While the public might relish a good crisis in the media now and then, most company board members would shiver at the thought of becoming the object of a stream of negative news stories. Becoming the center of turmoil is dreaded as it may damage the company’s reputation and result in reduced market share (see Herbig et al. 1994; Wrigley et al. 2003). A key variable that frequently suffers from negative news is company trustworthiness, a variable that is easily destroyed and painfully slowly restored (Nakayachi and Watabe 2005; Kim et al. 2006, 2009; Parayitam and Dooley 2009). Moreover, research suggests that negative events such as product malfunctions, negative rumors, transgressions, or acts of dishonesty also adversely influence company evaluations and overt behavior of consumers and stakeholder groups toward the company (see Cialdini 2009). Hence, the issue of how to respond to adverse events and limit its damaging impact is a pivotal one for company survival and has been an issue of ongoing interest in the literature (e.g., Dirks et al. 2009; Folkes 1984; Ferrin et al. 2007; Folkes and Kotsos 1986; Gillespie and Dietz 2009; Kim et al. 2006, 2009; Kramer and Lewicki 2010; Nakayachi and Watabe 2005 Wooten 2009; Xie and Peng 2009).

Previous research has typically focused on the question how organizations should deal with negative events that are widely known and which strategies would best help to repair the damaged reputation. Since companies are usually aware of negative events (long) before they become known to the general public, a much earlier decision for them is whether to disclose the negative event themselves or risk discovery and disclosure by a third party. The most important factor determining their decision is probably their confidence in being able to keep a lid on the disaster. However, since such estimates are difficult to make and notoriously unreliable, it would certainly be important to know the extent to which the blow of such negative information on company-related consumer responses could be softened by disclosing it oneself rather than running the risk of third-party disclosure. Indeed, an effective disclosure strategy might well lessen the damage that news about a negative event may otherwise do to consumer judgment and decision making. To the best of our knowledge, this question has not received any research attention in the business and management literatures.

In this article, three studies will be presented that demonstrate that a company doing nothing more than revealing the bad news itself, as opposed to a third party, can substantially reduce the damage that this information would otherwise do to company reputation. We propose that the positive impact of this strategy of mere self-disclosure is due to its effectiveness in lessening some of the damage of the negative information to the perceived trustworthiness of a company and that, paradoxically, this effect is stronger for companies whose reputation was already poor beforehand.

Disclosing Negative Information

In order to pinpoint the spheres where mere self-disclosure can and cannot lessen the damaging impact of bad news, we need to define closely the types of consumer responses that might be affected. In an influential theoretical article, Mayer et al. (1995) draw a distinction between trust, trustworthiness and risk-taking. Trust is defined as “the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party” (p. 712). Trust in another party is determined by the perceived trustworthiness of that party or more specifically by the belief that that party possesses certain attributes, namely competence, benevolence, and integrity. As a consequence of trust, trustors “make themselves vulnerable” by taking risks which could result in loss, if the trust was misplaced. For example, trusting a used car dealer can result in loss if the car one buys turns out to have serious problems that were not disclosed by the dealer (see also Colquitt et al. 2007).

The model of Mayer et al. (1995) can be conceived of as a special case of the theory of reasoned action. According to Fishbein and Ajzen (2010), people’s willingness or readiness to engage in a behavior reflects their behavioral intention to engage in that behavior. Similar to Mayer et al. (1995), the intention to engage in behavior is determined by three types of beliefs, namely the beliefs that the other party is competent, benevolent, and integer. These beliefs determine the intention to take risk and engage in trusting behavior such as choosing or buying a good from that party. These trustworthiness beliefs, which are central aspects of a company’s reputation, can be seriously undermined by information that reflects lack of competence (e.g., that a whole product line has to be recalled due to construction or production faults) or lack of integrity (e.g., that stocks were sold by a bank despite their knowledge that the relevant company was going to fail in the near future).Footnote 1 In sum, both the Mayer et al. (1995) and the Fishbein and Ajzen (2010) models highlight the critical role of perceived trustworthiness as a precursor to trust and trust-related judgment and decision making as a consequence of it. Consequently, we will assess both trustworthiness beliefs and their downstream consequences (company evaluations, choice behavior) as key target variables that may be affected by the (self) disclosure of negative information.

Companies that are confronted with adverse events that are likely to seriously impact on their trustworthiness have various options at their disposal to restore trust, ranging from unambiguous “stonewalling” to unambiguous acknowledgement (see Dawar and Pillutla 2000). For instance, one may simply remain silent, deny the event, apologize, pass the buck, or engage in defensive impression management (e.g., Kim et al. 2004, 2006; Wan and Pfau 2004).

