Further Information on Consumer Perceptions of Product Recalls

John C. Mowen, Oklahoma State University
ABSTRACT - An experiment was conducted to determine consumer reactions to a company making a product recall. In the 2 X 2 X 2 full factorial design the manipulations included: (a) familiar versus unknown company making the recall, (b) action or inaction by the Consumer Product Safety Commission to force the recall, and (c) whether or not other manufacturers had had similar problems. The results revealed that consumers perceived a familiar company as significantly less responsible for the defect than an unfamiliar company. It was also found unexpectedly that consumers perceived the company as more responsible when it acted prior to intervention by the Consumer Product Safety Commission.
[ to cite ]:
John C. Mowen (1980) ,"Further Information on Consumer Perceptions of Product Recalls", in NA - Advances in Consumer Research Volume 07, eds. Jerry C. Olson, Ann Abor, MI : Association for Consumer Research, Pages: 519-523.

Advances in Consumer Research Volume 7, 1980     Pages 519-523


John C. Mowen, Oklahoma State University


An experiment was conducted to determine consumer reactions to a company making a product recall. In the 2 X 2 X 2 full factorial design the manipulations included: (a) familiar versus unknown company making the recall, (b) action or inaction by the Consumer Product Safety Commission to force the recall, and (c) whether or not other manufacturers had had similar problems. The results revealed that consumers perceived a familiar company as significantly less responsible for the defect than an unfamiliar company. It was also found unexpectedly that consumers perceived the company as more responsible when it acted prior to intervention by the Consumer Product Safety Commission.


A corporate problem receiving increasing attention is that of product recalls. Among automobile manufacturers in the first half of 1978, recalls totaled 8.5 million and estimates are that in 1977 automobile makers spent $150 million to handle recalls (Tamarkin 1978). Because of the enormity of the recall problem, attention has been addressed to the development of managerial guidelines for making recalls (Fisk and Chandran 1975, Kerin and Harvey 1975, Gumbhir and Jamison 1975, Warner 1977, Snyder 1974). However, while substantial energy has been focussed on the normative issues of how managers "ought" or "should" plan for and conduct product recalls, little attention has gone to the question of how consumers perceive companies which make recalls.

Mowen (1979) performed the first experimental investigation into consumer perceptions of product recalls. In a 2 X 2 X 2 full factorial design, he manipulated the severity of injury resulting from the defective product, the length of time the company took to decide to make the recall, and the number of previous recalls made by the company. The results of the study revealed that the extent of injury, the number of previous product recalls, and the length of time to recall the product each influenced consumer perceptions of the corporation and their intention to purchase a replacement product made by the company.

A theoretical account of the results was made through an integration of signed-digraph analysis (Belk 1976) and attribution theory (Kelley 1967, Jones and Davis 1965). Figure 1 presents a signed-digraph analysis of a product recall. Signed-digraph analysis extends balance theory (Heider 1958) so that any number of cognitive elements may be analyzed. In the case of a product recall, the elements are consumer, the consumer's perception of himself as a consumer, the corporation making the recall, the defective product (DP), and a replacement product (RP). (In the experiment a defective percolator was the product.) The lines denoted by arrows represent sentiment relations which indicate the consumers affective perception (liking) for himself, the corporation, the defective percolator, and the replacement percolator. The sentiment relation connecting the corporation to the consumer's self represents the consumer's estimate of the corporation's liking for himself as a consumer.

In Figure 1 the double lines indicate unit relations. A unit relation occurs when the consumer perceives a strong association between two cognitive elements. As discussed by Rosenberg and Abelson (1960) a unit connection may be positive (strong association), negative (strong disassociation), or null (irrelevant association). The unit relation connecting XYZ to the defective percolator represents the consumer's perception of the company's responsibility or blame for the defective percolator. If the consumer perceives the company to have caused the defect through its actions, a strong association will exist between XYZ and DP. In contrast, if bad luck, rather than the company, were perceived as causing the defect, responsibility would be lower. (The DP-RP and XYZ-RP connections are assumed to be positive unit connections.)



A basic assumption of balance theory is that the multiplication of the signs of the relations, when arranged into semi-cycles, must result in a positive value in order for the grouping of elements to be in balance. (A semi-cycle is defined as a connected grouping of three or more elements.) If the resulting sign is not positive, cognitive consistency forces will tend to make the perceiver reorganize his perceptions so as to achieve balance. It has been proposed that the reorganization occurs through a change in the weakest of the cognitive relations (Abelson and Rosenberg 1958).

