Components of Consumer Reaction to Company-Related Mishaps: a Structural Equation Model Approach

ABSTRACT - The study addressed in this paper examines the relationship of the circumstances underlying a company-related mishap and company management's communicated response to the mishap as factors influencing consumer reaction. A structural equation model is run on experimental data to test the interrelationship of cognitive, emotional, and attitudinal components of consumers' reactions. The stated likely cause of an incident is found to have a more profound effect on the consumer than does company response. Consumers' judgments of company responsibility for an incident are shown to combine with consumers' emotional reactions in serving as predictors of attitudes and behavioral intentions.

Citation:

Brian K. Jorgensen (1996) ,"Components of Consumer Reaction to Company-Related Mishaps: a Structural Equation Model Approach", in NA - Advances in Consumer Research Volume 23, eds. Kim P. Corfman and John G. Lynch Jr., Provo, UT : Association for Consumer Research, Pages: 346-351.

Advances in Consumer Research Volume 23, 1996      Pages 346-351

COMPONENTS OF CONSUMER REACTION TO COMPANY-RELATED MISHAPS: A STRUCTURAL EQUATION MODEL APPROACH

Brian K. Jorgensen, Humboldt State University

ABSTRACT -

The study addressed in this paper examines the relationship of the circumstances underlying a company-related mishap and company management's communicated response to the mishap as factors influencing consumer reaction. A structural equation model is run on experimental data to test the interrelationship of cognitive, emotional, and attitudinal components of consumers' reactions. The stated likely cause of an incident is found to have a more profound effect on the consumer than does company response. Consumers' judgments of company responsibility for an incident are shown to combine with consumers' emotional reactions in serving as predictors of attitudes and behavioral intentions.

INTRODUCTION

While most companies are constantly striving to become better known to their customers, some types of publicity are clearly undesirable. When disaster or scandal hits a company, consumers may very well choose to head in another direction. In a June 1993 survey of more than 1,000 Americans conducted by public relations consultants Porter/Novelli, a majority of respondents indicated that wrongdoing by a company is likely to affect their purchase of the companies' products or services. Further, these respondents had good memories for companies involved in crisis situations, even when the crises were several years old. When asked to name specific incidents, 51% mentioned the Exxon Valdez oil spill of 1989, 22% mentioned Sears' auto repair difficulties of 1992, and 11% mentioned the Tylenol product tamperings that took place in 1982.

Although crisis prevention is now an important function for many companies, company-related mishaps of all types continue to happen. In fact, some suggest that the number of serious incidents appears to be growing (Carey 1991; Lerbinger 1986). In recent months or years, we have witnessed allegations of tainted food and beverage, crashes of trains and aircraft, and a major snafu involving the accuracy of a well-known computer chip. The ongoing nature of these types of incidents, the likelihood that they will receive high levels of media attention, and the seriousness of their impact on consumers suggest the need for continuing research on how consumers respond to companies in crisis. The focus of the study presented here is on understanding the interrelationship among consumers' cognitive, attitudinal, and behavioral reactions to a company in crisis and the role that the company's communicated response to the crisis may play in influencing consumers' processing of the incident.

BACKGROUND AND HYPOTHESES

While general treatments of crisis management are varied and abundant, empirical examinations of consumer response to company crisis situations are few and very recent. Since consumer response to company crisis is a fairly new field, no clear consensus has yet been reached as to what might be the most important features of these situations for study. Some suggestions include: the reputation of the firm or brand (Romeo, Weinberger, and Antes 1994; Siomkos and Malliaris 1992), the performance history of the company with regard to product-related problems (Griffin, Babin, and Attaway 1991), the nature of commentary by external parties (Siomkos and Malliaris 1992), credibility cues relating to the communication of the incident (Griffin, Babin, and Attaway 1991; Romeo, Weinberger, and Antes 1994), and the severity of the incident (Romeo, Weinberger, and Antes 1994). Although at least some support has been found for the relevance of each of these factors, this paper focuses on two other factors that have been found to have substantial impact on consumer reaction, namely, the causal circumstances underlying the company-related incident and the nature of the company's communicated response.

