Consumer Causal Reasoning on Product Failure

ABSTRACT - This study examines three factors that are believed to affect the level of consumer attribution; i.e., control, consistency, and consensus information. Subjects completed questionnaires describing a product failure with information about control, consistency and consensus. Each factor was manipulated into two levels, low or high, creating eight different scenario groups. Multiple comparisons of the means of each group were made. Findings indicate that regardless of other information, a high level of control invariably elicits external attribution. The external attribution was divided into five different levels; company, branch, employee, bad luck, and situation. The results provide partial evidence that with a high level of control and consensus, and both high and low levels of consistency, the consumer attributed the failure to the company. With a high level of control and consistency, and a low level of consensus, the consumer attributed the failure to the branch. A high level of control, and low levels of both consistency and consensus evoked attribution toward bad luck. Low control was hypothesized to cause internal attribution regardless of other information. The results did not thoroughly support this hypothesis, which is explained to be an outcome of self-serving bias. Marketing implications of the findings are discussed.



Citation:

Moonkyu Lee (2002) ,"Consumer Causal Reasoning on Product Failure", in AP - Asia Pacific Advances in Consumer Research Volume 5, eds. Ramizwick and Tu Ping, Valdosta, GA : Association for Consumer Research, Pages: 61-71.

Asia Pacific Advances in Consumer Research Volume 5, 2002      Pages 61-71

CONSUMER CAUSAL REASONING ON PRODUCT FAILURE

Moonkyu Lee, Yonsei University, Korea

Moonhee Cha, Yonsei University, Korea

ABSTRACT -

This study examines three factors that are believed to affect the level of consumer attribution; i.e., control, consistency, and consensus information. Subjects completed questionnaires describing a product failure with information about control, consistency and consensus. Each factor was manipulated into two levels, low or high, creating eight different scenario groups. Multiple comparisons of the means of each group were made. Findings indicate that regardless of other information, a high level of control invariably elicits external attribution. The external attribution was divided into five different levels; company, branch, employee, bad luck, and situation. The results provide partial evidence that with a high level of control and consensus, and both high and low levels of consistency, the consumer attributed the failure to the company. With a high level of control and consistency, and a low level of consensus, the consumer attributed the failure to the branch. A high level of control, and low levels of both consistency and consensus evoked attribution toward bad luck. Low control was hypothesized to cause internal attribution regardless of other information. The results did not thoroughly support this hypothesis, which is explained to be an outcome of self-serving bias. Marketing implications of the findings are discussed.

INTRODUCTION

In the occurrence of an unexpected failure of a certain goal, people often search for a reason to explain the failure (Weiner 1985). Some will blame it on the situation, some will blame it on themselves, and some will blame it on a third person that might have been involved in the achievement of that goal. In any case, the blame will ultimately land on the actor himself/herself or on anything or anyone besides the actor. That is to say, the actor perceives the cause of the failure to be either internal to himself/herself or external. In that case, on what grounds does the actor decide where the cause of the failure lies? What are the factors that influence the actor to believe that the cause of failure is internal or external? Also, when the cause of failure is thought to be external, and what factors determine the level of external attribution?

Importance of Consumer Attribution

When a product performs below expectation, consumers search for reasons why. "Why is my cable TV fuzzy?", "Why isn’t my cell phone working?", "Why did the waiter serve the wrong dish?" The consumer seeks to find an answer to these questions, and the 'who’ or 'what’ that the consumer decides to blame is the object of attribution. Attribution, therefore, refers to the action of placing the blame on a certain entity for the result of a certain occurrence (see Folkes 1984, 1987, 1988; Harvey and Weary 1984; Hastie 1984; McGill 1989; Weiner 2000).

The development of attribution theory plays an important part in the study of consumer behavior due to the fact that attributions made by consumers affect, to a great extent, how the consumer will subsequently behave. For instance, if a consumer is dissatisfied with a product that he/she has purchased, and decides to blame the manufacturer, the consumer will consequently switch to another brand or even go as far as spreading the word about his/her experience, ultimately creating negative publicity for the manufacturer. On the other hand, if the consumer blames himself/herself, or attributes the dissatisfaction to a situational factor, he/she would not have negative feelings toward the company, and would stay loyal by continuing to buy the company’s products.

Consumer attribution involves a complex process of a consumer having certain expectations about a product or service, experiencing failure and dissatisfaction, and placing blame on a certain entity, eventually leading to some sort of behavior. It is the behavior of the consumer that most affects the firm, and it is this behavior that the marketer is ut to control. This study contributes to the marketer’s insight into consumer behavior by illustrating how certain information will determine the level to which the consumer might attribute a purchase failure.

