Disobedience of Immoral Orders from Authorities: An Issue Construction Perspective

Abstract

The purpose of this study is to examine how disobedience to immoral orders from an authority emerges in organizations. Using organizational discourse analysis to analyze the verbal communication of the original participants in Stanley Milgram’s obedience to authority experiments, I show how participants constructed the same experimental situation differently by analyzing their communication. Disobedient participants were more likely to display two different communication patterns: assessing consequences or self-referential objections. In contrast, obedient participants were more likely to seek guidance on the experimental procedure and interrupt the learner’s protests. Overall, I present a process model of how disobedience emerges in situations. This study’s findings also expand our understanding of moral imagination, moral decision making, and employee voice in organizations primarily by demonstrating how people can exercise agency in equivocal situations by constructing the situations they face.

Keywords

ethics organizational discourse analysis qualitative sensemaking theory theoretical perspectives

Authority has been defined by some scholars as the ability of an individual, or group of individuals, to legitimately dictate the behavior of others by virtue of their position in an organizational hierarchy (Weber, 1924). Scholars have shown that authority emerges when cooperative work requires delegation to intermediary agents, who have different interests and levels of expertise (Okhuysen & Bechky, 2009; Osterman, 2006;). Although the basis of authority may differ (Weber, 1947), it can enable organizations to carry out their activities efficiently (Barnard, 1968; Weber, 1924).

Authority’s efficiency is also accompanied by risks. Specifically, authority can be used to perpetuate corrupt or unethical practices when there are differences between the actions the authority requests and those which the larger organization or society deems acceptable (Pina e Cunha, Rego, & Clegg, 2010; Warren, 2003). Managers feel pressure to do things that they suspect are morally wrong in their work (Ashforth, Gioia, Robinson, & Treviño, 2008; Brief, Buttram, & Dukerich, 2001). Yet individuals rarely speak up to stem illegal or immoral behavior (Detert & Edmondson, 2011; Ryan & Oestrich, 1998). This reticence may be explained in part by the idea that when individuals enter into a labor contract, they expect to put aside their personal preferences and focus on what their employers want (Simon, 1960). Additionally, in large complex organizations, authority relationships are nested within other authority relationships (Barnard, 1968). Thus individuals can become accustomed to accepting directives from impersonal authorities, and come to consider obeying requests from authorities as an organizational duty, not a matter of their personal values (Kelman & Hamilton, 1989; Posner & Schmidt, 1993). When individuals spend more time looking up for direction than looking around to discern a morally acceptable course of action, obedience to authority can function as a counter-productive cognitive short-cut (Ashforth & Anand, 2003; Greve, Palmer, & Pozner, 2010) that helps perpetuate organizational corruption and fraud (Ashforth et al., 2008; Bandura, 1997; Misangyi, Weaver, & Elms, 2008).

Individuals can prevent unethical practices from spreading in and around organizations by disobeying what they perceive as immoral orders from an authority. Stanley Milgram’s extensive work on obedience to authority (Milgram, 1963, 1974) demonstrates how difficult it can be to disobey immoral orders from an authority, even when that authority has no control over access to projects, promotions, or other resources. Control over these kinds of resources only increases the difficulty. For example, Jackall (1988) argues that the inherent ambiguity of managerial work coupled with constant political turmoil in organizations causes managers to look up and aim to be consistent with their direct supervisors when making decisions to protect their careers. Similarly, Detert and Edmondson (2011) demonstrate that employees hold several implicit theories that prevent them from speaking up, such as “don’t embarrass the boss in public” and “don’t bypass the boss upward.” Disobedience is made more unlikely by the fact that the costs of exercising voice are born by the individual, while the benefits are dispersed (Detert & Burris, 2007).

Early research on disobedience argues that disobedience is most likely when situational forces are less powerful than a subordinate’s level of personal or psychological strain (Milgram, 1974, p. 153). According to Milgram, situational elements that enable obedience in his experiments included the obligation to the authority, the sequential nature of the task, and the physical and psychological distance between the confederate and the experimenter. Sources of strain in his experimental conditions included the confederate’s cries of pain, the guilt participants felt in administering shocks to an innocent person, fear of retaliation, and any other actions that were incongruous with their own self-image (Milgram, 1974). In the case of the obedience experiments, considerable effort has been expended to find evidence for underlying personality factors in predicting disobedience, but the evidence has been mixed at best (Blass, 2009; Burger, 2009; Elms & Milgram, 1966). Milgram argued that when the situation allowed the sources of strain to be more salient than the elements that increased obedience, disobedience was more likely. A recent replication using a different experimental procedure confirms the effect of varying the salience of situational factors on disobedience (Bocchario & Zimbardo, 2010).

Building on Milgram’s work, social psychologists, Kelman and Hamilton argued that

The ability to redefine the authority situation and hence to challenge authoritative demands presupposes two psychological conditions that are not readily available. In the cognitive sphere, individuals must be aware of—or be able to conceive—alternative definitions of the situation and interpretations of its requirements. In the motivational sphere, they must be prepared to disrupt established social order and the smoothness of social interaction and to suffer the consequences of such disruption. These two psychological conditions are closely linked to access to material resources, without which individuals are often unable to learn about available options or before the risk of disruption. (Kelman & Hamilton, 1989, p. 138)

Examining participants across Milgram’s 18 conditions illustrates that situational resources do indeed shape behavior, yet it does not explain how participants disobeyed within the same experimental condition (Haslam & Reicher, 2012; Lutsky, 1995). This paper aims to fill this gap by examining the process by which disobedience emerges in participants’ verbal construction of the issues—which can be an indicator of how they make meaning (Fairclough, 1989). In particular, without access to other people, money, formal hierarchical power, and a clear choice of alternatives, how do individuals generate the interpretation of the situation and the will to act that Kelman and Hamilton (1989) argue are necessary for disobedience?

To answer these questions, I utilize organizational discourse analysis (Fairhurst & Uhl-Bien, 2012) to analyze the audio tapes and transcripts from Milgram’s original obedience experiments (Milgram, 1963, 1974) in order to understand the communication patterns that distinguish disobedient from obedient individuals within the same condition. The findings offer a different view of disobedience than the literature has assumed (Burger, 2009; Milgram, 1974; Reicher, Haslam, & Smith, 2012; Ross & Nisbett, 1991). I find that disobedient participants made the consequences of their actions and their personal preferences more salient through communication; the data suggest that these individuals engaged in cycles of sensemaking, until they were able to construct an interpretation of the situation that problematized the experiment and indicated a will to exit. In contrast, obedient participants were more likely to seek the authority’s directives, specifically by asking more questions about the experimental procedure. They also interrupted the learner’s protests—an indicator that they and the authority had similar interpretations of the situation. These findings suggest one way in which disobedience to immoral orders from authorities can emerge. By using language to extract different cues from an equivocal situation, disobedient participants were able adjust the salience of particular social and situational cues and stabilize their issue construction to trigger an alternate interpretation of the situation, and the will to disobey, even in a situation designed to elicit obedience.

Revisiting this historical and unique dataset also provides several implications for current theories in organizational studies. First, this study deepens our understanding of how individuals construct the situations they face, and thus expands our understanding of how individuals can exercise agency in the face of equivocal situational pressures (Levina & Orlikowsky, 2009). This processes is underexamined in the moral imagination literature (Werhane, 1999, 2008). Second, these results refine specific assumptions in the moral decision-making literature (Bazerman & Gino, 2012; Treviño, Weaver, & Reynolds, 2006) such as the presence of a discrete moment of choice, and the role of ambiguity in fostering unethical behavior. Additionally, this work speaks to research on employee voice (Burris, 2012; Detert & Burris, 2007), suggesting that scholars may benefit from broadening the kinds of voice behavior studied to include how employees come to form their views of morally questionable behaviors. Finally, the findings underscore a larger role for morality in theories of organizing (Heath & Sitkin, 2001; Weick, 1995), while current theory generally separates the two (Wicks & Freeman, 1998).

An Issue Construction Lens

One lens that is useful in understanding the dynamic process of disobedience is issue construction (Sonenshein, 2006, 2007). This area of research aims to understand how organizational members shape the meaning of their experiences and aim to influence the sensemaking of others (Bruner, 1990; Gioia & Chittipeddi, 1991). Sensemaking is “a process, prompted by violated expectations, that involves attending to and bracketing cues in the environment, creating intersubjective meaning through cycles of interpretation and action, and thereby enacting a more ordered environment from which further cues can be drawn” (Maitlis & Christianson, 2014). In this view language is an essential indicator of how organizational members make meaning and how they act in environments (Sonenshein, 2006). Issue construction is influenced by individual expectations and motivation, as well as other actors.