Causal Circumstances Surrounding a Company-Related Crisis

When a company-related crisis, disaster, or mishap takes place, the first question on everyone's mind is "Why?". Investigators comb the site, bodies are examined or autopsied, the "black box" is recovered and scrutinized so that the cause can be understood and future mishaps avoided. Members of the public seek causal information to guide their judgments, feelings, and behavioral reactions toward parties involved in the incidents (Weiner 1986).

Weiner's (1986) research indicates that once a person has attributed a particular cause to an incident, this causal attribution may be expected to have particular consequences based on the attribution's position along three dimensional continua. The three dimensions of causality are locus, controllability, and stability. The locus of an incident is the degree to which the incident, action, or occurrence is internal to or external a person or company. Controllability describes the extent to which the incident is controllable by the target person or company. An incident whose cause is internal to a company, such as a machinery malfunction, may not necessarily be considered completely controllable by the company. In general, however, company-related incidents with external causation are likely to be considered uncontrollable, while incidents with internal causation are more likely to be considered controllable. The third dimension, stability, concerns the extent to which an incident is considered likely to recur. Although issues surrounding the stability of the cause may be important in some company mishap situations, the stability dimension is not examined in the study presented here.

According to Weiner's (1986) theory, the point at which a particular cause lies along each of the causal dimensions gives rise to particular emotions and cognitions. Behavioral intentions flow from these thoughts and feelings. The controllability dimension, which is particularly important to cases of company mishap or wrongdoing, influences judgments of responsibility and emotions of anger and sympathy. Folkes (1984; Folkes, Koletsky, and Graham 1987) found that when product failure was viewed as controllable by the company, consumers expressed more anger than when the failure seen as uncontrollable. Causal attributions and feelings of anger combined to influence consumer intentions to complain about the failure and to repurchase the product.

As part of an experimental study of single versus multiple explanations for negative company-related events, Jorgensen (1994) compared consumer reactions to incidents based on internal/controllable causes and external/uncontrollable causes. The study showed that locus and controllability had a significant effect of judgments of responsibility, feelings of anger, and intentions to purchase. Griffin, Babin, and Attaway (1991) found a significant effect of what they termed "locus of responsibility" on attitude toward a (fictional) company in an experiment involving a food poisoning incident. They also found that locus of responsibility had a marginal effect on purchase intentions toward the company. Using an experimental factorial survey approach to examine factors influencing consumer response to negative product safety information, Romeo, Weinberger, and Antes (1994) found that the degree to which a company was said to have tested a product before introduction affected all three of their dependent measures: brand opinion, purchase risk, and purchase intention. Although not a manipulation of locus or controllability per se, this examination of company culpability evidences consumers' use of responsibility judgments in responding to companies.

H1: Consumers will find a company more responsible for a negative incident that is viewed as internal to and controllable by the company than for a negative incident that is viewed as external to and uncontrollable by the company.

H2: The more responsible a company is judged to be for a negative incident, the more negative consumer emotions will be toward the company.

Company Communicated Response to a Negative Incident

When a company finds itself involved in a negative incident, it is likely to use impression management to formulate a response (Ginzel, Kramer, and Sutton 1992; Russ 1991). The impression management literature describes five types of "account" that may be used to communicate a response to a difficult situation: confession, excuse, justification, denial, and silence (Sch÷nbach 1980; McLaughlin, Cody, and O'Hair 1983). Of these responses, denial and variations on confession have been most often examined in the company crisis literature. These two extremes on the response scale are also addressed in the study presented here, with the intention of pursuing other forms of response in future research. While confession embodies, at least to some degree, an acceptance of blame, denial is an attempt to shift blame as far from the suspected party as possible (Schlenker 1980). To the extent that the party is indeed not responsible or that the situation is sufficiently ambiguous with respect to responsibility, a denial may be an effective form of reducing blame (Weiner, Graham, Peter, and Zmuidinas 1991).

