The social implications of punishing unethical behavior: observers’ cognitive and affective reactions

Linda Klebe Trevino

Unethical behavior can be costly to organizations in terms of diminished public trust, tarnished organizational reputation, and lost profits (Simpson, 1987). Thus, management responses to unethical behavior are important. Previous research has suggested that punishment can reduce unethical behavior (Laczniak & Inderrieden, 1986). However, punishment can also have implications beyond the punished individual (Walters & Grusec, 1977). Arvey and Jones (1985) argued that researchers should examine the impact of discipline on individuals not punished. Recent research has begun to investigate these vicarious effects. Some studies have found that punishment can positively influence observers’ productivity and attitudes (O’Reilly & Puffer, 1989; O’Reilly & Weitz, 1980; Schnake, 1986). However, these reactions may depend upon workers’ evaluation of the legitimacy and fairness of the punishment (Arvey & Jones, 1985; Schnake, 1987).

In practical terms, observers’ reactions to punishment may be as important, if not more important than reactions of the punished individual. For example, an employee who is terminated is no longer an organizational member. Thus, his or her reactions to the punishment may be less significant than the reactions of workers who remain. This study investigates the effects of punishing unethical behavior on observers’ outcome expectancies, justice evaluations, and emotions.

Observer Reactions to the Punishment of Unethical Behavior

Punishment takes place in a social context. Thus, organizational members who are aware of the punishment may be affected. This study refers to these individuals as observers even though they may not “observe” the punishment directly. Theories of social learning and justice are used to understand observers’ cognitive and emotional reactions to the punishment of a coworker.

Outcome Expectancies

The concept of outcome expectancy has increased in importance as cognitive views of learning and behavior have become more prominent. In this cognitive view, learning is interpreted as the development of expectations that behavior will result in certain outcomes. These expectations are considered to be the major determinant of behavior (Bandura, 1986).

According to social learning theory, vicarious learning operates primarily through outcome expectancies (Bandura, 1977; Bandura, 1986; Manz & Sims, 1981). The observation of positive or negative outcomes for model influences the formation of outcome expectancies in the observer and either enhances or decreases the probability of the observer performing the behavior. In response to vicarious punishment, observers who notice and remember the punishment will expect similar management responses in the future. As a result of their expectation of future punishment, observers will alter their own behavior. Thus, outcome expectancies are thought to be important influences on future conduct.

Previous organizational research (Trevino & Youngblood, 1990) found that rewarding ethical behavior influenced observers’ outcome expectancies, which in turn positively influenced ethical behavior. However, moderate punishment of unethical behavior did not influence observers’ outcome expectancies. The researchers reasoned that, because organizational members may have a priori expectations that unethical behavior will be punished, only severe punishment is salient enough to alter their outcome expectancies. This study will test this hypothesis.

Hypothesis 1: A severe punishment response to unethical behavior will influence observers’ expectancies of future management responses more than a moderate punishment response or no management response.

Justice Evaluations and Punishment Severity

Observers may scrutinize management responses to misconduct through a justice lens (Arvey & Jones, 1985). Furby (1986) defined justice as evaluative judgments about the rightness of a person’s fate or treatment by others. Most previous justice theory and research has focused on distributive justice–the fair distribution of rights and benefits (Adams, 1963; Cook & Messick, 1983). However, the concept of retributive justice (Hogan & Emler, 1981) suggests that people are also concerned with the fairness of punishment distributions. When norms are breached, people want to know that the violator will be punished and that the social order will be maintained. This concept applies to organizational settings where workers can be expected to monitor the social environment to insure that violators of organizational rules and norms are disciplined. The retributive justice literature suggests that, in general, punishment of rule violators will be perceived as just whereas the failure to punish will result in injustice perceptions.

Management’s responses to unethical behavior may be particularly appropriate for studying coworkers’ justice evaluations. In unethical behavior situations, the punishment of an offender (or the failure to punish) may convey important symbolic meaning (Blau, 1964) to observers about the value of conformity to ethical behavior norms. Thus, deserved punishment of unethical behavior should be perceived as just to the organization’s good citizens whereas unethical behavior that goes unpunished should result in perceptions of injustice.

The potential influence of punishment severity on justice evaluations has been neglected in the justice and management literatures. The labor relations and criminal justice literatures suggest that observers’ reactions to the severity of punishment may depend upon their perception of the appropriateness of the punishment to the infraction (BNA, 1985; Markowich, 1989). This “just deserts” model argues that the severity of the sentence should be commensurate with the seriousness of the offense (Blumstein & Cohen, 1980; Emerson, 1981; Flanagan, 1987).

Alternatively, other evidence from the criminal justice literature suggests that observers may prefer severe discipline. For example, the public has been found to prefer punishments that are quite severe and more severe than those generally meted out to criminals (Blumstein & Cohen, 1980; Cullen, Clark, Cullen, & Mathers, 1985). Similarly, Trevino and Youngblood (1990) reported that, in response to specific unethical behaviors, organizational members expected the violator to be punished harshly. Because previous organizational research has not explicitly investigated the differential effects of varying punishment severity, we treat this question as exploratory.