Additionally, in the literature on sensemaking and issue construction, talk and conversation are one type of action that is used to make sense. “Actions are important for sensemaking because they create more raw ingredients for sensemaking by generating stimuli or cues: people can quickly learn more about a situation by taking action and paying attention to the cues generated by that action” (Weick, 1988; Maitlis & Christianson, 2014, p. 84). Additionally actions can test provisional understandings generated through prior sensemaking. This lens is useful for my research question for several reasons. First, it draws attention to the ongoing and dynamic sensemaking of participants and how they construct the issues they face. Second, it focuses on language as one important indicator of how individuals make meaning. Third, this lens can shed light on the implicit and explicit choices that individuals make in interpreting the situation, which may explain disobedience even in situations with few material resources.

Data

To explore the question of how disobedience emerges, I utilized audio recordings, transcripts, and the researchers’ notes from Stanley Milgram’s experiments on obedience to authority, conducted in early 1961 at Yale University. This data has not been systematically examined in this way before. Milgram and his team created a variety of manipulations, but the standard baseline experiment consisted of:

Two people [who] come to a psychology laboratory to take part in a study of memory and learning. One of the individuals is designated as a “Teacher,” and the other as a “Learner.” The Experimenter explains that the study is concerned with the effects of punishment on learning. The Learner is conducted into a room, seated in a kind of miniature electric chair; his arms are strapped to prevent excessive movement, and an electrode is attached to his wrist. He is told that he will be read lists of simple word pairs, and that he will then be tested on his ability to remember the second word of a pair when he hears the first one again. Whenever he makes an error, he will receive electric shocks of increasing intensity. The real focus on the experiment is the Teacher. After watching the Learner being strapped into place, he is seated before an impressive shock generator. The instrument panel consists of thirty lever switches set in a horizontal line. Each switch is clearly labeled with a voltage designation ranging from 15 to 450 volts. The surprising result is that in most conditions, over half of the participants obey to 450 volts. (Milgram, 1963)

Prior research has examined the role of personality traits such as locus of control on disobedience and found little evidence that static personality traits influence disobedience (Blass, 2009; Burger, 2009; Elms & Milgram, 1966). Therefore, Milgram and the obedience experiments played a large role in shifting the emphasis in social psychology from individual traits to situational forces (Blass, 2009; Elms, 1995). Additionally, the results of this experiment hold today; Burger (2009) conducted a truncated replication of the obedience experiments at Santa Clara in 2007 and found similar obedience rates.

I argue that there are several reasons to describe Milgram’s experimental situation also as an organizational one. First, Ross and Nisbett (1991) as well as Blass (2009) argue that the power of Milgram’s work is that it harnesses the embedded and pre-existing authority structures of laboratory experiments. This sample of participants from Milgram’s obedience experiments has the potential to extend theory about how disobedience to immoral orders emerges, because the laboratory environment creates a standardized situation and allows the exploration of factors beyond situation that can contribute to disobedience, which have not been given much attention, such as communication, talk, and issue construction.

Second, these experiments offer a lens into the process of organizing because authority is integral to organized social activities (Barnard, 1968; Weber, 1924, 1947). The teacher, learner, and experimenter in the obedience experiments have interlocking routines that require differentiated, albeit temporary, roles (Weick, 1993). There is also generic subjectivity that allows individuals to be easily replaced from hour to hour (Wiley, 1988) because of the formal rules and procedures that distinguish tasks and responsibilities. Therefore the interactional process by which the teacher, learner, and the experimenter coordinate—or fail to coordinate—their actions, while not as routinized as a large organization, can still shed light on the authority dynamics present in more complex organizations (Heath & Sitkin, 2001), such as when and how individuals can overcome the influence of authority on their decision making.

It is also important to note that the kinds of interactions displayed in the obedience experiments are not perfect replications of naturally occurring interactions. For example, the experimenter has scripted responses to participant questions, therefore this data provides an opportunity to study how an individual makes sense in a social context, but the data falls short of true social interaction. The data provide an opportunity to examine the participants’ unscripted response to an experimental situation and are therefore ideal to study how construction of moral issues occurs. Additionally, because the situation utilizes the naturally occurring authority of the laboratory (Blass, 2009; Ross & Nisbett, 1991), the phenomenon of interest is more readily observable in this extreme situation (Pettigrew, 1988) that is not so easily produced today. This kind of “unconventional” sample provides the opportunity to examine the emergence of disobedience and to compare how it emerges across participants (Bamberger & Pratt, 2010). This set of data might be an optimal way to isolate theoretical constructs of interest to explore how disobedience emerges.

I collected the audio files of all 40 participants from condition 2, the original baseline condition conducted at the interaction lab at Yale University between July and August 1962. In condition 2, the learner was placed in a separate room and responses to the shocks were audio recorded and standardized.¹ Each audio file was approximately one hour in length, and included both the experiment and the post-experimental debrief. I analyzed over 1,000 pages of transcripts and 200 pages of records kept by Milgram on the participants for condition 2. One participant’s original audio file was corrupted and thus could not be used.

Milgram and his team drew a random sample of individuals from the area surrounding New Haven, Connecticut. All participants volunteered to participate in a one-hour study on memory and learning, and were paid $4.50 (~$35 in 2015 dollars) for their participation. The teachers in this sample ranged from ages 23 to 50 with a mean of 32.9, and came from a variety of educational and vocational backgrounds (see Table 1 for a summary). All but one of Milgram’s experimental conditions comprised men. Thirty-nine of the participants in condition 2 were Caucasian males and there was one African American male. Previous research has validated the findings about obedience rates with women and in other cultures (Kilham & Mann, 1974; Shanab & Yahya, 1977). All data was anonymized by Yale University Library before any analysis was conducted.

Table 1.

Participant Overview.

Subject number	Voltage level	Education level	Age	Occupation
235	16	3	30	Construction Manager
239	16	3	29	Account Executive
208	17	3	31	Insurance and Real Estate Agent
217	17	4	31	Self-Employeed
229	17	3	44	Hardware Store
230	19	3	32	Instrument Engineer
210	20	3	34	Meter Reader
226	21	4	32	Physician
237	26	2	41	Tool Maker
228	27	2	32	Welder
202	28	2	42	Insurance Adjustor
236	28	3	47	Watch Repairman
224	29	4	43	Design Engineer
219	30	2	42	Photo Engraver
232	31	3	23	Electrical Engineer
201	39	2	41	Mailman
203	39	2	38	Electric Heating Engineer
204	39	data missing	data missing	data missing
205	39	3	45	Engineer
206	39	3	40	Credit Manager
207	39	2	32	Scuba Instructor
209	39	3	38	Salesman
211	39	2	23	Salesman
212	39	4	43	Experimental Technician
213	39	2	47	Machinist
214	39	3	28	Insurance Underwriter
215	39	3	36	Office Manager
216	39	2	49	Sales
218	39	3	47	Machinist
220	39	2	28	Draftsman
221	39	2	27	Hairdresser
222	39	4	39	Social Worker
223	39	4	34	Chemist
225	39	3	41	Economist
227	39	2	23	Factory Worker
231	39	3	28	Accountant
233	39	4	38	Draftsman
234	39	2	35	Drill Press Operator
238	39	4	50	Teacher
240	39	3	38	Pharmacist

Education level: 1= elementary school complete, 2= high school complete, 3= college complete, 4= graduate or professional school complete.

Occupation = Self-described.

Voltage level = number of times the shock buttons were pressed. 39 = fully obedient.

Method: Organizational Discourse Analysis

To examine how individuals disobey immoral orders from authorities, I utilized organizational discourse analysis (ODA) (Fairhurst & Uhl-Bien, 2012; Grant, Putnam, & Hardy, 2011; Putnam & Fairhurst, 2001). ODA is a broad collection of methods that draw from a variety of constructivist theories (Alvesson & Deetz, 2006; Fairclough, 1989; Garfinkel, 1967) and focus “on how organizational members use language and other material means to socially construct their relationships and other organizational realities” (Fairhurst & Uhl-Bien, 2012, p. 1045). ODA methods can be used to examine multiple levels of analysis including individual, dyads, and collectives, in both qualitative and quantitative form. This class of approaches views organizational reality as co-constructed in social interaction processes, therefore communication and language is an important indicator of how organizational actors make meaning (Fairhurst, 2007). This does not mean, however, that language is the only determinant of behavior, but that it plays an important role in framing the reality to which people react.

