Using a Bug-Killing Paradigm to Understand How Social Validation and Invalidation Affect the Distress of Killing

Abstract

Clinical evidence demonstrates that killing among soldiers at war predicts their experience of long-lasting trauma/distress. Killing leads to distress, in part, due to guilt experienced from violating moral standards. Because social consensus shapes what actions are perceived as moral and just, we hypothesized that social validation for killing would reduce guilt, whereas social invalidation would exacerbate it. To examine this possibility in a laboratory setting, participants were led to kill bugs in an “extermination task.” Perceptions of social validation/invalidation were manipulated through the supposed actions of a confederate (Study 1) or numerous previous participants (Study 2) that agreed or refused to kill bugs. Distress measures focused on trauma-related guilt. Higher levels of distress were observed when individuals perceived their actions as invalidated as opposed to when they perceived their actions as socially validated. Implications for posttraumatic stress disorder (PTSD) experienced by soldiers and the paradoxical nature of publicly expressing antiwar sentiments are discussed.

Keywords

killing social validation PTSD guilt war

Research on posttraumatic stress disorder (PTSD) paints a grim portrait of how war negatively affects psychological health. From a clinical perspective, one particularly debilitating component of war is that soldiers are placed into morally injurious situations that require them to act in ways that may run counter to previously held beliefs about what is personally acceptable (Litz et al., 2009). These actions include killing, being unable to help the wounded, firing on civilians, and participating in guerilla warfare, all of which can shake one’s foundations of right and wrong. Research has more specifically indicated killing at war as a major risk factor for psychological distress among soldiers. Vietnam veterans who killed, for instance, exhibited greater PTSD symptoms than those not directly involved in the killing of an enemy combatant, and killing remained a significant predictor of PTSD symptoms after controlling for battle intensity (i.e., fighting in light vs. heavy combat; MacNair, 2002). In fact, research identified killing as the best predictor of psychological dysfunction, after controlling for combat exposure in Iraq War veterans (Maguen et al., 2010) or controlling for witnessing killing among Gulf War veterans (Maguen et al., 2011). Despite these findings, however, soldiers eventually diagnosed with PTSD are in the minority. Within the same sample of Iraq veterans, although 40% had killed during combat, only 16% met PTSD criteria (Maguen et al., 2010). This implies that more than 50% of soldiers who killed during battle returned home in relatively good mental health. If killing at war is so often associated with psychological distress, why do many soldiers emerge unscathed?

One social variable that seems to influence the distress experienced by soldiers is an individual soldier’s perception of what the greater collective thinks about the validity of the war effort. In other words, a soldier’s level of distress should be influenced by the perception of social consensus approving and supporting a war effort—that is, social validation—or by the perception of social consensus condemning and disapproving of the war effort—that is, social invalidation. The presence of validation or invalidation may influence the degree to which soldiers experience trauma in response to killing because social opinion helps shape what actions are perceived as moral or just. At a broad level, various theorists argue that our worldviews or belief systems—including conceptions of right and wrong—are socially constructed and maintaining faith in them requires consensual validation (e.g., Becker, 1962/1971; Berger & Luckmann, 1966; Kelley, 1967). Indeed, Festinger (1954) advanced the idea that other people provide a primary basis for evaluating the appropriateness of one’s behavior and the correctness of one’s attitudes. From this perspective, one’s views of right and wrong are strongly affected by social consensus. This process is similar to the effect of informational influence on conformity, in which an individual conforms to the actions and opinions of the collective because that person assumes the collective to be more informed and accurate than an individual (Crutchfield, 1955; Deutsch & Gerard, 1955).

If moral attitudes about right and wrong are shaped by social consensus, then social validation may ease distress by bringing moral attitudes in line with killing behavior. This notion follows directly from cognitive dissonance theory (Festinger, 1957). According to the theory, dissonance should arise when, by killing, soldiers are asked to behave in a manner that runs counter to moral beliefs. Social validation should reduce this dissonance by changing the perception that killing is wrong. In other words, because moral attitudes are socially constructed, social validation creates the perception that, at least in the current context, killing is morally acceptable. As a result, consonance between moral attitudes and behavior is restored, thereby reducing dissonance and distress. Likewise, social invalidation should increase dissonance because it reinforces the perception that killing is immoral, thus increasing the discrepancy between moral attitudes and killing.