Motivation of Attribution

Attribution is, therefore, the answer to the question "Why?" If the answer is "Because of A," then the cause has been attributed to 'A.’ But we don’t always ask ourselves why an incident occurred. Many studies about what elicits causal reasoning have been conducted. Enzle and Schopflocher (1978) demonstrated that subjects only performed attributional reasoning when requested to. Several studies support the conclusion that unexpected events instigate attributional processing (Pyszczynski and Greenberg 1981; Wong and Weiner 1981). Lau (1984) reported that attributional activity was elicited more by failure than by success. A summary of the instigators of causal reasoning follows.

Control Motivation. A prominent assumption in attribution work is that people spontaneously engage in attributional activities. But what factors make more probable explicit attribution-type inquiries on the part of the attributor? According to Berscheid, et al. (1976), the perceiver’s outcome dependency on a target person may be such an instigator. Pittman and Pittman (1980) found evidence that attributions are instigated by control motivation and that attributional activity increases following an experience with lack of control.

Expectancy Disconfirmation. In other research on instigators, as mentioned above, Wong and Weiner (1981) and Pyszczynski and Greenberg (1981) found that expectancy disconfirmation promotes attributional search. Also, when a person fails to achieve a certain goal, for instance, a person fails to be elected mayor, or a person fails an exam, the person will dwell on the incident and try to figure out what caused the failure.

Direct Attributional Questions. A person will engage in attributional activity when asked the question "why" about something that would normally go unchecked.

Consequences of Attribution

When a consumer experiences a product failure, the attribution will be made to something related to the product or the purchase situation. Depending on where the consumer perceives the cause to be, his/her future behavior will change extensively (Folkes 1984). The following is a summary of some of the behaviors resulting from attributional activity (also see Figure 1).

Complaining to the Firm. This sort of behavior is prevalent when the consumer attributes the failure to the company, and feels the need to complain. The act of complaining is a rather troublesome act that takes time and effort. Therefore, it is not done in every incident of product failure (Hirschman 1970). According to Hirschman, a consumer will complain when he/she feels that the complaint will be worth it, when there is a good chance of achieving something by complaining, and when the consumer has the ability or the will to complain. The complaint is considered "worth it" when the company makes amends for its failure. A consumer perceives his/her complaint to be successful when he/she receives some sort of compensation, perhaps a refund, an exchange or an apology for the failure.

FIGURE 1

A MODEL OF SERVICE EVALUATIONS (BITNER, 1990)

Seeking Revenge. Few consumers actually go this far, but when the failure of the product proves to be harmful, and the cause of failure is attributed to the company, some consumers go out of their way to seek revenge. Also, when a consumer complains about the failure of a product and receives no physical or psychological compensation, the intentions of the consumer to seek revenge grow stronger. Spreading negative information about the company and creating bad publicity is one popular method of revenge. Some consumers go as far as suing, and some stop at informing a consumer organization or a government agency. All methods have the intention of ruining the reputation of the firm, which can be quite harmful due to the difficulty of recovering.

Warning Friends Through Word-of-Mouth Communication. Telling people about a bad experience with a product is not as aggressive as seeking revenge or complaining to the firm, but it can have quite an influence. For an expensive service or product that requires a high level of involvement and depends on the opinions of others, the spreading of negative information could be detrimental to the company’s reputation. Negative word-of-mouth is known to be much more harmful than positive word-of-mouth is beneficial.

Refraining from Using the Same Product (or Brand Switching). When the attribution is placed on the company, the least offensive action that a consumer might take is to simply refrain from using the same product, and switching to a different brand. According to the 20/80 rule, 20% of the consumers are loyal customers, and this minority creates 80% of the company’s margin. Therefore, the simple act of switching brands can have a deep impact on the company.

When the attribution is directed toward the branch or the employee rather than the company, the consumer is expected to use a different branch or ask for a different employee rather than switching brands. In some cases, switching brands can be very wearisome for the consumer, especially when the product is a service, and the consumer has a deep involvement with the company, such as a bank.

Shrugging off and Staying Loyal. When the consumer attributes the failure of the product to a situational factor, to bad luck, or to himself/herself, then his/her future purchasing behavior might differ in the sense that he/she might take more precautions in using the product at the next purchase. This type of attribution does not induce any negative feelings toward the company; therefore, the consumer will most likely remain a loyal customer.

THEORETICAL BACKGROUND

Attribution theory views people as rational information processors whose actions are influenced by their causal inferences. It is not merely the judgment that the product has failed that determines the consumer’s response, but the perceived reasons for the failure that influence how a consumer responds. The perceived reasons are based on a number of things; the person’s prior beliefs, the context in which the failure occurred, information related to the failure, and underlying motivational reasons of the person. When the failure of a product occurs, people try to determine why the product failed, and the type of reason inferred will influence their subsequent actions. For instance, suppose a person purchased and assembled a vacuum cleaner, and after using it discovered that there were still pieces of hair and dirt on the floor. The person would search for reasons why the vacuuming did not work properly and arrive at any of several explanations; e.g., the person assembled the vacuum cleaner incorrectly, the manufacturer made a lousy product, the salesman gave him/her a defective product, or the person incidentally missed vacuuming the area where the hair and dirt was. If the consumer were to attribute the failure of the vacuum cleaner to the manufacturer, the consumer would consequently form a negative image of the manufacturer and refrain from buying products made by that manufacturer or even spread the word about how that company makes products of poor quality. Therefore, a consumer’s attribution about the failure of a product or service can have important consequences for the firm; should a consumer attribute the failure to a firm-related reason, the result could range anywhere from refusing to repurchase the product to attempting some kind of revenge.