Alternatively, one may straightforwardly disclose the negative event without any refutation, bolstering, or other forms of “spin.” This is a risky strategy and it is important to know whether it would be effective in reducing the damage to perceived trustworthiness. Although to the best of our knowledge, this issue has not been studied in the context of disclosures of company failures, some relevant research has been conducted in a courtroom-setting. In a series of studies, Williams and colleagues (Dolnik et al. 2003; Williams et al. 1993) demonstrated that a defendant in a courtroom-setting revealing incriminating evidence about him/herself before the prosecution does, may benefit his/her cause. These researchers asked participants to act as mock jurors and to pass a verdict in both a criminal and a civil trial case. Participants were either exposed to a version of the trial in which the defendant himself (through a counselor) revealed incriminating evidence before the prosecution could (which Williams et al. 1993, term ‘stealing thunder’), or to one of two control versions of the trial (i.e., a ‘no thunder’ condition in which no negative information on the defendant was present, or a ‘thunder’ condition, in which the prosecution revealed something negative about the defendant). Results showed that employing a stealing thunder strategy positively affected juror perceptions of the defendant’s credibility and trustworthiness and resulted in reduced probabilities of a guilty verdict. In addition, such self-disclosure of negative information may not only help in the context of a trial, but also to control damage in political life. In one study by Baldwin (1992, reported in Williams and Dolnik 2001), a scandal involving a politician running for reelection was either disclosed by the politician himself, by a newspaper, or there was no mention of a scandal. Participants were asked how much they liked the candidate and how likely it would be that they would vote for him. Self-disclosure increased participant’s liking and voting willingness almost to the level of the control condition, where no scandal was mentioned.

These findings resonate well with research on the effectiveness of two-sided advertising. More in particular, Kamins and Assael (1987) contrasted a ‘classic’ two-sided message (where the negative information is subsequently refuted) with an alternative message where the advertiser presented negative information, but did not refute it (a message akin to mere self-disclosure in the present context). Both types of messages outperformed a one-sided message in terms of persuasion. More importantly, the two-sided messages did not differ in their effectiveness, suggesting that simply disclosing some negative attribute without refutation or any other kind of ‘spin’ may be effective in is own right (Fennis and Stroebe 2010). Hence, there is empirical evidence suggesting that mere self-disclosure of negative information may positively affect consumer judgment and decision making.

Why might the strategy of mere self-disclosure of negative information work? Interestingly, the findings reported above converge on the positive impact of mere self-disclosure on perceptions and behavior, but are less unequivocal on the underlying process responsible for that impact (see also Ward and Brenner 2006). Two complementary mechanisms might account for the positive effects of mere self-disclosure. First, speaking against one’s self-interest is likely to make the person appear more trustworthy and even likeable. More specifically, research by Eagly and colleagues (Eagly et al. 1978, 1981; Wood and Eagly 1981) points to an attributional account that suggests that message recipients maintain or develop expectations about a message based on the likely position the message source will advocate. In most cases, these expectations will be biased because (commercial and nonprofit) organizations are typically thought to take a self-serving position, or at least are thought to go at considerable lengths not to harm their own vested interests (cf. Campbell and Kirmani 2008). If the message confirms these expectations, recipients tend to form inferences that are in line with the biasing elements. Hence, they tend to discount both the message itself and the credibility of the message source as being low in trustworthiness. This explains why many of the crisis communication strategies outlined above –e.g., buck-passing, bolstering the negative news or apologizing—are sometimes ineffective, since they all serve to confirm the expectation of message recipients that the issue’s source is biased (but see Tomlinson et al. 2004; Kim et al. 2006). In contrast, if a company discloses information that clearly argues against its own self-interest, the message disconfirms the expected “reporting bias” (Eagly et al. 1978) and, as a result, people might infer that the company is honest. This might aid in protecting company trustworthiness and might positively affect evaluations of the company (see also Tomlinson and Mayer 2009 for an alternative perspective on the role of causal attributions in repairing trust).

Second, when a third party brings up damaging information about a company, people might infer that the company was trying to withhold this information. This does not only suggest lack of trustworthiness on their part but also indicates that the company thought the information was extremely damaging. Why else would the company try to hide it, when there was a considerable risk that others would find out anyway? Hence, message recipients might attribute more weight to the negative news when divulged by a third party, and less when disclosed by the company itself (cf. Dolnik et al. 2003; Skowronski and Carlston 1989). Interestingly, both mechanisms lessen the impact of damaging information by protecting the perceived trustworthiness of a company: The first mechanism (i.e., disconfirming the reporting bias) might buffer some of the damage to perceived integrity, whereas the latter (perceiving the issue as less severe) might buffer some of the damage to either perceived integrity or perceived competence, depending on the nature of the issue.