Research question 1: How will punishment of differing severity levels influence observers’ justice evaluations?

Emotional Responses

Emotions are intense positive and/or negative dispositions that are directed at a specific object, person, or situation (Clark & Isen, 1982; Fiske, 1982). Although little research has been conducted on emotional reactions to punishment (Miller & Vidmar, 1981), punishment has been associated with negative emotional responses in the punishment recipient (Skinner, 1953). Bandura (1986) suggested that observers would feel empathy for the punished individual and experience negative emotions. However, Walters and Grusec (1977) suggested that observers may not react with negative emotions because they do not experience the punishment directly.

The justice literature suggests that emotional reactions are likely to accompany perceptions of injustice. Injustice is thought to be experienced in a highly emotional way (Greenberg, 1984; Reis, 1984) and has been associated with negative emotions such as moral outrage (Bies, 1987) and anger (Mikula, 1986). Thus, management’s failure to impose deserved punishment for unethical behavior can be expected to result in negative emotions in observers. It is also possible that positive emotional reactions may accompany perceptions of fair punishment. Observers may feel pleased that the violator has gotten his or her due (Hogan & Emler, 1981). However, positive emotions may be mitigated if the observer feels empathy for the punished individual (Bandura, 1986).

These conflicting perspectives and the paucity of previous work make it difficult to predict observers’ emotional responses to punishment of varying severity. Thus, we treat this question as exploratory as well.

Research question 2: How will punishment of differing severity levels influence observers’ emotional reactions?


Sample and Procedure

Subjects were 75 students from two sections of the MBA Organizational Behavior course taught at a large university in 1988. Twenty-one of the students were women and 54 were men. Subjects had a mean age of 25. Work experience ranged from 1 to 20 years with a mean of 4.5 years.

In lieu of a regular class session, subjects participated in a decision-making exercise to familiarize them with in-basket assessment techniques and to provide them with an opportunity to practice managerial decision making. Subjects were told that their voluntary participation in the research component of the exercise would involve completing a brief questionnaire that would be distributed after the in-basket was completed and turned in. All but one of the students agreed to participate in the research.

Each subject chose an individual “office,” a partitioned-off section of the behavioral laboratory. Subjects were randomly assigned to one of three vicarious reinforcement conditions. Names were not recorded. Before leaving, each subject was asked not to discuss the exercise with classmates who had not yet participated. All subjects were debriefed in class later in the semester.

In-Basket Exercise

The major data collection vehicle in this experiment was an in-basket exercise adapted from previous research (Trevino & Youngblood, 1990). Three in-baskets representing three levels of vicarious punishment (no management response, appropriate punishment, harsh punishment) were created. Because observers’ reactions to punishment severity are thought to be related to their evaluations of the extent to which the punishment “fits the crime,” the severity manipulations were carefully developed based upon a pre-test.(1)

The authors would like to acknowledge the following colleagues for their critiques of earlier versions of this manuscript: Jerald Greenberg, Richard Jacobs, Mark P. Sharfman, Henry P. Sims, Charles C. Snow, and Stuart A. Youngblood. They would also like to acknowledge Gloria Hsieh for assistance with the statistical analyses.

Address all correspondence to Linda Klebe Trevino, Department of Management and Organization, Smeal College of Business Administration, The Pennsylvania State University, University Park, PA 16802.

1 For the pre-test, a questionnaire was administered to 67 MBA students in two different sections of the same Organizational Behavior course. These subjects were presented with the two in-basket unethical behavior situations (a sexual harassment situation and a manager’s substitution of unsafe substandard wiring in an electrical product). For each situation, subjects rated the “severity appropriateness” of a series of 10 potential organizational responses including no response, verbal reprimands, various types of pay cuts, demotion, termination, and termination with legal action. “Severity appropriatness” was rated on a 7-point scale from too lenient to too harsh. Responses with severity appropriateness means closest to 1 (too lenient), 4(just right), and 7 (too harsh) in both situations were used to create the three experimental in-basket conditions. Thus, no action, 1-week suspension without pay, and termination with legal action were used to create the no punishment, appropriate, and harsh vicarious punishment conditions respectively. The severity appropriateness means for these three conditions in the sexual harassment situation were: M = 1.36, SD = .68; M = 3.81, SD = 1.85; M = 6.56, SD =.88 respectively. For the substandard wiring situation, the severity appropriateness means for the three conditions were: M = 1.00, SD = .00; M = 3.90, SD = 1.92,; M = 6.35, SD = .80 respectively.

The in-basket designed for this experiment consisted of an organization chart, a company newsletter, and 13 letters, memos, or phone messages. Subjects were instructed to play the role of Pat Sneed, a national sales manager for a company in some financial trouble. They were to make decisions based upon the information provided in the in-basket. Although the in-basket materials appeared identical, three different in-baskets were distributed to subjects randomly assigned to the three conditions. In each condition, information was provided regarding management’s response to two instances of unethical behavior–an unspecified sexual harassment incident, and a manager’s substitution of unsafe substandard wiring in an electrical product. Both types of unethical behavior were noted to be first offenses and against company policy. When completed, the in-basket was collected.