ODA has several advantages that make it appropriate to examine how individuals disobey immoral orders from authorities. First, these methods are well suited to answer how or process questions, as opposed to the more post-positivist why questions (Fairhurst & Uhl-Bien, 2012). Second, ODA allows both qualitative and quantitative methods to make sense of data; therefore this multi-method flexibility allows one to get the most out of their dataset. Third, “By directly observing the sequential context and consequences of quick verbal exchanges, analysts can track quite distinct interactional processes that are simultaneously organizational” (Fairhurst & Uhl-Bien, 2012, p. 1048). Therefore these methods allow the analysis of both individual verbal behavior and the impact that it has on the organizational unit. Additionally, ODA methods are useful in conditions of ambiguity or uncertainty where the power of language to frame is more readily apparent. Since many have argued that the obedience experiment is a situation of ambiguity, the focus on how language frames the situation is useful (Milgram, 1974; Ross & Nisbett, 1991). Finally, ODA methods are consistent with my theoretical approach on issue construction and sensemaking (Gioia & Chittipeddi, 1991; Weick et al., 2005).

The goal of this inquiry is to create a plausible suggestive theory of how disobedience emerges that can further inform inquiry (Edmondson & McManus, 2007). Therefore, I conducted the analysis of my data in three broad phases (see Table 2 for an overview). In phase 1, I transcribed and repeatedly listened to the audio files from the 39 participants in the condition (Eisenhardt, 1989). I also read through Milgram’s participant notes, which included any physical and facial reactions. This step helped me understand the temporal sequence of the experiments, and the context in which participants engage, and therefore to assess the potential cause-and-effect relationships as the transcripts allowed me to create a time series of the events in the experiment (Yin, 1984). Repeatedly listening to each audio file allowed me to become familiar with each participant and how and when they asserted themselves in the experiment. I used journaling (Locke, 2001) in this early stage to keep track of my emerging sensemaking about the experiment and potential drivers of disobedience, but no formal analyses were conducted at this stage.

Table 2.

Summary of Methods: Organizational Discourse Analysis.

Phase 1:

• Repeatedly listened to and transcribed all audio files condition 2 (men)

• Journaled about interesting observations

• Grouped participants based on different axes

• Setttled on obedient/disobedient and high/low equivocality

Phase 2:

• Inductively generated a list of frequent communication patterns from transcripts

• Replication coding by naive coder (98% reliability)

• Calculated descriptive statistics and t-tests on communication patterns across groups

Phase 3:

• Listened to and transcribed audio from condition 20 (women)

• Coded communication patterns generated in phase 2

• Replication coding by two naive coders (94% reliability)

• Calculated descriptive statistics and t-tests on communication patterns

I noticed that obedient and disobedient participants were not homogenous groups, but that there were important differences within each of those groups. I began to try out several ways of grouping participants, until a stable taxonomy of four groups emerged, based on two axes: obedient/disobedient and high/low equivocality. Participants differed on how much their behaviors indicated equivocality. Equivocality is the presence of multiple simultaneous interpretations of a situation (Sonenshein, 2007; Weick, 1995). Equivocality is not just a feature of a situation, but a relationship between a specific situation and a specific individual, since the same situation can trigger different levels of equivocality for different individuals based on their experiences, knowledge, and their own actions (Dane & Pratt, 2007). Several scholars familiar with the experiments have argued that participants in the Milgram experiments confront an equivocal situation because they are placed between the experimenter’s interpretation of the situation and the one indicated by the learner’s protests and are uncertain about how to act (Milgram, 1963, 1974, Ross & Nisbett, 1991; Weick, 1995).

In the second phase of analysis, I then used this taxonomy as a basis for further inquiry. I sought to understand how these different groups coped with the level of ambiguity they experienced. Therefore I compared participants’ speech, by inductively identifying several types of communication patterns that participants exhibited during the course of the experiment. In keeping with grounded theory guidelines (Glaser & Strauss, 1967, Miles & Huberman, 1984), each unique communication that was not a part of the experimental protocol was hand coded and compared with an emerging set of categories (Locke, 2001). ODA methods can include the use of a priori coding schemes or inductively derived coding schemes that can be highly quantitative (Fairhurst & Uhl-Bien, 2012). An inductively derived coding scheme is useful in this context because it allows one to see the “influential acts” (Fairhurst & Uhl-Bien, 2012) that are key moves used by participants to shape the situation and allows more freedom for the data to speak for itself, than when using an a priori coding scheme.

In creating categories for the communication, I relied on the functional purpose of the communication, sorting thought length statements (Bakeman & Gottman, 1986) in response to the question, “What is the participant trying to accomplish with this verbalization?” (ten Have, 2007). As an illustration, I treated questions that participants asked of the experimenter, such as “Do you want me to read these fast or slow?” Or “do you want me to write down the ones he gets wrong” as procedural guidance seeking. From this process, a set of categories emerged such as “procedural guidance seeking,” “testing assumptions,” “interruptions,” and “self-referential objections.”

To strengthen the validity of these themes an independent naive coder also read all the transcripts to analyze communication patterns using the coding scheme I generated; the overall level of inter-rater agreement was 98%. I then counted the frequency of these communication patterns across the four types of participant groups I observed and calculated t-tests to examine whether the differences I found were statistically significant. ODA allows for patterns to be identified and tested through observed frequencies in the data. Coding enables interaction analysts to quantitatively analyze the sequences and stages of interaction, their redundancy and predictability (Fairhurst & Uhl-Bien, 2012). In phase 3, I replicated the analysis with data from a similar condition (20) to add further validity to the findings.

This data suggests a process where different communication strategies are more likely to trigger an alternate interpretation of a situation, and thus make disobedience more likely. I began to focus on writing and explaining my emerging theory when the findings seemed to converge on a set of plausible, supportable, and intriguing findings (Eisenhardt, 1989). I organize the remaining sections as follows: (1) I present my qualitative findings; and (2) I describe an emerging model of disobedience that these multi-methods corroborate.

A Grounded Typology of Participants

In this section, I introduce a grounded typology of the process and outcomes in the obedience experiments. The “outcome dimension” is whether participants were able to disobey and break off their engagement in the experiment. In the sample of 39, 25 participants (64%) were obedient, continuing to 450 volts and pressing the final switch three times, and 14 (36%) quit the experiment before reaching the end. The earliest exit was at 90 volts and the latest was at 315 volts, with an average of 225 volts.

The “process dimension” is the number of times a participant exhibited equivocality by vacillating between different interpretations of the situation during the experimental procedure. A vacillation occurred when a participant switched between making mention of the learner, or problematizing the experiment in any way during the course of the experiment (not including the debrief or experimental training) and then returning to the experimental protocol.² For example, high equivocality participants said, “We can’t hurt him in there (teacher 228),” or expressed a desire to stop the experiment, as when teacher 210 said “I don’t feel I should be doing this.” Returning to the experimental protocol after problematizing the experiment indicates two competing interpretations of the situation—the first, that the experiment is somehow problematic and should not continue, and the second, that the experiment is not problematic enough to discontinue. I split the sample at the mean level of vacillations 2. Low equivocality participants vacillated no more than twice during the experiment, whereas high equivocality participants vacillated more than twice, on average 4.7 times, as they continued to make sense of the unfolding situation. Table 3 depicts this typology.

Table 3.

Grounded Typology of Participants.

	Obedient	Disobedient
High equivocality	17 participants • Vacillated repeatedly between questioning the experiment and following orders	6 participants • Vacillated repeatedly between questioning the experiment and following orders • Emerging clarity around the inappropriateness of the experiment
Low equivocality	8 participants • Less than 2 vacillations • Mostly protocol-driven speech • Few if any signs about the inappropriateness of the experiment	8 participants • Less than 2 vacillations • Quick exit at first sign the learner was being harmed (~150 volts)

Communication Patterns

After generating, this typology, I began to look for more subtle differences across cases that could explain the variation I observed. I first examined the demographic factors that Milgram and his team had collected such as age, education level, occupation, marital status, military service, military combat, and number of siblings, and as confirmed in previous research (Milgram, 1974) there were no discernible differences in these variables across the four groups. Therefore I turned my attention to what the participants said and when they said it during the course of the experiment as indicators of their dynamic issue construction.

Equivocality reducing communication

In this section I illustrate four communication patterns that emerged from coding the transcripts across all 39 participants. These communication patterns are grouped into two types. The first are communication strategies that aim to reduce some kind of equivocality: (1) whether to continue the experiment based on the consequences; or (2) how to continue the experiment in terms of the mechanics of the procedure. The second two patterns are indicators of how participants interpret the situation, whether they see the situation as problematic or whether they accept the authority’s definition of the situation.

Assessing consequences

In an equivocal situation participants are left to their mental representations to decide what to do. People go beyond the information given by relying on their own expectations and experiences (Bruner, 1957; Ross & Nisbett, 1991). In organizational settings one’s mental representations are influenced and shaped by the larger role system as well as by leaders who share their own sense of the situation (Gioia, 1992).