Obviously, the extent to which self-disclosure of a negative event can lessen the damage to the company’s reputation as a trustworthy party depends both on company reputation and on the nature and seriousness of the event. If the company has a long-standing reputation of high competence and integrity, and if the event is not too serious, it is unlikely to do serious damage to company reputation. In this case, there is either only minimal damage (if the event is not too serious) or the company’s strong reputation can make up for the negative impact on trustworthiness. Hence, it will probably not matter whether the event is disclosed by the company itself or by a third party. On the other hand, if the company already has a shaky reputation due to negative events happening in the past, then there is no shield of a positive reputation thus increasing the negative impact of the event on perceived trustworthiness. In addition, also the source of disclosure will now become more important, because by disclosing the event itself, the company can somewhat reinforce its reputation as a trustworthy party. Thus, we would expect that the impact of type of disclosure will be moderated by the reputation of the company about which the event is being disclosed for two reasons: First, because the negative event will be most damaging for a company that already has a poor reputation and second, because in terms of the explanation suggested by Eagly and her colleagues (Eagly et al. 1978, 1981; Wood and Eagly 1981), for companies with a poor reputation, self-disclosure might be particularly unexpected and consequently more effective in reinforcing perceived trustworthiness (cf. Román and Ruiz 2005; Walster et al. 1966). In contrast, if a company is already considered credible and trustworthy, self-disclosure is less unexpected and therefore less likely to further improve this perception. In contrast, a reputation of great trustworthiness is likely to act as a shield and protects the company against the suspicion that they knowingly and willfully tried to hide damaging information from the public. Such suspicion is more likely for a company which already suffers from a poor reputation (see Ahluwalia et al. 2000; Dawar and Pillutla 2000; Wooten 2009). Therefore, third-party disclosure should be more damaging for such a company than one that enjoys a very positive reputation.

The Present Research

We conducted three experiments to test these hypotheses. Study 1 manipulated type of disclosure to test Hypothesis 1 that self-disclosure of negative information about a company has a less negative effect than third-party disclosure. Studies 2 and 3 added company reputation as a second factor to test Hypothesis 2 that the impact of type of disclosure is moderated by company reputation, with self-disclosure being beneficial (or third-party disclosure damaging) for companies with a poor rather than a good reputation. In addition, whereas Study 1 focuses on a key downstream consequence of trust: consumer choice behavior (cf. Cialdini 2009), Studies 2 and 3 examined the impact of type of disclosure on company trustworthiness beliefs. Finally, Study 3 extended Study 2 by also including a measure of company evaluations and by testing Hypothesis 3 that the impact of self-disclosure on company evaluations is mediated by perceived trustworthiness.

Study 1

This study provides a first test of the impact of the strategy of company mere self-disclosure of negative information on consumer choice behavior. We expect mere self-disclosure of negative information to positively affect consumer choice compared to a condition where the same information is disclosed by a third party.

Method

Design and Participants

A total of 65 undergraduates at a large Dutch university (34 females, 31 males) with a mean age of 22.6 years (SD = 4.03) participated voluntarily in the present study that used a single factor between-subjects factorial design (mere self-disclosure vs. third-party disclosure) as part of a larger study. The key dependent variable consisted of a measure of consumer choice.

Procedure

Participants were randomly assigned to conditions, and were shown a (fictitious) message ostensibly downloaded from the website of a large and well-known newspaper and were asked to give their opinion on this news item. The article featured a negative issue associated with a (fictitious) pharmaceutical company (“AescuMed”) that marketed a drug to control cholesterol-levels. The article stated that research had shown that this drug was prone to a noticeable side effect, such that consumers might experience a headache when using the drug.

After they had read the newspaper article, participants were asked to complete the questionnaire containing manipulation checks, filler questions and the choice behavior measure. Finally, in this and the next two studies, participants were asked to indicate what the true objective of the experiment was. In all studies, none of the participants guessed the true objective of the study.

Disclosure of Negative Information

In the mere self-disclosure condition, the message stated that the company itself had called a press conference where the results of the investigation demonstrating the side effects were presented by the board of directors of the company. In addition, the header of the news article explicitly stated that it was “AescuMed” itself that had revealed the negative news. In the third-party disclosure condition, the message was identical except that the issue was exposed by an external investigating agency that had discovered the side effects, the (fictitious) “National Agency for the Registration and Evaluation of Medication Side Effects.”

Choice Behavior

At the end of the questionnaire, participants were given the opportunity to order a free CD-ROM, containing a “health-scan,” developed by the same company, Aescumed. The extent to which participants filled in the coupon to order the CD-ROM served as our measure of choice behavior.

Manipulation Checks

In this and the next two studies, the manipulation of type of disclosure was checked by a 5-point Likert statement assessing the extent to which participants were convinced that the company itself had revealed the incident (administered directly after the manipulation in Study 3, and at the end of the study in Studies 1 and 2). In all studies, full factorial ANOVA’s on this item consistently revealed that the type of disclosure manipulation was successful (Study 1: F (1, 63) = 35.64, p < .001; Study 2: F (1, 67) = 7.08, p < .01; Study 3: F (1, 70) = 9.35, p < .01).

