Position Paper

WHO’S WITH ME? FALSE CONSENSUS, BROKERAGE, AND ETHICAL DECISION MAKING IN ORGANIZATIONS

FRANCIS J. FLYNN Stanford University

SCOTT S. WILTERMUTH University of Southern California

We propose that organization members overestimate the degree to which others share their views on ethical matters. Further, we argue that being a broker in an advice network exacerbates this false consensus bias. That is, a high level of “betweenness centrality” increases an individual’s estimates of agreement with others on ethical issues beyond what is warranted by any actual increase in agreement. We tested these ideas in three separate samples: graduate business students, executive students, and employees. Individuals with higher betweenness centrality overestimated the level of agreement between their ethical judgments and their colleagues’.

For members of organizations, ethical standards can help guide individual decision making by clar- ifying what the majority of others believe is appro- priate. But given that ethical standards often are tacitly held, rather than explicitly agreed upon (Haidt, 2001; Turiel, 2002), individuals may strug- gle to recognize the normative view—what most others believe is the “right” course of action. Peo- ple’s tendencies to project their own opinions can alter their judgments about what others think is ethical, perhaps giving them a sense of being in the majority even when they are not. The ramifications of this false consensus effect may be problematic: if members of organizations erroneously assume that their actions are in line with prevailing ethical principles, they may subsequently learn of their misjudgment when it is too late to avert the consequences.

In the present research, we examine whether bro- kers in a social network show evidence of false consensus in ethical decision making. Because bro- kers span structural holes (missing relationships that inhibit information flow between people [see Burt, 1992]), one might assume that these individ- uals possess greater insight into others’ attitudes and behaviors. But can acting as a broker (i.e., hav- ing “betweenness”) inform a focal individual about his or her peers’ ethical views? In interactions with colleagues, people generally refrain from initiating moral dialogue; rather, they prefer to discuss less sensitive attitudes and opinions (Sabini & Silver,

1982). We argue that this tendency to avoid moral discourse and instead discuss superficial connec- tions worsens the false consensus bias in ethical decision making, providing an illusion of consen- sus where none exists.

The notion that having an advantageous position in a social network might exacerbate, rather than mitigate, false consensus bias in ethical decision making represents a novel insight for those inter- ested in the link between social networks and in- dividual judgment. Prior work on identifying the determinants of false consensus has focused pri- marily on motivational drivers, such as ego protec- tion, or cognitive heuristics, such as “availability bias” (for a review, see Krueger and Clement [1997]). Yet, the nature of false consensus—a flawed view of one’s referent group—suggests that an individual’s set of social ties can also play an important role. In contrast to the authors of work on ethical decision making who have treated social networks as a means of social influence (e.g., Brass, Butterfield, & Skaggs, 1998), we propose that social networks can distort social cognition (see Flynn, Reagans, Amanatullah, & Ames, 2006; Ibarra, Kilduff, & Tsai, 2005), particularly the judgment of others’ ethical views.

We aim to make several contributions to the scholarly literatures on social networks and ethical decision making in organizations. First, we intro- duce the concept of false consensus bias in the context of ethical judgments, thereby adding to a growing literature on the psychological factors af- fecting organization members’ moral reasoning (see Mannix, Neale, & Tenbrunsel, 2006). Second, and more importantly, we explore whether one’s loca-

The authors would like to acknowledge helpful com- ments on drafts of this article offered by Dale Miller and Benoit Monin.

� Academy of Management Journal 2010, Vol. 53, No. 5, 1074–1089.

1074

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express written permission. Users may print, download or email articles for individual use only.

tion in an advice network can influence this false consensus bias; thus, we examine ethical decision making in organizations as a form of social judg- ment. Finally, we provide a counterpoint to re- search showing that many forms of centrality in social networks can improve social perception (e.g., Krackhardt, 1987), suggesting instead that an individual’s judgments of ethical standards (i.e., the ability to gauge a consensual position) may be impaired by occupying a broker role (i.e., by having more betweenness).

Ethical Judgments and False Consensus Bias

Ethical principles can be defined as consensually held positions on moral issues (e.g., Frtizsche & Becker, 1984; Kohlberg, 1969, 1981; Mackie, 1977; Payne & Giacalone, 1990; Toffler, 1986; Turiel, 2002). According to this conventional view of eth- ics, moral values are continually evolving and are shaped by patterns of behavior and discourse within a social group (Phillipps, 1992; Schweder, 1982; Wieder, 1974). Other approaches to ethics, such as the principles approach advocated by Locke’s utilitarianism, Kant’s categorical impera- tive, and higher-level stages of morality in Kohl- berg’s (1969) model of moral reasoning, do not con- tain such an assumption that ethics are socially determined. In the present research, we focus on the conventional approach and acknowledge that in other approaches, consensus may not be as im- portant in deciding which behaviors are ethical.

For members of organizations, socially shared ethical standards are important to recognize but often difficult to gauge (Treviño, 1986). Such un- certainty in diagnosing the conventional ethical view can invite various forms of cognitive bias. In particular, empirical studies of the “false consen- sus effect” (Marks & Miller, 1987) have consistently shown that “people’s own habits, values, and be- havioral responses . . . bias their estimates of the commonness of the habits, values, and actions of the general population” (Gilovich, 1990: 623). Peo- ple who are shy, for example, tend to think that more people are shy than do those who are gregar- ious (Ross, Greene, & House, 1977). In some sense, false consensus is akin to an “anchoring and adjust- ment” process, whereby people anchor on their own attitudes and insufficiently adjust for ways in which they are likely to differ from others (Davis, Hoch, & Ragsdale, 1986).

We propose that false consensus bias can play a critical role in ethical decision making by influ- encing how individuals see their decisions in relation to how others see the same decisions. Ac- cording to Haidt’s (2001) social intuitionist model,

when prompted to defend their moral reasoning, most individuals are motivated to see their choices and attitudes as consistent with others’ choices and attitudes. This desire for normative alignment may, in turn, lead them to interpret their own actions and beliefs as “common and appropriate” (Ross et al., 1977: 280). Conversely, the same people will see alternative responses (particularly those di- rectly opposed to their own) as deviant, or uncom- mon and inappropriate. In short, people are predis- posed to view their decisions as being more in line with the prevailing view than others’ decisions are (Krueger & Clement, 1997).

The concept of appropriateness plays a pivotal role in the domain of ethical decision making be- cause it provides a motivational driver for individ- ual judgment (Haidt, 2001; Haidt, Koller, & Dias, 1993). People are motivated to make moral judg- ments and avoid making immoral ones (Colby, Gibbs, Kohlberg, Speicher-Dubin, & Candee, 1980). But what happens when they are unclear about moral standards? When the ethical course of action is ambiguous (e.g., there is a dilemma in which one ethical principle stands opposed to another), mem- bers of organizations will be inclined to see their actions as normative rather than deviant. The cumulative effect of this motivated reasoning is straightforward: employees’ intuitions about whether others agree with their ethical judgment will be biased, so that they overestimate the prev- alence of their own views. We therefore put forth:

Hypothesis 1. People estimate that a majority of others share their views on ethical issues— even when their views are actually held by a minority of others.

Brokerage and False Consensus Bias in Ethical Decision Making

Can an individual’s location in a social network, particularly his or her centrality, affect ethical de- cision making? Brass et al. (1998) argued that more centrally located employees are less likely to per- form immoral acts because being well known makes individual behavior more visible and in- creases potential damage to one’s reputation. Ac- cording to Sutherland and Cressey (1970), the effect of network centrality on individual ethical judg- ment is a matter of social influence rather than reputation. To the extent that a focal individual is connected to many unethical colleagues, network centrality will likely be a strong predictor of uneth- ical behavior; to wit, a higher percentage of “bad apples” in one’s social circle can cloud one’s moral judgment.