Clarifying the Relative Impact of Validation and Invalidation

Though we theorize that validation should decrease distress from killing and that invalidation should increase this distress, each of these effects may be more or less present or visible depending on the situation. Specifically, when other existing factors are in place that help justify killing and thus reduce distress from killing, we may see little additional distress-reduction as a result of social validation, but a marked increase in distress due to invalidation. In contrast, when few or no other existing factors are present to help justify killing and thus mitigate distress, invalidation may have little noticeable effect on increasing distress, but validation may markedly decrease distress.

Based on this theorizing, in the context of killing at war, the presence of invalidation may have greater impact than validation because a host of factors frequently emerge in conflict situations that mitigate the perceived immorality of killing and thus mitigate soldiers’ distress. First, war is conducted under a mantle of governmental legitimacy (Archer & Gartner, 1992) where soldiers are provided believable evidence that annihilating the opposition is instrumental in achieving a higher purpose. Second, dehumanization—labeling and perceiving the enemy as subhuman—frequently emerges in conflict situations. In essence, war becomes a situation of “us” versus “it,” as can be seen through German propaganda presenting Polish people as an “Eastern-European species of cockroach” (Weinberg, 1995) or American soldiers treating captured enemies at the Abu Ghraib prison like dogs, even making them bark and leading them around on leashes (Fay, 2004; Zimbardo, 2007). Third, political figures and military leaders utilize euphemistic labeling to make the act of killing sanitized, detached, and mechanistic. Soldiers, for instance, do not kill, but “engage targets”; bombing missions are referred to as “clean surgical strikes”; and torture of enemy soldiers becomes “enhanced interrogation” (e.g., Bandura, 1999; Baum, 2004). Fourth, dissonance theory (Festinger, 1957) suggests that people are naturally inclined to justify actions that are out of line with their values (i.e., to devise their own rationalizations and defenses to feel that their actions are more ethically acceptable).

All these moral disengagement factors can assist in enhancing the moral acceptability of killing, and in turn make killing easier and less distressing (e.g., Aquino, Reed, Thau, & Freeman, 2007; Castano, 2008; Castano & Giner-Sorolla, 2006; Maoz & McCauley, 2008; McAlister, Bandura, & Owen, 2006). As a result, we surmise that in real-world conflict situations, which tend to foster these distress-diminishing factors, social invalidation may have a particularly powerful influence by way of reinforcing the perception that killing is immoral.

Indirect Evidence From the Clinical Literature

To our knowledge, research has yet to examine the effect of social validation/invalidation on PTSD. Clinical research, however, has identified social support—emotional and physical comfort provided by members of one’s social network—as protective against PTSD formation and important in PTSD remission (for a review see Charuvastra & Cloitre, 2008). The protective nature of social support received during battle, for instance, has been documented by examining feelings of connection and support from a soldier’s unit members (Brailey, Vasterling, Proctor, Constans, & Friedman, 2007). Similarly, social support received upon returning home, or during one’s homecoming reception, is a powerful protective factor in the development of PTSD (Bolton, Litz, Glenn, Orsillo, & Roener, 2002; King, King, Fairbank, Keane, & Adams, 1998; Koenen, Stellman, Stellman, & Sommer, 2003; Schnurr, Lunney, & Sengupta, 2004; Solomon, Waysman, & Mikulincer, 1990). For example, Somalia peacekeepers who indicated a warm homecoming reception 15 weeks after returning home also endorsed fewer PTSD symptoms 1.5 years later (Bolton et al., 2002). It seems plausible that social support frequently includes explicit or implicit social validation of having engaged in killing. For example, friends and family that provide emotional support also seem likely to support a soldier’s decision to fight in war and deem that fighting necessary. Indeed, the social validation inherent in social support may aid in its psychologically protective nature.