Appropriate vicarious punishment condition. A newsletter from the company president discussed a sexual harassment offense. Although the specifics were not discussed, the president reported that an investigation had provided convincing evidence that the harassment had occurred. The president announced that the violator was suspended for one-week without pay. In a second situation, Pat Sneed (the subject’s role) received a memo regarding the substitution of unsafe substandard wiring in products in one particular plant. The plant manager was responsible for this substitution and received a one-week suspension without pay.

Harsh vicarious punishment condition. In this condition, the newsletter announced that the harasser was dismissed from the company and legal action was begun. Similarly, in the substandard wiring situation, the plant manager was dismissed and legal action was taken.

No punishment condition. In this condition, Pat Sneed (the subject’s role) received a memo concerning a sexual harassment incident in the organization. The memo suggested that there had been no discernible management response. Pat also received a copy of a letter to the president of the company concerning the substitution of unsafe substandard wiring in electrical products at a particular plant. Again, there was no discernible management action.


Manipulation checks. Two multiple choice items about recall of management’s disciplinary actions, one for each of the two unethical behavior incidents, were included in the post-exercise questionnaire. Subjects were asked which of four possible actions (no action, one-week suspension without pay, termination and legal action, or don’t recall) “comes closest to representing management’s response to the sexual harassment (substitution of substandard wiring) incident?”

In addition, subjects in the appropriate and harsh vicarious punishment conditions completed a measure of discipline severity. Severity was measured on a five-item, 7-point bipolar scale (was management’s response mild/severe, harsh/lenient, gentle/cruel, hard/soft, and easy/strict). The mean of the five items was used to create a severity score. A higher severity score represents the belief that management’s response to the unethical behavior was more severe. Cronbach’s alpha was .95 in the sexual harassment situation and .92 in the substandard wiring situation.

Outcome expectancies. The outcome expectancy measure was adapted from previous research where outcome expectancies were found to predict ethical behavior (Trevino & Youngblood, 1990). Subjects were asked to rate five managerial behaviors in terms of how “you expect managers to be treated in the future if they engage in certain behaviors.” The behaviors were similar to those used in the in-basket. For example, “a plant manager, on his own, substitutes a cheaper but potentially unsafe product component to cut costs.” For each behavior, subjects were asked to mark a scale ranging from “1 — no action” to “7 — severe disciplinary action.” The mean of the five items was used to create a variable reflecting the subject’s expectancies of management’s future responses to unethical behavior. A higher score represents a higher expectancy that management would use severe punishment in response to unethical behavior in the future. Cronbach’s alpha was .84.

Justice evaluations. The questionnaire asked subjects to rate the justice of management’s response twice, once for the sexual harassment situation and once for the substandard wiring situation. The measure consisted of six items (was management’s response appropriate/inappropriate, proper/improper, legitimate/illegitimate, just/unjust, equitable/inequitable, and fair/unfair) on a 7-point bipolar scale. The mean of the six items was used to create a justice evaluation score. A higher score represents an evaluation of management’s action as more just. Cronbach’s alpha was .96 for the sexual harassment situation and .67 for the substandard wiring situation.

Emotions. Previous research suggests that it is appropriate to think of the evaluative aspect of emotions in terms of two broad dimensions, positive and negative (Russell, 1980; 1983). Thus, we measured both positive and negative emotional reactions to management’s actions using measures developed for a pilot study (Trevino & Ball, 1988). As with justice evaluations, positive and negative emotions were each measured twice, in response to the sexual harassment and substandard wiring situations. Six emotion words were used as indicators of positive emotions (content, pleased, comforted, agreeable, satisfied, happy). These words are similar to those used to measure the construct labeled “pleasantness” in Watson and Tellegen’s (1985) analysis of the structure of self-reported mood. Eight words were used as indicators of negative emotions (sad, discouraged, resentful, irritated, discontented, annoyed, angry, hostile). The emotion measures were designed to have face validity in that the words were selected to represent emotions that may be felt in response to the observation of management’s response to an unethical behavior situation (i.e., pleased, satisfied, annoyed, irritated). Each emotion word was accompanied by a 7-point scale anchored by the words low and high. The mean of the six positive emotion scores was used to create a measure of positive emotions for each of the two situations. A high score represents more positive emotions. Cronbach’s alpha for the positive emotions measure was .96 in the sexual harassment situation and .93 in the wiring situation. The mean of the eight negative emotion scores was used to create a measure of negative emotions for each of the two situations. A high score represents more negative emotions. Cronbach’s alpha for the negative emotions measure was .94 in the sexual harassment situation and .92 in the wiring situation.


The experimental design contained a treatment variable with three levels of vicarious punishment (no punishment, appropriate, harsh). The dependent variables were the outcome expectancy measure, justice evaluations, and positive and negative emotions. Justice evaluations and positive and negative emotions were measured twice, in response to both the sexual harassment and substandard wiring situations. Thus, questionnaire responses to these situations were treated as repeated measures in the analysis.