2010 1075Flynn and Wiltermuth

We suggest an alternative link between network centrality and ethical decision making— one that connects social networks with social projection. Centrality in a social network is often described as a form of power (or potential power) because hav- ing a central network position offers an individual “greater access to, and possible control over, rele- vant resources,” such as information (Brass, 1984: 520). One specific form of power in advice net- works, betweenness centrality, is closely associated with informational advantage (Burt, 1992). Be- tweenness captures the extent to which a point falls between pairs of other points on the shortest path connecting them. In other words, if two people, A and C, are connected only through another person, B, then B has some control over any resources that flow between A and C. In effect, B can act as a broker between A and C. As a measure of centrality, betweenness is well suited to capture the control of information in advice networks (Freeman, 1979). Thus, betweenness centrality may be a particularly relevant source of power that pertains to ethical decision making in organizations.

Given that brokers have an informational advan- tage, one might assume that these individuals have greater insight into their group’s shared moral atti- tudes and beliefs and therefore will be more accu- rate in estimating others’ ethical judgments. In con- trast, we argue that the opposite may be true. Individuals learn about others’ attitudes through ongoing conversation and casual observation, but the insight they gain from such interactions can often be superficial (Hollingworth, 2007). People are inclined to talk about “safe” subjects—sports, kids, current events—rather than sensitive subjects such as politics, religion, and morality (Kanter, 1979; Skitka, Baumann, & Sargis, 2005). Thus, little of the information that brokers gain from their so- cial ties may apply to personal bases of moral judg- ment because people are loath to discuss their moral values openly with their colleagues. Instead, such discussions of morality seem almost taboo (Sabini & Silver, 1982; Turiel, 2002).

Although people may be reluctant to discuss moral quandaries with their colleagues, those who broker social ties in an advice network may assume that they share their colleagues’ views on moral issues, even when this is not the case. Recent re- search suggests that powerful individuals, such as those who occupy powerful positions in social net- works, are prone to failures in perspective taking. They are less attentive to social cues and less sen- sitive to others’ views (Erber & Fiske, 1984; Keltner, Gruenfeld, & Anderson, 2003). In fact, according to Galinsky, Magee, Inesi, and Gruenfeld, perspective taking—“stepping outside of one’s own experience

and imagining the emotions, perceptions, and mo- tivations of another individual” (2006: 1068)— has been described as antithetical to the mind-set of the powerful, given that powerful individuals are less empathic, less considerate of others’ opinions, and less likely to take into account others’ information when making decisions.

Are powerful people, such as brokers, likely to assume that others’ ethical views are more like their own, or less? We propose that brokers may be more likely to assume their views are in line with the conventional standard because they possess an inflated sense of similarity. Brokers often have to negotiate across boundaries and manage people with diverse interests (Burt, 2007). As social con- duits, they identify, establish, or create bases of connection, communion, and correspondence, which, in turn, reinforce a sense of shared attitudes and beliefs (Stasser, Taylor, & Hanna, 1989). Bro- kers may assume that the agreement they share with others on explicit topics of conversation per- tains to unspoken attitudes, such as moral beliefs. Thus, for individuals who have high levels of be- tweenness (i.e., brokers), estimates of how their moral attitudes align with those of their colleagues may become exaggerated. They may overestimate the extent to which their ethical views are aligned with others’ because brokers are inclined to believe they are highly similar to their peers. Formally, we propose:

Hypothesis 2. People who are highly central within a social network are more likely than those who are less central to overestimate so- cial support for their ethical views.

Overview and Summary of Predictions

We propose that ethical decision making in or- ganizations is subject to the false consensus bias—a tendency for people to assume that others hold the same opinions as they do. We predict that individ- uals who are asked to evaluate whether a certain act is ethical or unethical will assume that more of their peers will provide a response similar to their own than is actually the case. One might predict that an advantageous position in an advice network (i.e., betweenness) would mitigate this false con- sensus bias, because brokers sometimes enjoy an informational advantage over others. However, we predict that having more betweenness centrality increases, rather than decreases, the false consen- sus bias in ethical judgments. Being a broker will not expose the unspoken differences in moral opin- ions that often exist; rather, it will strengthen an individual’s belief that her or his peers hold similar

1076 OctoberAcademy of Management Journal

ethical views (i.e., inflating estimates of agreement with others).

We tested these ideas by collecting judgments of ethical decision making in the workplace from master’s of business administration (MBA) stu- dents, students enrolled in a master’s of manage- ment program for executives, and employees in the marketing department of a manufacturing firm. Fol- lowing previous research on the false consensus bias (e.g., Ross et al., 1977), we asked each partici- pant in our study to consider several hypothetical scenarios that featured ethical dilemmas and to provide their opinion about whether the action de- scribed was ethical and what percentage of their colleagues held the same view. We also collected data from participants describing their advice net- works; specifically, we asked whom among their colleagues they would go to for help and advice (and who among their colleagues would come to them for help and advice).

METHODS

Participants

Marketing department sample. Thirty-four em- ployees (78 percent women; mean age, 34.3) in the marketing department of a large food manufactur- ing company participated in this study in exchange for $25 gift certificates to a major online retailer. Seventy-seven percent of eligible participants (i.e., all employees in the marketing department at the company’s headquarters) completed the survey. On average, participants had worked for the company 3.2 years (s.d. � 4.3) and for the department 2.3 years (s.d. � 2.5). All participants worked at one location and therefore had the opportunity to inter- act with each other frequently. In fact, the (former) head of the marketing department suggested that the group had a strong identity because (1) the employees were colocated and (2) they often did not need to contact employees from other depart- ments to complete their work (a large percentage of their communication was internal). People from all levels of the department completed the survey. Re- spondents’ titles ranged from administrative assis- tant to vice president of marketing.

MBA student sample. One-hundred-sixty-two master’s of business administration students (20 percent women; mean age, 29.4 years s.d. � 5.4) at a private East Coast university participated in this study as part of a required course in organizational behavior. Participants were split equally across three sections of the same class. During the first year of the MBA program, students were required to take courses with the same group of fellow stu-

dents. At the time of this study, the students in each class had been together for approximately seven months. These “clusters” constituting the MBA classrooms are meaningful to the students, given that they take all of their classes and organize many of their social activities together during their first year. Providing evidence of how insular these groups are, a separate survey revealed that more than 80 percent of respondents’ self-reported net- work ties were with people from their own cluster rather than other clusters. Ninety-five percent of eligible participants completed the online study questionnaire. Participants had an average of 5.6 (s.d. � 2.8) years of work experience.

Executive sample. Fifty-three students in a full- time master of management program for executives at a private West Coast university (25 percent wom- en; mean age, 35.6 years, s.d. � 3.9) participated in this study as part of a required course in organiza- tional behavior (again, students were required to take all of their courses with the same group of fellow students). Ninety-five percent of eligible participants completed the study questionnaire, which was administered eight weeks after the start of their program. Executive students in this pro- gram have a strong social identity. They take all their classes in the same room and socialize fre- quently outside of class. The executive students had an average of 12.7 (s.d. � 3.9) years of work experience.

Procedures

Participants were invited to complete an online survey. After following a link to the study website, they were presented with a series of hypothetical scenarios describing ethical dilemmas in a work- place setting. Following each scenario, participants were asked to indicate whether they viewed the action taken in each dilemma to be ethical (“yes” or “no”). In addition, they were asked to estimate the percentage of others within their class or depart- ment who would agree with their response. Partic- ipants then responded to a series of questions de- signed to assess their social network within the class or department. Finally, they were asked to provide basic demographic information on their race, sex, home country, and age. Participants in the marketing department sample also indicated their hierarchical status.

Materials and Measured Variables

Ethical dilemmas. Following other research on social projection (e.g., Ross et al., 1977; Sabini, Cosmas, Siepmann, & Stein, 1999), we created a set

2010 1077Flynn and Wiltermuth

of hypothetical scenarios to serve as stimuli. The Appendix summarizes the six scenarios. Each of the scenarios we derived drew on Kidder’s (1995) taxonomy of “right vs. right” ethical dilemmas, in which two moral values are placed in direct oppo- sition; following one moral value would lead an individual to make one decision and following the other moral value would lead the same individual to make a different decision (see also Badaracco [1997] and Toffler [1986] for a similar description of ethical dilemmas). Specifically, we employed Kidder’s classification of three different types of ethical dilemmas: individual/community, truth/ loyalty, and justice/mercy.