The purpose of this research is to investigate the aspects of a sitation that influence a consumer’s attribution of product failure, and to examine the levels of the attribution based on the information about the situational factors. The potential determinants of the attribution examined in this study are control, consistency, and consensus. These three aspects are treated as tools to establish the causal background when the causal question is rather ambiguous. The basis for the selection of these particular factors lies in Kelley’s attributional cube (1967, 1972, 1973) and Weiner’s classification system (1980).

Kelley’s Attributional Cube

Kelley’s model (1967, 1972, 1973) provides insight into the type of information that people may use to establish the causal background. Kelley proposed that attributions are a function of the three informational factors that covary with behavior; consensus, consistency, and distinctiveness.

Consistency information refers to whether or not a person’s response is consistent to the same stimulus in past circumstances. Consistency information is longitudinal, within person, information. Consistency would be high if the same individual exhibits this behavior frequently within this situation. If the behavior is unusual for this person in this situation, then consistency would be low. Consensus information is concerned with whether or not the same behavior is exhibited by others in the same situation. If other people act in the same manner in this situation, this information would be classified as high consensus. If the behavior is unique, or rare for other people, then consensus would be low. Distinctiveness information compares the behaviors of the individual in other situations.

There are two major differences in the use of these concepts in this study. First, Kelley’s usage of the three types of information referred to a person’s response toward a given stimulus. However, in this study, the three types of information illustrate different aspects of the stimulus itself (in this case, the failure of the product), eliciting causal inferences about what caused the stimulus to occur. For instance, in this study, consistency refers to whether or not the same product failure occurred to the person in the past, and consensus refers to whether or not other people experience the same product failure. The second difference is that this study excludes distinctiveness information from consideration. As mentioned above, distinctiveness information refers to the response of the individual toward different stimuli. In the process of forming a causal background and ultimately attributing the failure to a certain entity, many studies have proven that a person will act or think upon prior beliefs, underlying motivations, and available information. However, there is no existing literature that supports the theory that in the occurrence of product failure a person will consider an unrelated stimulus in the process of attribution. Therefore, distinctiveness information was replaced with a concept from Weiner’s categorization system, control, which is explained below.

Weiner’s Categorization System

There are many possible explanations for the failure of a product. For instance, if a pen dries out and is unable to write, the consumer might have left the cap off, or the ink could have leaked after having been dropped in the retailing process, or the manufacturer could have made a defective product to begin with. By analyzing underlying properties of causes, Weiner developed a 2x2x2 classification system based on locus of causality, controllability, and temporal stability.

Locus of causality refers to whether the cause is located in the consumer or in the seller or manufacturer. An internal attribution indicates that the individual perceives the self as the cause of the outcome. An external attribution indicates that the individual attributes the outcome to external environmental characteristics. Attributions of ability and effort are typically considered to e internal. For this reason, the term 'control’ replaces 'locus of causality’ in this study. Control, or 'the level of control,’ refers to the degree to which the consumer put in effort to prevent the failure of the product. As will be discussed in the next chapter, this study hypothesizes that when the level of control is high, for instance, when the consumer has done everything possible on his/her part to make the product work and yet the product fails, then the consumer will attribute the failure to an external reason. However, when the consumer has not done all he/she could to prevent the failure, then the consumer will attribute the failure internally.

Stability refers to whether causes are perceived as relatively permanent and unchanging or as temporary and fluctuating. There is a relationship between Weiner’s concept of stability and Kelley’s concept of consistency. Consistency information leads to attributions regarding the stability of the cause (Martinko and Thomson 1998). A high level of consistency indicates repetitive product failure in past circumstances, in which case, the occurrence of product failure can be referred to as stable.

Controllability is the third concept that Weiner discussed. Causes can be volitional or nonvolitional. In other words, choice can be involved or constraints may force a product failure. This term is different from the concept 'control’ used throughout this study. Weiner’s 'controllability’ concept is not examined in this study because it was not viewed as a factor that would determine the 'level’ of attribution.

HYPOTHESES

Based on the existing literature, the three types of information, control, consistency, and consensus, were chosen for examination. Before stating the hypotheses to be tested in this study, the expected effects of the respective types of information will be discussed.