Manipulation Checks

Subjects responded to multiple choice questions asking them to indicate the specific level of punishment they recalled, if any, in the two situations. Chi square analyses tested the independence of manipulation recall and experimental conditions. The Chi square was significant in the sexual harassment (Chi square = 70.6, p |is less than~ .0001) and the substandard wiring situations (Chi square = 61.1, p |is less than~ .0001).

For the sexual harassment situation, 65% of those in the no punishment group, 82% of those in the appropriate punishment group, and 72% of those in the severe group checked the punishment manipulation cues corresponding to their respective conditions. In the substandard wiring situation, 58% of those in the no punishment group, 68% of those in the appropriate group, and 68% of those in the severe group checked the exact punishment manipulation cues for their respective conditions. In the no punishment condition, a substantial number of respondents checked the “don’t recall” option (12% in the sexual harassment situation and 27% in the substandard wiring situation). We consider this “don’t recall” option to be accurate because the no punishment in-basket provided no information about management’s response that could be recalled.

The severity scores served as a further manipulation check. A t-test was used to compare the mean severity perceptions of subjects in the appropriate and harsh vicarious punishment conditions who responded to the severity question. The t-tests were significant in both situations. In the sexual harassment situation (t = -5.66, p = .0001) the mean severity scores for the appropriate and harsh groups were 3.31 and 5.25 respectively. In the substandard wiring situation (t = -5.40, p = .0001) the means for the appropriate and harsh groups were 3.63 and 5.44 respectively.

Overall MANOVA Result

MANOVA examines the relationship between a set of categorical predictor variables and two or more dependent response measures. This simultaneous test for the effect on the combination of dependent variables is important because in many cases, as in this study some of the dependent variables are correlated. As an initial omnibus test to control the overall alpha level at .05, multivariate analysis of variance procedures were used to test the hypothesis of no overall cell effect for the following dependent variables: justice evaluations, positive emotions, negative emotions, and outcome expectancies (Bray & Maxwell, 1985). The MANOVA was statistically significant, Wilk’s lambda = .355, F (14,116) = 5.63, p |is less than~ .0001.


Observers’ Outcome Expectancies

To test hypothesis one, ANOVA procedures with planned comparisons were used. The overall analysis of variance was statistically significant. The planned comparisons suggested that expectancies of management’s future use of severe punishment were lowest in the no punishment condition and highest in the harsh condition. As hypothesized, subjects in the harsh vicarious punishment group were significantly more likely to expect management to punish unethical behavior harshly in the future than those in the no punishment or appropriate vicarious punishment conditions.

Observers’ Justice Evaluations

Subjects were asked to report justice evaluations for both the sexual harassment and substandard wiring situations. Because these justice evaluations are not independent, SAS repeated-measures ANOVA procedures were used to test (a) the overall main effect of level of punishment across both situations and (b) the univariate main effect of level of punishment in each situation. Although specific differences among means were not hypothesized in advance, we manipulated punishment severity with the intent of creating and looking at differences across the three manipulated conditions. Therefore, planned contrasts were used to compare mean justice evaluations across the punishment conditions within each situation. Finally, the analysis provided information about the situation effect and the punishment by situation interaction.

Table 2

Observers’ Outcome Expectancies(a)

Under Different Punishment Severity Levels

Mean Outcome Expectancies

Univariate F No Pun Appropriate Harsh

(n=26) (n=22) (n=25)

|F.sub.2,64~= 4.24(*) 4.72 4.83 5.62

ns (*)


a A higher outcome expectancy score represents a higher

expectancy that management would use severe punishment in

response to unethical behavior in the future.

* p |is less than or equal to~ .05. ** p |is less than or equal

to~ .01.

Table 2

Observers’ Justice Evaluations(a)

Under Different Punishment Severity Levels

Mean Justice Evaluations

Univariate F No Pun Appropriate Harsh

(n=26) (n=21) (n=25)

Sexual Harrassment

|F.sub.2,68~= 26.88(**) 3.29 4.63 6.21

(**) (**)


(n=26) (n=21) (n=24)

Substandard wiring

|F.sub.2,68~= 6.35(**) 3.70 4.76 5.19

(*) ns


a A higher justice score represents an evaluation of

management’s actions as more just.

* p |is less than or equal to~ .05. ** p |is less than or equal

to~ .01.

The overall test for the main effect of level of punishment across both situations was (|F.sub.4,134~ = 11.72, p = .0001). Univariate between-subjects effects for level of punishment were significant for both the sexual harassment (|F.sub.2,68~ = 26.88, p = .0001) and the substandard wiring (|F.sub.2,68~ = 6.35, p = .003) situations. In both unethical behavior situations, the pattern of means was the same. The means were lowest in the no punishment condition (indicating an evaluation of management’s response as least just) and highest in the harsh punishment condition. The planned contrasts found that, in the sexual harassment situation, justice evaluations across the three punishment conditions were all significantly different from each other (p |is less than~ .01). In the substandard wiring situation, the contrast between the no punishment and appropriate conditions was statistically significant (p = .02), but the contrast between the appropriate and harsh conditions was not. Thus, with regard to research question one, subjects generally evaluated management’s response to be most just in the harsh vicarious punishment condition and least just in the no punishment condition.