H3: A company that confesses and apologizes for a negative incident will be judged to be more responsible than a company that denies responsibility for the incident.

Although judgments of company responsibility have generally not been addressed in the consumer behavior literature, attitudinal responses have been addressed (Griffin, Babin, and Attaway 1991; Romeo, Weinberger, and Antes 1994; Siomkos and Malliaris 1992). In the first two of these papers, the authors found that a proactive company response in which product safety problems are redressed has a more positive effect on brand attitudes than a denial or no comment response. Similarly, Siomkos and Malliaris found that voluntary recall of a potentially dangerous product leads to more positive brand attitudes than either denying responsibility or involuntarily recalling product. Impression management studies outside of the crisis management domain have found that confession, including apology and redress, has a softening effect on people's feelings and attitudes (Darby and Schlenker 1982; Weiner, Graham, Peter, and Zmuidinas 1991).

H4: Consumer emotions will be less negative and attitudes will be more positive when a company confesses and apologizes for a negative incident than when a company denies responsibility for the incident.

The crisis management literature recommends that a company's response to a crisis be prompt (Garbett 1988). Although consumer behavior research has not addressed the promptness of company response, the promptness recommendation seems sound in light of real-world experience. Johnson & Johnson was praised for its prompt action following the Tylenol poisonings of 1982, while Exxon received a critical drubbing for its slowness to act following the Valdez oil spill. More generally, Kremer and Stephens (1983) found that following a negative event, mitigating information delivered promptly was more effective in reducing retaliation than the same information delivered after a delay.

H5: Consumer emotions will be less negative and attitudes will be more positive when a company responds immediately to a negative incident than when a company delays its response.

Although the timing of mitigating information is addressed in both Kremer and Stephens (1983) and Weiner, Graham, Peter, and Zmuidinas (1991), the timing of potentially aggravating information, such as a denial of responsibility, is not addressed. While timing of response may have a different impact for denial of responsibility than confession of responsibility, no interaction is hypothesized here. Possible interactions are, however, investigated in the data analysis.

The Effects of Judgments and Feelings on Punitiveness and Behavioral Intentions

Consumer judgments, feelings, and attitudes are most interesting when shown to have an influence on consumer behaviors. While the studies of consumer reaction to company crisis cited above found effects of crisis situation and company response on measures of attitudes, they did not find significant effects on measures of purchase intentions. This failure to find significant results may have been due to the fact that the researchers examined only direct effects. Structural equation modeling allows for the investigation of both direct and indirect effects. The hypotheses in this section address the effects of judgments, feelings, and attitudes on punitiveness toward the company and purchase and investment intentions.

In her work on consumer attributions, Folkes (1984; Folkes, Koletsky, and Graham 1987) found that both attributional dimensions and emotions could have an effect on consumers' behavioral intentions toward a company. In the present study, traditional measures of attitudes, emotions, and intentions are supplemented by a measure of "punitiveness," the extent to which consumers feel that a company should be pardoned and/or punished for a mishap. Punitiveness is viewed here as an attitudinal construct with both affective and cognitive elements to it. Thus, judgments of responsibility along with negativity of emotions and attitudes should influence levels of punitiveness. In turn, the degree of punitiveness, combined with more general emotions and attitudes should impact consumer purchase and investment intentions. In fact, purchase and investment behaviors may be some of the most obvious means by which consumer can punish an errant company.

H6: The more responsible a company is considered to be for a negative incident and the more negative emotions and attitudes are toward the company, the more punitive consumers will feel toward the company.

H7: The more punitive consumers feel toward a company involved in a negative incident and the more negative their emotions and attitudes are toward the company, the lower will be intentions to purchase from or invest in the company.