According to Kidder (1995), an individual versus community dilemma refers to situations in which one option presents substantial costs to an individ- ual but the alternative option presents substantial costs to the community. In contrast, a justice versus mercy dilemma entails a choice between delivering punishment swiftly and surely or demonstrating compassion and leniency for a given transgression. Finally, a truth versus loyalty dilemma involves a situation in which the principle of honesty com- pels one to answer accurately but doing so would simultaneously break the confidence of another colleague. We drew on these three different classi- fications to generate six scenarios: two pitted the needs of the individual against the needs of the community, two pitted truth against loyalty, and two pitted justice against mercy. After reading each of these six scenarios, participants were asked to indicate whether the decision described in the sce- nario was ethical (“yes” or “no”).

Each dilemma was pretested to confirm that par- ticipants would treat the dilemma as an ethical one. Fifteen pretest participants rated how much they perceived each scenario to be an ethical dilemma (1 � “not at all,” 7 � “very much”). Participants gave five of the six vignettes a rating significantly higher than the midpoint of the scale (all p’s � .01), and they rated the vignette that described an em- ployee leaving a start-up company marginally higher than the midpoint (p � .06). For the set of six scenarios, the mean rating of how much partic- ipants perceived a scenario to be an ethical di- lemma was 5.4 (s.d. � 1.4).

Estimated agreement. In addition to collecting participants’ individual responses about the deci- sion made in each scenario, we collected their opinions about how others would respond. That is, after reading each of the six hypothetical dilemmas, participants were asked, “What percentage of other people within your department [class] would share your opinion about the ethicality of the decision?” Participants could provide any number ranging

from 0 through 100 in response to this question. We calculated the average response as part of our mea- sure of social projection (see de la Haye, 2000; Krueger, 1998).

Actual agreement. For each participant, we calculated the actual level of agreement for each individual for each dilemma by computing the per- centage of others in that individual’s class or de- partment who made the same choice as the focal participant. In other words, if a participant classi- fied a decision as unethical, actual agreement was defined as the percentage of others in the class or department who also classified that decision as unethical.

Perhaps participants’ estimates of agreement did not accurately reflect the actual percentage of oth- ers in their referent group (i.e., class or department) who agreed with their ethical judgments but did reflect the extent to which others in their network agreed with their ethical judgments. To test for this possibility, we created a second measure of actual agreement that pertained to each individual’s ad- vice network. Specifically, we calculated the per- centage of people within a participant’s advice net- work (this measure is described below) who classified a particular decision the same way as he or she did (ethical or unethical).

Network centrality. Network centrality has been measured in several different ways. We calculated three specific measures: degree centrality, close- ness centrality, and betweenness centrality. Degree centrality was the sum of the number of ties di- rected to a focal individual (i.e., the number of other individuals from whom she or he received advice) and emanating from the focal individual (i.e., the number of other individuals to which she or he gave advice). Closeness centrality was a mea- sure of the shortness of paths between a focal indi- vidual and all other members of a network. Finally, betweenness centrality was the fraction of shortest paths between dyads that passed through a focal individual. Although we considered all three mea- sures of centrality in our analysis, we expected betweenness centrality to have the most direct con- nection to false consensus because betweenness centrality captures the potential influence that an individual has over the spread of information through a network. In line with our research ques- tion, we investigated whether individuals who had higher levels of power based on informational in- fluence overestimated the overlap between their own and others’ ethical views.

To calculate these measures of centrality, we col- lected ego (self) and alter (peer) reports of network ties, particularly those ties that refer to the ex- change of help and advice (Krackhardt, 1987). Par-

1078 OctoberAcademy of Management Journal

ticipants in each of the three samples were pre- sented with a complete list of colleagues (classmates for the business school students and executive students, coworkers for the marketing department employees) and asked, “To whom would you go for help or advice if you had a ques- tion or a problem?” Participants checked off the names of colleagues who met this criterion (for the members of the MBA student sample, the names were limited to the other students enrolled in their particular class); there was no limit on the number of names participants could select. On the next page of the questionnaire, participants were asked to repeat the exercise, but in this case they indi- cated which individuals might “come to them for help or advice.” Thus, participants were asked to describe both sides of each dyadic relation.

To reduce the influence of egocentric bias, we focused on confirmed ties (Carley & Krackhardt, 1996) as the basis for our measures of centrality. In a confirmed advice-seeking tie, a focal participant reports that he/she receives advice from the listed alter and the listed alter reports that he/she gives advice to the ego. We calculated a “directed” mea- sure of betweenness centrality following the steps outlined by White and Borgatti (1994). We also calculated a measure of closeness centrality based on directed graphs (Freeman, 1979). Finally, we calculated degree centrality by summing the num- ber of confirmed ties for each participant. Degree centrality represented not only the number of in- tragroup ties an individual possessed, but also the proportion of the referent group to which the indi- vidual was connected in this manner.

Hierarchical status. To account for the possibil- ity that employees in high-status positions estimate higher levels of agreement than employees in low- status positions (cf. Flynn, 2003), participants in

the marketing department sample were asked, “How would you describe your position in the organization?” They responded using a seven-point Likert scale (1 � “entry level,” 7 � “top level”).

Demographic variables. Given that sex differ- ences have appeared in measures of network cen- trality (Ibarra, 1992) and false consensus bias (Krueger & Zeiger, 1993), we controlled for partic- ipant sex in each of our analyses. We also con- trolled for age, which varied widely in the market- ing department sample and, to a lesser extent, in the executive student sample. We used dummy variables to control for the regions of participants’ home countries in the MBA student sample and the executive student sample (an Asian country, a Lat- in-American country, or another country outside of the United States). Because only three people in the marketing department sample had a home country that was not the U.S., we did not control for region in that sample.

RESULTS

Tables 1, 2, and 3 report means, standard devia- tions, and correlations for all variables in the marketing department sample, the MBA student sample, and the executive student sample, respec- tively. We analyzed each of these samples sepa- rately, including the three separate sections of the MBA student sample. Preliminary graphical and statistical analyses of the MBA student sample re- vealed that two outliers whose responses were three standard deviations away from the mean were strongly influencing our results. We excluded these two outliers from the final data set. No outliers were found in the other samples.

TABLE 1 Descriptive Statistics and Correlations, Marketing Department Sample

Variable Mean s.d. 1 2 3 4 5 6 7 8 9

1. Estimated consensus 63.22 13.31 2. Actual consensus 61.96 8.09 .24 3. Actual network consensus 0.50 0.08 �.09 .23 4. In-degree centrality 1.36 1.50 .33† .26 �.14 5. Degree centrality 2.71 2.69 .46 .48 .07 .84** 6. Betweenness centrality 0.11 0.02 .37* .35* .06 .75** .84** 7. Closeness centrality 0.25 0.17 .41 .34 �.13 .69** .78** .50** 8. Status 2.58 1.17 .26 �.05 .19 �.26 .02 �.02 .05 9. Gender 1.74 0.45 .14 �.15 �.16 .32 .27 .27 .08 �.11

10. Age 34.25 7.36 .32 .22 .13 �.37 �.26 �.23 �.11 .56** �.41*

† p � .10 * p � .05

** p � .01

2010 1079Flynn and Wiltermuth

False Consensus Bias

Measures of false consensus. We posit in Hy- pothesis 1 that participants who were asked to judge the ethicality of a decision would demon- strate false consensus by assuming that others made choices similar to their own. In keeping with the false consensus effect (Ross et al., 1977), people holding one view of the ethicality of a decision estimated the popularity of that view (i.e., the pro- portion of others who made the same choice) to be higher than did those not holding that view. This effect appeared for each of the six dilemmas in the marketing department (all p’s � .01), MBA students (all p’s � .01), and executive students (all p’s � .01) samples.