Although not hypothesized, we also investigated the effect of the situation and found no significant situation effect (|F.sub.1,68~ = .86, p = .3579). Finally, the overall punishment condition by situation interaction was statistically significant (|F.sub.2,68~ = 5.35, p = .0058). Graphing the means for each situation across conditions illustrates that the profiles diverge with reactions to harsh punishment. In the substandard wiring situation, justice evaluations level off between appropriate and harsh punishment. However, respondents evaluated harsh punishment as significantly more just than appropriate punishment in the sexual harassment situation.

Observers’ Emotional Responses

Subjects were asked to report their positive and negative emotional responses for both the sexual harassment and substandard wiring situations. Thus, SAS repeated-measures MANOVA procedures were used to test for (a) the main effect of level of punishment on the two dependent variables across the two situations (b) the univariate effect of level of punishment in each situation (c) planned contrasts to compare mean emotion scores across the punishment conditions in each situation (d) the overall situation effect and (e) the overall punishment by situation interaction.

The multivariate test of between subjects effects of level of punishment was statistically significant (|F.sub.8,124~ = 6.91, p = .0001). In the sexual harassment situation, the univariate F-tests for the effect of punishment level were significant for positive (|F.sub.2,65~ = 26.25, p = .0001) and negative (|F.sub.2.68~ = 19.36, p = .0001) emotional responses. Planned contrasts were used to compare mean positive and negative emotions across the three punishment conditions and found that the means were all significantly different from each other (p |is less than~ .01) with the exception of the means for negative emotions in the no punishment and appropriate conditions (p |is less than~ .06). In response to research question two, positive emotions were most positive in the harsh condition and least positive in the no punishment condition. Negative emotions were most negative in the no punishment condition and least negative in the harsh condition. However, the substandard wiring situation did not produce similar results. The univariate F tests of differences among mean emotion scores for subjects in the three conditions were not significant for either positive TABULAR DATA OMITTED (|F.sub.2,65~ = .95, p = .3925) or negative (|F.sub.2,65~ = .59, p = .5592) emotions.

We also tested for the overall effect of situation and a punishment condition by situation interaction. There was no statistically significant situation effect (|F.sub.3,63~ = 2.15, p = .10). However, there was a statistically significant punishment condition by situation interaction (|F.sub.6,126~ = 8.934, p = .0001), suggesting that the pattern of means across the two situations differs by punishment condition. Similar to the finding for justice evaluations, observation of the plotted means in Figure 2 suggests that emotional reactions diverge under harsh punishment. In the substandard wiring situation, emotional responses level off between appropriate and harsh punishment. However, the increase to harsh punishment produces more positive and less negative emotions in the sexual harassment situation.

Finally, although we did not hypothesize a difference, we conducted a post hoc analysis to explore whether women might experience more negative emotional reactions to the sexual harassment situation than men. However, t-tests resulted in no significant differences between men and women for either negative emotions (t= 1.16, p|is greater than~.25) or positive emotions (t = -.90, p|is greater than~.37).


The study findings suggest that punishing unethical behavior at varying severity levels has differential social consequences. First, as hypothesized, only a severe punishment response to the in-basket unethical behaviors influenced outcome expectancies. Compared to observers in the other two conditions, observers in the harsh vicarious punishment condition perceived that management’s future response would be more severe. In addition, subjects in the appropriate and harsh vicarious punishment conditions generally evaluated management’s response as more just and reported more positive emotions than subjects in a condition where management did not punish unethical behavior. Emotions and justice evaluations were generally most positive in the most severe punishment condition.

Study Limitations and Implications for Future Research

An in-basket study using graduate student subjects may have limited external validity. However, in-basket exercises have been used effectively in much management research and properly designed in-basket exercises have been found to correspond to actual on-the-job work performance (Brass & Oldham, 1976). In addition, these study subjects had considerable work experience. Future research conducted in field settings can enhance generalizability of the findings.

The in-basket punishment was somewhat removed from the subject’s personal experience. Thus, the study results may be generalizable only to situations where organizational members are aware of disciplinary action taken against someone they do not know. However, these situations are not rare. The organization’s grapevine is likely to carry information about disciplinary actions that have been taken against organizational members outside one’s immediate work group. Future research should investigate differences in reactions when the observer identifies with or knows the discipline recipient (Brockner, Grover, Reed, DeWitt, & O’Malley, 1987; Schnake, 1987).

