Consumers Reaction to Product Failure: Impact of Product Involvement and Knowledge

T. N. Somasundaram, University of San Diego
ABSTRACT - This paper investigates differences in how consumers explain why a product is a "failure" i.e. falls short of expectations. It extends the Attribution Theory framework that has been used by other researchers (e.g., Folkes, 1984 and Folkes and Kostos, 1986) by incorporating the concepts of consumers' product involvement and causal complexity stemming from product knowledge. It is suggested that consumers who are more involved are likely to be more motivated to engage in causal search following a determination of product failure. Further, consumers with higher levels of product knowledge or experience with the product class are likely to be more causally complex i.e., assign blame for the failure over a greater number of reasons and are therefore likely to be less certain as to the cause of failure. Hence, they are likely to form less extreme beliefs and attitudes about the product. These differences in how consumers explain why a product has failed are likely to impact subsequently, their choice of remedial actions.
[ to cite ]:
T. N. Somasundaram (1993) ,"Consumers Reaction to Product Failure: Impact of Product Involvement and Knowledge", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 215-218.

Advances in Consumer Research Volume 20, 1993      Pages 215-218

CONSUMERS REACTION TO PRODUCT FAILURE: IMPACT OF PRODUCT INVOLVEMENT AND KNOWLEDGE

T. N. Somasundaram, University of San Diego

ABSTRACT -

This paper investigates differences in how consumers explain why a product is a "failure" i.e. falls short of expectations. It extends the Attribution Theory framework that has been used by other researchers (e.g., Folkes, 1984 and Folkes and Kostos, 1986) by incorporating the concepts of consumers' product involvement and causal complexity stemming from product knowledge. It is suggested that consumers who are more involved are likely to be more motivated to engage in causal search following a determination of product failure. Further, consumers with higher levels of product knowledge or experience with the product class are likely to be more causally complex i.e., assign blame for the failure over a greater number of reasons and are therefore likely to be less certain as to the cause of failure. Hence, they are likely to form less extreme beliefs and attitudes about the product. These differences in how consumers explain why a product has failed are likely to impact subsequently, their choice of remedial actions.

CONSUMER SATISFACTION JUDGMENTS

The satisfaction of consumer needs lies at the very heart of the marketing concept. Satisfaction with a given brand is supposed to lead to a higher likelihood of it being repurchased, brand loyalty, positive word of mouth and higher profitability for the firm. Dissatisfaction is supposed to reduce the likelihood of attaining these goals. Further, it may generate negative word of mouth, complaints, demand for substitute goods, refunds and even litigation (Day, 1977; Richins, 1982; Bearden and Teel, 1983 and Folkes, 1984). In extreme cases, consumer dissatisfaction is likely to encourage the imposition of new legislative controls on an entire industry as was the case recently with the airline industry and its poor 'on time' performance.

Folkes (1984) proposed an Attribution theory framework for predicting consumer responses to product failure. Attribution theory pertains to the processes by which people make causal inferences from the information they receive. Within the context of consumer dissatisfaction and complaining behavior, attribution theory suggests that the perceived reason for a product's failure will influence how the consumer will respond.

Causes of product failure were classified by Folkes in terms of their underlying properties. These were 1) Stability, i.e., whether the cause is temporary or will reoccur, 2) Locus, i.e., is the cause located in the seller/ manufacturer or in the consumer? and 3) Controllability i.e., did the party responsible for the failure have any control over the outcome? Folkes suggested that the underlying causal properties influenced three types of consumer responses: (1) expectancy reactions, (2) marketplace equity reactions, and (3) anger reactions. It was shown that stable attributions lead to certainty about a product's failure and a preference for a refund rather than an exchange. When product failure is firm-related, the consumer is perceived to be owed a refund and an apology but when a failure is perceived to be consumer-related, neither a refund nor an apology is expected. When the failure is firm-related and it is perceived to have control over the reason for product failure, consumers feel angry and seek to hurt the firms business.

INTENSITY OF CAUSAL SEARCH AND PATTERNS OF CAUSAL ATTRIBUTIONS

Two issues neglected by the foregoing research relate to: 1) The intensity of causal search and 2) The patterns of causal attributions. By intensity of causal search we mean the volume of cognitive activity as measured by the number of causal thoughts evident in the subjects' thought protocols. By patterns of causal attributions we mean the locus of the attributions i.e. buyer or seller and subjects' judgments as to the controllability and stability of the factors causing failure. This paper reports preliminary results of an investigation currently in progress into these two issues by incorporating the concept of consumers' product involvement and product knowledge.