Dawes (1989) pointed out that existence of the traditional false consensus effect does not necessar-

ily provide evidence of a judgmental bias. Noting this, we examined whether our participants dem- onstrated Krueger and Clement’s (1994) “truly false consensus effect” (TFCE) by examining within-in- dividual correlations between item endorsements and estimation errors. If participants exhibited these within-individual correlations, we could in- fer that they were not merely engaging in the sta- tistically appropriate Bayesian reasoning process of generalizing their own views to the population at large. We would know instead that there was a nonrational component to their social projection (see Krueger and Clement [1994: 596 –597] for a full discussion of this point). We found strong evidence across samples that participants did demonstrate this bias. The average TFCE within-subject correla- tion was positive and differed significantly from

TABLE 3 Descriptive Statistics and Correlations, Executive Student Sample

Variables Mean s.d. 1 2 3 4 5 6 7 8 9 10 11

1. Estimated consensus 68.60 13.24 2. Actual consensus 57.95 7.12 .02 3. Actual network consensus 0.44 0.12 .16 �.04 4. In-degree centrality 3.55 3.41 .25† �.05 �.04 5. Degree centrality 7.08 5.94 .31 �.02 �.15 .88** 6. Betweenness centrality 0.03 0.04 .46** �.06 �.11 .73** .82** 7. Closeness centrality 0.41 0.07 .36 �.06 �.08 .74** .90** .76** 8. Gender 1.25 0.43 �.07 �.21 .04 �.21 �.26† �.12 �.27†

9. Age 35.58 3.95 �.14 �.03 .01 �.15 �.08 �.16 .08 �.21 10. Asian 0.28 0.45 �.07 �.19 .02 .17 .11 .06 .15 �.07 .22 11. Latin 0.21 0.41 .05 .11 .08 �.10 �.15 �.03 �.14 .03 .08 �.31* 12. Other non-U.S. 0.12 0.32 �.10 �.03 .04 .07 .06 �.02 �.02 �.07 .18 �.22 �.17

† p � .10 * p � .05

** p � .01

TABLE 2 Descriptive Statistics and Correlations, MBA Student Sample

Variables Mean s.d. 1 2 3 4 5 6 7 8 9 10 11

1. Estimated consensus 64.66 10.18 2. Actual consensus 57.70 7.39 .14 3. Actual network consensus 0.53 0.09 .00 �.04 4. In-degree centrality 2.34 2.06 .19* .03 .11 5. Degree centrality 4.65 3.55 .21 .09 .11 .85** 6. Betweenness centrality 0.04 0.06 .23* .02 �.08 .55** .57** 7. Closeness centrality 0.27 0.14 .10 .08 .05 .60** .61** .49** 8. Gender 1.20 0.40 �.20* �.11 .00 �.02 �.04 �.02 .03 9. Age 29.39 5.43 .13 �.03 �.10 .07 .04 .07 .03 �.15

10. Asian 0.23 0.42 .05 .07 .06 �.12 �.09 �.14† �.10 �.01 .04 11. Latin 0.06 0.24 .02 .09 .05 .06 .10 .12 .07 �.13 �.05 �.14 12. Other non-U.S. 0.23 0.42 �.07 �.10 �.01 .10 .02 �.06 .19* .10 .06 �.30** �.14†

† p � .10 * p � .05

** p � .01

1080 OctoberAcademy of Management Journal

zero (r � .26, p � .001). Further, for 161 of the 239 participants in our samples, the correlation be- tween endorsement and the difference between es- timated and actual consensus was positive. The binomial probability of obtaining 161 or more pos- itive correlations out of 239 total correlations is less than .001.

We chose to follow the suggestion of de la Haye (2000), who argued that the partial correlation be- tween endorsement and estimated consensus with actual consensus controlled for better captures the existence of false consensus than does Krueger and Clement’s (1994) truly false consensus effect met- ric. As she pointed out, both the error and endorse- ment component of the correlation used to calcu- late the TFCE also correlate with the actual popularity of a given belief or behavior. To create an unbiased, within-individual measure of false consensus, it is therefore necessary to partial out the popularity of the belief from the correlation between endorsement and estimated consensus (see de la Haye [2000: 572–573] for a full discus- sion). We conducted this test in each of our three samples. In keeping with our expectations, the av- erage within-individual partial correlation was positive and significantly different from zero (r � .68, p � .001). Further, 85 percent of the partial correlations were positive. The binomial probabil- ity of obtaining this frequency of positive correla- tions by chance alone is less than .001. Taken to- gether, these results provide strong evidence of false consensus.

Overestimation. One might read these results and wonder if they apply to those individuals whose ethical judgments were inconsistent with the majority view (i.e., less than 50 percent of their colleagues or classmates agreed with their choice), given that the potentially negative consequences of false consensus bias in ethical decision making per- tain mainly to those who hold the minority opin- ion. In keeping with Hypothesis 1, people holding the minority view for each of the six dilemmas significantly overestimated the popularity of their viewpoints in the MBA (all p’s � .01) and executive (all p’s � .05) student samples. In the marketing department sample, the effect was significant for four dilemmas (all p’s � .05) and marginally signif- icant in the remaining two. Table 4 is a comparison of actual with estimated consensus.

The presence of the false consensus effect or even the truly false consensus effect (Dawes, 1989; Krueger & Clement, 1994) does not necessarily in- dicate that an individual believes he or she is in the majority. Nevertheless, it is worth noting that for all six dilemmas in each of the three samples (includ- ing all three MBA subsamples), participants who

were in the minority camp estimated that the ma- jority of their peers (i.e., greater than 50 percent) held the same view as they did.

We did not expect that people whose ethical judgments were in line with the majority view would overestimate the degree to which others held their beliefs (this effect has not been shown in previous work, such as the original studies con- ducted by Ross et al. [1977]). Indeed, people who held the majority view did not consistently overes- timate consensus. As seen in Table 4, people hold- ing the majority view overestimated the popularity of their response in only two of six dilemmas in each of our samples. For some dilemmas, those holding the majority view showed evidence of un- derestimation (this finding is consistent with those of other research on the false consensus bias).

Network Centrality Hypothesis

We tested Hypothesis 2, which posits a positive link between network centrality and false consen- sus bias, by examining the impact of multiple forms of network centrality. To understand the true indi- vidual effects of degree, betweenness, and close- ness centrality, Kilduff and Tsai (2003) recom- mended that researchers simultaneously control for the other two centrality measures. Unfortunately, including the three measures of centrality simulta- neously in the same regression equations in our samples revealed unacceptable levels of multicol- linearity (a variance inflation factor [VIF] � 2.5, tolerance � .40) (see Allison, 1999). We therefore entered data from all samples into one three-level hierarchical linear modeling (HLM) analysis that controlled for sample to determine the relative im- pacts of these different measures of network cen- trality.1 Once we had done so, the level of multi- collinearity became acceptable (VIF � 2.0, tolerance � .50).

To test our main hypothesis regarding the link between network centrality and false consensus

1 We also tested whether degree centrality (number of ties), when entered on its own, predicted the extent to which people believed more of their colleagues agreed with their decisions. Indeed, the more confirmed ties an individual had, the higher the estimated level of agree- ment in the marketing department, MBA student, and executive student samples. We then controlled for actual levels of agreement to test whether people with higher levels of degree centrality were more likely to demon- strate false consensus. Individuals’ degree centrality scores positively predicted estimated levels of agreement even given control for the actual level. This effect was significant in all three samples.

2010 1081Flynn and Wiltermuth

bias, we used three-level hierarchical linear ran- dom intercept models with estimated consensus as the dependent variable (Raudenbush & Bryk, 2002). This analysis allowed us to predict the degree of consensus each participant estimated for each di- lemma (level 1) using characteristics of the individ- ual (level 2) while controlling for each sample (level 3). Because the MBA student data were col- lected from three different sections of the same

course, we included separate dummy variables to capture whether the assigned section influenced the impact of network centrality on false consensus.

Following the suggestion of de la Haye (2000), we determined whether an individual-level variable influenced the size of the false consensus effect by testing whether the estimated levels of agreement were affected by the individual-level variables (i.e.