METHODOLOGY

The data for this analysis were collected experimentally, using a convenience sample of adult consumers who were contacted at their places of employment and various other public places. A total of 129 usable questionnaires were filled out. Average age of the subjects was 36, and 60% of the subjects were female. Each questionnaire consisted of a cover page with instructions and a short vignette describing a company-related mishap. The vignette was followed by two pages of dependent measures, manipulation checks, and sample description questions. Each subject saw only one vignette, so the study design was completely between subject.

To increase external validity, variations on two company-related situations were used for the experimental vignettes. Thus, the study had a 2x2x2x2 design, the four factors being: company situation, causal attribution supplied, company response, and timing of company response. One of the company situations that was used addressed an airline accident in which all passengers and crew had been killed, while the other situation addressed a string of illnesses and deaths resulting from the taking of an over-the-counter drug. The study was designed to be collapsed across scenario if the pattern of results was the same for each scenario. Various scenarios had been pretested to use for the vignettes. The chosen situations had one relatively internal and controllable cause and one relatively external and uncontrollable cause. Causes that were judged as very near the endpoints of the locus and controllability scales were rejected so that some ambiguity with respect to the degree of company responsibility could be preserved. Real-world incidents are generally ambiguous in that while they are generally not intentional, one can always conceive of how by doing more the company might have forestalled the mishap.

Both the airline accident and the drug poisoning vignettes were printed and presented like short newspaper stories. Although subjects were informed that the companies and situations were fictional, they were asked to respond as though the incidents were real. For all vignettes, a description of the incident was followed by a statement to the effect that "investigations into the [incident] indicate that it was most likely caused by [cause]." Internal/controllable airline accident and drug scare causes were, respectively, "serious error made by a poorly trained pilot" and "machinery malfunction resulting in the wrong combination of chemicals." External/uncontrollable causes were "a severe, unpredictable storm that hit the aircraft mid-flight" and "product tampering that took place while the product was on store shelves."

Two elements of company response were examined: the content of the response and the timing of the response. In terms of the timing of response, each vignette reported either that the company had made its response immediately or after a delay of ten days. With respect to the content of the response, each vignette included either a confession condition or a denial condition. The confession condition stated, "We at [company] fully accept our responsibility in this matter. We are very sorry and express our deep-felt apology to the victims and their families. We will do all that we can to compensate them for their loss." The denial condition stated, "We at [company] do not feel that we are responsible for this incident. These kinds of incidents seem to happen periodically regardless of what kinds of precautions are taken."

After reading the vignette, subjects responded to a number of dependent measures and manipulation checks measured on seven-point scales. Multiple questions were asked for each construct with the exception of the behavioral intentions. Two separate question orders were used. The dependent measures addressed the following constructs: judgment of degree of company responsibility ("not at all responsible - very responsible," "very much to blame - not at all to blame," "very controllable - not at all controllable"), anger toward the company ("very annoyed - not at all annoyed," "not at all angry - very angry"), sympathy toward the company ("very sorry - not at all sorry," "not at all sympathetic - very sympathetic"), attitude toward the company ("very favorably - very unfavorably," "very bad - very good"), extent to which the company should be punished ("heavily fined - not fined at all," "no punishment - heavy punishment"), extent to which the company should be forgiven ("not at all forgiving, - fully forgiving," "full pardon, no pardon"), purchase intention ("very unlikely to choose - very likely to choose"), and investment intention ("very likely to invest in - very unlikely to invest in"). Eventually, measures of anger were combined with reverse-scored measures of sympathy to arrive at a measure of negative emotion, and measures of punishment were combined with reverse-scored measures of forgiveness to arrive at a measure of punitiveness. Other multiple measures were also designed to be averaged for the analysis.