We were also concerned about the possibility that subjects would respond with socially desirable responses in ethics or punishment situations. For this reason, the “punishment for unethical behavior” manipulation was embedded in a fairly complex in-basket exercise and a cover story was provided, describing the exercise as a general decision-making experience. In addition, subjects’ responses were completely anonymous and confidentiality was promised. Orne (1962) suggested that subjects in psychological experiments play the role of “good subject.” Therefore, their perceptions of the purpose of the research should be considered a potential influence on the results. If the purpose of the experiment is relatively ambiguous, subjects will consider many different hypotheses and the results will not be systematically affected. In an attempt to study the potential role of demand characteristics in this study, the first question on the post-exercise questionnaire asked subjects to state their perception of the exercise’s purpose. Content analysis of the open-ended responses revealed that only 6.9% of the sample perceived an ethics or punishment focus. In accordance with the cover story, most subjects believed that the exercise was designed to provide general decision-making experience. Nevertheless, despite multiple efforts to reduce demand characteristics, it remains possible that subjects responded in a socially desirable manner.

Further, this study focused on the influence of one important variable, observer reactions to the severity of punishment outcomes. Reactions to actual punishment situations are likely to be influenced by many additional variables not studied here. For example, recent research suggests that punishment procedures may be important influences on justice evaluations (Ball, 1991; Leventhal, 1980; Thibaut & Walker, 1975). Future research should also explore other variables that may influence observers’ reactions to discipline. For example, the unethical behaviors used in the in-basket assume consensus about the inappropriateness of the behavior based upon laws (sexual harassment) and organizational policies (regarding product safety). Responses may differ in situations where observers and management disagree about the impropriety of the violator’s behavior. Thus, unethical behavior that coworkers agree is wrong may be particularly prone to strong retributive reactions. Attributional processes may also play an important role in observers’ reactions to punishment (Cullen et al., 1985; Miller & Vidmar, 1981). If coworkers ascribe responsibility and intentionality to the offender, they may evaluate management’s punishment response positively. However, if management is blamed, observer reactions to the punishment are more likely to be negative.

Finally, patterns of responses were somewhat different in the two situations, particularly in response to harsh punishment. For example, in the sexual harassment situation, justice evaluations and positive emotions increased substantially from the appropriate to the harsh punishment condition, but they remained more stable in the substandard wiring situation. These findings suggest that observers’ reactions to harsh punishment may be situation-specific and may be related to moral intensity of the situation. Jones (1991) proposed that moral intensity comprises six components: magnitude of consequences, social consensus, probability of effect, temporal immediacy, proximity, and concentration of effect. A comparison of the sexual harassment and substandard wiring situations from the in-basket suggests that the sexual harassment situation is higher in moral intensity. Sexual harassment is illegal, suggesting a certain degree of social consensus that the behavior is wrong. The harm has already been done (suggesting higher temporal immediacy and probability of effect) compared to the substandard wiring situation, where there exists only the possibility of future harm. The sexual harassment also occurred within the organization (proximity). Finally, concentration of effect is difficult to discern because the consequences of the substitution of substandard wiring are potential rather than real consequences. On balance, this analysis suggests that the sexual harassment situation is higher in moral intensity and may, therefore, provoke a stronger retributive response favoring harsh punishment. Future research should investigate more systematically whether observer reactions to managerial responses to unethical behavior differ depending upon the moral intensity or other characteristics of the situation.

Social Learning and Justice Implications for the Manager

The study results suggest that managers should consider the social learning implications of their disciplinary actions and the important role that discipline severity may play. We found that discipline influences observers’ outcome expectancies only when it is relatively harsh. Given the previous research finding that outcome expectancies significantly and directly influenced ethical decision-making behavior (Trevino & Youngblood, 1990), our findings suggest that the application of harsh discipline to one individual may influence the future ethical behavior of other individuals. Future research is needed to definitively establish the links between discipline severity, outcome expectancies, and observers’ future ethical behavior.

Punishing unethical behavior also appears to raise retributive justice concerns (Hogan & Emler, 1981). Punishing or failing to punish an individual for misconduct takes on important social meaning. Others in the social setting pay close attention and react to the outcome in terms of its meaning for them and the values of their group. In this study, organizational members who were aware that unethical behavior had occurred, reacted positively when the misconduct (especially sexual harassment) was punished and punished harshly. Some subjects offered written reactions on the questionnaire. For example, one subject from the harsh condition said, “The organization must take an aggressive stance on values to protect individual employees.” Subjects’ reactions were most negative when management did not respond to the misconduct: “Receiving an unclear response does not show any direction from top management.” “Endangering people to save money is immoral and unethical.” In the appropriate punishment condition, many subjects felt that management’s response was too weak.

It is good that the person was punished–it is conceivable that the punishment should have been more severe. Action must be punished to send a clear message, both to possible offenders and possible victims, that this will not be tolerated. I hesitate to say that management’s response was fair only because I think it is possible that a more fair response would be to fire the person.

The latter response points to the social meaning of punishment. Punishment sends a message to potential offenders and victims that norm violation will not be tolerated. The practical message suggests that managers who avoid disciplinary situations may be sending a powerful unintended signal to their other subordinates and may be sowing the seeds of discontent (Arvey & Jones, 1985). Previous research also suggests that evaluations of punishment as unfair may affect coworkers’ productivity and satisfaction with the supervisor (O’Reilly & Puffer, 1989; Schnake, 1987). Future researchers may wish to investigate the relationship between justice evaluations and these other important outcomes (Fryxell & Gordon, 1989).