Role of Involvement

The search for the causes of product failure is a type of information processing activity undertaken following a determination that a product is unsatisfactory. The intensity of such information processing is likely to be a function of the consumers motivation to process the information. Consumers' product involvement is conceptualized as perceived personal relevance or importance the consumer assigns to the product. To the extent that a product is viewed as personally relevant in that it is perceived in some way to be instrumental in achieving their personal goals and values, the consumer is likely to be more motivated to process information about it (Mitchell, 1981 and Celsi and Olson, 1988). In the context of causal attribution search, consumers with higher involvement are more likely to ponder over the causes of a product's failure to live up to expectations. They are therefore likely to expend more cognitive effort in their causal search.

Role of Causal Complexity

There is a substantial body of research that suggests that the level of knowledge a consumer has about a product affects the amount and type of information processing (e.g. Bettman and Park, 1980; Johnson and Russo, 1981 and 1984). Sujan (1985) suggested that knowledgeable consumers were more likely to be able to discern discrepancies between a stimulus and prior expectations. Somasundaram (1989) showed that consumers with more knowledge were able to better discern when a product's performance did not match expectations for a product of that type. It is suggested in this paper that the number of plausible reasons (causes) advanced to explain product failure is likely to be a function of the consumers' knowledge or experience with a product class. For example, an avid photographer (expert) as compared to an amateur (novice) may be able to advance a wider range of reasons as to why a particular set of photographs received from a film processing facility turned out to be bad.

Further, the amount or proportion of total blame the expert as compared to the novice assigns to a given cause is likely to be different. The isolation of one or few causes is referred to as low causal complexity by Mizerski (1978) who showed that individuals who were low in causal complexity (causally simple individuals), as compared to causally complex individuals, tended to be more confident in their attributions and formed more extreme beliefs and attitudes about a stimulus.

Therefore, it is suggested here that less knowledgeable consumers are likely to be causally simple while more knowledgeable consumers are likely to be causally complex. In our photography example, more knowledgeable consumers may conclude that poor color fidelity in photograph prints received from the processing facility were the result of inherent color bias associated with that brand of film; incorrect tonal balance settings by the facility personnel; weak chemicals; or defective printing paper. Given the number of plausible causes and in the absence of further information, they may conclude that each of the causes was equally likely. Less knowledgeable consumers on the other hand might lump the majority of the blame on one or few causes. Consequently, following Mizerski (1978), it is expected that less knowledgeable consumers are likely to be more confident in their judgements as to why a product has failed and more likely to form extreme beliefs and attitudes about the product.

In addition, knowledgeable consumers and less knowledgeable consumers may differ in the locus of their attributions as well as their judgments as to the stability and controllability of the causes. Knowledge is also likely to impact the accuracy of the attributions.

HYPOTHESES

Based on the foregoing literature review, the following hypotheses are proposed:

Effect of Involvement:

Since involvement has been shown to have motivational properties,

H1: Consumers who have higher product involvement, as compared to consumers who are less involved, are more likely to engage in causal search following product "failure".

Effect of Knowledge:

Since greater knowledge translates into a greater ability to process information,

H2: Among consumers who do engage in causal search, more knowledgeable consumers (experts) as compared to less knowledgeable consumers (novices) are likely to advance a larger number of separately identified 'causes' for the failure.

Further,

H3: Since experts are likely to allocate blame for product failure to more reasons as compared to novices they are likely to be less confident about their attributions and likely to form less extreme beliefs and attitudes about the product.

PILOT EXPERIMENT

An experiment was designed to examine the differences in how consumers explain why a product has failed. The experiment involved subjects evaluation of a pair of photographs purported to have been received by them from a processing facility. Knowledge of photography and Involvement were not manipulated. Knowledge was measured using a 15 item multiple choice test designed to assess objective knowledge about photography. Involvement was measured using the Personal Involvement Inventory (PII) developed by Zaichkowsky (1984). The method of analysis was analysis of variance with involvement and knowledge as covariates.

Data was obtained from 80 male and female graduate business students at a small private university. They were presented with a pair of photographs purportedly processed at a local film processing facility. One photograph in the pair represented the "expected" photograph that they were supposed to receive while the other represented the "actual" photograph. Half the sample was assigned to the Success treatment and the other half to the Failure treatment. In the Failure treatment, the "actual" photograph suffered from several flaws: some caused by the picture taker such as poor composition and exposure selection and other flaws caused by the processing facility such as incorrect tonal balance and color proportions. In the Success treatment, both photographs in the pair were identical.