The dependent measures were followed by manipulation checks investigating perceptions of the extent to which the company's response was "prompt" versus "delayed" and the extent to which the company accepted "full responsibility" versus "no responsibility." Finally subjects were asked to respond to manipulation checks regarding the locus, controllability, and stability of the likely cause given for the incident. These manipulation check items were anchored by "wholly internal to [company] - wholly external to [company]," "entirely controllable by [company] - entirely uncontrollable by [company]," and "a type of cause that is very likely to occur again" - "a type of cause that is very unlikely to occur again."

RESULTS AND DISCUSSION

Examination of the data using scenario as a factor in analysis of variance (ANOVA) computations showed the pattern of results to be substantially the same across the two company scenarios. Therefore , the results reported here are collapsed across the two companies. Direct relationships between manipulated variables and dependent measures were preliminarily investigated using ANOVA. A structural equation model was then developed, using EQS, so that both direct and indirect relationships among variables could be examined.

Manipulation Checks

Analyses of the manipulation checks using ANOVA showed the checks to all be successful. The response given by the company (confession or denial) had a significant effect on the manipulation check item for how much responsibility the company had accepted (F(9,123)=498.05, p<.001). The timing of management's response significantly effected the promptness manipulation check item (F(9,124)=89.50, p<.001). The causal attribution given had a significant effect on locus (F(9,121)=130.24, p<.001) and controllability (F(9,121)=103.96, p<.001). No manipulation of the stability construct was attempted in the study, and none of the manipulated variables had a significant effect on the item measuring stability.

TABLE

SIGNIFICANT EFFECTS OF THE CAUSAL ATTRIBUTION MANIPULATION

Reliability of Aggregated Measures

Acceptable coefficient alpha scores were found for each of the groups of measures that were aggregated. The score for the responsibility variables was .87, for the negative emotion variables was .80, for the attitude variables was .81, and for the punitiveness variables was .87.

Analysis of Variance Results

Prior to running the structural equation model, ANOVA computations were performed for each of the dependent measures to investigate direct effects and the possibility of interactions. Only one significant interaction was found, as described below. No significant main effects of the timing of company response were found. The likely cause of the incident given in the vignette had a significant impact on each of the dependent variables, as shown in the Table.

Significant main effects for company response were found with respect to the measures of negative emotion (F(1,129)=5.09, p<.05) and attitude (F1,129)=9.10, p<.01). The mean level of negative emotion in the confession conditions was 4.64 and in the denial conditions was 5.13. A significant interaction of company response and timing of the response was found for the attitude measure (F1,129)=5.96, p<.05). While confession led to more positive attitudes than did denial, delayed confession led to the most positive attitudes (mean=3.58). The mean attitudes for immediate confession, immediate denial, and delayed denial were 2.70, 2.58, and 2.46 respectively.

Hypotheses

The results of the structural equation model run on the data are summarized in the Figure. Overall, the model has a good fit with a chi-square value of 35.03 based on 27 degrees of freedom. The probability value for this chi-square statistic is 0.138, indicating that the model cannot be rejected at the .05 level. Further, the model has a Bentler-Bonett Normed Fit Index value of 0.946 and a Comparative Fit Index value of 0.987. Fit indices above the .90 level are generally considered as indicators of a good fit (Hu and Bentler 1995).

Each of the significant paths is labeled with its standardized path coefficient. All paths that are shown are significant at the .05 level. Non-significant paths, including paths that had originally been hypothesized, are not shown. Also, because the timing of management response did not have any significant effect, it was dropped from the model. Each of the individual hypotheses is briefly addressed below.

Hypothesis 1. Because degree of controllability and perceived company responsibility are practically synonymous, the strong path between causal attribution given and responsibility is essentially a manipulation check. The more internal/controllable the incident, the more responsible a company is considered to be.

Hypothesis 2. As predicted by Weiner's (1986) model, the degree of controllability and responsibility is highly linked to negative emotions. Anger toward a company is higher and sympathy is lower when the company is considered to be highly responsible.