Results and Discussion

Overall, 37 % of participants (24/65) chose to order the CD-ROM containing the health-check developed by the organization. A logistic regression with the (dichotomous) choice behavior measure as criterion and type of disclosure (mere self-disclosure vs. third-party disclosure) as predictor, showed that type of disclosure strongly affected overall choice rates. That is, 59 % of participants (20/34) exposed to a mere self-disclosure condition chose to order the CD-ROM, whereas 13 % of participants in the third-party disclosure condition did so (4/31; Wald(1) = 12.57, p < .001).

These results provide a first empirical demonstration of the effects of mere company self-disclosure of negative information affects actual consumer choice behavior. In support of Hypothesis 1, it was found that, compared to third-party disclosure, active self-disclosure lessened the damaging impact of negative news on a key consequence of trust: consumer choice behavior.

The next study will extend the present findings along two lines. First, in addition to the behavioral outcome measure employed in the present study, Study 2 will assess whether company trustworthiness beliefs are affected by type of disclosure of negative information. In addition, we will assess the moderating role of the company’s reputation in the effectiveness of mere self-disclosure.

Study 2

This second study was designed to assess whether the strategy of mere self-disclosure would affect the perceived trustworthiness of a company and whether this effect was qualified by the reputation of a company. We expected the impact of type of disclosure on company trustworthiness beliefs to be mainly observed for companies with a negative rather than positive reputation.

Method

Participants

Seventy-four undergraduate students were initially recruited to participate voluntarily in the study conducted at a mid-sized Dutch university as part of a larger research project. Three participants who failed to answer all questions were excluded from the analyses. This resulted in a total sample of 71 students (23 men and 48 women) with a mean age of 25.5 years (SD = 3.04).

Design and Procedure

Participants were randomly assigned to conditions in the 2 (self-disclosure vs. third-party disclosure) × 2 (reputation: negative vs. positive) between-subjects factorial design. The focal dependent variable consisted of a measure of company trustworthiness beliefs (cf. Mayer et al. 1995; Kim et al. 2006). Participants were requested to give their opinion about the contents of a printed version of an Internet site, hosted by a stock-trading bank and providing news and background information on various companies. Participants only read the information with regard to one (existing) company, presented as a manufacturer of paints and chemical products. The contents of the website included both types of disclosure manipulation and the manipulation of the reputation of the company. The negative information pertained to an environmental incident involving the company. After they had read the website, participants were asked to complete the questionnaire containing manipulation checks, filler questions and the trustworthiness measure.

Independent Variables

Disclosure of Negative Information

In both disclosure conditions, the nature of the negative news about the company was identical, only the source that revealed the critical incident varied. Participants read that the focal company was responsible for spilling a large amount of chemical waste into the Rhine, a large European river, resulting in considerable damage to the ecosystem. Importantly, no attempt at discounting the seriousness of the critical incident was made in either the self-disclosure or the third-party disclosure condition. In the self-disclosure condition, the message head explicitly stated that the company itself reported the incident. The body of the message continued by stating that the CEO of the company had revealed the news at a press conference, and announced that an investigation would be started to examine the causes of the incident. In the third-party condition, the message was identical except that the issue was exposed by an (existing) environmental action group as indicated by the header of the message and body copy that reported about a press conference held by a spokesperson of the action group.

Company Reputation

Following previous research on company reputation (Wooten 2009; Purohit and Srivastava 2001), this independent variable was manipulated in the present study by presenting participants with a section of the fictitious website (on the same webpage) that contained older news items about the focal company. Under this section, several headers were included, revealing either a history of corporate success or failure. In the positive reputation condition the news items pertained to satisfied shareholders, satisfactory results in line with expectations, winning a prize for innovation and becoming a market leader in a new geographical market. In the negative reputation condition, participants learned that a past shareholder meeting resulted in shareholder protests, that profit estimates had to be adjusted downward, that future financial prospects were dim, and that the appointment of a new CEO after dismissal of the current one was problematic and presented the company with the risk of a continuity problem.

Dependent Variable

Company Trustworthiness

In line with previous research (Gardberg and Fombrun 2002; Kim et al. 2006; Mayer et al. 1995), we measured trustworthiness beliefs about the company using eight 5-point Likert statements (1 = totally disagree, 5 = totally agree). These statements were as follows: 1. “≪name organization≫ is an honest company”; 2. “≪name organization≫ is a sincere company”; 3. “≪name organization≫ is a trustworthy company”; 4. “≪name organization≫ is a credible company”; 5. “I trust the quality of products of ≪name organization≫”; 6. “≪ name organization≫ is a company that regularly renews and innovates its products”; 7. “≪name organization≫ really believes in the quality of its products” and 8. “the products of ≪name organization≫ really show that ≪name organization≫ is up to date.” A factor analysis using varimax rotation yielded a 2-factor solution (R 2 = 65 %), with the first four items loading on the first factor and the remaining four on the second (all factor loadings >.50). Reliability of the 8-item measure was satisfactory, Cronbach’s α = .75. An index of perceived trustworthiness was created by averaging the scores on the items.