Research Overview

Research addressing phenomena involving the trauma of war and PTSD typically involves interviewing veterans returning from war in an attempt to isolate variables associated with PTSD symptoms. While this research has been invaluable, it is retrospective and correlational in nature, and limits the ability to draw causal conclusions. To examine causation, an experimental paradigm is ideal. Thus, the present research approximated the experience of soldiers within a laboratory setting by adapting a bug-killing paradigm used in prior research (Martens, Kosloff, Greenberg, Landau, & Schmader, 2007; Martens, Kosloff, & Jackson, 2010). In this paradigm, participants “become” exterminators and complete a bug-extermination task. Although we fully recognize that killing bugs in the lab is not identical to killing people at war, using this paradigm as a proxy for a soldier’s experience allowed for the manipulation of social validation/invalidation, an impossible feat under real-world constraints.

Social validation and invalidation can be determined through both verbal expressions of opinion (i.e., agreement/disagreement with killing bugs) and behavior that is consistent or inconsistent with the specific act of killing (i.e., agreeing/refusing to kill bugs). After inducing participants to ostensibly kill a number of bugs, social validation and invalidation were manipulated through the behavior of other ostensible participants. In Study 1, this was manipulated through a participant–confederate interaction in which a supposed student agreed or refused to complete the study. In Study 2, participants were given a participation record book that provided access to the actions of supposed previous participants that either killed bugs or not. We then assessed various indices of psychological distress. Clinical researchers suggest that guilt and shame are critical in understanding PTSD among soldiers at war, and may be psychological markers that distinguish PTSD among soldiers from PTSD resulting from other traumatic events (Litz et al., 2009). As such, we adapted a clinical scale designed to assess these constructs and utilized measures of guilt more typical in experimental social psychology.

Across both studies, we were most interested in the comparison between the validation and invalidation conditions mentioned above (i.e., that invalidation should lead to greater distress than validation). Nevertheless, we included a control condition to explore whether validation or invalidation have greater respective influences on distress. The bug-killing paradigm has several parallels to the type of killing perpetrated at war: killing to achieve a higher purpose (i.e., gaining knowledge through research) legitimized by institutional authority, killing of a subhuman target (literally, an insect), and killing in a manner that is mechanistic and that uses euphemistic labeling (i.e., participants “exterminate” instead of kill by “depositing” insects into an “extermination machine”). Furthermore, we expected participants to engage in some form of dissonance reduction as they should be motivated to reduce the tension that arises from the act of killing. We therefore tentatively predicted that relative to the control condition, invalidation may have more of an effect on increasing distress than validation does on reducing it.

Study 1

Participants were randomly assigned to one of three conditions after ostensibly killing bugs. In the social invalidation condition, perceptions that killing insects was wrong in the eyes of college students were induced by having participants interact with a confederate who refused to complete the study on moral grounds. In the social validation condition, participants interacted with a confederate who readily agreed to participate, thus presenting the bug-killing task as acceptable to one’s peer group. In the control condition, participants had no confederate interaction, and were thus given no information concerning the social acceptance of their actions. The study ended after participants completed a packet containing various measures of distress.

Providing participants with social validation—information that others agree with and support the killing behavior—should reduce the behavior-standard discrepancy and therefore reduce distress. In contrast, social invalidation should highlight the behavior-standard discrepancy and increase distress. As such, our main prediction was that social invalidation should lead to higher levels of distress than social validation. We expected levels of distress for control participants to fall somewhere in between these two means, but to be at a similar level as the validation condition.