Level of Control

The level of control refers to the degree to which the consumer put effort into preventing the product failure. In every purchase situation, the consumer must meet certain requirements in order to retrieve the expected satisfaction. For instance, if a consumer were to purchase a VCR, he/she would have to know how to work it in order to watch a video or record a program. If he/she does not know how the functions of the machine work, then it is up to him/her to read the instructions to learn how to use it effectively. Likewise, the consumer has certain 'responsibilities’ in the purchase of a product to achieve satisfaction. A high level of control means that the consumer put great effort into making the product work, whereas a low level of control means that the consumer did not put much effort into making the product work. Based on these premises, it is hypothesized that the level of control will affect whether the failure of a product is attributed internally or externally. Therefore,

H1: The level of control will determine the internal or external locus of attribution.

H1-a: When the level of control is high, the failure of a product will be attributed externally.

H1-b: When the level of control is low, the failure of a product will be attributed internally.

Consistency and Consensus

Consistency information refers to the stability of the product failure. From a longitudinal poit of view, has this type of product failure occurred repeatedly? Has the consumer experienced the same kind of product failure in the past? When a certain product failure occurs to the same person several times, the person will seek to find out why the same failure keeps on occurring. However, this information in itself is not sufficient for a person to attribute the failure to any one entity.

FIGURE 2

RESEARCH FRAMEWORK

Consensus information informs the consumer about whether or not other people have experienced similar product failure. This information, combined with information about consistency and level of control is hypothesized to affect the level in which the consumer attributes the failure of the product. The 'level’ refers to the different possible entities of external attribution. Internal attribution means placing blame in oneself, which is a single entity. The reason can vary within the entity, but the entity is merely the consumer himself/herself. In external attribution, the different entities, or the different levels can refer to anyone or anything besides the consumer. In this study, the levels of external attribution were divided into company (or manufacturer), branch, employee, bad luck, and situational factors.

The following hypotheses rely on the previous hypothesis that a high level of control will lead to external attribution. The main idea is that attribution is a result of different types of information processed. In this case, a combination of information regarding control, consistency, and consensus are expected to affect the attribution. When the level of control is high, depending on the level of consistency and consensus, the final attribution will result in different levels of external attribution. More specifically,

H2: The level of control, consistency, and consensus will determine the consumer’s attribution toward the manufacturer.

H2-a: When the level of control is high, and information regarding the product failure indicates that both consistency and consensus are high, the consumer will attribute the failure to the manufacturer.

H2-b: When the level of control is high, and information regarding the product failure indicates that consistency is low, but consensus is high, the consumer will attribute the failure to the manufacturer.

Consequently, Hypothesis 2-a and 2-b test the interaction effects of control, consistency, and consensus on the consumer’s attribution toward the manufacturer. If the results show that both are supported, then it can be inferred that consistency information has little to do with attribution toward the manufacturer.

When the level of control is high, consistency is high, but the level of consensus is low, the consumer is expected to blame the branch or employee. To rephrase, when a consumer has done everything on his/her part and yet the product keeps failing, and the consumer finds out that a friend who purchased the same product never experienced such failure, then it is anticipated that the consumer will find fault with the branch that he/she goes to, or the employee that he/she deals with. Therefore,

H3: When the level of control is high, and information regarding the product failure indicates that consistency is high, but consensus low, the consumer willattribute the failure to the branch or employee.

On the other hand, when the consumer has done everything on his/her part, and the product has never failed until this occasion, and other people who have purchased the same product have not experienced any kind of failure, then the consumer is expected to blame this odd failure to bad luck.

H4: When the level of control is high, and information regarding the product failure indicates that both consistency and consensus are low, then the consumer will attribute the failure to bad luck or situational.

Figure 2 illustrates the hypothesized results of the different levels of information.

METHOD

The effects of the three types of information were tested by presenting subjects with a hypothetical product failure and including information on which the subjects were meant to base their attributions.

Procedure

Subjects were 226 undergraduate business students (231 questionnaires were collected out of 248 distributed, 5 cases were unfit for analysis and were therefore excluded). Each subject completed a questionnaire describing the same product failure, but with different types of information. A scenario was designed to vary in low or high levels of control, consistency, and consensus, creating eight different scenarios. The scenario depicted a hypothetical situation where the subject purchased a sports cream for sore muscles that required refrigeration. The product failure was described as a rotting odor that came from the cream. Control was manipulated by stating that the subject did (high control) or did not (low control) read the instructions to refrigerate the product prior to using it. Consistency was manipulated by indicating whether this had happened before (high consistency) or whether it was the first time (low consistency). Consensus was manipulated by indicating whether a friend who purchased the same cream at a different store experienced the same rotting odor (high consensus) or whether the friend had no problem with his/her product (low consensus). Following the scenario, there were questions asking if the subjects understood the situations correctly, regarding the three dimensions (control, consistency, and consensus) manipulated in the scenario. In other words, the subjects were asked whether or not the person in the scenario read the instructions (manipulation of control), this had happened before (manipulation of consistency), and this occurred to another person (manipulation of consensus). Following were six items measuring the locus of attribution and eight items regarding future behavior. Each question was answered by a 7-point rating scale.