Manipulation Checks

In addition to the manipulation check for type of self-disclosure, in this and the next study the manipulation of company reputation was checked with a 5-point Likert statement assessing the extent to which participants perceived the company to have a history of turmoil and problems. Full factorial ANOVA’s on this item revealed that the company reputation manipulation was successful (Study 2: F (1, 67) = 6.64, p < .01; Study 3: F (1, 70) = 13.74, p < .001).

Results and Discussion

To test the hypothesis that self-disclosure of negative information contributes to lessening the damage done to the perceived trustworthiness of a company and that its impact is mainly observed when company reputation is negative, rather than positive, a full factorial ANOVA was conducted. This analysis yielded a significant main effect for type of disclosure (F (1, 67) = 4.90, p = .03, η 2 = .07) and a significant interaction effect between type of disclosure and reputation (F (1, 67) = 5.14, p = .03, η 2 = .07). In line with the previous results, self-disclosure promoted company trust (M = 2.81, SD = .53) compared to third-party disclosure (M = 2.54, SD = .47). More importantly, the interaction effect that qualified this main effect indicates that, in line with Hypothesis 2, the impact of type of disclosure is only observed when reputation is negative, rather than positive (see Table 1, 2 for means, standard deviations and correlations). For a company with a negative reputation, self-disclosing the negative information shielded company trustworthiness to a larger extent than third-party disclosure. For a company with a positive reputation, type of disclosure did not affect trustworthiness ratings, suggesting that the negative event did not damage the reputation of that company. Indeed, additional simple main effect analyses confirmed that the first contrast was significant (F (1, 67) = 11.49, p = .001, η 2 = .15), whereas the latter was not (F < 1).

Table 1 Means, standard deviations, and intercorrelations for Study 2
Full size table
Table 2 Means and standard deviations of company trustworthiness as a function of disclosure strategy and company reputation (Study 2)
Full size table

These findings extend the previous results and Hypothesis 1 by showing that active self-disclosure of negative news positively affects company trustworthiness in addition to consumer choice behavior. Moreover, these findings also show that the previous findings were not attributable to the specific type of company featured in Study 1. Of particular, import for our reasoning was the finding that the impact of type of disclosure strategy was moderated by the reputation of the company. Consistent with Hypothesis 2, the impact of self-disclosure was observed when the company’s reputation was negative but not when it was positive. The fact that type of self-disclosure appears to have made no difference for the company with the positive image suggests the negative event did not damage the reputation of that company. The finding that self-disclosure only affected company trustworthiness of the company with the initially negative reputation is consistent with the attribution account suggested by the research of Eagly and colleagues (Eagly et al. 1978, 1981; Wood and Eagly 1981). The fact that such a company decided to disclose the information reinforced a perception of trustworthiness compared to third-party disclosure and in this instance compensated for the negative impact of the damaging information.

The third study replicates and extends these results in three ways: First, the news story in Study 2 did not only mention an adverse event, but also stated that—in response to the event—an investigation would be started to examine the causes of the incident, which could have been perceived as a trust repair intervention. To rule out this message element as an alternate explanation of our findings, and to establish that the effects are indeed attributable to mere self-disclosure of negative information, the next study omitted any cues to a possible trust repair intervention. In addition, we used a more extreme manipulation of company reputation to test the limits of self-disclosure in repairing reputational damage. We also assessed self-disclosure effects on company evaluations (in addition to trustworthiness ratings) and examined whether any effects on evaluations are driven by company trustworthiness (Hypothesis 3).

Study 3

The third study was designed to replicate and extend the previous findings. First, the study included a different type of adverse event, committed by yet another (existing) company to provide converging evidence on the key finding that mere self-disclosure of negative information has positive (i.e., less damaging) effects compared to a situation where the same information is divulged by a third party (Hypothesis 1). However, in this study, we gave a more negative description of the company in the negative reputation condition than we did in Study 2, and we omitted any references to cues that might be perceived as trust repair strategies. This was intended to allow us to assess the limits of the effects of mere self-disclosure in buffering reputation damage but also to replicate the previous result that the impact of mere self-disclosure is mainly observed for companies with a negative, rather than positive reputation (Hypothesis 2). Third, in addition to a measure of company trustworthiness, the present study also included a measure of company evaluations to assess whether the presumed effects spill-over to affect evaluations. Finally, we extend previous findings by testing whether the any effects on evaluations are mediated by company trustworthiness (Hypothesis 3).