Hypothesis 3. Confessing responsibility for an incident was hypothesized as leading to higher levels of perceived responsibility. Although the path between company response and judgments of responsibility was not significant at the .05 level, and thus does not appear in the Figure, it was marginally significant and in the expected direction. Therefore, confessing responsibility may be a two edged sword, on the one hand leading to higher levels of perceived responsibility while, on the other hand, leading to a softening of negative emotional reactions, as addressed by Hypothesis 4.

Hypothesis 4. It was expected that a company's confessing and apologizing would reduce the level of negative emotions felt toward the company while improving attitudes toward the company. However, the path to emotions was significant, while the path to attitude was not. Still, the path between emotions and attitudes was very strong, suggesting a strong indirect effect of confession and apology on attitudes.

Hypothesis 5. As mentioned above, the paths between timing of company response and emotions and attitudes were not significant.

Hypothesis 6. Punitiveness of consumers toward a company involved in a serious mishap was expected to be affected by judgments of responsibility along with emotions and attitudes. Although responsibility and negative emotions were found to play a role, general attitudes did not have an effect on punitiveness. In fact, as illustrated in the Figure, attitudes proved to be a dead-end variable, with no significant effects on punitiveness, purchase intentions, or investment intentions.

Hypothesis 7. Emotions, attitudes, and punitiveness were hypothesized as influencing consumers' purchase and investment intentions. Although degree of consumer punitiveness did influence purchase and investment intentions, emotions played a direct role only with regard to purchase intentions. As mentioned above, attitudes did not play a major role in either intention measure.

Summary and General Discussion

The results of this structural equation analysis lend credence to the general thrust of Weiner's (1986) attribution theory, which predicts that attributional judgments give rise to feelings and attitudes which, in turn, give rise to behavioral intentions. The research also shows the extreme importance of causal information on how consumers react to negative company-related incidents. Even in the face of somewhat ambiguous causal information, widely different company responses played only a minor, secondary role in influencing consumer reactions.

FIGURE

STRUCTURAL EQUATION WITH STANDARDIZED PATH COEFFICIENTS FOR SIGNIFICANT PATHS

Although timing of a company's response did not play its expected role in this study, one should not necessarily conclude that it is an irrelevant factor. Given the design of the study, the manipulation of response timing was probably the weakest of the manipulated elements. Subjects may have been able to envision a causal scenario with a particular management response, but they may have had a hard time imagining how a delay of response would influence their reactions. Also, in a real-world situation, a delay in response would probably be accompanied by the arrival of various other pieces of information that might influence consumer reaction to a greater extent than the delay.

The failure of the attitude measure to have an impact on the intention measures is initially unsettling to those who expect that feelings should lead to attitudes, which should then lead to intentions. However, the attitude measures here were very general measures of attitude toward the company. The measures of punitiveness that were taken here could be viewed as more specific measures of attitudes and, as such, more predictive of behavioral intentions (Ajzen and Fishbein 1977). Punitiveness was the only variable directly predictive of investment intention. This is understandable, given that an investment is more likely to be based on a rational analysis of a company's value and future than on more visceral concerns for personal safety. A company seems a less attractive investment if it is likely to be punished by regulatory agencies or groups of victimized consumers.

CONCLUSIONS AND FUTURE DIRECTIONS

The circumstances underlying a company-related mishap along with the company's communicated response to the mishap have implications for consumers' feelings, attitudes, and intentions toward the company. Through a structural equation model, the study presented here was able to demonstrate that previous studies' failure to find effects with regard to purchase intentions may have been due to these studies' non-examination of indirect effects. The study presented here was also able to place the relative effects of causal attribution and company response into perspective. Finally, this study compared the differences in factors influencing purchase intentions versus investment intentions following a negative company-related incident.

The research presented here still leaves a number of questions unanswered. One particular question that suggests the need for continued research is how much more important the role of company response is in situations that are more causally ambiguous. When causation is unclear, to what extent might a company's response be effective in changing consumers' perceptions of locus, causality, and responsibility? Other company responses that were not addressed here, such as excuses and justifications, should also be examined, particularly in light of more ambiguous situations, where blame might be more easily manipulated. Additionally, real crisis situations and real company names should be used to further test the external validity of the findings here.