Dependent Measures

The scenario was followed by six items measuring the locus of attribution, to which subjects responded by placing a check on a 7-point scale (1="strongly disagree"; 7="strongly agree"). The six questions were:

(1) The manufacturer is making poor quality goods (manufacturer or company).

(2) The pharmacy where I purchased the cream was not refrigerating it properly (pharmacy or branch).

(3) It is the pharmacist’s fault for not telling me to refrigerate the product (pharmacist or employee).

(4) It is my own fault for not reading the instructions to refrigerate the product (my fault).

(5) I had bad luck and purchased an inferior good (bad luck).

(6) Several different situational factors caused the rotting of the product (situation).

The same questions were asked for all eight scenarios. Figure 3 illustrates the types of manipulation that each group received. Starting from the top and then following the path down to each group, the figure shows how each treatment (control, consistency, and consensus) goes in two directions, high or low. Group 1 follows the path of high level of control, high consistency, and high consensus, which is a description of the scenario that the subjects in Group 1 read. Likewise, Group 2 received the scenario type that was manipulated by high level of control, high consistency, and low consensus. Group 3 received the scenario type that was manipulated by high level of control, low consistency, and high consensus. Group 4 received the scenario type that had a high level of control, low consistency, and low consensus. Groups 5, 6, 7, and 8 were all exposed to a scenario where the level of control was low. Despite the expectations stated in Hypothesis 1-b, each group was exposed to manipulations of consistency and consensus as well, shown in Figure 3.

The mean values of how each group rated the possible causes for the product failure were then contrasted. According to the hypotheses, Groups 1 and 3 were expected to rate 'company’ highest, Group 2 was expected to rate 'branch (pharmacy)’ or 'employee (pharmacist)’ highest, Group 4 was expected to rate 'bad luck’ or 'situational factors’ highest, and Groups 5 to 8 were expected to attribute the failure internally by rating 'my fault’ highest.

RESULTS

Since subjects did not have any problem understanding the situation described in the scenario, ScheffT’s tests of multiple comparisons were conducted to determine which group(s) differed in each dependent measure. The mean values of each dependent measure for the respective groups are shown in Table 1. A vertical comparison of the mean values of the respective dependent measures in Table 1 shows how each respective group rated the dependent variables in regards to the product failure. A horizontal comparison of the mean values of the groups shows primarily which group(s) rated highest in each respective dependent variable, and also which groups are homogeneous.

Vertical Analysis

A vertical analysis entails a comparison of the two highest means, or a comparison of the hypothesized mean(s) and the highest mean. ScheffT’s tests and paired sample t-tests were conducted to determine whether the means were significantly different. The results of the analysis provide evidence that partially supports most of the research hypotheses.

Group 1 rated 'company’ highest (M=5.45), followed by 'situation (M=4.38),’ and the difference was not significant, which partially supports Hypothesis 2-a that when the level of control is high, and information regarding the product failure indicates that both consistency and consensus are high, the consumer will attribute the failure to the manufacturer.

Group 2 rated 'pharmacy’ highest (M=5.52). This is consistent with Hypothesis 3 that when the level of control is high, and information regarding the product failure indicates that consistency is high, but consensus low, the consumer will attribute the failure to the branch or employee. The second highest rating was 'situation (M=4.41),’ and the mean difference proved to be marginal (p=.072). Thus, Hypothesis 3 was also partially supported.

Group 3 rated 'pharmacy’ highest (M=5.21) followed by 'company (M=4.75),’ 'emplyee (M=4.75)’ and 'situation (M=4.71).’ Paired sample t-tests indicated that these means were not significantly different from each other. This result again partially supports Hypothesis 2-b that when the level of control is high, and information regarding the product failure indicates that consistency is low, but consensus is high, the consumer will attribute the failure to the manufacturer.

FIGURE 3

EXPERIMENTAL DESIGN

Group 4 rated 'bad luck’ highest (M=5.34) followed by 'branch (M=5.31)’ and 'situation (M=5.24).’ The mean values of 'bad luck,’ 'branch,’ and 'situation’ were not significantly different, whereas the mean difference between 'situation’ and the next highest mean ('company,’ M=3.72) was significant (p<.000). This result again partially supports Hypothesis 4 that when the level of control is high, and information regarding the product failure indicates that both consistency and consensus are low, then the consumer will attribute the failure to bad luck or situational factors.

Group 5 rated 'employee’ highest (M=5.39) followed by 'branch (M=4.48).’ The difference between the two means proved to be significant (p<.000); therefore, the result shows that when the level of control is low, and information regarding the product failure indicates that both the consistency and consensus are high, then the consumer will attribute the failure to the employee.

Groups 6, 7, and 8 rated 'employee’ highest. In all three groups, the next highest rating was 'my fault,’ and in all three cases, the mean differences were not significant. Therefore, except for when both the consistency and consensus were high, in cases when the level of control was low, the attribution was directed toward either the employee or the self. Thus, Hypothesis 1b was also partially supported.

