Rumor Psychology

The Watercooler Effect

Rumor Psychology

Social and Organizational Approaches

by Nicholas DiFonzo and Prashant Bordia

Rumor Psychology Chapter Excerpts

Copyright © 2007 by the American Psychological Association. Reprinted with permission.


Rumors are an enduring feature of our social and organizational landscapes. They attract attention, evoke emotion, incite involvement, affect attitudes and actions—and they are ubiquitous. A small sample: Groundless rumors that McDonald’s uses worm-meat in its hamburgers grounded sales in Atlanta (Goggins, 1979). Sober reports that Paul McCartney was dead were discussed with sadness and snowballed even after a photo and interview with—a very much alive—McCartney was published in LIFE magazine (Rosnow, 1991). Office scuttlebutt often eats away at trust—and feeds upon distrust—among organizational members (DiFonzo, Bordia, & Rosnow, 1994). False rumors that a Haitian coup leader was to be set free spurred angry riots that killed ten people (“10 die in Haiti,” 1991). Seven million people heard the incorrect claim that Coca-Cola contains carcinogens (Kapferer, 1989). Two bizarre and fallacious rumors, widespread in Africa, were that the AIDS’ HIV virus was developed in a western laboratory, and that a World Health Organization team inoculated 100,000 Africans with an untested vaccine that caused the continent’s pandemic of AIDS (Lynch, 1989). Harmful or potentially harmful rumors harangue the ears of top corporate public relations personnel nearly once per week on average (DiFonzo & Bordia, 2000). Email hoaxes of the “Good Times” virus that will rewrite your hard drive and the “teddy bear” icon that destroys your whole system regularly alarm novice Internet users (Bordia & DiFonzo, 2004; “JDBGMGR.EXE,” 2002). The catalog continues in abundance; rumors flourish, fascinate, and frustrate.

Chapter 1: Defining Rumor

Rumors arise in situations that are ambiguous or threatening in some way (G. W. Allport & Postman, 1947b; Rosnow, 1991; Shibutani, 1966). Ambiguous situations are those in which the meaning or significance of events is unclear, or where the effects of events are not certain. Ambiguity is problematic for people. Why? In any context, humans have a core social motive to understand and to act effectively (Fiske, 2004). Culturally defined categories ordinarily help individuals do this (R. H. Turner, 1964). But sometimes events don’t fit well together or fail to convey meaning. In these cases, individuals refer back to the group to understand the situation and act (Asch, 1955; Sherif, 1936). This referring back to the group—or group thinking—is rumor discussion. Thus, rumors occur when a group is attempting to make sense of ambiguous, uncertain, or confusing situations. Sociologist Tamotsu Shibutani (1966) proposed that when formal information is absent, people compensate by informally interpreting the situation (cf., Bauer & Gleicher, 1953; R. H. Turner & Killian, 1972). Rumor discussion is here a process of group interpretation; rumor is a product of that process (Rosnow, 1974).

Chapter 2: Forms, Frequency, and Fallout of Rumors

Rumor effects may be classified as behavioral or attitudinal. In business settings, behavioral effects of rumor include those that affect purchase behaviors. The false rumor that Tropical Fantasy Soda Pop was owned by the Ku Klux Klan and made black men sterile reportedly caused sales to drop by 70% and incited attacks on delivery trucks (Freedman, 1991). Unger (1979) reported similar losses in sales due to false product rumors: Bubble Yum bubble gum is contaminated with spider eggs, and Pop Rocks candy, ingested with soda pop, explodes in your stomach. Rumors have also affected stock purchase behaviors—and thus stock values (Lazar, 1973; Rose, 1951). For example, prior to publication of takeover rumors published in the “Heard on The Street” column of The Wall Street Journal, price runups occurred, indicating that the takeover rumors pushed prices up as they diffused through the financial community (Pound & Zeckhauser, 1990); investors often “buy on the rumor.” Workplace productivity has also been affected—usually negatively—by internal rumors (DiFonzo & Bordia, 2000). Tangible effects such as these are often mediated through rumor’s impact on attitudes. One such attitude is reputation; clearly, rumors can wreak havoc on a company’s public standing (Zingales, 1998). Koenig (1985) documented the case of Continental Bank whose reputation was besmirched by rumors of impending bankruptcy. Rumors during organizational change episodes such as restructuring and layoffs may also have damaging effects on organizational attitudes such as morale and trust (DiFonzo & Bordia, 1998; DiFonzo et al., 1994; Smeltzer & Zener, 1992). These effects are but a small sampling of the many outcomes that rumors can result in or contribute to. The remainder of this chapter summarizes systematic research from several recent studies conducted to investigate the scope and nature of organizational rumor effects.