Method

Participants

In return for partial course credit, a sample of 77 undergraduate students was initially recruited to participate in the study conducted at a mid-sized Dutch university. After excluding three students of non-Dutch nationality who could not understand Dutch or otherwise failed to comply with experimental instructions (cf. Oppenheimer et al. 2009), the final sample consisted of 74 students (24 male, 50 female) with a mean age of 20.4 years (SD = 2.07).

Design and Procedure

The experiment formed part of a larger research project embedded in a sequence of unrelated studies conducted by other researchers in the lab facilities at the university. We used a 2 (mere self-disclosure vs. third-party disclosure) × 2 (reputation: negative vs. positive) between-subjects factorial design.

Participants were presented with a news website featuring a lead article and were requested to give their opinion on this news item. The article featured a report of considerable financial losses suffered by clients of a mid-sized insurance company because damage claims had been unduly rejected by the company. Participants were randomly exposed to either the mere self-disclosure or third-party disclosure condition. Furthermore, participants were either exposed to information conveying a negative or a positive company reputation. After they had read the article, participants were asked to complete a questionnaire containing manipulation checks, filler questions and the dependent measures of company trustworthiness beliefs and company evaluations.

Independent Variables

Disclosure of Negative Information

In all conditions, the news message stated that an investigation, spurred by complaints of a large number of consumers, had revealed that the company had rejected almost all claims for compensation of damage it had received, including reasonable and justifiable claims. This had led to considerable financial losses on the part of the clients of the company. In the mere self-disclosure condition, it was emphasized that the company itself had called a press conference where the results of the investigation were presented by the board of directors of the insurance company. In addition, the headline of the news article explicitly stated that it was the board of the insurance company itself that had revealed the incident. In the control condition, the message was identical except that the issue was exposed by the media source that reported the news item rather than the responsible company.

Company Reputation

The reputation of the company responsible for the negative information was varied by presenting participants (on the same website as the disclosure manipulation) with information on the corporate history that either spoke of success or failure. In the negative reputation condition, participants read that the company had a tainted history, with the largest share of dissatisfied customers in the market, and in the previous year had been checked and warned by the “Authority for Financial Markets” for financial misconduct. In the positive reputation condition, participants learned that the company had a pristine corporate history, with the largest share of satisfied customers in the market. In addition, the message stated that the company was awarded with the “Prix d ‘Or” for corporate excellence the previous year.

Dependent Variables

Company Trustworthiness

In line with Study 2, we used a similar 8-item measure to assess perceived trustworthiness. Because the company and the type of industry in the present study were different from Study 2 (in the present study, a financial service provider and in Study 2, a manufacturer of physical products), we adapted the measure to fit the specific company and type of industry. We measured trustworthiness beliefs about the company using 5-point Likert statements (1 = totally disagree, 5 = totally agree). These statements were as follows: 1. “≪name organization≫ appears honest to me”; 2. “≪name organization≫ appears sincere in its actions”; 3. “≪name organization≫ seems like a trustworthy company”; 4. “≪name organization≫ seems to have a lot of experience”; 5. “≪name organization≫ appears to have a lot of expertise”; 6. “≪name organization≫ is a company that regularly renews and innovates its services”; 7. “≪name organization≫ really believes in the quality of its services” and 8. “the services of ≪name organization≫ really show that ≪name organization≫ is up to date.” A factor analysis using varimax rotation yielded a 2-factor solution (R 2 = 59 %), with the first three items loading on the first factor and the remaining five on the second (all factor loadings >.57). Reliability of the 8-item measure was satisfactory, Cronbach’s α = .80. An index of perceived trustworthiness was created by averaging the scores on the items.

Company Evaluations

To prevent shared method variance, we used an alternative procedure to assess company evaluations. In line with dual process frameworks (Chaiken and Trope 1999), we administered a thought-listing procedure. Participants were asked to list the thoughts they had during message exposure in ten separate boxes (cf. Karmarkar and Tormala 2010). After having established inter-rater reliability, two coders blind to experimental hypotheses, independently and separately rated the valence of all message-related thoughts as positive (range: 0–4, M = .35, SD = .69; e.g., “it is good the company has had the guts to reveal the problem itself”) or negative (range: 0–5, M = .96, SD = 1.02; e.g. “how awful that such a company is capable of hurting so many clients”). Percentages agreement between both coders were satisfactory (positive thoughts = 81 %, negative thoughts = 92 %). A large body of research has reliably shown that the thought-listing procedure yields valid measures of evaluations for a host of objects, events and entities (e.g., Chaiken and Trope 1999; Eagly and Chaiken 1993; Fennis and Stroebe 2010; Petty and Cacioppo 1986). An index of company evaluations was created by subtracting negative from positive message-related thoughts, with higher scores indicating more positive evaluations.