A key theme that emerged in coding the data was that some participants used communication to fuel their own sensemaking by explicitly assessing the consequences of continuing the experimental situation, rather than taking the experimenter’s definition of the situation for granted. This kind of communication can weaken the experimenter’s role as authority because the participant is asking questions whose answers they will use to decide whether to continue the experiment or not. In this way, the participant becomes another decision maker in the situation and contests the role of the authority by acting as more than just a vehicle for the authority’s intentions. Communication was counted as “assessing consequences” when a participant asked a question or requested information that would allow them to make a judgment about whether continuing the experiment was acceptable or not to them because of the outcomes. General questions about the experimental procedure were not counted as assessing consequences, for example, “How fast do you want me to read these word pairs?”

For example, teacher 207 asks to experience the highest shock level before the experiment begins to get a sense of what the learner might go through when he says, “Can I see what it [the shock] is way up at the top? So I know what kind of punishment I’m giving him?” Similarly, teacher 226 wants to know what might happen in the worst case scenario, “Suppose he gets all these wrong, and I get up to a level where it’s going to be extremely painful?”

Teacher 208 also tries to understand the consequences of using the highest voltage shocks when he says, “What happens when we uh, get up to the top of the uh machine here?” Later on he refuses to continue, unless his assumption that the learner is willing to continue is validated. He says at the 120 volt mark, “Should we ask him if we should go on? Well, I wouldn’t continue without asking. I still won’t go on unless he says it’s alright.” When the experimenter urges him to continue, he asks to personally verify the learner’s cooperation before continuing when he says, “I mean if I can walk in there and he says to go on, I’ll go on, but I’m not going to go on with him hollering at me to stop, you know?”

There were several types of assumptions that participants most frequently tested such as: (1) whether the learner was able and willing to continue, (2) whether the amount of voltage was safe, (3) the duration of the shocks and the experimental protocol, and (4) whether it was routine to use the higher voltage switches.

Procedural guidance seeking

Like assessing consequences, procedural guidance seeking aims to resolve equivocality. Unlike assessing consequences, where participants communicate to learn about what will happen in the future, procedural guidance seeking demonstrates a concern about what to do in the immediate moment in regard to the mechanics of the experimental procedure. Participants aim to clarify confusion about how to best accomplish the experimental procedure, not about resolving whether the experiment is appropriate. Thus procedural guidance seeking preserves the experimenter’s role as authority in the experimental situation, because the participant assumes a subservient role and does not exercise his own potential as a decision-making agent in the situation.

A communication was counted as procedural guidance seeking when a participant asked a question or made a statement that aimed to clarify some mechanics of accomplishing the experiment. For example, teacher 230 clarifies what he is to say to the learner when he asks, “Do I say, ‘correct answer’ or what?” Teacher 219 repeats the directions to confirm that he has understood the procedure: “I give him the answers, tell him how many volts he gonna get when he does get it wrong.”

Indicators of interpretations during the experiment

In addition to the two types of equivocality reducing communication, I found two verbal behaviors that indicate how participants interpret the experimental situation at a specific point in time: self-referential objections, and interruption of the learner’s protests.

Self-referential objections

Another pattern that emerged in the data was the differential presence of statements that conveyed the teacher’s own intentions about continuing the experiment. For example, teacher 202 says, “I can’t keep doing this to him,” and teacher 208 says, “I don’t think I want to be a part of this any longer.” Similarly, teacher 217 says, “I find it very difficult to go on, if it is actually hurting him.” Self-referential objections were counted when a participant made any statement using or implying the pronoun “I” in conjunction with a statement that problematized the experiment, or with statements that conveyed his intention to discontinue the experiment. These statements indicate that participants are constructing an alternate definition of the situation—one which casts the experiment in a problematic light. The use of the “I” pronoun also conveys the participant’s own intentions and preferences, and thus can be an indicator of their willingness to disrupt the established social order and weakens the authority of the experimenter.

Interruptions of protests

Communication research has shown that those who perceive they have power over others are more likely to interrupt and speak out of turn when conversing with seemingly less powerful individuals (DePaulo & Friedman, 1998; Hutchby, 1996). When participants in the experiment interrupted the learner’s protests, by continuing to read the word pairs over their cries, it indicated that they accepted the authority’s definition of the situation: that the shocks were painful but not harmful, and that the experiment must continue. Therefore, I examined the number of times that a specific teacher would interrupt the protests of a learner.

Distribution of communication patterns

Table 4 summarizes the differential distribution of these patterns for obedient and disobedient participants, and Table 5 displays the means and standard deviations for each of the four types of participants I identified in the previous section. Where t-tests confirm statistically significant differences, I have included the group number and significance level. Several insights emerged from examining the data in Tables 4 and 5. First, the data suggest that while it has been difficult to distinguish between obedient and disobedient participants based on demographic and personality factors (Blass, 2009; Burger, 2009; Elms & Milgram, 1966) there are statistically significant differences in communication patterns for obedient and disobedient participants. Obedient participants engage in more interruptions of the learner’s protests and procedural guidance seeking, as well as seeking guidance later in the temporal sequence of the experiment. Disobedient participants were more likely to test assumptions, as well as make more self-referential objections. Second, making self-referential objections is not the same thing as being disobedient, because obedient participants also exhibit that communication pattern, but do it much less frequently and much later in the temporal sequence of the experiment. Third, when examining the differences between the four groups I identified earlier, it seems that a key difference between low and high equivocality disobedient participants is that low-equivocality disobedient participants tested assumptions and made self-referential objections earlier in the experiment. They also made fewer interruptions, which suggests that high-equivocality disobedient participants adhered to the experimenter’s interpretation longer and vacillated between interpretations as they tested enough assumptions to ultimately exit.

Table 4.

Means, Standard Deviations, and t-Test Results for Differences Between Obedience and Disobedient Participants in Communication Patterns.

		M	S TDEV	t	Sig. (2-tailed)
Assessing consequences (count)	DOB	3.79	3.96	2.70	0.010
	OB	1.20	2.04
Mean time of assessing consequences	DOB	165.24	131.80	2.13	0.040
Mean time of assessing consequences	OB	72.77	129.27
Self-referential objections (count)	DOB	5.00	2.63	5.55	0.000
	OB	0.80	2.04
Mean time of self-referential objections	DOB	200.22	68.65	–4.64	0.000
Mean time of self-referential objections	OB	348.75	54.45
All procedural guidance seeking (count)	DOB	4.07	2.34	–1.80	0.079
All procedural guidance seeking (count)	OB	6.36	4.40
Mean time of all procedural guidance seeking	DOB	36.63	50.91	–4.31	0.000
Mean time of all procedural guidance seeking	OB	171.27	110.04
Procedural guidance seeking only during experiment (count)	DOB	0.21	0.58	–4.59	0.000
Procedural guidance seeking only during experiment (count)	OB	1.84	1.25
Mean time of procedural guidance seeking only during experiment	DOB	43.93	102.83	–5.29	0.000
	OB	275.00	143.64
Interruptions of learner’s protests (count)	DOB	2.50	2.24	–6.39	0.000
Interruptions of learner’s protests (count)	OB	7.16	2.15
Voltage level	DOB	252.86	100.51	–9.91	0.000
	OB	450.00	0.00

Mean time is measured by the average voltage level a communication pattern occurred.

OB = Obedient participants.

DOB = Disobedient participants.

Table 5.

Differences in Verbal Behavior Across Four Types of Participants.

	Means, STDEV, and t-test Results for Differences Across Four Groups of Participants
	1		2		3		4
	Low equivocality disobedient		High equivocality disobedient		High equivocality obedient		Low equivocality obedient
	N = 8		N= 6		N=17		N= 8
	M	STDEV	M	STDEV	M	STDEV	M	STDEV
Assessing consequences (count)(Significantly different from group number(s))	2.88	3.48	5.00	4.56	1.47	2.27	0.63	1.41
			3, 4		2*		2*
Mean time of assessing consequences (by voltage level)	90.13	107.91	265.38	88.31	87.50	147.44	41.47	77.16
	2**		1*, 3, 4**		2*		2**
Self-referential objections (count)	4.50	2.00	5.67	3.39	1.18	2.40	0.00	0.00
	3, 4		3, 4		1, 2		1,2
Mean time of self-referential objections (by voltage level)	165.94	63.44	255.08	31.32	348.75	54.45	0.00	0.00
	2, 3*		1, 3		1, 2
All procedural guidance seeking (count)	3.38	1.51	5.00	3.03	7.12	4.91	4.75	2.60
All procedural guidance seeking (count)		3, 4*			1*		1**
Mean time of all procedural guidance seeking (by voltage level)	23.42	35.71	54.26	65.59	167.14	114.28	180.06	107.39
	3, 4		3, 4		1, 2		1*,2
Procedural guidance seeking only during experiment (count)	0.00	0.00	0.50	0.84	2.00	1.37	1.50	0.93
	3, 4		3*		1*, 2		1**
Mean time of procedural guidance seeking only during experiment (by voltage level)	5.63	15.91	95.00	147.17	254.41	142.54	318.75	145.21
	3, 4		3, 4		1*, 2		1*,2
Interruptions of learner’s protests (count)	1.13	0.64	4.33	2.34	6.82	2.46	7.88	1.13
	2, 3, 4**		1*, 3, 4**		1*, 2		1,2
Exit voltage level	183.50	59.85	292.50	76.35	450.00	0.00	450.00	0.00
	2, 3, 4*		1, 3, 4*		1, 2		1,2

Correlation is significant at the 0.01 level (2-tailed).