Horizontal Analysis

The differences among means were analyzed from a different perspective, i.e., a horizontal point-of-view. A horizontal analysis entails a comparison of means by each dependent variable.

Company’s Fault. According to Table 1, blame on the company was placed highest in Group 1. In other words, high control, high consistency, and high consensus information formed a causal inference toward the company. Table 2 illustrates the homogeneous subsets of blame on the 'company.’ According to Table 2, there was no single completely separate subset for blame on the company. However, the fourth subset, which included the groups with the highest means, was dissimilar to the first subset except for Group 4, which was included in every subset. Therefore, Groups 1, 3, and 5 were considered significantly different from Groups 2, 6, 7, and 8.

Table 3 supports the above by showing how significantly different Groups 1, 3, and 5 were from Groups 2, 6, 7, and 8. This provides additional evidence that supports Hypotheses 2-a and 2-b. However, the attribution of Group 5 to company was unexplained for.

Branch’s Fault. The highest mean value for attribution toward the pharmacy (or branch) was made by Group 2 (M=5.52). Table 4 shows the homogeneous subsets. Although vague, judging from subset 1, Groups 6, 8, 7, 5, and 1 were significantly different from Groups 3, 4, and 2. Therefore, even though Group 2 rated 'pharmacy’ highest on the attribution, Groups 3 and 4 were not significantly different. This finding supports Hypothesis 3 to the extent that a high level of control, a high level of consistency, and a low level of consensus cause the subject to attribute the failure to the branch. The fact that Group 2 could not be distinguished from Groups 3 or 4 contradicted Hypotheses 2-b and 4.

Bad Luck. Group 4 rated 'bad luck’ highest for the attribution of the product failure. The homogeneous subset turned out quite clearly, placing Groups 2 and 4 in the same subset. Table 5 illustrates the significance of differences in the mean values. The mean difference between Groups 2 and 4 was not signiicant, and thus these groups were placed in the same subset. This result supports Hypothesis 4 to the extent that a high level of control, combined with low consistency and low consensus leads to attribution toward bad luck.

TABLE 1

MEAN OF EACH DEPENDENT MEASURE FOR THE RESPECTIVE GROUPS

TABLE 2

MEANS FOR GROUPS IN HOMOGENEOUS SUBSETS

DEPENDENT VARIABLE: COMPANY

TABLE 3

MULTIPLE COMPARISON

DEPENDENT VARIABLE: COMPANY

TABLE 4

MEANS FOR GROUPS IN HOMOGENEOUS SUBSETS

DEPENDENT VARIABLE: BRANCH

TABLE 5

MEANS FOR GROUPS IN HOMOGENEOUS SUBSETS

DEPENDENT VARIABLE: BAD LUCK

My Fault or Employee’s Fault? As can be seen in Table 6, attribution toward oneself ('my fault’) was rated highest by Groups 6, 7, and 8 (Ms=5.57, 5.00, and 5.10, respectively). The homogeneous subsets also indicated that groups 6, 7, and 8 were not significantly different. However, the three same groups and also Group 5 rated highest, and were homogeneous, in attributing the failure to the pharmacist ('employee’). The difference between the mean value of 'employee’ and that of 'my fault’ was not significant, which is encouraging in testing Hypothesis 1. However, an explanation for the attribution toward employee is still required. Also, for Group 5, attribution toward oneself was low and part of a different subset from Groups 6, 7, and 8. Instead, blame was placed primarily on the employee (or the pharmacist), followed by the branch (or the pharmacy).

The homogeneous subsets of the dependent variable 'employee’ are illustrated in Table 7. Groups 5, 6, 7, and 8 were of a homogeneous subset, and were completely different from Groups 1, 2, and 4 (Group 3 was included in all subsets). Paired sample t-tests of the mean values of Groups 6, 7, and 8 in terms of the dependent variables 'employee’ and 'my fault’ showed that the differences were not significant, which implied that the subjects did actually blame their own negligence. This provides an additional support for Hypotheses 1-b.

Situational Factors. The dependent measure 'situation’ turned out to interfere with the clarity of the study rather than contribute to explaining the effects of the three treatments. An ANOVA showed that none of the eight groups was significantly different from the rest in terms of attributing the product failure to situational factors (F=1.174, p>.319). This dependent measure maintained the means that ranged from 4.19 to 5.24 across different groups, and was therefore thought to have been a 'nonpolitical answer choice’ for those who were not certain about their direction of attribution.