REFERENCES

Ajzen, Icek and Martin Fishbein (1977), "Attitude-Behavior Relations: A Theoretical Analysis and Review of Empirical Research," Psychological Bulletin, 84 (5), 888-918.

Carey, John (1991), "Getting Business to Think about the Unthinkable," Business Week, June 24, 104-107.

Darby, Bruce W. and Barry R. Schlenker (1982), "Children's Reactions to Apologies," Journal of Personality and Social Psychology, 43 (October), 742-753.

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

Folkes, Valerie S., Susan Koletsky, and John L. Graham (1987), "A Field Study of Causal Inferences and Consumer Reaction: The View from the Airport," Journal of Consumer Research, 13 (March), 534-539.

Garbett, Thomas (1988), How to Build a Corporation's Identity and Project Its Image, Lexington, MA: Lexington Books.

Ginzel, Linda E., Roderick M. Kramer, and Robert I. Sutton (1992), "Organizational Impression Management as a Reciprocal Influence Process: The Neglected Role of the Organizational Audience," in Research in Organizational Behavior, Volume 15, eds. L. L. Cummings and Barry M. Staw, Greenwich, CT: JAI Press, 227-266.

Griffin, Mitch, Barry J. Babin, and Jill S. Attaway (1991), "An Empirical Investigation of the Impact of Negative Public Publicity on Consumer Attitudes and Intentions," in Advances in Consumer Research, Vol. 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT: Association for Consumer Research, 334-341.

Hu, Li-Tze and Peter M. Bentler (1995), "Evaluating Model Fit," in Structural Equation Modeling: Concepts, Issues, and Applications, ed. Rick H. Hoyle, Thousand Oaks, CA: Sage, 76-99.

Jorgensen, Brian K. (1994), "Consumer Reaction to Company-Related Disasters: The Effect of Multiple Versus Single Explanations," in Advances in Consumer Research, Vol. 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT: Association for Consumer Research, 348-352.

Kremer, John F. and Laura Stephens (1983), "Attributions and Arousal as Mediators of Mitigation's Effect on Retaliation," Journal of Personality and Social Psychology, 45 (August), 335-343.

Lerbinger, Otto (1986), Managing Corporate Crises: Strategies for Executives, Boston: Barrington Press.

McLaughlin, Margaret L., Michael J. Cody, and H. Dan O'Hair (1983), "The Management of Failure Events: Some Contextual Determinants of Accounting Behavior," Human Communication Research, 9 (Spring), 208-224.

Romeo, Jean, Marc G. Weinberger, and David Antes (1994), "An Investigation of the Communication Cues that Affect Consumers' Responses to Negative Product Safety News," working paper, Research Centre for Consumer Behaviour, The University of Birmingham, England.

Russ, G. S. (1991), "Symbolic Communication and Image Management in Organizations," in Applied Impression Management: How Image-Making Affects Managerial Decisions, eds. R. A. Giacalone and P. Rosenfeld, Newbury Park, CA: Sage.

Schlenker, Barry R. (1980), Impression Management, Belmont, CA: Wadsworth, Inc.

Sch÷nbach, P. (1980), "A Category System for Account Phrases," European Journal of Social Psychology, 10 (April-June), 208-224.

Siomkos, George J. and Peter G. Malliaris (1992), "Consumer Response to Company Communications During a Product Harm Crisis," Journal of Applied Business Research, 8 (Fall), 59-65.

Weiner, Bernard (1986), An Attributional Theory of Motivation and Emotion, New York: Springer-Verlag.

Weiner, Bernard, Sandra Graham, Orli Peter, and Mary Zmuidinas (1991), "Public Confession and Forgiveness," Journal of Personality, 59 (June), 281-312.

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