Product Recall Communications: the Effects of Source, Media, and Social Responsibility Information

ABSTRACT - The experimental study investigated three factors that may affect consumer perceptions of companies making product recalls. When information indicating that the company acted in a socially responsible manner was present, student consumers held more favorable feelings toward the company. A government press release was viewed as more objective than a company advertisement discussing the recall. Finally, the print medium was viewed as more trustworthy and somewhat more objective than the sound medium. Managerial implications are provided.


David W. Jolly and John C. Mowen (1985) ,"Product Recall Communications: the Effects of Source, Media, and Social Responsibility Information", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 471-475.

Advances in Consumer Research Volume 12, 1985      Pages 471-475


David W. Jolly, Oklahoma State University

John C. Mowen, Oklahoma State University


The experimental study investigated three factors that may affect consumer perceptions of companies making product recalls. When information indicating that the company acted in a socially responsible manner was present, student consumers held more favorable feelings toward the company. A government press release was viewed as more objective than a company advertisement discussing the recall. Finally, the print medium was viewed as more trustworthy and somewhat more objective than the sound medium. Managerial implications are provided.


Product recalls have become a major problem for companies over the past decade because of the increased complexity of products and of the involvement of such agencies as the National Highway Traffic Safety Administration. While it has been charged that under the Reagan administration, enforcement of product safety standards has become lax, product recalls have continued to be made and in some cases cost companies millions of dollars, as with the 1982 Tylenol recall by Johnson and Johnson. In 1974 almost 25 percent of all consumer goods' companies appearing in Fortune's 500 had initiated recall campaigns (Tamarkin 1978). When this article was written, General Motors was in court arguing against being forced to make what could become the largest recall in corporate history for possible defective brakes on its X-body automobiles.

Much of the written work on product recalls by academicians has been aimed at increasing the effectiveness of the recall program and the organization in handling the recall (e.g., Fisk and Chandran 1975). A few empirical pieces on product recalls have been written. Wynne and Hoffer (1976) found that recalls didn't appear to influence market share unless a series of recalls occurred involving the same product model. Simpson and Mowen (1981) found that recalls had little measurable impact on the stock prices of publicly traded companies. These studies indicate that markets in some way adjust to the news of product recalls, making it difficult to measure their impact on market share or stock prices.

Another approach to the study of the impact of recalls is to identify how consumers react to various types of information about the company and the recall. Much of this work has been conducted by Mowen and his colleagues. In the first of a series of studies, Mowen (1979) manipulated whether or not the company was known to consumers, whether or not the Consumer Product Safety Commission (CPSC) directives were given to the company prior to the recall, and whether or not other companies had experienced a similar product problem. The results revealed that company familiarity and CPSC action influenced how responsible people felt the company was for the defect. Further, a well-known company was perceived as less responsible for a defect than an unknown company.

Mowen, Jolly, and Nickell (1980) conducted a survey to obtain consumer reactions to recalls which had recently occurred. A stepwise regression was employed to examine the possible effects of the recall on perceptions of the companies. The subjects' perceptions of the perceived danger of the defect, of a company's social responsibility, and of a company's perceived responsibility for the defect were the best predictors of the subjects' favorability toward the company.

In a third study, Mowen and Ellis (1981) conducted an experiment in which three variables were manipulated the injury resulting from the defective product, the length of time it took the company to decide to make the recall, and the number of previous recalls mate by the company. Each variable was found to influence subject perceptions of the company and their intentions to purchase a replacement product.

The theoretical rationale for the studies by Mowen and his colleagues was attribution theory (Kelley 1967). Indeed, the work of attribution theorists suggests that product recalls, like other types of negative information, can severely damage a company's corporate image. For example, Mizerski (1975) found that subjects tended to attribute favorable information about a product to external causes (i.e., that a person likes products like that), and unfavorable information to internal causes (i.e., it's the company's fault it makes a poor product). People appear to weight negative information more heavily than positive information. Similarly, Kanouse and Hanson (1972) proposed that people are cost oriented in their evaluations and place greater weight on negative information than on positive content.

The findings of Mizerski (1975) and Kanouse and Hanson (1972) have important implications for companies recalling products. Because the news of a product recall is a type of negative information, consumers are likely to weight the information more heavily than advertising revealing product strengths. Thus, companies have a need to mitigate such potential negative information. One approach could be to inform consumers of the socially responsible actions of the company in recalling the product. As Mowen, Jolly, and Nickel (1980) found, companies perceived as more socially responsible are liked more. One approach to creating such an impression could be simply to tell consumers that the company is acting in a socially responsible way during the product recall. Thus, one purpose of the present research was to test whether information on the socially responsible nature of a recall influences consumer liking for a company.