Correlation is significant at the 0.05 level (2-tailed).

Replication using the condition of women

To corroborate the inductively generated insights from the first analysis, I sought to replicate the findings by examining a similar condition of participants. Condition 20 utilized the same protocol and experimental set-up as condition 2, only it was conducted with 40 women. Milgram and several other researchers have noted no significant differences in obedience rates between men and women (Bock & Warren, 1972; Milgram, 1974); indeed the number of obedient participants in this sample was virtually the same (24 vs 25 for condition 2), therefore this condition offers an ideal opportunity to replicate the initial inductive findings. Five participant audio files were corrupted or incomplete, therefore only 35 out of 40 participants were analyzed using the same methods as outlined earlier.

Just as in the first analysis, I transcribed and listened to all the audio files. Two additional naive coders then read through the transcripts and hand coded for the four types of statements found in the original inductive analysis. The inter-rater reliability was 94%. In this analysis, we did not examine the timing of different communication.

The findings from the condition of women corroborate the same pattern found in the condition of men (see Table 6). Disobedient women assessed consequences more often, as well as made more self-referential objections than obedient women. Obedient women asked more procedural questions and interrupted the learner’s protests more often than disobedient women. When compared to the men, the condition of women was slightly more verbal overall, but the same communication pattern between obedient and disobedient participants was found, suggesting that how participants dynamically constructed the situation was integral to disobedience.

Table 6.

Comparing the condition of men and women
		Men	STDEV	Women	STDEV
Assessing consequences (count)	DOB	3.79	3.96	4.54	4.66
	OB	1.20	2.04	2.16	3.22
Self-referential objections (count)	DOB	5.00	2.63	8.45	5.87
	OB	0.80	2.04	2.00	2.68
Procedural guidance seeking only during experiment (count)	DOB	0.21	0.58	3.09	2.91
Procedural guidance seeking only during experiment (count)	OB	1.84	1.25	5.83	6.48
Interruptions of learner’s protests (count)	DOB	2.50	2.24	1.00	1.79
	OB	7.16	2.15	6.71	4.90
Voltage level	DOB	252.86	100.51	210.91	68.12
	OB	450.00	0.00	450.00	0.00

OB = Obedient participants.

DOB = Disobedient participants.

A Model of How Disobedience to Immoral Orders from Authorities Emerges

Figure 1 depicts the process by which an individual’s communication and behavior can trigger the psychological resources for disobedience. I will illustrate the general model with examples from the obedience experiments and from the financial services industry.

Figure 1.

A Model of How Disobedience to Immoral Orders from Authorities Emerges.

The model begins with an equivocal situation where multiple interpretations are triggered for an individual by different cues and social anchors in the environment such as the experimenter and the learner. For example, in the obedience experiments the point at which the learner first refuses to continue and the experimenter urges the participant to continue is one such equivocal situation (Ross & Nisbett, 1991). In an organizational setting another equivocal situation could be when a supervisor of financial advisors asks a requests that financial advisors push particular products to clients because they generate more fees for the firm, but do not necessarily benefit (and in some cases could even harm) the client.³

Next individuals engage in communication which builds upon specific extracted cues from the equivocal situation. The focal individual’s communication can vary in how much it supports or challenges the authority’s definition of the situation. For example, some participants exercised their agency in the obedience experiments by challenging the authority’s definition of the situation by problematizing the experimental situation or assessing consequences about continuing the experimental protocol. In contrast, participants could have chosen to simply continue the experimental protocol or sought procedural guidance about how to do it more effectively. These actions support and perpetuate the authority’s definition of the situation and the experimenter’s role as authority; they also shape the very situation that they participants experience by making particular situational cues including the sensemaking of others more salient. In the financial services example, the advisor has a choice about how to respond: she may speak or act in ways that support her supervisor’s definition of the situation, or she may ask questions or remind him of the corporate values involving serving the clients’ interests. Each of her actions subsequently alters the situation she faces.

These verbal and physical acts subsequently create outcomes in the situation which also must be made sense of by individuals. These outcomes are verbal, social, physical, and can also be equivocal, with conflicting cues about whether the behavior was acceptable or not. In the experiment, continuing the protocol was met with more conflicting cues of acknowledgment from the experimenter and continued protests from the learner. For the financial advisor, asking questions to her peers about the corporate policy validated the supervisor’s interpretation, when co-workers verbally expressed the positive outcomes of selling particular products, particularly for their annual bonuses.

Constructing issues in equivocal situations involves cycles of action and sensemaking until stable interpretations emerge. In constructing the issues, individuals again have the choice about what kinds of cues to extract from the potentially equivocal outcomes of their initial actions. They may choose to focus on similar cues or change course and now focus on cues that represent an alternate interpretation. If originally they communicated in a way that questioned the authority’s definition of the situation, and then again questioned the outcomes of their initial action, they will start to form a stable interpretation of the situation. For example, when the learner first refuses to continue, and the experimenter urges the participant to continue, the participant could again ask why, or what would happen if we continued? Each subsequent communication that questions the authority’s definition of the situation helps the participant to construct a stable interpretation, which can be indicated by their communication. Individuals will differ in the number of iterations they will need to form a stable interpretation of the situation, based on their past experiences and how easily particular interpretations are available to them.

After speaking with her colleagues the financial advisor also begins to see the sales situation similarly to her supervisor and is no longer conflicted about her effects on clients. If the orders are immoral from the perspective of a broader society, then the corrupt practices start to be normalized (Ashforth & Anand, 2003) as individuals reframe their meaning.

Additionally, individuals can vacillate between communicating in ways that support or undermine the authority’s definition of the situation. This vacillation continues as they switch from communicating in ways that challenge the authority’s definition and continue to enact it, until they are unintentionally complicit or until they start to form stable interpretive cycles. These vacillations should not be seen only as indicators of indecisiveness, but experiments from which individuals test the outcomes of how they will feel and how the world will look if they make sense and act in a particular fashion. These experiments are not independent. Research suggests that along the way, individuals also make sense of their own behavior and thus make it harder to switch interpretations after several cycles in a particular interpretation (Freedman & Fraser, 1966).

The data suggest that by choosing which cues to pay attention to through their communication and behavior in an equivocal situation and examining the emotional, social, and physical outcomes of a particular set of actions, is one way that individuals can iteratively build stable interpretations of an equivocal situation and disobey an immoral authority even in conditions with limited access to material resources, or can make them complicit in immoral practices.

Constructing situations by extracting cues

The data presented in this study suggest one way in which disobedience to immoral orders from authorities can emerge. By using language to extract different cues from an equivocal situation, disobedient participants were able adjust the salience of particular social and situation cues in their environment and to stabilize their issue construction to trigger an alternate interpretation of the situation, and the will to disobey, even in a situation designed to elicit obedience.

Implications for current organizational theories

The findings of this study have implications for contemporary theory in organization science. First, the data provide empirical support for research on how people overcome situational pressures by constructing the situations they face (Sonenshein, 2006, 2007). This study offers a different perspective by showing that by using language and attention to adjust the salience and meaning of particular situational features, individuals can shape how they make sense of situation and trigger psychological resources for disobedience. This is relevant to literatures in business ethics, such as the literature on moral imagination. Werhane (2008) defines moral imagination as the ability to discover, evaluate, and act upon possibilities not merely determined by a particular circumstance, or limited by a set of operating mental models, or merely by a framed set of rules (Werhane, 2008, p. 466). In short it is the ability to step back from a situation and reflect on one’s behavior. In this sense, it can be argued that disobedient participants exercised moral imagination in deviating from the determined social roles during the experiment, which presumed obedience. This study adds to the literature on moral imagination by unpacking the communicative process by which individuals struggle with ambiguity, construct issues, and test assumptions to “envision and evaluate” a new situation—a process assumed and underexamined in the literature on moral imagination. By paying attention to alternate cues and testing assumptions, disobedient participants constructed a situation that was focused on the problems that the dominant mental model dismissed.