Results and Discussion

Company Trustworthiness

To assess whether the reputation of a company qualified the effect of the type of disclosure on company trust, a 2 × 2 factorial ANOVA was performed. This analysis paralleled the previous findings. In addition to a main effect of company reputation that showed that a positive reputation promoted higher levels of company trustworthiness beliefs (M = 3.09, SD = .40) than a negative reputation (M = 2.45, SD = .53; F (1, 70) = 33.57, p < .001, η 2 = .32), a significant interaction between type of disclosure and reputation emerged (F (1, 70) = 4.25, p = .04, η 2 = .06). Similar to the previous study and in line with hypotheses, additional simple main effects analyses indicated that the impact of type of disclosure on company trustworthiness beliefs was only observed when the company had a negative, but not when it had a positive reputation (see Table 3, 4 for means, standard deviations and correlations). More specifically, compared to third-party disclosure, for a company with a negative reputation, self-disclosing the negative information positively influenced company trustworthiness perceptions (F (1, 70) = 4.13, p = .05, η 2 = .06), whereas this strategy was inconsequential for a company with a positive reputation (F < 1).

Table 3 Means, standard deviations, and intercorrelations for Study 3
Full size table
Table 4 Means and standard deviations of company trustworthiness and company evaluations as a function of disclosure strategy and company reputation (Study 3)
Full size table

Company Evaluations

To examine whether this qualified effect spills over to affect company evaluations, a second ANOVA was performed. The results provided converging empirical evidence for our notions. First, a main effect of type of disclosure (F (1, 70) = 10.32, p = .002, η 2 = .13) paralleled earlier findings and showed that participants exposed to the mere self-disclosure message had relatively more favorable company evaluations (M = − .50, SD = 1.13) than participants exposed to the third-party disclosure message (M = − 1.32, SD = 1.38). Furthermore, a main effect of reputation was observed (F (1, 70) = 36.15, p < .001, η 2 = .34) indicating that a company with a good reputation produced relatively more positive evaluations (M = −.12, SD = .96) than a company with a negative reputation (M = −1.56, SD = 1.23). In addition, and of more interest, these main effects were qualified by a two-way interaction (F (1, 70) = 4.94, p = .03, η 2 = .07). In line with the previous results, additional simple main effect analyses to probe the interaction confirmed that self-disclosing negative information was only effective in improving company evaluations when company reputation was negative (F (1, 70) = 16.56, p < .001, η 2 = .19), but not when it was positive (F < 1, see Table 2).

Mediation Analysis

To assess the mediating role of company perceived trustworthiness in driving the interaction effect on company evaluations (Hypothesis 3), we conducted a mediated moderation analysis, in line with the procedure suggested by Muller et al. (2005). That is, we assessed whether company trust mediated the type of disclosure x reputation interaction on company evaluations controlling for all main effects. As noted previously, the two-way interaction between type of disclosure and company reputation was significant for both company trustworthiness beliefs (β = −.19, t(70) = −2.06, p = .04) and company evaluations (β = −.20, t(70) = −2.22, p = .03). Furthermore, perceived trustworthiness predicted company evaluations (β = .51, t(72) = 5.02, p < .001). Finally, in a simultaneous regression analysis treating the two-way interaction, and all main effects as predictors of company evaluations, company trustworthiness beliefs continued to be a significant predictor (β = .22, t(69) = 1.97, p = .05) whereas the interaction effect between type of disclosure and company reputation was reduced and no longer significant (β = −.16 t(69) = −1.73, p < .10). A bootstrapping procedure as outlined by Preacher and Hayes (2004) to confirm the pattern of mediated moderation showed that the 95 % confidence interval around the indirect effect ranged from .08 to .71. The fact that zero fell outside this interval indicates a mediated moderation effect significant at p < .05.

In sum, these results replicate and extend the previous findings but also show the limits of self-disclosure as a strategy of reputation repair. First, in line with Hypothesis 1, self-disclosing negative information not only affects consumer choice behavior and perceptions of company trustworthiness, but also improves company evaluations, compared to third-party disclosure. Moreover, in support of Hypothesis 2 and paralleling the results of Study 2, mere self-disclosure proved to be a suitable strategy only for companies with a negative rather than positive reputation. Indeed, for a company with a poor reputation, the damage of the negative information was considerably lessened if the company disclosed the negative information itself, whereas this strategy was not effective for a company with a good reputation, which apparently was not affected by the negative event. Moreover, these findings were observed while there were no other intervention cues present in the news messages, strongly suggesting that the beneficial effects were indeed attributable only to the source that divulged the negative event. Finally, and consistent with Hypothesis 3, the moderated effect of type of disclosure was found to be mediated by company trustworthiness beliefs. Hence, the beneficial effects of self-disclosure for companies with a poor reputation appear to be a function of the trustworthiness-protecting function of this strategy.