When combined with previous research, these findings may create an interesting paradox for the manager. First, observers appear to react positively to harsh discipline of unethical behavior. However, previous literature has argued that the disciplined individual may react negatively to harsh punishment (Beyer & Trice, 1984; Skinner, 1953). If the offender and observers differ in their reactions, with whom should managers be most concerned? Is it possible to make punishment a constructive experience for all? These unanswered questions present a challenge for future research.

Second, organizational behavior textbooks prescribe that punishment should be carried out in private (e.g., Hellriegel, Slocum, & Woodman, 1989). In fact, publicity can be considered additional punishment for the offender. However, the research reported here suggests that, for punishment to have social meaning, the discipline may need to be made known to observers in some way. We found that observers’ justice perceptions and emotional reactions were most positive when they knew that misconduct had been punished. Thus, the manager may be faced with an additional dilemma. How private or public should punishment be? We speculate that a potential solution may be to conduct discipline in private, but to publicize information about offenses and managerial responses without identifying the offender or the specifics of the incident. For example, West Point regularly publishes a list of honor code violations and the disciplinary actions without disclosing the offenders’ identities. In this way, the organization informs organizational members of disciplinary actions, thus contributing to vicarious learning and justice perceptions while keeping the specific offender and incident details confidential. Future research should investigate the effectiveness of such systems.

This study contributes to the growing empirical work on punishment and the management of unethical organizational behavior. It suggests that punishing unethical behavior is not only acceptable to observers, but that relatively harsh punishment seems to be preferred. In this study, harsh punishment had positive outcomes in terms of observers’ outcome expectancies, justice evaluations, and emotional responses. However, many factors not studied here may also contribute to the observer’s reactions to a punishment incident. This study should be considered only one step in a research area that needs much additional work. Future studies should consider the influence of severity within a more complex punishment context.


Adams, J.S. 1963. Toward an understanding of inequity. Journal of Abnormal and Social Psychology, 67: 422-436.

Arvey, R.D., & Jones, A.P. 1985. The use of discipline in organizational settings: a framework for future research. In B. Staw & L.L. Cummings (Eds.), Research in Organizational Behavior, 7: 367-408. Greenwich, CT: JAI Press.

Ball, G. 1991. Outcomes of punishment incidents: The role of subordinate perceptions, individual differences, and leader behavior. Unpublished doctoral dissertation, The Pennsylvania State University.

Bandura, A. 1977. Social learning theory. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. 1986. Social foundations of thought and action: a social-cognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Beyer, J.M., & Trice, H.M. 1984. A field study of the use and perceived effects of discipline in controlling work performance. Academy of Management Journal, 27: 743-764.

Bies, R.J. 1987. The predicament of injustice. In B.M. Staw & L.L. Cummings, (Eds.), Research in Organizational Behavior, 9: 290-318. Greenwich, CT: JAI Press.

Blau, P.M. 1964. Exchange and power in social life. New York: John Wiley and Sons.

Blumstein, A., & Cohen, J. 1980. Sentencing of convicted offenders: An analysis of the public’s view. Law and society review, 14: 223-261.

BNA. 1985. BNA Employee Discipline and Personnel Policies Forum. 139:10, Washington, DC: Bureau of National Affairs.

Brass, D.J., & Oldham, G.R. 1976. Validating an in-basket test using an alternative set of leadership scoring dimensions. Journal of Applied Psychology, 61: 652-657.

Bray, J.H. & Maxwell, S.E. Multivariate analysis of variance. Beverly Hills: Sage.

Brockner, J., Grover, S., Reed, T., DeWitt, R.L., & O’Malley. 1987. Survivors’ reactions to layoffs: we get by with a little help for our friends. Administrative Science Quarterly, 32: 526-541.

Clark, M.S., & Isen, A.M. 1982. Toward understanding the relationship between feeling states and social behavior. In A. Hastorf & A. M. Isen (Eds.), Cognitive social psychology: 73-108. New York: Elsevier Science Publishing.

Cullen, F.T., Clark, G.A., Cullen, J.B., & Mathers, R.A. 1985. Attribution, salience, and attitudes toward criminal sanctioning. Criminal justice and behavior, 12: 305-331.

Emerson, R.M. 1981. On last resorts. American Journal of Sociology. 87: 1-22.

Fiske, S.T. 1982. Schema-triggered affect: Applications to social perception. In M.S. Clark & S.T. Fiske (Eds.), Affect and cognition: 55-78. Hillsdale, NJ: Erlbaum.

Flanagan, T.J. 1987. Change and influence in popular criminology. Journal of Criminal Justice, 15: 231-243.

Fryxell, G.E., & Gordon M.E. 1989. Workplace justice and job satisfaction as predictors of satisfaction with union and management. Academy of Management Journal. 32: 851-866.

Furby, L. 1986. Psychology and Justice. In R.L. Cohen (Ed.), Justice: views from the social sciences: 153-203. New York: Plenum.