TABLE 4 Actual Versus Estimated Consensus by Classification of Decision Means

Sample and Dilemma

People Classifying Decision as Ethical People Classifying Decision as Unethical

Actual Percentage Classifying Decision

as Ethical

Estimated Percentage Classifying Decision

as Ethical (s.d.) t df p

Actual Percentage Classifying Decision

as Unethical

Estimated Percentage Classifying Decision

as Unethical t df p

Marketing department 1. Shifts 52.9 65.3 (21.0) 2.5 17 0.02 47.1 58.4 (22.4) 2 15 0.06 2. Supplies 81.8 75.6 (15.5) �2.1 26 0.05 18.2 52.0 (27.0) 3.1 5 0.03 3. Hire 60.6 61.3 (16.6) 0.2 19 0.86 39.4 56.7 (21.8) 2.9 12 0.01 4. Side business 20.6 57.9 (9.1) 10.9 6 �.001 79.4 57.5 (19.4) �5.9 26 �.001 5. Leave start-up 57.6 69.8 (20.2) 2.7 18 0.02 42.4 50.0 (16.2) 1.8 13 0.10 6. Layoffs 12.1 60.0 (8.2) 11.7 3 �.001 87.9 66.4 (18.7) �6.16 28 �.001

MBA students Section 1

1. Shifts 71.2 67.4 (16.3) �1.4 36 0.17 28.8 67.0 (17.0) 8.7 14 �.001 2. Supplies 78.8 67.3 (16.6) �4.4 40 �.001 21.2 51.4 (23.7) 4.2 10 �.01 3. Hire 48.1 64.4 (15.2) 5.4 24 �.001 51.9 65.2 (20.5) 3.4 26 �.01 4. Side business 36.5 60.3 (18.7) 5.5 18 �.001 63.5 62.0 (20.7) �0.4 32 .670 5. Leave start-up 63.5 67.1 (14.6) 1.4 32 0.16 36.5 60.5 (18.8) 5.6 18 �.001 6. Layoffs 21.2 56.4 (15.2) 7.7 10 �.001 78.8 65.4 (17.2) �5 40 �.001

Section 2 1. Shifts 68.1 73.4 (19.4) 1.6 31 0.13 31.9 52.0 (18.1) 4.3 14 �.001 2. Supplies 76.6 73.4 (17.2) �1.1 35 0.27 23.4 66.8 (25.6) 5.6 10 �.001 3. Hire 48.9 61.6 (18.1) 4.2 22 �.001 51.1 67.3 (17.2) 4.6 23 �.001 4. Side business 31.9 54.0 (13.8) 6.2 14 �.001 68.1 63.0 (16.8) �1.7 31 0.10 5. Leave start-up 67.4 73.5 (15.0) 2.3 30 0.03 32.6 59.0 (13.3) 7.7 14 �.001 6. Layoffs 14.9 51.4 (17.7) 5.5 6 �.001 85.1 55.5 (16.7) �7.0 39 �.001

Section 3 1. Shifts 67.9 75.6 (17.8) 2.7 37 0.01 32.1 51.4 (17.9) 4.6 17 �.001 2. Supplies 73.7 69.1 (19.9) �1.5 41 0.15 26.3 51.7 (22.1) 4.5 14 �.01 3. Hire 50.9 61.6 (18.1) 3.2 28 �.01 49.1 60.2 (19.6) 3.0 27 �.01 4. Side business 41.8 57.6 (19.9) 3.8 22 �.01 58.2 63.9 (19.3) 1.7 31 0.11 5. Leave start-up 66.1 69.1 (15.9) 1.2 36 0.25 33.9 60.0 (12.5) 9.1 18 �.01 6. Layoffs 30.9 62.6 (24.4) 5.4 16 �.001 69.1 63.3 (18.6) �1.9 37 0.06

Executive students 1. Shifts 54.7 77.6 (15.8) 7.8 28 �.001 45.3 57.1 (23.1) 2.5 23 0.02 2. Supplies 69.8 71.3 (17.9) 0.5 36 0.61 30.2 63.1 (26.5) 5 15 �.001 3. Hire 43.4 63.0 (18.4) 5.1 22 �.001 56.6 68.8 (17.5) 0.7 29 0.49 4. Side business 20.8 59.2 (21.2) 5.8 10 �.001 79.2 72.7 (22.5) �1.9 41 0.07 5. Leave start-up 60.4 70.1 (16.1) 3.4 31 �.01 39.6 64.0 (20.2) 5.6 20 �.001 6. Layoffs 17.3 57.2 (10.3) 11.5 8 �.001 82.7 73.8 (20.9) �2.8 42 �.01

1082 OctoberAcademy of Management Journal

in-degree centrality, betweenness centrality, close- ness centrality, and demographic variables) after controlling for the actual level of agreement. To facilitate interpretation of these results, we cen- tered all continuous predictors (Cohen, Cohen, West, & Aiken, 2003). In addition to controlling for age and gender, we ran exploratory analyses within HLM to identify whether our measure of hierarchi- cal status in the marketing department sample pre- dicted estimated or actual levels of consensus. Given that status did not predict either estimated or actual consensus and given that we did not have such status differences in the other samples, we did not include it in the final models.

Table 5 presents the random intercept models for the combined samples. As shown in model 1 of Table 5, when all three measures of network cen- trality were included simultaneously in the same regression equation (e.g., Oh & Kilduff, 2008), be- tweenness centrality positively predicted esti- mated levels of consensus, whereas the effects of degree centrality and closeness centrality were not significant. The results of model 2, which regressed actual levels of consensus on the three measures of

centrality, indicated that no measure of centrality affected actual levels of consensus. Most impor- tantly, model 3 showed that individuals’ between- ness centrality scores positively predicted esti- mated levels of agreement even after we controlled for the actual level of consensus. In contrast, degree centrality and closeness centrality did not signifi- cantly predict estimated consensus when actual consensus was included in the HLM equation. Al- though not conclusive, these results suggest that the link between false consensus bias and network centrality may be driven by betweenness (being a broker who can control the flow of information in a network), rather than by network size or closeness centrality.

Supplementary Analyses

One alternative explanation for our results could be that people with higher levels of betweenness accurately estimate agreement among those indi- viduals with whom they have ties, but not among the entire group. To test this possibility, we calcu-

TABLE 5 Results of Random Intercept Analysis of Combined Samples

Dependent Variables Model 1: Estimated

Consensus Model 2: Actual

Consensus Model 3: Estimated

Consensus Model 4: Estimated

Consensus

Fixed effects Intercept 66.48*** 61.91*** 65.64*** 66.11*** MBA sample 2 1.17 0.59 1.03 1.17 MBA sample 3 �0.52 �3.16 0.24 �0.36 Executive student sample 3.07 2.34 1.95 2.42 Marketing department sample 2.52 �1.39 3.39 3.99 Degree centralitya 0.16 �0.11 0.19 0.16 Betweenness centrality 50.99** �1.63 51.35** 51.87** Closeness centrality 5.49 6.73 3.89 4.66 Age �0.09 0.14 �0.12 �0.09 Gender �2.36 �1.81 �1.93 �2.29 Asian 0.83 �1.08 1.08 0.90 Latin 0.51 �0.93 0.73 0.71 Other non-U.S. 1.49 �1.57 1.86 1.63 Actual consensus in class 0.24*** Actual consensus in help/advice network 0.05**

Variance Component (R2)

Random effects Level 1 288.96 (.01) 344.18 (.02) 268.80 (.08) 287.11 (.02) Level 2 58.94 (.18) 0.25 (.06) 59.47 (.17) 57.82 (.20)

Deviance 11,453 11,509 11,368 11,443 �2 492.79*** 186.50*** 515.72*** 489.37*** Intraclass correlation 0.17 0.00 0.18 0.17

a Network size; a standardized measure of degree centrality was used in the analysis. ** p � .01

*** p � .001

2010 1083Flynn and Wiltermuth

lated another measure of actual agreement using only the responses from a participant’s set of con- firmed ties. We then ran a separate model using this alternative measure of actual consensus. As shown in model 4 in Table 5, only betweenness centrality remained a significant predictor of social projec- tion when degree, betweenness, and closeness cen- trality were all included in the equation.