Information Sources

Much has been written on the impact of sources of information. (For example, see the excellent review of Sternthal, Phillips, and Dholakia 1978). Attribution theorists (e.g., Kelley 1973) have argued that sources which provide expected information are relatively uninformative and are therefore discounted." In contrast, sources which give unexpected information are relatively more informative and the impact of such information is therefore augmented.

One issue of relevance to a product recall is consumer perceptions of the trustworthiness and objectivity of the government. Recalls ordered by the government have not always been popular, as in the case with the saccharin recall made in the late 1970s. Sandage and Barban (1970) found that farmers felt the government to be too involved in the area of consumer protection. The respondents felt that protection from government bureaucrats could become as vital as protection from unsavory business practices. A company needs to consider the relative impact of information coming from itself or from the government. Indeed, because of the ambivalence felt by some Americans towards the government, it could be beneficial to simply advertise a company's social responsibility rather than hope that the government will say something positive about the company. Thus, a second purpose of the study was to investigate how consumers react to messages sponsored either by the government or by the company.

Media Influences

The degree to which the medium affects the message has been a topic of some interest to those in the mass communications field. The results of such research, however, have been inconsistent. Dommermuth (1974) investigated the effect of four different media (television, motion picture, print, and radio) on audience perceptions of the communicator and his presentation. Subjects who received the presentation in print thought the presentation was better, fairer, and stronger than either sound or television. The radio mote produced the greatest mean attitude changes with the print mode a close second. Television produced the least amount of attitude change.

Other researchers, however, have obtained results inconsistent with Dommermuth's. Lee (1978) found that television was believed three times more often than newspapers when conflicting reports were given. Clarke and Ruggels (1970) found that a correlation between education and preferences for print media existed. In addition, the types of news mate a difference. The preferred medium to obtain international news was television. However, for national news subjects preferred newspapers. The third purpose of the study, then, was to assess the relative impact of a print or a simulated radio source on consumers learning about a product recall.

Design Overview

Based on the three major goals of the study, a 2x2x2 between groups factorial experiment was developed. Factors investigated included the source of information(government or company), the type of media used to convey the information (print or sound), and the inclusion or deletion of social responsibility information. Major dependent variables of interest included subject perceptions of the company (favorable-unfavorable) and the company's objectivity and trustworthiness.

Based on the literature review, the following hypotheses were developed.

H1: The inclusion of information that the company acted socially responsible will lead consumers to perceive the company more favorably than if no such information is given.

H2: The government source will be perceived as more trustworthy and objective than company advertisements.

H3: The print medium will be viewed as more trustworthy and objective than the sound medium.

No prediction was made concerning interactions among the independent variables.


A 2x2x2 between subjects factorial design was employed. The three factors were source of information (a company advertisement or government news release), media type (radio or print), and favorable social responsibility information either presented or not presented. Because a company would not place unfavorable social responsibility information in its own advertisements, only two levels of social responsibility information were included in the design--social responsibility information given or not given.

The defective product scenario used in the study was a modification of an actual product recall case involving a hair dryer manufactured by Conair Corporation. In this instance, the hair dryer deposited asbestos (a possible carcinogen) onto the heads of hair dryer users. In the experimental situation, the defect was changed to an electrical short circuit that could cause the user to receive an electrical shock when the unit was in operation. The rationale for changing the scenario was to create a defect which would be perceived as serious enough to cause a consumer to become concerned. (Subjects were debriefed after the study as to the true nature of the recall.)

A second reason for choosing a hair dryer as the defective product lies in its relevance to the students that made up the sample. Many students of both sexes own hair dryers and would be able to picture themselves as owning one of these defective products. In addition, the defect needed to occur in a consumer product that had little ownership documentation associated with it, so that ownership could not be traced. It would then be reasonable to expect the recall information to be imparted via the mass media.

In previous experimental work the independent variables have been operationalized in script form. That is, participants read a scenario about a product that had been recalled and some background information leading to the recall. One purpose of the present study was to reoperationalize the social responsibility variable so that a greater sense of realism is obtained. Fictional advertisements and radio newscasts were developed, with the experimental manipulations embedded within them, and presented to subjects.


