jr. and Brian Sternthal (eds.), 19, 870-877.

Johnson, Joel T. (1986), "The Knowledge of What Might Have Been: Affective and Attributional Consequences of Near Outcomes," Personality and Social Psychology Bulletin, 12 (1), 551-562.

Kahneman, Daniel and Dale T. Miller (1986), "Norm Theory: Comparing Reality to its Alternatives," Psychological Review, 93 (2), 136-153.

Kahneman, Daniel and Amos Tversky (1982), "The Simulation Heuristic," in Judgment under Uncertainty: Heuristics and Biases, D. Kahneman, P. Slovic, and A. Tversky (eds.), New York: Cambridge University Press, 201-208.

Kahneman, Daniel and Carol A. Varey (1990), "Propensities and Counterfactuals: The Loser That Almost Won," Journal of Personality and Social Psychology, 59 (6), 1101-1110.

McGill, Ann L. (1990), "Predicting Consumers' Reactions to Product Failure: Do Responsibility Judgments Follow from Consumers' Causal Explanations?" Marketing Letters, 2 (1), 59-70.

Miller, Dale T. and Saku Gunesagaram (1990), "Temporal Order and Perceived Mutability of Events: Implications for Blame Assignment," Journal of Personality and Social Psychology, 59 (6), 1111-1118.

Miller, Dale T. and Cathy McFarland (1986), "Counterfactual Thinking and Victim Compensation: A Test of Norm Theory," Personality and Social Psychology Bulletin, 12 (4), 513-519.

Miller, Dale T., William Turnbull, and Cathy McFarland (1990), "Counterfactual Thinking and Social Perception: Thinking About What Might Have Been," in Advances in Experimental Social Psychology, ed. Mark P. Zanna, San Diego, CA: Academic Press, 305-331.

Scott, Carol A. (1976), "The Effects of Trial and Incentives on Repeat Purchase Behavior," Journal of Marketing Research, 13 (August), 263-269.

Wells, Gary L., Brian R. Taylor, and John W. Turtle (1987), "The Undoing of Scenarios," Journal of Personality and Social Psychology, 53 (3), 421-430.

Wells, Gary L. and Igor Gavanski (1989), "Mental Simulation of Causality," Journal of Personality and Social Psychology, 56, 161-169.

----------------------------------------