Second, this study questions fundamental assumptions made in the moral decision-making literature, such as the presence of a discrete moment of choice. The study of ethical decision making has largely presupposed a discrete moment of choice (Bazerman & Gino, 2012; Kohlberg, 1971; Rest, 1989; Treviño et al., 2006;) an assumption that is useful for testing decision making in the laboratory, and better understanding the contextual factors that shape choice, but this assumption may not always translate to practice. The data presented here demonstrates that the decision to disobey was not a discrete moment, but emerged from a process of interaction that made an alternative interpretation of the experiment salient. Disobedient participants constructed a situation over time in which disobedience was the only option.

Third, most research on moral decision making also ignores equivocality, under the assumption that individuals generally have perfect knowledge of the moral situations they face—which they do when researchers give them scenarios with perfect information (Sonenshein, 2006). When equivocality is mentioned, it is cast as a situational feature which diminishes the likelihood of ethical behavior (Bandura, 1999; Ho, Vitell, Barnes, Desborde, 1997; Jackall, 1988; Jones & Ryan, 1998; Ross & Nisbett, 1991; Shminke, Ambrose, Neubaum, 2005; Waters & Bird, 1987). This study suggests a more nuanced view. Specifically, that equivocality is an important feature of organizational moral issues and provides the opportunity for individuals to exercise their agency—without multiple competing interpretations about the situation, participants would not have had the opportunity to interpret the situation differently and act ethically. Equivocality and ambiguity can provide opportunities to perpetuate corrupt practices or to engage in moral courage (Quinn & Worline, 2008); without understanding the process of how people exercise their agency and construct issues, the effects of equivocality on ethical behavior is hard to determine.

Fourth, related to research on moral decision making in organizations, this study contributes to the literature on voice, specifically to the types of employee voice exercised (Burris, 2012; Detert & Burris, 2007). For example, the literature on voice has focused predominantly on low equivocality cases, where employees who exercise voice have stable representations of the situation at hand. Burris (2012) argues that the voice literature can be understood as focusing on two types of voice behaviors: supportive or challenging voice. Challenging voice is proactive, critical, and can include disagreement and confrontation (Burris, 2012; Grant & Ashford, 2008; Grant, Gino, & Hoffman, 2011). In contrast, supportive voice is stabilizing, defensive of the status quo, reactive, and may also be ingratiatory (Burris, 2012; Grant & Ashford, 2008). The work on both supportive and challenging voice suggests that the employees who exercise voice have a stable, relatively unambiguous interpretation of the situation. This study suggests that researchers should pay attention to the exercise of “learning voice”—or statements and questions that are delivered in a supportive way, but that can also exert considerable influence over organizational action by catalyzing group sensemaking about potential harms and benefits of different alternatives. Like disobedient participants who assess the consequences of continuing the experiment, learning voice can allow employees to engage in the decision-making process earlier, make their voice behaviors less personally threatening to their managers, and increase their influence on organizational outcomes. If employees have access to specific conversations, voice can be exercised in subtle ways as groups make sense together (Gioia & Chittipedi, 1991). The ways in which subordinates subtly communicate could mitigate the costs of speaking up while maximizing their influence over the situation at hand. Exploring these strategies in more detail and in the field is an important next step to help make exercising of voice in organizations more effective.

Finally, Karl Mannhiem said that “it is precisely when a social order begins to fall apart that one can discern what held it together in the first place” (Jackall, 1988). The results of this study underscore a larger relationship between the process of organizing and morality. Organizing refers to a specific kind of social order by which people “solve the dynamic problems of aligning goals and coordinating action” (Heath & Sitkin, 2001). During the course of the experiment, the experimenter, teacher, and learner constitute a temporary organizational unit, with differentiated roles and responsibilities. The act of disobedience disrupts that social order. Scholars have argued that ethics and organizational theory have been separated, with ethics seen as peripheral to organizational theory (Wicks & Freeman, 1998). This study demonstrates how obedient and disobedient participants typify different relationships between organizing and ethics. Obedient participants show how even temporary role relationships can override individual morality (Bandura, 1997; Milgram, 1974), particularly when the moral implications of the situation are obfuscated. Less well documented is how disobedient participants use morality—through assessing consequences—to dissolve temporary organizational relationships. By focusing on how disobedient participants communicate in order to construct the experimental situation in a way that renders exit the only viable moral alternative; this study resurfaces an understudied and important relationship between ethics and the process of organizing.

Social actors need to create and maintain inter-subjectively binding normative structures that sustain and enrich their relationships (Reed, 1991). Thus, organizations become important because they can provide meaning and order in the face of environments that impose ill-defined, contradictory demands. However, when organizations fail to express some conception of moral behavior that threatens the self-concept of its members, and individuals believe they have the rights and ability to exit, organizations can unravel because the meaning and order they provide become unnecessary illusions (Weick, 1993). Framed this way, ethics is not peripheral to organizational theory; it is central to the process of organizing and maintaining institutions that express some conception of the good life (Reed, 1991).

Implications for research on disobedience

In addition to implications for current organization theory, this study has implications for the study of disobedience. First, the data demonstrate how participants are able to trigger an alternate interpretation of the situation through communication, when few material and psychological resources are available to them. Previous theory has highlighted the importance of situational and material resources (Kelman & Hamilton, 1989; Milgram, 1963, 1974) as well as individual factors (Burger, 2009; Packer, 2008; Reicher et al., 2012) in disobedience, and has not been able to explain how participants within the same experimental condition with access to the same material and situational resources were able to disobey. I show that by asking questions to assess the consequences of their actions, disobedient participants were more likely to construct an interpretation of the situation and the will to exit that would eventually lead to disobedience.

Second, by making the dynamic process of communication and issue construction a central mechanism for disobedience, this model integrates various competing explanations of Milgram’s obedience experiments, by showing how each of the four styles of participant interaction is connected to how the participants communicate during the experiment. Until now, explanations of obedience have focused on only one group of participants (Milgram, 1974; Packer, 2008; Ross & Nisbett, 1991). For example, Milgram’s explanation about displaced agency leading to obedience focused on low-equivocality obedient participants. Ross and Nisbett (1991) focus predominantly on high-equivocality obedient participants when they argued that people obey because of uncertainty, and Packer (2008) generalizes from low-equivocality disobedient participants when he argues that people exit because they respect the learner’s rights. This study shows that these different styles were in part outcomes of how participants addressed ambiguity.

Implications for practice

If moral decision making is only about individual factors such as personality or situational factors such as incentives, then organizations would do well to recruit the “right” people and incentivize “good” behavior. These are useful strategies and examining the dynamic and unfolding process that individuals use to make sense of equivocal moral issues can suggest additional strategies for organizations to use. Specifically, by training people to (1) ask questions that uncover the effects of their actions on various stakeholders and (2) engage in conversation around moral issues, organizations can increase the likelihood that people will form multiple interpretations of moral issues that are supported by data and experience, rather than only an authority’s interpretation. It is because talk constructs and constitutes situations (Wood & Kroger, 2000) that these communicative strategies can be effective. When people ask questions or engage in conversation around moral issues, they help constitute a situation and create an opportunity for others to make sense of what is happening rather than rely on taken-for-granted assumptions that could lead to unethical behavior. By opening opportunities to perceive a situation differently, people can exercise their agency. Second, this study shows that obedient participants were more likely to make cues salient that confirmed the authority’s definition of the situation and disobedient participants assessed the consequences of their actions and highlighted data that disconfirmed the authority’s interpretation. Managers would be well served to augment their moral decision making by looking at both confirmatory and disconfirmatory data before acting. Finally, many times in organizations unethical practices are perpetuated because organizational actors perceive that they do not have a choice, or cannot act otherwise. Organizations can reduce the likelihood of perpetuating unethical behaviors if they make clear to their employees what their choices are in situations where they may disagree with their supervisors. By setting expectations well in advance about what will happen and what choices employees have, organizations can reduce the likelihood that employees obey morally questionable orders from authority because they perceive that they do not have a choice.