Subjects evaluated the two photographs in the pair and indicated in a thought elicitation task if and why they perceived the actual picture to be different from the expected picture. These thoughts were coded and content analyzed for distinct causal dimensions. The responses were categorized as "Causal Thoughts" and "Other Product Related Thoughts" and "Irrelevant Thoughts". The sum of "Causal thoughts and "Other Product Related Thoughts", "Total Thoughts", was taken to indicate intensity of causal search. After the open-ended responses, subjects responded to likert type statements relating to the extent the actual results matched their expectations and their level of satisfaction. They were then asked to allocate blame on a constant sum scale to the photographer, or the processing facility or to chance/ random factors. They also responded to a scale designed to determine their level of confidence in their attributions. An on-going study refines the foregoing approach and will be explained in the Modified Experiment section.

RESULTS

Manipulation Check:

Subjects responded to the statement: "How would you rate the actual picture as compared to the expected picture" on a 5-point scale anchored by "Much Worse than Expected" represented by 1 and "Much Better than Expected" represented by 5. An Analysis of Variance (ANOVA) with Success/Failure as the main factor and Involvement and knowledge as covariates revealed a significant main effect for the success/failure treatment (F=118.35, p<0.000). The cell means show that the actual picture is much worse than expected in the failure condition (mean=1.34) than in the success condition (mean=3.00). Subjects also responded to the statement measuring satisfaction with the outcome. The ANOVA results again show a significant main effect for the Success/failure treatment (F=55.91, p<0.000) with higher satisfaction scores in the success condition (mean=3.58) than in the failure condition (mean=1.82). Taken together, the results suggest that the manipulation of success and failure was successful.

Intensity of Causal Search

In H1 it was suggested that subjects motivation to engage in causal search would increase as a function of their level of involvement. Therefore the total number of thoughts generated when taken as a measure of cognitive activity would be greater as involvement increased. The ANOVA results for Total Thoughts as the dependant variable show a significant main effect for the success/failure treatment (F=3.55, P<0.06). The cell means suggest that the total number of thoughts subjects generated was greater in the failure condition (mean=5.89) as compared to the success condition (mean=4.97). The covariate involvement was not significant, thus H1 was not supported. However, the covariate Knowledge was significant (F=5.99, p<0.017) suggesting that higher levels of knowledge results in a greater ability to engage in cognitive activity.

Patterns of causal search

In H2 it was suggested that more knowledgeable subjects would have a greater ability to generate a large number of plausible causes to explain the product failure. Thus we would expect the thought protocols of these subjects should contain statements that pertain to distinct causal dimensions or Causal Thoughts. The thought protocols should also contain proportionately, fewer Other Product Related Thoughts. The ANOVA results with causal thoughts as the dependant variable showed a highly significant main effect for the success/failure treatment (F=27.97, p<0.000). The covariate for knowledge was also significant (F=4.80, p<0.03) suggesting that subjects with higher levels of product knowledge were more capable of generating a large number of plausible causes. Thus H2 was supported. The main effect for success/failure condition supports the notion that a discrepancy between expectations and actual performance in more likely to generate cognitive activity (mean=4.13 versus 1.48) than one in which expectations and performance are more congruent (Sujan, 1985; Somasundaram, 1989). The results for Other Product Related Thoughts show a main effect for the success/failure treatment (F=12.29, p<0.001). More Other Product Related Thoughts were in evidence in the success condition (means=3.48 versus 1.76) than in the failure condition. The covariate knowledge, however, was not significant.

Extremity of beliefs

In H3 it was suggested that subjects who advanced a large number of plausible causes when compared to subjects who advanced fewer causes, would be less certain that they have correctly determined the locus of blame for the failure. Our analysis for the failure condition alone, suggests that neither the number of causes advanced, nor the level of knowledge or the pattern of allocation of blame to oneself or the processing facility was significant in explaining subjects confidence in the accuracy of their blame assignment. Thus H3 was not supported.

DISCUSSION

The results that have been reported in this paper were based on a pilot experiment that has subsequently been refined and is currently being rerun. The preliminary results are encouraging in that they suggest that involvement and product knowledge are important variables that might impact the extent and pattern of causal search conducted by a consumer when a product has failed to perform as expected.

MODIFIED EXPERIMENT

Based on results obtained from the pilot study, the experiment has been made considerably more elaborate. Each subject will view two pairs of photographs where one pair will depict a prominent and well liked building that is on campus. The other pair will depict an architecturally similar off-campus building that is only moderately well known. Thus following Celsi and Olson (1988) we will be manipulating situation specific involvement.

Subjects will be put into three treatment conditions where one-third of the subjects will see two pairs of pictures both of which are successes. One-third of the subjects will be in a failure treatment where both pairs of pictures they see will represent failure. Finally, one-third of the subjects will be in an ambiguous treatment where they will see one pair of pictures that represent success and one pair that represents a failure.