Having inflated estimates of agreement may be most problematic for people whose ethical judg- ments differ from the prevailing view. Noting this, we examined whether network centrality corre- lated with estimates of consensus for participants in the minority (less than 50 percent of respon- dents). These results can be seen in Table 6. When we examined only those cases in which an individ- ual held the minority view, we found once again that betweenness centrality positively correlated with estimated consensus when actual consensus was controlled for. Neither degree centrality nor closeness centrality significantly predicted esti- mated consensus when under control for actual consensus.

DISCUSSION

We must not say that an action shocks the con- science collective because it is criminal, but rather that it is criminal because it shocks the conscience collective. We do not condemn it because it is a crime, but it is a crime because we condemn it (Durkheim, 1972: 123–124).

Ethical standards are derived from socially agreed upon principles and values (Turiel, 2002) that can change over time (McConahay, 1986) and vary among institutions (Weaver, Treviño, & Coch- ran, 1999) and national cultures (Haidt et al., 1993). Given that ethical standards often are clandestine, ethical decision making in organizations can be described as a problem of social judgment— deter- mining where the majority of others stand on issues of moral concern. Individuals are motivated to con- sider the attitudes and opinions of their peers in order to assess where they stand relative to the norm and to help gauge others’ reactions to their decisions. However, we found strong evidence in each of our samples that people are not very good at

TABLE 6 Results of Random Intercept Analysis of Projection of Minority Views in Combined Samples

Dependent Variables Model 1: Estimated

Consensus Model 2: Actual

Consensus Model 3: Estimated

Consensus Model 4: Estimated

Consensus

Fixed effects Intercept 56.55*** 37.93*** 55.91*** 56.94*** MBA sample 2 �2.01 �4.07 �0.90 �2.49 MBA sample 3 �4.27 �0.78 �3.91 �5.35 Executive student sample �5.50 �3.79 �4.43 �3.71 Marketing department sample �2.38 �0.88 �1.95 �0.76 Degree centralitya 0.09 �0.21 0.13 0.07 Betweenness centrality 67.81** 6.99 66.00** 72.74** Closeness centrality 0.67 3.42 �0.04 �0.03 Age 0.04 0.13 0.00 �0.14 Gender 2.96 �0.29 3.11 2.79 Asian 0.36 0.14 0.32 0.55 Latin 5.43 0.53 5.17 3.16 Other non-U.S. 2.32 0.47 2.04 1.94 Actual consensus in class 0.24** Actual consensus in help/advice network 0.00

Variance Component (R2)

Random effects Level 1 282.02 (.00) 96.89 (.02) 270.34 (.04) 278.73 (.01) Level 2 62.39 (.26) 0.05 (.11) 69.87 (.17) 59.22 (.29)

Deviance 4,068 3,484 4,060 3,897 �2 317.63*** 143.93*** 234.71*** 148.60*** Intraclass correlation 0.18 0.00 0.21 0.18

a Network size; a standardized measure of degree centrality was used in the analysis. ** p � .01

*** p � .05

1084 OctoberAcademy of Management Journal

estimating the percentage of others who agree with their moral choices. Instead, they fall victim to a false consensus bias, wanting to believe that others are more like them than not.

Our findings challenge the idea that more central individuals in a social network tend to be more in line with shared ethical standards (Brass et al., 1998). Instead, we found that having higher levels of betweenness in an advice network (i.e., acting as a broker) may put people in a dangerous position when determining what they should do and whether their actions are in line with the prevailing moral view. When responding to an ethical di- lemma, these individuals increased their estimates of agreement with their colleagues on ethical deci- sions, going significantly above and beyond any actual increase in agreement. In short, individuals who were positioned as brokers (i.e., had control over information flow in the network), including those who held the minority view, showed greater evidence of false consensus, wrongly assuming that their ethical judgments were in line with the ma- jority of their colleagues.

Theoretical Implications

Research on ethical decision making in organiza- tions has become increasingly focused on the psy- chological drivers of ethical judgment (Tenbrunsel & Messick, 2004). We add to this growing field of research by highlighting the influence of false con- sensus bias and suggesting that social projection might play a critical role in employees’ judgments of ethical standards. We note the depth of existing research on false consensus (for a review, see Gross and Miller [1997]), which has identified situational and cognitive factors that reduce the magnitude of the bias, including differential construal (e.g., Gilovich, 1990), issue objectivity (e.g., Biernat, Ma- nis, & Kobrynowicz, 1997), and attitude importance (e.g., Fabrigar & Krosnick, 1995). These moderating variables may be of interest to organizational ethics scholars, especially those attempting to identify po- tentially effective interventions.

A second, and relatively more noteworthy, con- tribution of the present research is the link between false consensus bias and social networks. Part of what shapes individuals’ understanding of social norms, or consensus views, is their set of social relations. One might reasonably expect individuals with greater betweenness centrality to be less vul- nerable to false consensus bias because they are more closely connected to other members of their referent group and enjoy an information-based advantage. By being well-connected, these actors appear to be in a better position to acquire infor-

mation that may be useful in formulating social judgments.

However, contrary to this straightforward as- sumption and some findings from previous re- search (e.g., Krackhardt, 1987), our results suggest that those individuals who have more advanta- geous positions in an advice network, specifically in terms of betweenness, are not necessarily more accurate in their judgments of others’ views.

We suggest that part of what drives the connec- tion between betweenness centrality and false con- sensus in ethical decision making is an underlying link between power and social judgment. However, we recognize that this link may not apply to other domains of decision making. Although ethical stan- dards are socially shared, they often remain im- plicit. Brokers may be prone to overestimate their ability to diagnose ethical standards because per- sonal moral values are not often shared; the same individuals may be much more accurate in diag- nosing other social standards (e.g., dress code, speaking order at meetings) that are evident and perhaps explicit. Noting this, theorizing about the influence of network centrality on decision making in organizations should account for the type of decision being made and the availability of the data needed to make an informed decision.

Practical Implications

Overestimating support for one’s ethical judg- ments may lead to costly decision making errors in organizations. Theories of moral reasoning empha- size the notion of consensus (Haidt, 2008), which Kohlberg (1969) described as the conventional ap- proach to ethical decision making. According to Treviño, most managers adopt this conventional approach, looking “to others and to the situation to help define what is right and wrong, and how they should behave in a particular situation” (1986: 608). But what if their views of others are mistaken? Our results suggest an intriguing possibility: that some people who perform unethical acts could be confident that their actions are ethical because they hold a false expectation that others share their eth- ical views.

An intuitive solution to the problem of faulty social projection is developing more expansive ad- vice networks (Ickes, 1997; Ickes & Aronson, 2003). However, our findings suggest that this solution might have the opposite effect—worsening individ- uals’ social judgments. If having an advantageous position in an advice network actually predisposes people to be more susceptible to false consensus, ethics scholars need to account for this problem in offering useful advice to practitioners. Specifically,

2010 1085Flynn and Wiltermuth

greater insight into peers’ ethical attitudes may be gained from having more meaningful interpersonal conversations rather than a more influential network location. Though acting as a broker can undoubtedly provide insight, such insight is likely limited to what people discuss or what they can easily observe. Moral attitudes often are unspoken and therefore require further investigation and interrogation.

Limitations and Directions for Future Research

In operationalizing ethical dilemmas, we chose to focus on right versus right decisions, rather than on unambiguously immoral acts. But perhaps false consensus bias is mitigated when an act in question is clearly unscrupulous. In a separate study, we attempted to investigate this possibility by using a set of ten “moral temptations,” ranging from steal- ing office supplies to downloading files illegally at work. Following the same procedure described in the present study (but with this alternative opera- tionalization), we found strong evidence of the false consensus bias once again. In each case, peo- ple overestimated the percentage of others who be- haved similarly to them. Although this result is reassuring, additional research is needed to test whether our operationalization of ethical dilemmas influenced our results.