Method

Participants

Participants were 45 undergraduate students participating for partial course credit. Data from five participants were excluded.¹ The data from the remaining 40 students (22 female, 17 male, 1 unreported) were used in all reported analyses (M_age = 19.18, SD_age = 1.68).²

Materials

Trauma-related guilt inventory (TRGI)

To remain consistent with the trauma experienced by soldiers at war, we adapted the TRGI (Kubany et al., 1996) to the experimental context. The TRGI captures various affective and cognitive components of traumatic guilt derived from clinical experience, interviews with trauma survivors, and a review of the clinical literature (e.g., Kubany et al., 1995). The inventory is divided into two scales: Global Guilt and Guilt Cognitions. The Guilt Cognitions scale is further divided into four subscales—Distress, Lack of Justification, Hindsight Bias/Responsibility, and Wrongdoing—each capturing distinct components of traumatic guilt, as experienced in relation to some specific traumatic event. The Distress subscale assesses negative emotional arousal. The Wrongdoing subscale captures if the traumatic event violated moral standards about right and wrong. The Hindsight Bias subscale examines perceptions of personal responsibility in causing the traumatic event, and cognitions about the changeability of the event. Finally, the Lack of Justification subscale measures the existence of justifiable reasons explaining why the event occurred.

The TRGI was revised, a priori, to adapt it to the experimental context. The Global Guilt scale was removed entirely because it asked participants to indicate how frequently they experienced guilt since the occurrence of some traumatic event (e.g., Indicate how frequently you experience guilt that relates to what happened). These items were deemed unnecessary as the traumatic event reported on had occurred merely minutes before completing the inventory. Within the Guilt Cognition scale, only the Distress subscale was revised. Two items were removed because they also dealt with the frequency of distress (e.g., When I am reminded of the event, I have strong physical sensations such as sweating, tense muscles, dry mouth, etc.). The remaining four items were revised to reflect the type of distress expected from the killing paradigm. For example, “I feel sorrow or grief about the outcome” was changed to “I feel troubled and concerned when I think about what happened.” The hindsight, justification, and Wrongdoing subscales were completely unrevised. However, another four items comprising the Guilt Cognition scale, but not comprising any of the four subscales, were removed to reduce redundancy. The resulting scale consisted of 22 items rated on a 5-point scale (1 = not at all true, 5 = extremely true).

State Shame and Guilt Scale (SSGS)

We also utilized a measure of guilt not specific to traumatic experience. The SSGS (Marschall, Sanftner, & Tangney, 1994) consists of 15 brief phenomenological depictions of shame (e.g., I want to sink into the floor and disappear; I feel worthless, powerless), guilt (e.g., I feel remorse, regret; I feel bad about something I have done), and pride (e.g., I feel good about myself; I feel proud). Statements are rated on a 5-point scale based on how participants feel at the present moment (1 = not feeling this way at all; 5 = feeling this way very strongly).

Raffle ticket measure

Participants also completed a behavioral measure of guilt. Nelissen and Zeelenberg (2009) reasoned that although guilt has been documented to increase prosocial behavior in an effort to amend any pain caused by one’s actions (e.g., De Hooge, Zeelenberg, & Breugelmans, 2007; Nelissen, Dijker, & De Vries, 2007), not every situation allows for compensatory action because the harm caused is irreversible. In such cases, the researchers found that participants will attempt to alleviate any guilt through self-punishment. A raffle ticket measure was designed to measure this. To reward participants for completing the extermination, we provided the option of entering a raffle for $50. Participants were given a roll of raffle tickets and instructed to enter the raffle as many times as they liked. If participants felt guilty after killing bugs, we expected them to punish themselves for their actions by taking fewer raffle tickets, thus limiting their chances at the supposed $50 reward.

Procedure

Participants were run one at a time. Upon arriving at the laboratory, each participant was greeted in the hall by the experimenter who explained that another participant was expected to arrive. However, he explained, there was no time to wait any longer, and began the study without this second student. This was a ruse designed to set up the impending confederate interaction. The participant was then instructed to complete a study examining how people in various roles deal with different animals, specifically, how exterminators deal with insects.