Limitations and future directions

As with most inductive work, my aim is to generate plausible new explanations based on close correspondence with the data. The nature of my data does not allow me to assess whether there are underlying personality drivers of the communication patterns that I observed. There has been a battery of dispositional studies that try to explain the behavior of Milgram’s participants, to little avail (Blass, 1991; Burger, 2009; Elms & Milgram, 1966). Milgram’s doctoral student, Alan Elms, contacted several participants of the original experiments and gave them a battery of personality scales, the results of which did not make a strong case for personality-based explanations (Blass, 1991; Elms & Milgram, 1966). As a result, Milgram became increasingly stalwart in his belief that situation trumps personality as he examined the obedience data and personality traits of his participants (Blass, 2009). Yet, this work makes possible the idea that there may be more fine-grained relationships between personality traits and particular kinds of communication patterns that can be tested. To date, I could not find research connecting personality traits and communication patterns in ethics, but there exists work on relationships (Smith, Heaven, & Ciarrochi, 2008) and self-descriptions (Schmitt, Allik, McCrae, & Benet-Martínez, 2007) that could offer helpful starting points. Blending the work on moral capacities (Hannah, Avolio, & May, 2012) with communication practices may be a fruitful first step in this direction.

Additionally, future research should more deeply examine the sequence and social interaction of organizational actors in constructing moral issues. The experimental nature of my data does not allow me to deeply examine social construction of issues between the experimenter, learner, and teacher. Using unscripted data from the field, through audio and video tapes, may yield interesting further insights about the social and sequential aspects of this process that were held constant in this analysis.

Conclusion

Disobedience to immoral orders from an authority is a critical mechanism for preventing organizational corruption and fraud from becoming pervasive. Situational forces and material resources such as access to power, alternatives, and relationships can certainly help individuals disobey immoral orders. This study also demonstrates that individuals can disobey even without these resources when they use their agency to make sense of an unfolding situation by communicating differently, specifically by assessing consequences and testing assumptions. Understanding the communicative processes by which authority relationships are co-constructed can help organizations more consciously attend to these relationships and unleash the positive aspects of authority while minimizing the risks.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

Author biography

Bidhan (Bobby) Parmar is an Assistant Professor of Business Administration at the Darden School of Business at the University of Virginia. His research interests focus on how managers make decisions and collaborate in uncertain and changing environments to create value for stakeholders. His work helps executives better handle ambiguity in their decision making.

References

Ashforth

B. E.

Anand

(2003). The normalization of corruption in organizations. Research in Organizational Behavior, 25, 1–52.

Ashforth

B. E.

Gioia

D. A.

Robinson

S. L.

Treviño

L. K.

(2008). Re-viewing organizational corruption. Academy of Management Review, 33, 670–684.

Alvesson

Deetz

(2006). Critical theory and postmodernism approaches to organizational studies. In The Sage handbook of organization studies (p. 255).

Bakeman

Gottman

J. M.

(1986). Observing behavior: An introduction to sequential analysis. Cambridge: Cambridge University Press.

Bamberger

P. A.

Pratt

M. G.

(2010). Moving forward by looking back: Reclaiming unconventional research contexts and samples in organizational scholarship. Academy of Management Journal, 53, 665–671.

Bandura

(1997). Self-efficacy: The exercise of control. New York: Freeman.

Bandura

(1999). Moral disengagement in the perpetration of inhumanities. Personality and Social Psychology Review, 3(3), 193–209.

Bandura

(2001). Social cognitive theory: An agentic perspective. Annual Review of Psychology, 52, 1–26.

Barnard

C. I.

(1968). The functions of the executive. Vol. 11. Cambridge, MA: Harvard University Press.

10.

Bazerman

M. H.

Gino

(2012). Behavioral ethics: Toward a deeper understanding of moral judgment and dishonesty. Annual Review of Law and Social Science, 8, 85–104.

11.

Blass

(1991). Understanding behavior in the Milgram obedience experiment: The role of personality, situations, and their interactions. Journal of Personality and Social Psychology, 60, 398.

12.

Blass

(2009). The man who shocked the world: The life and legacy of Stanley Milgram. New York: Basic Books.

13.

Bocchario

Zimbardo

P. G.

(2010). Defying unjust authority: an exploratory study. Current Psychology, 29, 155–170.

14.

Bock

D. C.

Warren

N. C.

(1972). Religious belief as a factor in obedience to destructive commands. Review of Religious Research, 185–191.

15.

Brief

A. P.

Buttram

R. T.

Dukerich

J. M.

(2001). Collective corruption in the corporate world: Toward a process model. In Turner

(Ed.), Groups at work: Theory and research (pp. 471–499). Mahwah, NJ: Lawrence Erlbaum Associates.

16.

Bruner

J. S.

(1957). Going beyond the information given. In Gruber

Terrell

Wertheimer

(Eds.), Contemporary approaches to cognition vol. 1 (pp. 119–160). Hillsdale, NJ: Lawrence Erlbaum Associates.

17.

Bruner

J. S.

(1990). Acts of meaning (Vol. 3). Cambridge, MA: Harvard University Press.

18.

Burger

J. M.

(2009). Replicating Milgram: Would people still obey today? American Psychologist, 64, 1.

19.

Burris

E. R.

(2012). The risks and rewards of speaking up: Managerial responses to employee voice. Academy of Management Journal, 55, 851–875.

20.

Dane

Pratt

M. G.

(2007). Exploring intuition and its role in managerial decision making. Academy of Management Review, 32, 33–54.

21.

DePaulo

B. M.

Friedman

H. S.

(1998). Nonverbal communication. In Gilbert

D. T.

Fiske

S. T.

Lindzey

(Eds.), The handbook of social psychology (pp. 3–40). Boston, MA: McGraw-Hill.

22.

Detert

J. R.

Burris

E. R.

(2007). Leadership behavior and employee voice: Is the door really open? Academy of Management Journal, 50, 869–884.

23.

Detert

J. R.

Edmondson

A. C.

(2011). Implicit voice theories: Taken-for-granted rules of self-censorship at work. Academy of Management Journal, 54, 461–488.

24.

Edmondson

A. C.

McManus

S. E.

(2007). Methodological fit in management field research. Academy of Management Review, 32, 1155–1179.

25.

Eisenhardt

K. M.

(1989). Building theories from case study research. Academy of Management Review, 14, 532–550.

26.

Elms

A. C.

(1995). Obedience in retrospect. Journal of Social Issues, 51, 21–31.

27.

Elms

A. C.

Milgram

(1966). Personality characteristics associated with obedience and defiance toward authoritative command. Journal of Experimental Research in Personality.

28.

Fairclough

(1989). Language and power. Harlow, UK: Longman.

29.

Fairhurst

(2007). Discursive leadership: In conversation with leadership psychology. London: SAGE Publications.

30.

Fairhurst

G. T.

Uhl-Bien

(2012). Organizational discourse analysis (ODA): Examining leadership as a relational process. Leadership Quarterly, 23, 1043–1062.

31.

Freedman

J. L.

Fraser

S. C.

(1966). Compliance without pressure: the foot-in-the-door technique. Journal of Personality and Social Psychology, 4(2), 195.

32.

Garfinkel

(1967). Studies in ethnomethodology. Engelwood Cliffs, NJ: Prentice-Hall.

33.

Gioia

D. A.

(1992). Pinto fires and personal ethics: A script analysis of missed opportunities. Journal of Business Ethics, 11, 379–389.

34.

Gioia

D. A.

Chittipedi

(1991). Sensemaking and sensegiving in strategic change initiation. Strategic Management Journal, 12, 433–448.

35.

Glaser

B. G.

Strauss

A. L.

(1967). The discovery of grounded theory: Strategies for qualitative research. New York: Aldine de Gruyter.

36.

Grant

A. M.

Ashford

S. J.

(2008). The dynamics of proactivity at work. Research in Organizational Behavior, 28, 3–34.

37.

Grant

A. M.

Gino

Hofmann

D. A.

(2011). Reversing the extraverted leadership advantage: The role of employee proactivity. Academy of Management Journal, 54, 528–550.

38.

Grant

Putnam

L. L.

Hardy

(Eds.) (2011). History, key challenges, and contributions of organizational discourse studies. In: Organizational discourse studies. Thousand Oaks, CA: SAGE Publications.

39.

Greve

H. R.

Palmer

Pozner

(2010). Organizations gone wild: The causes, processes, and consequences of organizational misconduct. Academy of Management Annals, 4, 53–107.

40.

Hannah

S. T.

Avolio

B. J.

May

D. R

. (2011). Moral maturation and moral conation: A capacity approach to explaining moral thought and action. Academy of Management Review, 36, 663–685.

41.

Haslam

S. A.

Reicher

S. D.

(2012). Contesting the “nature” of conformity: What Milgram and Zimbardo’s studies really show. PLoS Biology 10.11: e1001426.

42.

Heath

Sitkin

S. B.

(2001). Big-B versus Big-O: What is organizational about organizational behavior? Journal of Organizational Behavior, 22, 43–58.

43.

F. N.

Vitell

S. J.

Barnes

J. H.

Desborde

(1997). Ethical correlates of role conflict and ambiguity in marketing: The mediating role of cognitive moral development. Journal of the Academy of Marketing Science, 25(2), 117–126.