Chapter 3: Psychological Factors in Rumor Spread

In general, situations that threaten a goal will activate the associated motivation. When our ability to cope effectively with the environment is threatened, fact-finding is activated: we search for accurate information. For example, upon hearing about merger talks between their organization and a rival company, employees are likely to seek accurate information about the consequences to the structure of their organization as a result of the merger. Similarly, when the self or ingroup is threatened, self-enhancement motivation may be activated and we are likely to favor information that makes us feel good about ourselves. In such situations, we are less likely to be concerned about the accuracy of the rumor and more about its self-enhancing value. At other times, characteristics of the person or the situation may make a particular goal (and the associated motivation) more salient. For example, a young adult keen on developing romantic relationships is likely to be driven by relationship-enhancement motivations and is likely to share a rumor that helps attract or hold attention of the audience.

Sometimes, motivations will work in tandem. In particular, the relationship-enhancement and self-enhancement motivations are easily served by the same rumor. Spreading a rumor that sheds positive light on the ingroup is likely to be self-enhancing (it will boost my self-esteem) and will also help form a good impression in other ingroup members (it will enhance our relationship). In a similar way, rumors that are derogatory of the outgroup are self-enhancing and also aid in the development of solidarity and cohesion (Kakar, 2005) among ingroup members. At other times, however, the motivations may compete. For example, when the recipient is an outgroup member, sharing a positive ingroup rumor may be self-enhancing but may not serve relationship-enhancement goals as well as sharing a positive outgroup rumor would.

Chapter 4: Factors Associated with Belief in Rumor

Some attitudes are so deeply rooted in a group that they are more rightly considered as culture. One of R. H. Knapp’s (1944) characteristics of a thriving rumor was that it accorded with the cultural traditions of the group. Prasad (1935) noted that rumors gained currency when they were consistent with local superstitions. In a later (1950) study, he characterized rumor as a picture completion task with attitude (largely made up of cultural traditions) painting the picture. Similarly, London and London (1975) speculated that widespread false variations of a rumor that President Nixon had stolen a tea cup from Chairman Mao during his visit to China sprang from elements of Chinese national character. Finally, Jung (1959) ventured beyond culture as the mold with which some rumors are cast; he viewed rumor generation as a kind of projective mechanism giving evidence of presuppositions that are handed down from our ancestors through a collective unconscious.

Chapter 5: Rumor as Sense Making

Stable-cause attributions have produced illusory associations—the erroneous judgment that two characteristics are associated (Jennings, Amibile, & Ross, 1982). Undergraduates and clinicians, after viewing random pairings of Draw-A-Person test drawings and a psychological diagnosis, thought that an emphasis on the eyes (in the drawing) covaried with paranoia; this happened even though the diagnosis of paranoia was not associated with large or salient eyes in drawing (Chapman & Chapman, 1969). Why then did participants associate the two? A stable cause attribution—drawing large eyes is caused by a paranoid sensitivity to being watched—led subjects to associate these variables. Stable cause attribution leads people to see relationships—even when they don’t exist.

Similarly, rumors lead people to see relationships even if they are non-existent; that is, stable cause attributions embedded in many rumors lead to illusory associations. For example, rumor has long been associated with the formation and maintenance of racial stereotypes (G. W. Allport & Postman, 1947b; R. H. Knapp, 1944; Knopf, 1975; P. A. Turner, 1993). Rumors embedded with stable-cause race-characteristic attributions encourage these illusory associations. Children hearing multiple variations of such rumors as “I heard that Johnny Black stole a car” (because blacks are thieves) or “I heard that Officer White beat him unnecessarily with his Billy-club” (because whites are brutal) will be led to associate African Americans with theft and European Americans with aggression. Experimental evidence accords with these ideas. The investors in the computerized stock market simulation game described above were exposed to price changes that were uncorrelated with rumors; despite this they thought that these price changes were associated with rumors (DiFonzo & Bordia, 2002b). Investors perceived relationships that were non-existent.

Chapter 6: Rumor Accuracy: Patterns of Content Change, Conceptualization, & Overall Accuracy

Although adamant that snowballing is a “misconception” even G. W. Allport and Postman (1947b, p. 153) observed some invention (i.e., adding) in their laboratory [serial transmission] demonstrations and offered one possible situation in which rumors might snowball: After highly emotionally straining events, people may tend to perseverate, “…mull it over, talk about it endlessly, [and] explore in fantasy all possible consequences” (p. 154). Further, Schachter and Burdick’s (1955) field experiment strongly supports this idea. This study manipulated importance while creating a highly ambiguous situation: The percentage of girls reporting new rumors in the high-importance conditions was much higher (70%) than in the low-importance conditions (15%). Additionally, the diversity of rumors was also much greater in the high-importance conditions (average of 12 different rumors in each of these conditions versus average of 1.5 different rumors in the low-importance conditions). The high-importance condition groups were composed of friends of the girls who had been suddenly removed from a classroom setting; these friends were undoubtedly earnestly and interactively collaborating about the event and thus prone to invention. The low-importance groups were comprised of girls who were not acquainted with the accomplices; these girls were presumably passing along (serially transmitting) interesting bits of information. Again, adding seems to occur in real life situations that are interactively collaborative in character.