The participant was led into an extermination cubicle containing the extermination machine and 10 woodlice (the bugs measured approximately 1 cm in length) placed in individual plastic cups. The machine consisted of a coffee grinder with PVC tubing attached to the side of the grinder assembly, giving the illusion of a chute leading directly to the grinder chamber (Martens et al., 2007). The experimenter explained that the grinding apparatus would be used for the extermination because the use of poison sprays and chemicals was not permitted within the building. He then mimed the extermination procedure with the following instructions: “dump each bug into the chute, and after all the bugs are deposited within, press and hold the activation button for at least three seconds.”

The participant returned to his or her original cubicle, and, to aid him or her in assuming the exterminator role, was instructed to read neutral information describing the insects (e.g., scientific classification, identifying features, and habitat). After reading this information, the participant returned to the extermination cubicle and completed the extermination in private. In actuality, the PVC tubing ran directly into the side of the grinder, and when dumped into the chute, the bugs merely rested inside the tubing against the outside edge of the grinding apparatus. The grinder itself was filled with bits of Styrofoam. Thus, although no bugs were killed, when the activation button was depressed, participants heard the blades engage and heard what they believed to be the grinding of the deposited insects.

The experimenter reentered the cubicle once the participant was finished, acting frustrated. The experimenter explained that the second participant had arrived late and the extermination machine needed to be reset so this participant could also complete the task. The experimenter then asked the participant to explain the study to this new participant and collect her informed consent, thus allowing him time to reset the machine. All participants agreed to this request. The participant was given a copy of the introduction script and a blank consent form. The late-arriving participant was a confederate instructed to act in one of two ways. In the social validation condition, the confederate acted interested and readily agreed to participate. When completing the consent form, she wrote the following note at the bottom of the form: “Sounds cool! I get to kill bugs ☺.”³ After finishing her consent form, she got up from her desk, walked toward the participant, submitted the completed form, and stated: “Here you go. I’m looking forward to it.” She then went back into the cubicle and took her seat. In the social invalidation condition, the confederate acted bothered by the prospect of exterminating bugs, and refused to participate. The confederate wrote the following note at the bottom of the consent form: “I refuse to do this study because it violates my beliefs about right and wrong.” After finishing, she got up from her desk, walked to the participant, submitted the form, and stated “I don’t think I can do this study because it goes against what I believe in. I’m just gonna go.” She then immediately exited the lab.

After the interaction had transpired, the participant returned the consent form to the experimenter. In the social validation condition, the experimenter thanked the participant and continued with the study. In the social invalidation condition, the experimenter reassured the participant that the confederate’s refusal was not his or her fault, stating that many other students had refused to complete the study. This was added to the procedure to prevent participants from attributing the refusal to a single outlying individual and reduce chances that any subsequently measured distress resulted from feelings of failing the experimenter. In the control condition, there was no participant–confederate interaction, and participants moved directly to the final task after completing the extermination.

The participant returned to his or her original cubicle to complete a questionnaire packet, which contained, in order (a) the revised TRGI, (b) a filler scale, (c) the SSGS, (d) a second filler scale, and (e) the reward raffle.⁴ Upon completion, the participant was probed for suspicion and thoroughly debriefed. The participant was informed about the confederate, who then reentered the lab, introduced herself to the participant, and proceeded to debrief. This allowed the experimenter to remain blind to our hypotheses, and limited possible experimenter bias.

Results

The main hypothesis was that social invalidation would lead to greater distress than social validation. Mean scores were calculated for each of the subscales of the TRGI (i.e., Guilt Cognitions, Distress, Wrongdoing, Justification, and Hindsight) and SSGS (i.e., Guilt, Shame, and Pride), and the number of raffle tickets taken was summed. A distress composite score was calculated by reverse-coding the raffle and pride scores, calculating standardized scores (i.e., z scores) for each variable, and taking the mean of all items. When subjected to a one-way ANOVA, significant differences were found between the three validation conditions, F(2, 37) = 3.30, p < .01, η² = .22. Three pairwise contrasts were conducted to probe the pattern of results—one comparing invalidation and validation conditions, one comparing invalidation with control, and one comparing validation with control. Supporting our main prediction, validation (M = −.45, SD = .60) led to less distress than that experienced in the invalidation condition (M = .27, SD = .60); t(37) = 2.85, p < .01. Validation also led to less distress than the control condition (M = .25, SD = .74); t(37) = −2.81, p < .01. The invalidation and control conditions did not differ on distress; t(37) = .09, p = .93 (see Figure 1).