But the findings of Study 3 also differed from those of Study 2. Although self-disclosure positively affected the reputation of the company in the negative reputation condition, it did not raise reputation to the level of that of the company with the positive reputation. Since unfortunately we did not assess company reputation before participants were given the information about the negative event, we cannot differentiate between self-disclosure buffering a negative impact of negative information or actually restoring trust. On the other hand, it is quite striking that type of disclosure proved to be beneficial for a company that had a history of integrity violations rather than just one violation.

General Discussion

The present research extended findings on the impact of mere self-disclosure in a new and yet untested “playing field.” We found that companies with a poor reputation might profitably use an influence strategy of mere self-disclosure of negative information to promote positive effects on consumers’ judgment and decision making, compared to a situation where the same negative information is revealed by a third party. The effect of the self-disclosure strategy appears to be robust: it was demonstrated across three studies, for various types of negative events, for both fictitious and existing companies, and on both consumer judgment and behavioral outcomes. More specifically, the present series of studies found evidence that mere self-disclosure affects message recipients’ choice behavior (Study 1), perceptions of company trustworthiness (Study 2 and 3) and company evaluations (Study 3). These findings are in line with the results Williams et al. (1993) reported in their study on stealing thunder as a courtroom strategy. It is also consistent with key findings in the related fields of persuasion and two-sided advertising (Kamins and Assael 1987). Because self-disclosure may function to limit the damaging impact of negative information on persuasion, this strategy falls under the broad rubric of what Knowles and Riner (2007) have termed “omega strategies” designed to reduce consumer resistance to persuasion.

As alluded earlier, the pattern of findings showed that the effect is not unconditional. More specifically, and in line with early research on attribution processes in persuasion (e.g., Eagly et al. 1978, 1981), Study 2 demonstrated that self-disclosure proved to be an effective strategy when it disconfirms consumer expectations. This proved to be the case when the strategy was employed by companies with a poor reputation. In contrast, companies with a positive reputation, the type of disclosure (self vs. other) did not differentially affect levels of perceived trustworthiness. Importantly, though, in Study 1, where company reputation was not manipulated, the mere self-disclosure effect was also observed. This suggests that unless a company has an established positive reputation, it might be well-advised to disclose negative events itself, rather than leaving disclosure to others. However, future research might systematically address whether reputation salience -in addition to reputation valence- may qualify the present results.

Study 3 replicated and extended the key findings of Study 1 and 2 to yet another type of consumer judgment, company evaluations, and demonstrated that the effectiveness of the self-disclosure strategy spills over to affect company evaluations in addition to perceived trustworthiness, again only for companies with a poor reputation. Moreover the results of the moderated mediation analysis in this study suggest that company trustworthiness beliefs is indeed the driver behind the beneficial effects of self-disclosure of negative information for companies with a poor reputation.

The present findings point to a buffering effect of self-disclosure to shield or protect a company from damaging effects that would otherwise ensue to its reputation, trustworthiness and related consumer responses had the information been divulged by a third party. As noted earlier though, the present findings do not directly speak to the possibility of self-disclosure actively restoring or repairing damage that negative information might cause. In terms of research design, testing this possibility requires at least a set-up in which (existing) trust is first violated and then either or not restored. Although trust repair strategies have been documented in the literature (e.g., Kim et al. 2009; Kramer and Lewicki 2010), it remains an open question whether mere self-disclosure qualifies as one of them. Conceptually, the present theoretical reasoning based on attribution processes would not rule this possibility out per se. However, when delving into the underlying process, it seems fair to suggest that such repair or restoration effects are particularly likely when message receivers do not critically scrutinize the negative news (in which case mere self-disclosure might function as a simple trust-restoring cue, cf. Chaiken and Trope 1999). If such negative information is deeply processed, though, it may take more than just this simple strategy to set the record straight. Translated to the various stakeholder groups with which a company has to interact (e.g., customers, shareholders, employees), this reasoning might tentatively imply that trust restoration effects of mere self-disclosure become more plausible to the extent that the direct involvement of the stakeholder group with the company becomes lower and so might be most pronounced for consumers and least observable among employees and/or stakeholders. This may constitute a viable avenue for future research. Furthermore, future studies might systematically explore the role of a negative versus positive company reputation compared to a neutral reputation, and might include a no-news condition in addition to a negative news condition.

The findings of the three studies reported in this article demonstrate that self-disclosure is a viable, “lean” consumer influence strategy, particularly for companies that are not protected by a positive prior reputation. In contrast to discounting bad news, buck-passing or ignoring negative events, mere self-disclosure’s effectiveness underscores that communicating about negative issues does not require putting a “spin” on the information the company discloses. It appears that consumers appreciate companies for disclosing negative information as it is: negative information. They reward the company with more favorable ratings and choosing its products (compared to third-party disclosure). Whether this amounts to a differential preference over competitors is an issue future research might address.