In the Mowen (1979) research the extent of injury was shown to influence consumers' sentiment regarding the defective percolator. Looking at Figure 1, a semi-cycle exists of consumer (C), defective percolator (DP), and company (XYZ). According to balance theory, if consumers perceive a unit relation between XYZ and the DP, changes in the perception of the DP will lead to changes in the perception of XYZ. Thus, with the sign of the DP-XYZ relation positive, a more negative view of the defective percolator (resulting from serious injuries caused by the percolator) consequently results in a more negative view of the company. In a similar manner, a large number of previous recalls was shown to influence negatively the perception of XYZ because it created the impression of greater responsibility for XYZ for the defective product (XYZ-DP relation). In the same way, the "length of time to recall" influenced negatively the XYZ-Self sentiment relation, which in turn influenced the Consumer-XYZ relation within the Consumer-Self-XYZ semi-cycle. Thus, as the time taken to make the recall increased, subjects perceived the company as less consumeristic. The company's perceived lack of interest in the subject as a consumer resulted in a negative XYZ-Self relation. Consequently, with a positive Consumer-Self bond and a negative XYZ-Self bond, the Consumer-XYZ relation would become negative.

The results of the Mowen (1979) research strongly supported the signed-digraph analysis. The research also supported hypotheses derived from attribution theory concerning the impact of the "number of previous recalls" on the XYZ-DP relation and the "length of time to recall" on the XYZ-Self relation. In the study subjects learned that the XYZ Corporation either had made none or five previous recalls. Kelley's (1967) consistency principle states that the cause for action can be more confidently judged to the extent that two elements covary together over time. When a series of recalls is made, covariation occurs between the recall and corporation. It was predicted and shown that high consistency resulted in more attribution of responsibility for the defective product than did low consistency.

The prediction of the effects of varying the "length of time to make the recall" was based upon Correspondence Inference Theory (Jones and Davis 1965). Correspondence Inference Theory suggests that a dispositional tendency of another person (here the XYZ Corporation) is inferred if an act occurred without coercive outside forces pressuring the person into the act and/or if the behavior had a low prior probability of occurring. The application of the theory suggests that consumers may attempt to determine if the causes for the recall were internal to the company or if factors external to the company caused the act. When a corporation makes a recall more rapidly than other corporations (low prior probability) and prior to intervention by governmental agencies (low external pressure), according to correspondence theory, the company's act and behavior will correspond; and, it will be perceived as dispositionally inclined towards consumerism. Conversely, by delaying the recall, external pressures exist, correspondence will not be perceived, and the company will not be seen as acting because of its consumeristic disposition but because of its fear of governmental action. As hypothesized, the results revealed that as the length of time taken to recall the product increased, the estimates of the corporation's concern for consumer welfare decreased.

While the results of the Mowen (1979) research were clear-cut, a confounding of variables occurred in one instance. In manipulating the length of time to recall, a statement that the recall was made prior to or after a Consumer Product Safety Commission investigation was varied simultaneously. Consequently, the variations of the perceptions of the company's consumerism could have resulted from either the manipulation of the length of time to recall or the presence or absence of an investigation. One purpose of this research is to determine whether length of time to recall or the actions of the CPSC accounted for the significant effect.

An aspect of the previous research which limited its external validity was the fact that a fictitious (XYZ) corporation was used as the company. Quite possibly when consumers know and recognize the company, their perceptions are ingrained and to an extent immune to change. A second purpose of this research is to compare consumer perceptions of a known versus an unknown company making the product recall.

A third variable selected for investigation in the present research was derived from the attribution literature. Kelley (1967) discussed the impact of consensus information on judgments of causality. Consensus refers to the extent with which other individuals act in the same way as the target individual. To the extent that others behave similarly to the target person, one can assume that the cause for action (or possibly inaction) is in the stimulus entity rather than in the specific actions of the target. Extending these attributional concepts to product recalls, one can conceive of consumers questioning whether a company's deliberate actions caused the defective percolator or whether some problem in making percolators could have caused the defect. If other companies have a problem with the percolator similar to the corporation's, according to attribution theory consumers should perceive the corporation as less responsible for the defect than if no other companies had the defect.