Additional research is also needed to clarify the psychological mechanisms that help explain our findings. Recent research by social psychologists offers compelling evidence that powerful actors tend to be poor perspective takers and that they tend to assume that others have views that are similar to their own—more similar than is actually the case (Galin- sky et al., 2006; Gruenfeld, Inesi, Magee, & Galinsky, 2008). Future research might test whether this con- nection between power and perspective taking can help explain the link between network centrality and false consensus bias by measuring a focal individual’s sense of power and examining whether it mediates between network centrality and false consensus bias or by manipulating the individual’s sense of power (e.g., Gruenfeld et al., 2008) and examining whether it strengthens false consensus bias in ethical decision making.

We also proposed that members of organizations are susceptible to false consensus bias in making ethical judgments because people generally refrain from conversations about morality. A future study could test this idea directly by asking participants to estimate agreement on some topics they rarely discuss with others (ethics, sex) and some they regularly discuss (sports, kids). If our logic holds, we would expect to find less evidence of false consensus in estimates of the popularity of opin-

ions that are readily disclosed than in estimates of the popularity of those that are often withheld. One could also conduct an intervention in which co- workers are asked to meet and discuss ethical is- sues on a routine basis. Such an intervention might be effective in undermining false consensus bias if the content of the coworkers’ conversations is rel- evant to their ethical decision making.

Finally, in our study design, we did not leave it up to the participants to select their referent groups. Instead, we identified a group for each participant to consider. As a consequence, we know that partici- pants made their estimates with the same group of peers in mind. We chose to have people focus on the opinions of other members of their department or class because initial interviews indicated that the par- ticipants worked with these individuals closely and cared about their opinions deeply. However, if people were left to their own devices, would they consider a different referent group, and could this affect our results? We think it might. In a tight-knit circle of close colleagues, the exchange of moral opinions may be relatively more common because the sense of psy- chological safety is elevated, and this, in turn, may undermine any false consensus effect. In future re- search, this issue of referent group selection certainly deserves closer scrutiny.

Conclusion

Ethical decisions in organizations are calibrated according to what people believe to be the majority view, but having a more central position in the social structure seems to do little to help employees accurately calibrate their ethical judgments. Ethical pundits predict that “loners” are more likely to violate ethical standards because they lack insight into others’ attitudes and opinions and have fewer colleagues to whom they can turn for help and advice. But is this true? Our research suggests that the social butterfly, rather than the social outcast, may be the more likely to misjudge whether an ethical judgment is in line with the normative view. The potential danger here is that employees who act in unethical ways could, in some cases, erroneously assume that their actions are in line with the socially shared ethical standard and only learn of their misjudgment when it is too late to avert the consequences.

REFERENCES

Allison, P. D. 1999. Multiple regression: A primer. Lon- don: Pine Forge Press.

Badaracco, J. L., Jr. 1997. Defining moments: When

1086 OctoberAcademy of Management Journal

managers must choose between right and right. Boston: Harvard Business School Press.

Biernat, M., Manis, M., & Kobrynowicz, D. 1997. Simul- taneous assimilation and contrast effects in judg- ments of self and others. Journal of Personality and Social Psychology, 73: 254 –269.

Brass, D. J. 1984. Being in the right place: A structural analysis of individual influence in an organization. Administrative Science Quarterly, 29: 518 –539.

Brass, D. J., Butterfield, K. D., & Skaggs, B. C. 1998. Relationships and unethical behavior: A social net- work perspective. Academy of Management Re- view, 23: 14 –31.

Burt, R. 1992. Structural holes: The social structure of competition. Cambridge, MA: Harvard University Press.

Burt, R. 2007. Secondhand brokerage: Evidence on the importance of local structure for managers, bankers, and analysts. Academy of Management Journal, 50: 119 –148.

Carley, K. M., & Krackhardt, D. 1996. Cognitive inconsis- tencies and non-symmetric friendship. Social Net- works, 18: 1–27.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. 2003. Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Erlbaum.

Colby, A., Gibbs, J., Kohlberg, L., Speicher-Dubin, B., & Candee, D. 1980. The measurement of moral judg- ment, vol. 1. Cambridge, MA: Center for Moral Edu- cation, Harvard University.

Davis, H., Hoch, S., & Ragsdale, E. 1986. An anchoring and adjustment model of spousal predictions. Jour- nal of Consumer Research, 13: 25–37.

Dawes, R. M. 1989. Statistical criteria for a truly false consensus effect. Journal of Experimental Social Psychology, 25: 1–17.

de la Haye, A. M. 2000. A methodological note about the measurement of the false-consensus effect. Euro- pean Journal of Social Psychology, 30: 569 –581.

Durkheim, E. 1972. Anomie and the moral structure of industry, and the social bases of education. In A. Giddens (Ed.), Emile Durkheim: Selected writings. New York: Cambridge University Press.

Erber, R., & Fiske, S. T. 1984. Outcome dependency and attention to inconsistent information. Journal of Personality and Social Psychology, 47: 709 –726.

Fabrigar, M. E., & Krosnick, J. A. 1995. Attitude impor- tance and the false consensus effect. Personality and Social Psychology Bulletin, 21: 468 – 479.

Flynn, F. J. 2003. How much should I help and how often? The effects of generosity and frequency of favor exchange on social status and productivity. Academy of Management Journal. 46: 539 –553.

Flynn, F. J., Reagans, R., Amanatullah, E., & Ames, D.

2006. Helping one’s way to the top: Self-monitors achieve status by helping others and knowing who helps whom. Journal of Personality and Social Psy- chology, 91: 1123–1137.

Freeman, L. 1979. Centrality in social networks: Concep- tual clarification, Social Networks, 1: 215–239.

Fritzsche, D. J., & Becker, H. 1984. Linking management behaviour to ethical philosophy—An empirical in- vestigation. Academy of Management Journal, 27: 166 –175.

Galinsky, A. D., Magee, J. C., Inesi, M. E., & Gruenfeld, D. H. 2006. Power and perspectives not taken. Psy- chological Science, 17: 1068 –1074.

Gilovich, T. 1990. Differential construal and the false consensus effect. Journal of Personality and Social Psychology, 59: 623– 634.

Gross, S., & Miller, N. 1997. The “golden section” and bias in perceptions of social consensus. Personality and Social Psychology Review, 1: 241–271.

Gruenfeld, D. H., Inesi, M. E., Magee, J. C., & Galinsky, A. D. 2008. Power and the objectification of social targets. Journal of Personality and Social Psychol- ogy, 95: 111–127.

Haidt, J. 2001. The emotional dog and its rational tail: a social intuitionist approach to moral judgment. Psy- chological Review, 108: 814 – 834.

Haidt, J. 2008. Morality. Perspectives on Psychological Science, 3: 65–72.

Haidt, J., Koller, S., & Dias, M. 1993. Affect, culture, and morality, or is it wrong to eat your dog? Journal of Personality and Social Psychology, 65: 613– 628.

Hollingworth, M. 2007. Re-engineering today’s core business process—Strong and truthful conversation. Ivey Business Journal Online, http://www.ivey businessjournal.com/article.asp?intArticle 10-716.

Ibarra, H. 1992. Homophily and differential returns: Sex differences in network structure and access in adver- tising firm. Administrative Science Quarterly, 37: 422– 447.

Ibarra, H., Kilduff, M., & Tsai, W. 2005. Zooming in and out: Connecting individuals and collectivities at the frontiers of organizational network research. Orga- nization Science, 16: 359 –371.

Ickes, W. 1997. Empathic accuracy. New York: Guilford.

Ickes, W., & Aronson, E. 2003. Everyday mind-reading: Understanding what other people think and feel. New York: Prometheus.

Kanter, R. M. 1979. Power failure in management cir- cuits. Harvard Business Review, 57(4): 65–75.

Keltner, D., Gruenfeld, D. H., & Anderson, C. 2003. Power, approach, and inhibition. Psychological Re- view, 110: 265–284.

Kidder, R. M. 1995. How good people make tough choic-

2010 1087Flynn and Wiltermuth

es: Resolving the dilemmas of ethical living. New York: Fireside.

Kilduff, M., & Tsai, W. 2003. Social networks and or- ganizations. London: Sage.

Kohlberg, L. 1969. Stage and sequence: The cognitive- developmental approach to socialization. In D. A. Goslin (Ed.), Handbook of socialization theory and research: 347– 480. Chicago: Rand McNally.