Chapter 7: Mechanisms Facilitating Rumor Accuracy and Inaccuracy

We have discussed how individual trait and state anxiety may exacerbate cognitive processes that usually inhibit accuracy. Collective excitement—situations in which many or all individuals in a situation are anxious—may intensify such effects by increasing suggestibility (distortion of perception) and diminishing critical ability. Crowd milling, for example, may catalyze restless individuals into an excited mob where an inaccurate rumor is immediately acted upon (R. H. Turner & Killian, 1972). Sinha (1952) observed that suggestibility after catastrophic landslides was heightened because of high anxiety: people set aside their critical set, rumors were not scrutinized, and people did not desire to verify. Shibutani (1966) proposed two general sorts of rumor deliberation patterns that hinge upon anxiety: deliberative: “If unsatisfied demand for news is moderate, collective excitement is mild, and rumor construction occurs through critical deliberation” (p. 70, italics in original); and extemporaneous: in situations of intense collective excitement, rumor construction becomes behavioral contagion. Mausner and Gezon (1967) provide an example of extemporaneous rumor construction: A grade school temporarily closed due to unfounded fears of an outbreak of vaginal gonorrhea among girls (only 3 out of 173 girls had it, but many exhibited some symptoms); this rumor exemplified contagion in that it was marked by high collective excitement and suggestibility.

Chapter 8: Trust and Organizational Rumor Transmission

In summary, these results suggest that trust is a key variable in rumor transmission and is likely to play a central role in organizational rumor activity in at least two ways. First, distrust in the organization is likely to fuel rumor activity. For example, if I perceive the company to be uncaring and dishonest, I am unlikely to rely upon their explanations to account for recent changes in personnel that affect the quality of my job. Second, trust is likely to alter the relationships between uncertainty, anxiety, and rumor. When trust in the company is low, members may be especially prone to engage in rumor discussions regardless of their levels of uncertainty or anxiety; when trust is high, such rumor discussions are only necessary under conditions of high uncertainty or anxiety. For example, if I perceive the company as uncaring and dishonest, even small amounts of uncertainty and anxiety are enough to make me concerned. I am then likely to participate in rumor discussions because I think that my co-workers in the rumor mill—but not the management—have my best interests at heart. Even rumors appearing during times of quiescence and stability would receive lots of my attention because they might protect me from dreaded consequences that the company did not care about. However, when I trust the company, there is no need to pay much attention to rumors because the company explanation can be relied upon; I need only turn to rumors when the company is unable to quell my uncertainty or anxiety. Future research should seek to replicate these patterns in other arenas—both field and experimental—and should seek to further clarify the nature of the relationship between trust and transmission. Future transmission research—in whatever venue—should routinely measure trust in formal communication sources.

Chapter 9: Rumor Management

For a denial to be effective, it should be reliably recalled with the target and the accusation. Given that recall of negations take more cognitive effort, Wegner et al. (1985) reasoned that if the denial (“I am not a cheat”) is proposed in the affirmative (“I am honest”), it has higher likelihood of being recalled. In other words, victims of rumors should strive to be associated with positive impressions to replace the negative impressions created by the rumor. A similar idea was proposed by Tybout et al. (1981) who found that attitudes towards the McDonald’s restaurants in response to the rumor that the burgers contain worm meat were improved when worm meat was relabeled as a French delicacy or when participants were asked to think of certain details of the McDonald’s restaurant they visited (e.g., whether it had indoor seating or not). In other words, rumor control should be a multi-pronged approach that may include an outright rebuttal and other reputation-enhancing strategies. Indeed, most effective rebuttals adopt this approach. Consider the example of Stroh Beer (Koenig, 1985). In 1983, the Stroh Brewery Company was plagued by rumors that the company had donated money to Jesse Jackson’s presidential campaign. The company took out advertisements in the Chicago Tribune and in addition to stating that the rumors were “completely, totally false”, the brewery company provided information on its donations to the Statue of Liberty renovation effort.

Chapter 10: Summary, Model, and Research Agenda

With regard to situational features, to our knowledge, no experimental work has followed up on R. H. Turner and Killian’s (1972) idea that high collective excitement results in less stringent norms about what information is acceptable, especially in close—versus diffuse—groups. In addition, a group’s corporate capacity to check rumor verity is an area in need of greater conceptual clarity and measurement. One goal of this work would be to reliably measure a group’s capacity to check, and to relate this to overall rumor accuracy levels during any given rumor episode. One interesting question here: How does the capacity to check fare against anti-accuracy motivations? That is, how accurate are rumors in groups that have a high capacity to check rumor veracity but are strongly motivated to believe/disbelieve them? This question seems highly amenable to experimentation. We speculate that capacity to check trumps motivation when they are in conflict; rumors that are clearly false will not thrive, even among a group that strongly desires them to be true. We reason that in situations where the rumor can be objectively checked, even relationship- and self-enhancing motivations come to act in service of accuracy because sending true information is a deeply engrained social rule. This research question has important practical implications; as previously discussed, many troubling rumor effects arise in situations of conflict where people are strongly motivated to believe the worst about rival groups.