Design Overview

A 2 X 2 X 2 full factorial design was employed with a control group. The factors were: (a) name of company (XYZ versus Corning Ware), (b) Consumer Product Safety Commission action or inaction prior to recall, and (c) whether or not other manufacturers had had a similar defect. The control group consisted of subjects who only rated their perceptions of Corning Ware, receiving no manipulations and no information about product recalls.


One hundred thirty-nine subjects participated in the experiment. Four failed to complete the questionnaire and were discarded from the analysis, leaving a total of 135 subjects. Subjects were obtained from three sources. Forty-four questionnaires were distributed to the secretarial staff working in the College of Business and other departments at a large midwestern university. Thirty-five questionnaires were distributed to housewives living near campus using a door-to-door canvassing procedure. The remainder of the questionnaires were distributed to individuals attending a League of Women Voters meeting. The procedure was intended to gain a diverse sample of women. (Three men attending the League of Women Voters meeting slipped into the sample and were retained for analysis.) The ages of the subjects varied widely with 1.5 percent less than 21, 18.0 percent 21 to 25, 27.8 percent 26 to 35, 32.3 percent 36 to 50, and 20.3 percent 51 and over.

A comment on the sampling procedure should also be made. As stated by Sternthal, Dholakia, and Leavitt (1978), the use of a convenience sample is appropriate if the purpose of the research is to detect the relation among variables of theoretical significance. The present research meets such criteria. The convenience sample of women was selected in order to insure that the general issues investigated in the study (i.e., defective perceptions of companies recalling percolators) would have relevancy to the subjects.


Subjects were told that the information which they would receive was taken from an article in Newsweek magazine appearing in 1976. They then read a paragraph containing filler information, but within which the Corning or XYZ Corporation manipulation was made. The filler information was based upon an article appearing in Business Week magazine which described a recall of a percolator by Corning.

Corning Glass Works (the XYZ Corporation), the upstate New York company that has built a billion-dollar business out of glass, is recalling one of its products. What is involved are some 360,000 electric coffee percolators, manufactured in 1974 with handles that may come unstuck because of faulty epoxy. Words of pots falling apart first began reaching Corning (XYZ) a year ago, and this past June, when 373 persons had reported receiving burns from the pot, 12 of them seriously enough to require medical attention, the company proceeded with the recall.

In the next paragraph, action or inaction by the CPSC was varied.

A check with the Consumer Product Safety Commission by Newsweek staff members indicated that the government organization hasn't determined yet whether the pot (has determined that the pot) presents a serious enough hazard to warrant a recall. However, Corning (XYZ) decided to go ahead anyway with the recall. Says Robert L. McCleary, manager of consumer affairs for Corning (XYZ), "We decided . . . to get back every pot we could, and do it in a way that would make consumers happy."

Next "consensus" information was varied. In the condition in which subjects learned that other companies were not having problems, they read:

Checks with the Consumer Product Safety Commission also revealed no evidence of other manufacturers of percolators having difficulties similar to Corning (XYZ), The faulty percolator handles seem to be isolated in the Corning (XYZ) product.

In the conditions in which other companies were having problems, subjects read:

Checks with the Consumer Product Safety Commission also revealed that other manufacturers of percolators were having difficulties similar to Corning (XYZ). Investigations are proceeding on four separate manufacturers for producing percolators with faulty handles.

The last paragraph contained additional filler material which was taken from an article appearing in Business Week magazine.

To publicize the recall, Corning (XYZ) spent just short of $1 million on display kits for 90,000 retailers. It also planned a media campaign that included slide presentations on prime-time television, announcements in nearly every major newspaper, and mats for smaller inserts, to allow editors to insert the news in their columns without even having to set type. So far, some 15,000 pots have been returned, and more are currently arriving at the rate of about 750 a day. The recall, Corning (XYZ) says, "is costing in the millions," but the company will not be more specific. McCleary's target is to retrieve 35 percent of the defective models.

Subjects next responded on seven-point rating scales to a number of questions: (1) Rate your general impression of the corporation (very favorable--very unfavorable), (2) If you needed a new coffee pot, how interested would you be in buying the new pot which replaced the defective percolator? (highly interested--highly disinterested), (3) To what extent was bad luck or chance, rather than the actions of the company, responsible for the coffee pot being defective? (chance completely responsible--company completely responsible), (4) How concerned with consumer welfare do you perceive the corporation to be? (highly concerned--not at all concerned), (5) How foreseeable was it on the corporation's part that the percolator could have been defective? (highly foreseeable--impossible to foresee), and (6) How much danger does the coffee pot represent to the public? (extreme danger--no danger).