TABLE 6

MEANS FOR GROUPS IN HOMOGENEOUS SUBSETS

DEPENDENT VARIABLE: MY FAULT

TABLE 7

MEANS FOR GROUPS IN HOMOGENEOUS SUBSETS

DEPENDENT VARIABLE: EMPLOYEE

Additional Findings: Future Behavior

The respective means of the dependent measures for each treatment group were compared. Group 1, which chose 'company,’ among other dependent variables, as the entity to blame for the product failure rated "engage in negative word-of-mouth" highest for future behavior. Group 2 attributed the cause of failure to the pharmacy, and quite logically stated "switch to a different pharmacy" as the most probable future behavior. Both the pharmacy and the company were thought to be responsible for the product failure by Group 3, and consistent to those results, the future behavior included "engage in negative word-of-mouth" and "switch to a different company." Group 4 placed most blame on bad luck, but also attributed the failure to the pharmacy and situational factors. The expected future behavior included "simply exchange the product and stay loyal" and "switch to a different pharmacy." Group 5, which placed the attribution primarily on the employee, and then on the company, put down "switch to a different pharmacy" and "engage in negative word-of-mouth" as their most likely future behavior. Groups 6, 7, and 8 attributed the failure to both the employee and to their own negligence, and their future behavior entailed "switch to a different pharmacy" and "simply exchange product and stay loyal." Table 8 is a summary of the findings.

DISCUSSION

Apart from the results of this study that were consistent with most of the hypoteses, there were some unexpected outcomes that require much deliberation. First, the attribution as a result of low control was expected to be directed internally (H1-b); however, the attribution was placed on an external entity, the employee, as well as on oneself. The tendency of the subjects to blame the employee over their own negligence is understood as a consequence of the self-serving, or ego-defensive bias, which is the tendency for individuals to accept more causal responsibility for their positive outcomes than for their negative outcomes (Greenwald 1980; Miller and Ross 1975; Sedikides, et al. 1998).

TABLE 8

FUTURE BEHAVIOR OF EACH TREATMENT GROUP

Second, although Group 3 turned out to be a homogeneous subset of Group 1 in placing the attribution on the company, within Group 3, the highest rating was on branch, followed by company. This result is difficult to explain without going directly against the hypotheses that have been supported. The reasoning behind Hypothesis 2-b was that when the consensus information is high, or when the consumer is informed that the product failure occurred in other purchases from different branches, the consumer would attribute the cause to a larger entity than the branch at which the consumer purchased his/her own product. Therefore, the only way of understanding this result is to infer that the subject believed the branch was at fault in both his/her own case, and the case of the other person.

The results of this study offer important insights for marketers. First, attribution toward the company prevailed in all instances of high consistency. In other words, repeated failure of a product is a strong factor that causes attribution toward the company. Especially in the case of Group 5, where the level of control was low, much of the blame was placed on the company, which tells us that regardless of whose fault the failure really is, any kind of repeated failure will cause the consumer to blame the company. In every instance where the company was attributed to for the failure, negative word-of-mouth was stated as the probable future behavior. This is fatal to the company because engaging in negative WOM entails the switching of both the consumer himself/herself and other existing customers, not to mention potential customers. Therefore, the manufacturer must predict every possible scenario that a consumer could perceive their product to be defective, whether due to the product itself, or due to something that the consumer might have overlooked. For instance, in the scenario used in this study, the manufacturer should make sure that the respective pharmacies tell their customers about the need to refrigerate the product, or make the instructions more noticeable by using a large red font and printing the instructions on the product itself.

Second, a low level of consensus affects the consumer by not making it possible to generalize the product failure, and thus causing the consumer to suspect the factors pertaining to his/her purchase(s). For instance, when the consumer finds out that his/her product is the only one that is defective, he/she will search for a reason within the boundaries of his/her particular purchase and find fault with the branch where he/she bought the product, the employee that helped him/her, or himself/herself for misusing the product. Therefore, finding a way to cut off the spreading of consensus information could be helpful in preventing the consumer from instantly attributing the failure to the company, whether it be by eliminating the need to search for consensus information, or by influencing the consumer’s way of thinking so as to create a more "consumer responsible" attitude toward his/her own purchases.

One of the limitations of this study is that it does not consider many of the existing factors that are known to affect attribution, such as underlying motivation, prior beliefs, or other types of information that are thought to have influence. For instance, although two people may have experienced the same exact product failure, the context of the situation prior to failure or the purchase situation could be entirely different, not to mention the pre-existing differences in personality, personal background, or other individual caracteristics. However, no single model would be able to completely explain such a complicated thought process as attribution, and in terms of information that influences the process of attribution, the three variables that were studied, control, consistency, and consensus are doubtlessly important factors that have been studied numerous times by previous scholars and have been proven to have a significant effect.

Based on the implications and limitations of this study, future research should explore the relationship between the attribution and the anticipated future behavior, and aim to discover a way for manufacturers to affect the relationship after the attribution has been made. Since individual characteristics are too complex and have such an inconsistent effect on the attribution process as well as subsequent behavior, marketers should focus on finding a common ground across purchase situation and prior beliefs. Also, in light of the fact that this study was conducted using a scenario that included a low-involvement product, future studies should test the generalizability of these findings to high-involvement products. Future research should also examine potential moderating roles of individual differences and situational variables such as locus of control and situational involvement.