Figure 1.

Standardized distress scores for each validation condition (Study 1).

We then analyzed each distress variable independently to determine which facets of distress were most affected by the manipulations. Means and statistics are reported for all variables in Table 1. Raffle ticket scores failed normality assumptions with a skewness of 2.32 (SE = .37) and kurtosis of 6.39 (SE = .73). A constant of one was added to each person’s ticket total, and that number was logarithmically transformed (skewness = .37, SE = .37; kurtosis = −.27, SE = .73). All variables were subjected to a one-way ANOVA. Significant differences emerged among four of the variables (i.e., reward raffle, wrongdoing, hindsight, and guilt cognitions), marginal effects (p < .10) were found for three of the variables (i.e., guilt, shame, and distress), and no differences were found for two variables (i.e., lack of justification and pride). Pairwise contrasts revealed that four of the variables (i.e., raffle, wrongdoing, hindsight, and guilt cognitions) followed the same pattern as the distress composite. The remaining three variables (i.e., distress, guilt, and shame) only differed when comparing validation and control conditions.

Table 1.

Contrasts of Invalidation, Validation, and Control Conditions on All Distress Measures (Study 1).

	Condition means			Test statistics
	Invalid	Valid	Control	ANOVA	Invalid vs. valid	Valid vs. control	Invalid vs. control
Measure	n = 13	n = 13	n = 14	F(2, 37)	t(37)	t(37)	t(37)
Raffle	2.15 (1.63)	7.54 (7.29)	3.21 (3.26)	3.47**	−2.50**	2.00**	−0.54
Guilt cognition	2.94 (0.70)	2.19 (0.58)	2.81 (0.70)	4.72**	2.87***	−2.41**	0.51
Hindsight	3.04 (0.77)	2.10 (0.62)	2.69 (0.72)	5.96***	3.41***	−2.19**	1.29
Distress	2.62 (1.07)	1.92 (1.14)	3.09 (1.42)	3.09*	1.44	−2.48**	−1.01
Wrongdoing	2.25 (0.83)	1.49 (0.59)	2.64 (1.06)	6.24***	2.25**	−3.49****	−1.20
Justification	3.63 (1.08)	3.19 (0.97)	3.43 (1.08)	0.58	1.08	−0.59	0.51
Shame	1.83 (0.82)	1.28 (0.52)	2.04 (1.13)	2.75*	1.62	−2.28**	−0.63
Guilt	2.42 (1.25)	1.55 (0.90)	2.49 (1.19)	2.81*	1.96*	−2.15**	−0.16
Pride	1.80 (0.80)	2.02 (0.85)	1.99 (0.95)	0.24	−0.63	0.09	−0.56

Note: Condition means for the reward raffle are presented in untransformed form, but the test statistics are reported for logarithmic transformed data. Standard deviations are presented in parentheses.

p < .10. **p < .05. ***p < .01. ****p < .001.

Discussion

The results of Study 1 demonstrated that social invalidation led to higher levels of guilt and distress than validation, as measured by the distress composite score. This suggests that the degree of social validation/invalidation for the act of killing influences the amount of distress experienced. Social validation, relative to invalidation, presumably helps people view killing behavior as morally acceptable, as measured by the low levels of guilt, and thereby eases distress. Contrary to our predictions, however, we found that invalidation did not significantly increase distress relative to the control condition. The small sample size, however, likely makes the more fine-grained comparisons with the control condition less reliable than the key overall invalidation versus validation comparison. Thus, we viewed the control condition comparisons in Study 1 with caution and sought to double our sample size in Study 2 to more reliably analyze the pattern of means. Another limitation of Study 1 is the potential for experimenter bias. Although the experimenter was blind to the hypotheses, he was not blind to conditions. Study 2 also addressed this limitation.