After completing the questionnaire, subjects were debriefed concerning the nature of the experiment. The actual actions of Corning were detailed, and it was explained that Corning acted in a highly socially-responsible manner. In the case of the secretaries, written statements were provided to them.


The results of the analysis of variance performed on the 2 X 2 X 2 factorial design revealed only two significant effects, both for the "responsibility" dependent variable. Contrary to expectations, when consumers obtained information that the CPSC had not yet determined if a serious enough hazard existed to force a recall, they rated the company significantly more responsible than if such a determination had been made, M CPSC action = 4.45, no CPSC action = 5.16, F(1,119), p < .02. The second effect revealed that Corning was perceived as less responsible for the defect than the XYZ Corporation, M Corning = 4.49, M XYZ = 5.12, F(1,119) = 5.33, p < .03.

An analysis of variance was also performed to compare the control condition to the Corning and XYZ conditions. No differences were found among the conditions in subjects' general impression of the companies or with subjects' views on the consumerism of the companies. Such results indicate that information that a company has made a recall does not necessarily lower the general impression of the consumer regarding the company.

The overall results revealed that consumers perceived the company (whether Corning or XYZ) relatively favorably (M = 3.19 with 1 equals very favorable) and as relatively consumeristic (M = 3.03). Consumers were essentially neutral in their intention to buy the replacement percolator (M = 3.67). They viewed the defect as neither highly foreseeable nor impossible to foresee (M = 3.66) and as somewhat dangerous (M = 3.21).


One issue raised by the results of the study involves explaining why consumers perceived the company as more responsible for the defect when the CPSC had not acted on the recall. A tentative explanation is that after the CPSC makes a decision to recall a product, responsibility is perceived as shared by the CPSC and the corporation. Conversely, when no CPSC action has taken place, full responsibility for the defect is placed on the company. Future research should focus on replicating this unexpected finding and attempt to identify the operative mediator.

The unexpected finding, however, did begin to settle one issue of the Mowen (1979) research. The manipulation of whether or not the CPSC had taken action had no measurable impact on the perception of the corporation's consumerism. These results tend to indicate that it was the effects of manipulating the length of time to recall which influenced consumers' perceptions in the Mowen (1979) research rather than the actions of the CPSC.

The research also revealed that familiarity with the company may have some impact on consumer perceptions of corporate responsibility. The results revealed that Corning was perceived as less responsible for the defect than the XYZ Corporation. Such results indicate that care should be taken in future research to utilize the names of known corporations in order to develop as much external validity as possible. They also reveal that corporations who act responsibly and possess sound reputations may be perceived as less responsible for defective products than less well known companies.

According to the balance model, the significant effects obtained for the responsibility variable should have resulted in concomitant changes in consumers' perception of the corporations. However, no significant effects were obtained in consumers' favorability towards either company. In the Mowen (1979) research strong variations in perceptions of the company were obtained when the extent of injury, length of time to recall, and number of previous recalls were varied. The probable cause of the failure to find the expected effects in the present research was the relatively favorable view of the companies by the subjects. In the Mowen (1979) research the mean rating of favorability was 4.01. As noted in social judgment theory (Sherif and Hovland 1961) neutral attitudes tend to be those most amenable to change. Thus, the manipulations of the independent variables in the Mowen (1979) study could readily impact upon the attitude towards the corporations through the operation of cognitive consistency forces. However, with consumers viewing the corporation relatively favorably in the present research (M = 3.19), it may have been more difficult to influence the attitude.

The above line of reasoning forces the question, why did subjects perceive the corporations more favorably in the present experiment? The major difference in the scenarios presented to the subjects in the two experiments, outside of the manipulations, was statements made by an officer of the corporation. A possibility exists that by personalizing the recall, indicating that the company was trying to make consumers happy, and mentioning the percentage of percolators which the corporation hoped to retrieve, consumers may have reacted relatively more favorably to the recall in the present research than in the Mowen (1979) study. Future research must vary orthogonally the approach of straight-forwardly presenting the facts of the recall versus personalizing it to test the hypothesis.