The subjects that participated in this study were drawn from two sources. Forty-eight of them were randomly selected from a listing of all undergraduate students enrolled in the Colleges of Business and Arts and Sciences at a large midwestern university. The remainder of the sample was drawn from introductory management and consumer behavior classes at the same university. Two of the questionnaires were unusable, leaving a total of one hundred nineteen usable responses.


Subjects were asked to sit at a table and either listen to the radio or read the newspaper. All subjects who received the government press release condition were presented with the following material:

Some models of hair dryers manufactured by Conair Corporation in May 1979 are being recalled because of a possibility of electric shocks, a government spokesman for the Consumer Product Safety Commission said today.

Conair officials said they have discovered a wiring defect in the handle of the company's Warp-9 model hair dryer during additional performance tests on the Warp-9. Company officials said that anyone who had purchased a Warp-9 should return the hair dryer to the place of purchase or any store which carries Conair products. Consumers have a choice between either receiving another model hair dryer or receiving a full refund.

At this point, subjects in the manipulation offering positive social responsibility information read or heard:

The CPSC spokesman said, "Conair should be commended for initiating this prompt recall as soon as the defect was discovered. Socially responsible actions such as these make our job to insure the safety of consumers a little easier."

All subjects assigned to news sources then read or heard:

The defective models can be detected by checking the serial number on the bottom of the handle. If the first number is a 6, and the last number is either a 2, 3, or 5, then the hair dryer is defective and should be returned.

Those subjects assigned to advertising conditions were presented with the following material:

Did you just buy a hair dryer? A Conair hair dryer? If you did, Conair Corporation needs to hear from you. Conair has discovered that some of the hair dryers it made in 1979 may have developed a short in the electrical system. This problem may cause an electrical shock to the user when the unit is turned on.

How can you tell if you own one of these hair dryers? Just look at the serial number on the bottom of the handle. If the first number is a 6, and the last number is a 2, 3, or 5, then you will know you have purchased a defective hair dryer. If you own one of these hair dryers, please take it to the store where you purchased it or any store where Conair products are sold. There they will either replace the hair dryer or refund Your money in full.



Subjects receiving the social responsibility material were presented with an advertisement that contained this addition:

Although the government has not yet decided if this error is serious enough to force a recall, we at Conair feel it is our responsibility to you to offer only safe, quality products, and to act quickly to correct any problems that do occur.

Filler material surrounded each advertisement or news story to add a sense of realism and make the purpose of the study less obvious to subjects. This filler material was the same for all tape conditions and for all print conditions, but it differed between media. In the tape medium the filler material consisted of an advertisement for ladies' clothing and a news brief of local and national events. The local news was concerned with a car accident, and the national views related congressional response to a presidential budget request. The filler material on the print medium consisted of a story about a missionary's work in Africa and an advertisement for a political candidate. While the different type of filler material was confounded across the media, no reason existed to expect that it would influence the manipulations.

Following the presentation of the experimental materials, each subject was given a self-report questionnaire containing multiple choice and Likert-type questions. Table 1 presents the dependent variables. The scales were six-point scales anchored by Favorable/Unfavorable, Dangerous/Not Dangerous, and so an.


Because multiple dependent variables were employed, a procedure was followed to combine them into several distinct indices. In the first step a within-cell correlation matrix was constructed This matrix was analyzed using a principle components factor analysis with varimax rotation. Three factors were obtained with eigen-values greater than one. The factors accounted for 72 percent of the total variance.

The first factor was composed of three dependent variables--the favorability of the perception of Conair, the social responsibility of the company, and the extent with which the recall would deter the person from buying another Conair product. The factor was labeled perception of the company." Coefficient alpha was computed for the index and was .84.

The second factor was composed of two dependent measures--the objectivity of the presentation of the information and the trustworthiness of the source of information. The coefficient alpha was somewhat low for the index, 1 s .48. The factor was labeled "objectivity of message and source."

The third factor was composed of three scales--the certainty of the perception of the company, the danger of the defect and the responsibility of the company for the defect. The coefficient alpha for the scales was .54. The factor was labeled perception of the defect.While the coefficient alphas for factors two and three were lower than desired, for exploratory research they were marginally satisfactory.