Greenberg, J. 1984. On the apocryphal nature of inequity distress. In R. Folger (Ed.), The sense of injustice; social psychological perspectives: 167-186. New York: Plenum.

Hellriegel, D., Slocum, J.W., Jr., & Woodman, R.W. 1989. Organizational behavior (2nd ed.). New York: West Publishing Company.

Hogan, R., & Emler, N.P. 1981. Retributive justice. In M.J. Lerner & S.C. Lerner (Eds.), The justice motive in social behavior: 125-143. New York: Plenum.

Howarth, E., & Schokman-Gates, K. 1981. Self-report multiple mood instruments. British Journal of Psychology, 72:421-441.

Jones, T. 1991. Ethical decision-making by individuals in organizations: An issue-contingent model. Academy of Management Review, 16: 366-395.

Laczniak, G.R., & Inderrieden, E.J. 1986. The influence of stated organizational concern upon ethical decision-making. Journal of Business Ethics, 6: 297-307.

Leventhal, G.S. 1976. Fairness in social relationships. In J.W. Thibaut, J.T. Spence, & R. Carson (Eds.) Contemporary topics in social psychology: 211-239. Morristown, NJ: General Learning.

Leventhal, G.S. 1980. What should be done with equity theory? New approaches to the study of fairness in social relationships. In K. Gergen, M. Greenberg, & R. Willis (Eds.) Social exchange: Advances in theory and research: 27-55. New York: Plenum.

Manz, C.C., & Sims, H.P. 1981. Vicarious learning: The influence of modeling on organizational behavior. Academy of Management Review. 6: 105-113.

Markowich, M.M. 1989. A positive approach to discipline. Personnel, August: 60-63.

Mikula, G. 1986. The experiment of injustice. In H.W. Bierhoff, R.L. Cohen, & J. Greenberg (Eds.), Justice in social relations: 103-123. New York: Plenum.

Miller, D.T., & Vidmar, N. 1981. The social psychology of punishment reactions. In M.J. Lerner & S.C. Lerner (Eds.), The justice motive in social behavior: 145-172, New York: Plenum.

O’Reilly, C.O., III, & Puffer, S. 1989. The impact of rewards and punishments in a social context: a laboratory and field experiment. Journal of Occupational Psychology, 62: 41-53.

O’Reilly, C.O., III, & Weitz, B.A. 1980. Managing marginal employees: the use of warnings and dismissals. Administrative Science Quarterly, 25: 467-483.

Reis, H.T. 1984. The nature of the justice motive: some thoughts on operation, internalization, and justification. In J.C. Masters & W.P. Smith (Eds.), Social comparison, social justice, and relative deprivation: theoretical, empirical, and policy perspectives: 131-150, Hillsdale, NJ: Erlbaum.

Russell, J.A. 1980. A circumplex model of affect. Journal of Personality and Social Psychology, 39: 1161-1178.

Russell, J.A. 1983. Pancultural aspects of the human conceptual organization of emotion. Journal of Personality and Social Psychology, 45: 1281-1288.

Schnake, M.E. 1986. Vicarious punishment in a work setting. Journal of Applied Psychology. 71: 343-345.

Schnake, M.E. 1987. Vicarious punishment in a work setting: A failure to replicate. Psychological Reports, 71: 379-386.

Simpson, J.C. 1987. Wail Street’s courting of MBAs proceeds apace despite scandals. Wall Street Journal, January 28, Section 2: 1.

Skinner, B.F. 1953. Science and human behavior. New York: Macmillan.

Thibaut, J., & Walker, J. 1975. Procedural justice: A psychological analysis. Hillsdale, NJ: Erlbaum.

Trevino, L.K., & Ball, G. 1988. The use of vicarious punishment to manage unethical behavior: Affective and attitudinal outcomes. Paper presented at the meeting of the Academy of Management, Anaheim, CA.

Trevino, L.K., & Youngblood, S.A. 1990. Bad apples in bad barrels: a causal analysis of ethical decision making behavior. Journal of Applied Psychology, 74: 378-385.

Walters, G.C., & Grusec, J.E. 1977. Punishment. San Francisco: W.H. Freeman.

Watson, D., & Tellegen, A. 1985. Toward a consensual structure of mood. Psychological Bulletin, 98 (2): 219-235.

COPYRIGHT 1992 Sage Publications, Inc.

COPYRIGHT 2008 Gale, Cengage Learning

You May Also Like

Structured conflict and consensus outcomes in group decision making

Structured conflict and consensus outcomes in group decision making Richard L. Priem Pressures for early consensus during group dec…

Sources of CEO power and firm financial performance: a longitudinal assessment

Sources of CEO power and firm financial performance: a longitudinal assessment Catherine M. Daily The chief executive officer (CEO)…

Implications for human resource management

The knowledge, skill, and ability requirements for teamwork: implications for human resource management Michael J. Stevens Deutsch, …

Simple question, not so simple answer: interpreting interaction terms in moderated multiple regression

Simple question, not so simple answer: interpreting interaction terms in moderated multiple regression – Research Methods & Analysis Arthu…