Kohlberg, L. 1981. The philosophy of moral develop- ment: Moral stages and the idea of justice. San Francisco: Harper & Row.

Krackhardt, D. 1987. Cognitive social structures. Social Networks, 9: 109 –134.

Krueger, J. 1998. On the perception of social consensus. In M. P. Zanna (Ed.), Advances in experimental social psychology, vol. 30: 163–240. New York: Ac- ademic Press.

Krueger, J., & Clement, R. W. 1994. The truly false con- sensus effect: An ineradicable and egocentric bias in social perception. Journal of Personality and Social Psychology, 67: 596 – 610.

Krueger, J., & Clement, R. W. 1997. Estimates of social consensus by majorities and minorities: The case for social projection. Personality and Social Psychol- ogy Review, 1: 299 –313.

Krueger, J., & Zeiger, J. S. 1993. Social categorization and the truly false consensus effect. Journal of Person- ality and Social Psychology, 65: 670 – 680.

Mackie, J. L. 1977. Ethics: Inventing right and wrong. New York: Penguin.

Mannix, E., Neale, M., & Tenbrunsel, A. (Eds.). 2006. Research on managing groups and teams: Ethics in groups, vol. 8. Oxford, U.K.: Elsevier Science.

Marks, G., & Miller, N. 1987. Ten years of research on the false-consensus effect: An empirical and theoretical review. Psychological Bulletin, 102: 72–90.

Marsden, P. 1988. Homogeneity in confiding relations. Social Networks, 10: 57–76.

McConahay, J. B. 1986. Modern racism, ambivalence, and the modern racism scale. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and rac- ism: 91–126. New York: Academic.

Oh, H., & Kilduff, M. 2008. The ripple effect of person- ality on social structure: Self-monitoring origins of network brokerage. Journal of Applied Psychology, 93: 1155–1164.

Payne, S. L., & Giacalone, R. A. 1990. Social psycholog- ical approaches to the perception of ethical dilem- mas. Human Relations, 43: 649 – 665.

Phillips, N. 1992. The empirical quest for normative meaning: Empirical methodologies for the study of business ethics. Business Ethics Quarterly, 2(2): 223–244.

Raudenbush, S. W., & Bryk, A. S. 2002. Hierarchical

linear models: Applications and data analysis methods (2nd ed.). Newbury Park, CA: Sage.

Ross, L., Greene, D., & House, P. 1977. The false consen- sus effect: An egocentric bias in social perception and attribution processes. Journal of Experimental Social Psychology, 13: 279 –301.

Sabini, J., Cosmas, K., Siepmann, M., & Stein, J. 1999. Underestimates and truly false consensus effects in estimates of embarrassment and other emotions. Ba- sic and Applied Social Psychology, 21(3): 223–241.

Sabini, J., & Silver, M. 1982. Moralities of everyday life. Oxford, U.K.: Oxford University Press.

Schweder, R. 1982. Review of Lawrence Kohlberg’s “Es- says in moral development,” (vol. I, D). Contempo- rary Psychology, 27: 421– 424.

Skitka, L. J., Bauman, C. W., & Sargis, E. G. 2005. Moral conviction: Another contributor to attitude strength or something more. Journal of Personality and So- cial Psychology, 88: 895–917.

Stasser, G., Taylor, L. A., & Hanna, C. 1989. Information sampling in structured and unstructured discussions of three- and six-person groups. Journal of Person- ality and Social Psychology, 57: 67–78.

Sutherland, E., & Cressey, D. R. 1970. Principles of crim- inology (8th ed.). Chicago: Lippincott.

Tenbrunsel, A. E., & Messick, D. M. 2004. Ethical fading: The role of rationalization in unethical behaviour. Social Justice Research, 17: 223–236.

Toffler, B. L. 1986. Tough choices: Managers talk eth- ics. New York: Wiley.

Treviño, L. K. 1986. Ethical decision making in organi- zations: A person-situation interactionist model. Academy of Management Review, 11: 601– 617.

Turiel, E. 2002. The culture of morality: Social devel- opment, context, and conflict. New York: Cam- bridge University Press.

Weaver, G. R., Treviño, L. K., & Cochran, P. L. 1999. Integrated and decoupled corporate social perfor- mance: Management commitments, external pres- sures, and corporate ethics practices. Academy of Management Journal, 42: 539 –553.

White, D. R., & Borgatti, S. P. 1994. Betweenness central- ity measures for directed graphs. Social Networks, 16: 335–346.

Wieder, D. L. 1974. Language and social reality: The case of telling the convict code. The Hague, Neth- erlands: Mouton.

APPENDIX

Ethical Dilemma Scenarios

Dilemma 1: Shifts

You are in charge of testing a new software package that your company has recently developed. It will be

1088 OctoberAcademy of Management Journal

launched in a week, which means you will need to set up round-the-clock testing before then. You have to assign people to two teams— one daytime shift and one grave- yard shift. You decide to let your married employees off of the graveyard shift because many of them have kids.

Dilemma 2: Supplies

You notice one of your best employees taking printer paper, highlighters, and post-it notes home in her laptop bag. This employee has worked at the firm for many years, but there is a rule against this and clear procedures for providing employees with supplies if they choose to work at home. According to company policy, you are required to fire this employee on the spot. You decide not to fire her.

Dilemma 3: Hire

Your colleague, who you consider to be a friend, is looking to hire a new manager in her department. She has identified an external candidate she would like to hire, but company rules require her to consider internal can- didates first. She has asked you not to disclose to people within the company that she has already picked out an external candidate for the position. However, you know two employees in your area who would like to have this job, and each has asked you directly if your colleague has already picked someone for this position. You decide to tell them that she has not picked anyone yet.

Dilemma 4: Side Business

You work in a small division of a large company. Two of your colleagues, whom you are friends with outside of work, have been working on a new business venture together. Although it is against company policy, you notice that they have been spending a significant amount of time at work making plans for this new business. Despite their involvement in this side business, these colleagues have always made time to help you with the

issues you encounter at work. Your boss, who is con- cerned by the declining performance of your group, asks you if these colleagues are using company time to pursue interests not related to the company. You decide to cover for your colleagues and tell your boss that they are not using company time to pursue their own interests.

Dilemma 5: Leave Start-up

You manage a small company that is trying to secure an additional round of venture-capital financing. The firm employs five people, each of whom has an irreplace- able set of skills. If any of the five were to leave, the company would struggle to secure additional financing. One of the principal employees, whom you consider a friend, has recently informed you that he has received an extremely appealing offer from another company that is much more likely to succeed. The employee must make a decision in the next two days. Out of respect for you, this employee has told you that he will go to the other com- pany only if you offer your blessing. You decide to dis- courage this employee from leaving out of concern for the group.

Dilemma 6: Layoffs

You manage a medium-sized company that is located in a small town. Unfortunately, you are forced to lay off a third of your workforce in six months time. You know that as soon as you announce the layoffs property prices in the small town will fall off considerably, as will the effort of the company’s employees. One of your favorite employees, whom you admire very much, has been going through some hard times financially. You would like to give this employee some advance notice so that he could sell his house for a reasonable price. However, you know that if you tell him to sell the house there is a chance the rest of the company would read the sale as a sign that layoffs are imminent long before the planned announce- ment date. If this were to happen, not only would prop- erty prices drop, so too would firm productivity. None- theless, you decide to tell this employee to sell his house.

Francis J. Flynn ([email protected]) is an associate professor of organizational behavior at Stan- ford University. His research investigates how employ- ees develop healthy patterns of cooperation, how ste- reotyping in the workplace is mitigated, and how leaders in organizations acquire power and influence. He received his doctorate from the University of Cali- fornia, Berkeley.

Scott S. Wiltermuth ([email protected]) is an assistant professor of management and organizations at the Uni- versity of Southern California. His research investigates how interpersonal dynamics, such as dominance and submissiveness, influence cooperation and coordination. He also studies ethics and morality. He received his doctorate from Stanford University.

2010 1089Flynn and Wiltermuth

Copyright of Academy of Management Journal is the property of Academy of Management and its content may

not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.