Each of the factors was then analyzed via multivariate analysis of variance. (Table l presents the means of each of the scales across the main effects of the three factors design.) A number of significant effects were found. For Factor l (the perception of the company) a main effect was found for whether or not social responsibility information was provided in the message (Wilk's Criterion, F[3, 109] - 4.4, p < .001). The univariate analyses revealed significant effects for both the social responsibility of the company and the favorability of the impression of the company (ps < .01 ) . The means showed that when social responsibility information was given, the company was perceived more favorably and as more socially responsible than when it was not given. No other significant main effects or interactions were obtained for Factor 1. These results supported Hypothesis 1.

A number of multivariate effects were found for Factor 2 (Objectivity of message and source). A main effect for media occurred (Wilk's Criterion, F[2, 110] = 3.2, p < .05) as well as a main effect for the source of information (Wilk's Criterion, F[2, 110] = 5.4, p < .01). In addition, an interaction was found between the media and social responsibility information (Wilk's Criterion, F[2, 110] = 3.6, p < .05).

The univariate analysis revealed that the significant source effect occurred on the objectivity of source dependent variable (F = 10.9, p < .005). As was predicted, the government was viewed as more objective in the presentation of the recall information (p < .001). This outcome supported Hypothesis 2 of the study.

Univariate analyses also slowed that on the dependent variables of source trust and source objectivity a media effect was found (p < .05 for trust, p < .07 for source objectivity). The means revealed that in each instance the print media outperformed the sound medium. These results supported Hypothesis 3.

A significant interaction was found for Factor 2 in which whether or not social responsibility information was given interacted with the medium over which the information was received. The significant MANOVA effect appeared to result principally from the dependent variable of trust (p < .02). The pattern of means revealed that the trust in the source did not change significantly when social responsibility information was varied, if the print medium was used. However, if the sound medium was used, trust decreased when no social responsibility information was given in the message.

For Factor 3 (Perception of Defect) the only significant effect was a media by source of information interaction (Wilk's Criterion F[3, 108] - 3.5, p < .02). The overall MANOVA effect appeared to largely be influenced by the "certainty' dependent variable. The certainty of the perception of the company did not change when the source of information was varied, if the print medium was used (p < .01). However, when the sound medium was used, the certainty of the perception of the company was lower if the company sponsored the message. While univariate analyses revealed that the interaction was not significant on the other dependent variables making up the index (Danger of defect and responsibility for defect), the patterns of means were similar to those found for "certainty of perception." That is, differences across levels of source were minimal when the print medium was used. However, possible differences existed when the sound medium was used. Thus, the danger of the product and the responsibility of the firm were viewed as marginally higher when the company was the source of information rather than the government.


The results generally supported the hypotheses. When the company was described as socially responsible, whether by itself or by the government, more favorable feelings were held toward the company. This result indicates that a company involved in a product recall may be able to "pat-itself-on-the-back" in corporate advertisements and achieve a favorable response from consumers. Such action could act to mitigate some of the potential negative reactions expressed by consumers after receiving unfavorable corporate information. The second hypothesis stated that the government news source would be viewed as more trustworthy and objective than the company advertisement. This hypothesis also was supported. The MANOVA combining the dependent variables of objectivity and trust of source was significant. The univariate analysis revealed a strong effect such that subjects perceived the recall information to be presented in a more objective manner when it was presented as a government news release than as an advertisement. From a corporate perspective, these results indicate that, all else equal, the government is a better source of 'positive" information about a company. However, if the government cannot be persuaded to give positive information, the company can "pat-itself-on-the-back," as shown through support for Hypothesis l.

The third hypothesis was also supported by the data. The print medium was viewed as more trustworthy and marginally more objective than the tape medium. In addition, the print medium produced more positive responses concerning Conair's social responsibility than the tape medium. The print medium seemed to be more effective than the tape medium, a finding that does have support in the mass communications literature. These results suggest the use of corporate advertisements in news magazines as a potential medium for the message. Of course, television should also be examined. While not investigated in the present research, Johnson and Johnson, Inc., made effective use of television in defusing the negative effects of the Tylenol recall.


The study has several limitations. One is that respondents were required to role play. The use of role playing in experiments has been criticized. However, researchers have argued that role playing can be effective in capturing the decision process of individuals (Forward, Canter, and Kirsch 1976). If the situation cannot easily be created in an experiment and if the subjects have familiarity with the situation under investigation, role playing can be a useful methodology (Hansen 1972). The use of the hair dryer as the product was felt by the authors to he Lp insure subject familiarity with the role playing situation.

A second limitation of the study is that student subjects were used, caution is advised in generalizing the results to other populations. However, the sample did represent a relevant population due to the frequent use of hair dryers by student 3 of both sexes.