44.

Hutchby

(1996). Power in discourse: The case of arguments on a British talk radio show. Discourse & Society, 7, 481–497.

45.

Jackall

(1988). Moral mazes: The world of corporate managers. International Journal of Politics, Culture, and Society, 1, 598–614.

46.

Jones

T. M.

Ryan

L. V.

(1998). The effect of organizational forces on individual morality: Judgment, moral approbation, and behavior. Business Ethics Quarterly, 8, 431–445.

47.

Kelman

Hamilton

V. L.

(1989). Crimes of obedience: Toward a social psychology of authority and responsibility. New Haven, CT: Yale University Press.

48.

Kilham

Mann

(1974). Level of destructive obedience as a function of transmitter and executant roles in the Milgram obedience paradigm. Journal of Personality and Social Psychology; 29, 696–702.

49.

Kohlberg

(1971). Stages of moral development. In Moral education (pp. 23–92). Boston, MA: Harvard University Press.

50.

Levina

Orlikowski

W. J.

(2009). Understanding shifting power relations within and across organizations: A critical genre analysis. Academy of Management Journal, 52, 682–703.

51.

Locke

K. D.

(2001). Grounded theory in management research. London: SAGE Publications.

52.

Lutsky

(1995). When is “obedience” obedience? Conceptual and historical commentary. Journal of Social Issues, 51, 55–65.

53.

Maitlis

Christianson

(2014). Sensemaking in organizations: Taking stock and moving forward. Academy of Management Annals, 8, 57–125.

54.

Miles

M. B.

Huberman

A. M.

(1984). Qualitative data analysis: A sourcebook of new methods. Beverly Hills, CA: SAGE Publications.

55.

Milgram

(1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67, 371–378.

56.

Milgram

(1974). Obedience to authority. New York: Harper & Row.

57.

Misangyi

V. F.

Weaver

G. R.

Elms

(2008). Ending corruption: The interplay among institutional logics, resources, and institutional entrepreneurs. Academy of Management Review, 33, 750–770.

58.

Modigliani

Rochat

(1995). The role of interaction sequences and the timing of resistance in shaping obedience and defiance to authority. Journal of Social Issues, 51, 107–123.

59.

O’Fallon

M. J.

Butterfield

K. D.

(2005). A review of the empirical ethical decision-making literature: 1996–2003. Journal of Business Ethics, 59, 375–413.

60.

Okhuysen

Bechky

B. A.

(2009). Coordination in organizations: An integrative perspective. Annals of the Academy of Management, 3, 463–502.

61.

Osterman

(2006). Overcoming oligarchy: Culture and agency in social movement organizations. Administrative Science Quarterly, 51, 622–649.

62.

Packer

D. J.

(2008). Identifying systematic disobedience in Milgram’s obedience experiments: A meta-analytic review. Perspectives on Psychological Science, 3, 301–304.

63.

Pettigrew

(1988) Longitudinal field research on change: Theory and practice. Paper presented at the National Science Foundation Conference on Longitudinal methods in organizations. Austin, TX.

64.

Pina e Cunha

Rego

Clegg

S. R.

(2010). Obedience and evil: From Milgram and Kampuchea to normal organizations. Journal of Business Ethics, 97, 291–309.

65.

Pitesa

Thau

(2013). Compliant sinners, obstinate saints: How power and self-focus determine the effectiveness of (un) ethical social influences. Academy of Management Journal, 56, 636–658.

66.

Posner

B. Z.

Schmidt

W. H.

(1993). Values congruence and differences between the interplay of personal and organizational value systems. Journal of Business Ethics, 12, 341–347.

67.

Putnam

L. L.

Fairhurst

G. T.

(2001). Discourse analysis in organizations. In Jablin

F. M.

Putnam

L. L.

(Eds.), The new handbook of organizational communication: Advances in theory, research and methods (pp. 78–136). Thousand Oaks, CA: SAGE Publications.

68.

Quinn

R. W.

Worline

M. C.

(2008). Enabling courageous collective action: Conversations from United Airlines flight 93. Organization Science, 19, 497–516.

69.

Reed

(1991). Organizations and rationality: The odd couple? Journal of Management Studies, 28, 559–567.

70.

Reicher

S. D.

Haslam

S. A.

Smith

J. R.

(2012). Working toward the experimenter: Reconceptualizing obedience within the Milgram paradigm as identification-based followership. Perspectives on Psychological Science, 7, 315–324.

71.

Rest

J. R.

(1989). Development in judging moral issues. Minneapolis: University of Minnesota Press.

72.

Ross

Nisbett

R. E.

(1991). The person and the situation. Philadelphia, PA: Temple University Press.

73.

Ryan

K. D.

Oestreich

D. K.

(1998). Driving fear out of the workplace: Creating the high-trust, high-performance organization. San Francisco: Jossey-Bass.

74.

Schminke

Ambrose

M. L.

Neubaum

D. O.

(2005). The effect of leader moral development on ethical climate and employee attitudes. Organizational Behavior and Human Decision Processes, 97, 135–151.

75.

Schmitt

D. P.

Allik

McCrae

R. R.

Benet-Martínez

(2007). The geographic distribution of Big Five personality traits: Patterns and profiles of human self-description across 56 nations. Journal of Cross-Cultural Psychology, 38, 173–212.

76.

Shanab

M. E.

Yahya

K. A.

(1977). A behavioral study of obedience in children. Journal of Personality and Social Psychology, 35, 530–536.

77.

Simon

H. A.

(1960). The new science of management decision. New York: Harper & Row.

78.

Smith

Heaven

P. C.

Ciarrochi

(2008). Trait emotional intelligence, conflict communication patterns, and relationship satisfaction. Personality and Individual Differences, 44, 1314–1325.

79.

Sonenshein

(2006). Crafting social issues at work. Academy of Management Journal, 49, 1158–1172.

80.

Sonenshein

(2007). The role of construction, intuition, and justification in responding to ethical issues at work: The sensemaking-intuition model. Academy of Management Review, 32, 1022–1040.

81.

Ten Have

. (2007). Doing conversation analysis. London: SAGE Publications.

82.

Tenbrunsel

A. E.

Smith-Crowe

(2008). Ethical decision making: Where we’ve been and where we’re going. Academy of Management Annals, 2, 545–607.

83.

Treviño

L. K.

(1986). Ethical decision making in organizations: A person-situation interactionist model. Academy of Management Review, 11, 601–617.

84.

Treviño

L. K.

Weaver

G. R.

Reynolds

S. J.

(2006). Behavioral ethics in organizations: A review. Journal of Management, 32, 951–990.

85.

Warren

D. E.

(2003). Constructive and destructive deviance in organizations. Academy of Management Review, 28, 622–632.

86.

Waters

J. A.

Bird

(1987). The moral dimension of organizational culture. Journal of Business Ethics, 6, 15–22.

87.

Weber

(1924/1971). Legitimate authority and bureaucracy. In Pugh

D. S.

(Ed.) Organization theory: Selected readings (pp. 3–15). Harmondsworth, UK: Penguin.

88.

Weber

(1947). The types of authority and imperative coordination. Max Weber: The theory of social and economic organization (pp. 324–385). Glencoe, IL: Free Press.

89.

Weick

K. E.

(1988). Enacted sensemaking in crisis situations. Journal of Management Studies, 25, 305–317.

90.

Weick

K. E.

(1993). The collapse of sensemaking in organizations: The Mann Gulch disaster. Administrative Science Quarterly, 38, 628–652.

91.

Weick

K. E.

(1995). Sensemaking in organizations. Vol. 3. Thousand Oaks, CA: SAGE Publications.

92.

Weick

K. E.

Sutcliffe

K. M.

Obstfeld

(2005). Organizing and the process of sensemaking. Organization Science, 16(4), 409–421.

93.

Werhane

P. H.

(1999). Moral imagination and management decision-making. New York: Oxford University Press.

94.

Werhane

P. H.

(2008). Mental models, moral imagination and system thinking in the age of globalization. Journal of Business Ethics, 78, 463–474.

95.

Wicks

A. C.

Freeman

R. E.

(1998). Organization studies and the new pragmatism: Positivism, anti-positivism, and the search for ethics. Organization Science, 9, 123–140.

96.

Wiley

(1988). The micro-macro problem in social theory. Sociological Theory, 6, 254–261.

97.

Wood

L. A.

Kroger

R. O.

(2000). Doing discourse analysis: Methods for studying action in talk and text. Thousand Oaks, CA SAGE Publications.

98.

Yin

(1984). Case study research. Beverly Hills, CA: SAGE Publications.

99.

Zhong

(2011). The ethical dangers of deliberative decision making. Administrative Science Quarterly, 56, 1–25.