REFERENCES

Berscheid, E., W. Graziano, T. Monson and M. Dermer (1976) "Outcome Dependency: Attention, Attribution, and Attraction," Journal of Personality and Social Psychology, 34, 978-989.

Bitner, M. J. (1990), "Evaluating Service Encounters: The Effects of Physical Surroundings and Employee," Journal of Marketing, 54 (April), 69-82.

Enzle, M. E., and D. Schopflocher (1978), "Instigation of Attribution Processes by Attributional Questions," Personality and Social Psychology Bulletin, 4, 595-599.

Folkes, V.(1984), "Consumer Reactions to Product Failure: An Attributional Approach," Journal of Consumer Research, 10 (March), 398-409.

Folkes, V. (1987), "The Role of Causal Inferences in Postpurchase Processes," Research in Consumer Behavior, 2, 137-160.

Folkes, V. (1988), "Recent Attribution Research in Consumer Behavior: A Review and New Directions," Journal of Consumer Research, 14 (March), 548-565.

Greenwald, A. G. (1980), "The Totalitarian Ego: Fabrication and Revision of Personal History," American Psychology, 35, 603-618.

Harvey, J. H. and G. Weary (1984), "Current Issues in Attribution Theory and Research," Annual Review of Psychology, 35, 427-459.

Hastie, R. (1984), "Causes and Effects of Causal Attribution," Journal of Personality and Social Psychology, 46, 1, 44-56.

Hirschman, A. O. (1970), Exit, Voice and Loyalty: Responses to Decline in Firms, Organizations and States, Cambridge, MA: Harvard University Press.

Kelley, H. H.(1967), "Attribution Theory in Social Psychology," Nebraska Symposium on Motivation, ed. D. Levine, Lincoln, NE: University of Nebraska Press, 192-238.

Kelley, H. H. (1972), "Attribution in Social Interaction," in Attribution: Perceiving the Causes of Behavior, eds. E. E. Jones, D. E. Kanouse, H. H. Kelly, R. E. Nisbett, S. Valins, and B. Weiner, Morristown, NJ: General Learning Press.

Kelley, H. H. (1973), "The Process of Causal Attribution," American Psychologist, 28, 107-28.

Lau, R. R. (1984), "Dynamics of the Attribution Process," Journal of Personality and Social Psychology, 46, 1017-1028.

Martinko, M. J. and N. F. Thomson (1998) "A Synthesis and Extension of the Weiner and Kelley Attribution Models," Basic and Applied Social Psychology, 20, 4, 271-284.

McGill, A. (1989), "Context Effects in Judgments of Causation," Journal of Personality and Social Psychology, 57, 2, 189-200.

Miller, D. T., and M. Ross (1975), "Self-Serving Biases in Attribution of Causality: Fact or Fiction?" Psychological Bulletin, 82, 213-225.

Pittman, T. S. and N. L. Pittman (1980), "Deprivation of Control and the Attribution Process," Journal of Personality and Social Psychology, 35, 49-55.

Pyszczynski, T. A. and J. Greenberg (1981), "Role of Disconfirmed Expectancies in the Instigation of Attributional Processing," Journal of Personality and Social Psychology, 40, 31-38.

Sedikides, C., W. K. Campbell, G. D. Reeder and A. J. Elliot (1998), "The Self-Serving Bias in Relational Context," Journal of Personality and Social Psychology, 74, 2, 378-386.

Weiner, B.(1980), Human Motivation, New York, NY: Holt, Rinehart & Winston.

Weiner, B. (1985), "Spontaneous Causal Thinking," Psychological Bulletin, 97, 1, 74-84.

Weiner, B. (2000), "Attributional Thoughts about Consumer Behavior," Journal of Consumer Research, 27 (December), 382-387.

Wong, P. T. P. and B. Weiner (1981), "When People Ask 'Why’ Questions, and the Heuristics of Attributional Search," Journal of Personality and Social Psychology, 40, 650-663.

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

Authors

Moonkyu Lee, Yonsei University, Korea



Volume

AP - Asia Pacific Advances in Consumer Research Volume 5 | 2002



Share Proceeding

Featured papers

See More

Featured

M13. Keep Consistency in Good Old Days: The Effect of Nostalgia on Consumers' Consistency Seeking Behavior

Yafeng Fan, Tsinghua University
Jing Jiang, Renmin University of China

Read More

Featured

B1. Dynamic Pricing in Stationary Retailing - The Role of Consumer's Trust

Maximilian Clemens Pohst, Heinrich-Heine-University
Caspar Krampe, Heinrich-Heine-University
Peter Kenning, Heinrich-Heine-University

Read More

Featured

P4. Consistent Price Endings Increase Consumers Perceptions of Cheapness

Shih-Chieh Chuang, National Chung Cheng University
Yin-Hui Cheng, National Taichung University of Education

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.