A third problem in the study concerned the low coefficient alphas for factors two and three of the study. Additional work is required to develop internally reliable measures of how people view product defects (Factor 3) and of the manner in which people assess objectivity, trustworthiness, and such of companies (Factor 2).

Future Research

Much work remains on investigating the consumer impact of product recalls. Most needed is a longitudinal study of consumer perceptions of companies which must make product recalls. How attitudes vary over time after a product recall occurs has not been reported in the academic literature.

A second area of need is an overall model of how recalls affect consumers. Such a model should be fit into the broader literature on the effects of negative communications and on the attribution of responsibility for negative outcomes. Previous research has not done a satisfactory job of integrating the various findings into the marketing or psychological literature. Finally, more attention needs to be given to measurement issues.


Clarke, Peter, and Lee Ruggels (1970), Preferences Among News Media for Coverage of Public Affairs," Journalism Quarterly, 47, 464-471.

Dommermuth, William P. (1974), "How Does the Medium Affect the Message," Journalism Quarterly, 51, (2),

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

Forward, John, Rachelle Canter, and Ned Kirsch (1976), "Role-Enactment and Deceptive Methodologies, Alternative Paradigms?" American Psychologist, 31 (August), 595-604.

Hansen, Flemming (1972), Consumer Choice Behavior: A Cognitive Theory, New York: Free Press.

Kanouse, David E., and L. Reid Hanson (1972), "Negativity in Evaluations," In E. Jones et al. (Eds.), Attribution: Perceiving Causes of Behavior. Morristown, N.J.: General Learning Press.

Kelley, H. H. (1972), "The Process of Causal Attribution," American Psychologist, 28, 107-128.

Lee, Raymond H. (1978), "Credibility of Newspaper and TV News," Journalism Quarterly, 55, 282-287.

Mizerski, Richard W. (1975), "A Test of the Relationship Between Trait and Causal Attribution," in Mary Jane Schlinger (Ed.), Advances in Consumer Research, Vol. 3, Proceedings of the Association for Consumer Research, 471-479.

Mowen, John C. (1979), "Further Perceptions on Consumer Perceptions of Product Recalls," in H. Keith Hunt (Ed.), Advances in Consumer Research, Vol. 7, Proceedings of the Association for Consumer Research, 519-523.

Mowen, John and H. W. Ellis (1981), "The Product Defect: Managerial Considerations and Consumer Implications," Review of Marketing 1981, Ben Enis and Kenneth Roering (Eds.), American Marketing Association, Chicago, 158-172.

Mowen, John E., David W. Jolly, and Gary C. Nickell (1980), "Factors Influencing Consumer Responses to Product Recalls: A Regression Analysis Approach," Will appear in Advances in Consumer Research, Proceedings of the Association for Consumer Research.

Sandage, C. H., and Arnold M. Barban (1970), "What Farmers Think About Consumer Protection," Journalism Quarterly, 47, 47-48.

Simpson, W. Gary and J. C. Mowen (1981), "The Reaction of Stock Prices to Information on Product Recalls,' Paper presented at the Southwestern Finance meetings.

Sternthal, Brian, Lynn Phillips, and Ruby Dholakia (1978), "The Persuasive Effect of Source Credibility," Public Opinion Quarterly, 286-314.

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

Warner, H. W. (1977), "Guidelines for Product Recall," Public Relations Journal, 33, (July), 11-13.

Wynne, James and G. E. Hoffer (1976), "Auto Recalls: Do They Affect Market Share?" Applied Economics, 8, 157-163.



David W. Jolly, Oklahoma State University
John C. Mowen, Oklahoma State University


NA - Advances in Consumer Research Volume 12 | 1985

Share Proceeding

Featured papers

See More


P1. Constructed Preferences in Time-Money Tradeoffs: Evidence for Greater Violation of Procedural Invariance for Time as Opposed to Money Elicitations

Nazli Gurdamar Okutur, London Business School, UK
Jonathan Zev Berman, London Business School, UK

Read More


M1. How Rewarding is Your Rewards Program? Experiential vs. Material Rewards

Ayalla Ruvio, Michigan State University, USA
Farnoosh Khodakarami, Michigan State University, USA
Clay Voorhees, Michigan State University, USA

Read More


P5. Can(Can’t) Control, thus Try to Save (Earn): The Joint Effect of Perceived Control and Financial Deprivation on Financial Decisions

Min Jung Kim, Manhattan College

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.