Rather than have failure represented by one perfect picture labelled "Expected" paired with another flawed in a multitude of ways, labelled "Actual", failure in the modified experiment is more complex. Specifically, failure will be represented by one perfect picture paired with one that is flawed in only one of four possible dimensions. Two flaws are attributable to the photographer and two are attributable to the processing facility. Photographer attributable flaws are either controllable (poor composition) or uncontrollable (poor light owing to dusk and overcast skies). Processing facility flaws are likewise controllable (incorrect color balance) and uncontrollable (smears and blotches because the developer jammed). All failure pictures composing the failure condition are completely randomized. As before, success is represented by a pair of identical, perfect pictures.

With respect to the dependent measures, in addition to likert-type scales measuring disconfirmation of expectations and satisfaction, subjects will respond to five 7-point bipolar scales measuring the subject's locus and controllability and stability judgments as well as the accuracy and their confidence in their attributions.

The coding scheme for the thought protocols has also been considerably refined. In addition to classifying thoughts as "Causal Thoughts", "Other Product Related Thoughts" and "Irrelevant Thoughts", the "Causal Thoughts" will be subclassified into locus, controllability and stability dimensions following Folkes (1984).

Data will be collected from 144 male and female graduate and undergraduate students. Data collection is about to commence.

CONCLUSIONS

The implications of our findings for the marketer are that consumers predisposition to ponder over the outcomes of a purchase and consumption episode will be contingent on their level of involvement and knowledge of the product class. To the extent that any causal search occurs, subjects who are more knowledgeable may be better able to correctly determine the underlying causes. This suggests that firm personnel need to be in a position to assist less knowledgeable consumers in arriving at an accurate causal determination. Alternatively, a firm's marketing communication aimed at prospective purchasers needs to educate consumers to better deduce reasons underlying unsatisfactory performance. For example, a customer of a weight-loss program who has witnessed unsatisfactory results may need to know that the "failure" could have been caused by a multitude of factors. In the absence of such knowledge the customer may erroneously conclude that program was at fault or that it was inappropriate for their circumstances. Again, a firm seeking to break the loyalty of a rival brand may not be able to trigger problem recognition by consumers unless they are motivated to examine, more closely, instances when their chosen brand fails to live up to expectations. To the extent that it is difficult to heighten involvement in the field, the firm seeking to break into an established market may be better off targeting, in its introductory efforts, consumers who are already highly involved.

REFERENCES

Bearden, William O. and Jesse E. Teel (1983), "Selected Determinants of Consumer Satisfaction and Complaint Reports," Journal of Marketing Research, 20,21-28.

Bettman, John and C. Whan Park (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes: A Protocol Analysis," Journal of Consumer Research, 7, 234-248.

Celsi, Richard L. and Jerry C. Olson (1988), "The Role of Involvement in Attention and Comprehension Processes," Journal of Consumer Research, 15, 210-224.

Day, Ralph L. (1977), "Toward a Process Model of Consumer Satisfaction," in H.Keith Hunt, ed., Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, Cambridge, Mass.:Marketing Science Institute, 153-186.

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

Folkes, Valerie and Barbara Kostos (1986), "Buyers' and Sellers' Explanation for Product Failure: Who Done it?," Journal of Marketing, 50, 74-80.

Johnson, Eric and J. Edward Russo (1981), "Product Familiarity and Learning New Information," in Advances in Consumer Research, Kent B. Monroe ed., Ann Arbor, MI: Association for Consumer Research, 151-156.

Johnson, Eric and J. Edward Russo (1984), "Product Familiarity and Learning New Information," Journal of Consumer Research, 11, 542-550.

Mitchell, Andrew A. (1981), "The Dimensions of Advertising Involvement," in Kent B. Monroe ed., Advances in Consumer Research, Ann Arbor,MI: Association for Consumer Research, 25-30.

Mizerski, Richard W. (1978), "Causal Complexity: A Measure of Consumer Causal Attribution," Journal of Marketing Research, 15, 220-228.

Richins, Marsha L. (1982), "An Investigation of Consumers' Attitudes Towards Complaining," Andrew Mitchell ed., Advances in Consumer Research, Ann Arbor,MI: Association for Consumer Research, .

Somasundaram, T.N.(1989), "An Investigation of Alternative Consumption Evaluation Processes Explaining Consumer Satisfaction Judgments," unpublished doctoral dissertation, University of Wisconsin-Milwaukee.

Sujan, Mita (1985), "Consumer Knowledge: Effects on Evaluation Strategies Mediating Consumer Judgments," Journal of Consumer Research, 12, 31-46.

Zaichkowsky, Judith L.(1984), "Conceptualizing and Measuring the Involvement Construct in Marketing," Unpublished doctoral dissertation, University of California, Los Angeles.

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