The manipulation of consensus information failed to influence perceptions in any measurable way. Previous researchers have found consensus information to have relatively little impact on attributions. For example, Nisbett and Borgida (1975) found that consensus information given in the form of base rates for behavior failed to influence attributions. While later researchers have found significant effects for consensus information (Hansen and Stonner 1978, Wells and Harvey 1977) the precise mediator of the variability in the effects of consensus has not been established. The question of whether consumers consider the actions of competitors when evaluating a specific company is important and deserves further research.

An important point raised by the study concerns the ethics of using the "Corning" name in the present research. The author believes that the use of a corporate name is justified when the research involves an incident or situation within which the company was involved. In the present experiment, Corning Glass Works was contacted and permission gained to use the "Corning" name. As indicated by a Corning spokesperson, however, several caveats should be observed. The company utilized should in no way be represented as in "collaboration" with the research or as responsible for any statements or conclusions drawn in the research. In addition, the researcher is obligated to debrief subjects and give them the information necessary to insure that they possess full knowledge of the actual actions of the company.

Future research on consumer reactions to product recalls should focus on assessing the generalizability of the present study and the Mowen (1979) research. The research should employ a survey research approach and utilize econometric techniques to focus on the extent with which the variables investigated to date influence consumer reactions, i.e., such variables as the life threatening quality of the defect, the length of time to recall, the number of previous recalls, the consumer's experience with competitor's products, and the corporate reputation. Further, the research should assess the extent with which consumers are aware of product recalls. Quite possibly in the "clutter" of everyday activities, the variables shown to be causally related to consumer perceptions in the laboratory will lose their potency.


Abelson, R. P. and Rosenberg, M. J. (1958), "Symbolic Psycho-logic: A Model of Attitudinal Cognition," Behavioral Science, 3, 1-13.

Belk, R. W. (December 1976), "It's the Thought That Counts: A Signed-Digraph Analysis of Gift-Giving," Journal of Consumer Research, 3, 155-161.

Fisk, G. and Chandran, R. (November-December 1975), "How to Trace and Recall Products," Harvard Business Review, 90-96.

Gumbhir, A. K. and Jamison, M. C. (November 1975), "Assessing the Effectiveness of Drug Recall in Pharmacies,'' Drug and Cosmetic Industry, 17, 41-43, 113-116.

Hansen, R. D. and Stonner, D. M. (1978), "Attributions and Attributions: Inferring Stimulus Properties, Actions, Dispositions and Causes," Journal of Personality and Social Psychology, 36, 657-667.

Heider, Fritz (1958), The Psychology of Interpersonal Relations, New York: John Wiley and Sons, Inc.

Jones, E. F. and Davis, K. E. (1965), "From Acts to Dispositions: The Attribution Process in Person Perception,'' in L. Berkowitz (ed.), Advances in Experimental Social Psychology, New York: Academic Press.

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

Kerin, Roger A. and Harvey, Michael (1975), "Contingency Planning for Product Recall," MSU Business Topics, 23, 5-12.

Mowen, J. C. (1979), "Consumer Reactions to Product Recalls: An Empirical and Theoretical Examination," submitted for publication.

Nisbett, R. E. and Borgida, E. (1975), "Attribution and the Psychology of Prediction," Journal of Personality and Social Psychology, 32, 932-943.

Rosenberg, M. J. and Abelson, R. P. (1960), "An Analysis of Cognitive Balancing," in M. J. Rosenberg et al., Attitude Organization and Change, New Haven, Conn.: Yale University Press, 112-163.

Sherif, M. and Hovland, C. J. (1961), Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Changes, New Haven, Conn.: Yale University Press.

Snyder, J. D. (October 1974), "How to Survive a Product Recall," Sales Management, 23-27.

Sternthal, Brian, Dholakia, Ruby, and Leavitt, Clark (1978), "The Persuasive Effect of Source Credibility: Tests of Cognitive Response," Journal of Consumer Research, 4, 252-260.

Tamarkin, Bob (July 10, 1978), "Recalls--The Costs Soar," Forbes, 79-80.

Warner, H. W. (July 1977), "Guidelines for Product Recall," Public Relations Journal, 11-13.

Wells, G. L. and Harvey, J. H. (1977), "Do People Use Consensus Information in Making Causal Attributions?" Journal of Personality and Social Psychology, 35, 279-293.