Study 2

Study 2 conceptually replicated Study 1 with a different method of manipulating social validation and invalidation. Participants were asked to record their participation in a “record book.” In the control condition, this record book contained no information about the progress of previous participants. In the remaining conditions, the record book gave the perception that all previous participants either refused to complete the extermination (invalidation) or willingly completed the extermination (validation). This procedure eliminated the need for a participant–confederate interaction—thus removing a potential source of error—and allowed us to keep the experimenter blind to conditions—thus removing the potential for experimenter bias. After completing the same distress measures used in Study 1, we added a measure of participants’ perception of social validation (i.e., that their peers—the supposed prior participants—completed the extermination task and viewed killing the bugs as morally acceptable). This additional measure allowed us to assess the effectiveness of our manipulations and to obtain some indication as to participants’ assumptions about social consensus in the absence of any clear social information (i.e., in the control condition). Moreover, this measure allowed us to examine whether cognitions regarding social validation/invalidation are indeed the driving force behind the experience of distress. We had several predictions. First, we expected to replicate the main finding of Study 1 that social invalidation will lead to greater levels of distress than social validation. Second, our reasoning prior to conducting Study 1 led us to expect levels of distress in the control condition to be similar to those expressed in the validation condition. Third, we expected the influence of the social validation/invalidation manipulations on levels of distress to depend on participants’ cognitions of social validation. Specifically, we predicted that perceptions of validation/invalidation would mediate the effect of the validation manipulation on levels of distress.

Method

Participants

Participants were 133 undergraduate students participating for partial course credit. Data from 11 participants were excluded from analyses. The data from the remaining 122 participants (100 female, 21 male, 1 unreported) were included in all analyses (M _age = 19.00, SD _age = 2.90).

Materials and Procedure

The procedure only differed to that reported in Study 1 after the bug-extermination task had transpired. After exterminating, participants filled out a participation record book in which they indicated if they completed the extermination (yes/no), how many bugs they killed, and provided their initials. In the validation and invalidation conditions, each participant had to flip through three completed pages in the record book to find the first free spot to record their progress. Social validation and invalidation were manipulated through the actions of 34 bogus previous participants. In the social validation condition, all previous participants completed the extermination task and indicated killing all 10 bugs. In the social invalidation condition, all previous participants indicated killing zero bugs or that they withdrew from the study. In the control condition, participants recorded their progress in the same manner, but were not provided information about previous participants.

Participants then completed the same distress measures used in Study 1 before answering two validation perception items on a 7-point scale (1 = strongly disagree, 4 = neither agree nor disagree, 7 = strongly agree). The items read as follows: “Recording my progress in the folder led me to believe that most other students completed the extermination task” and “The students that participated before me must have thought killing bugs was wrong” (reverse coded). Once the study was over, each participant was probed for suspicion and thoroughly debriefed.

Results

The Effect of Condition on Perceptions of Validation

To examine whether the validation/invalidation manipulations influenced perceptions of validation as intended, we averaged the two validation perception items to form a composite (r = .54, p < .001) and then subjected this composite to a one-way ANOVA, F(2, 119) = 113.13, p < .001, η² = .66. As intended, the validation condition (M = 5.44, SD = .96) led to greater perceptions of validation than the invalidation condition (M = 1.80, SD = 1.19); t(119) = −14.79, p < .001, and control (M = 4.18, SD = 1.15); t(119) = 5.11, p < .001, and invalidation decreased validation perceptions relative to control; t(119) = −9.74, p < .001.

The Effect of Condition on Distress

To examine the predicted effect that social invalidation would lead to greater distress than social validation, a distress composite was created in the same manner as Study 1 and subjected to a one-way ANOVA. The ANOVA revealed significant differences between the conditions, F(2, 119) = 14.12, p < .001, η² = .19 (see Figure 2). As predicted, pairwise contrasts revealed that invalidation (M = .44, SD = .70) led to significantly greater distress than validation (M = −.32, SD = .57); t(119) = 5.07, p < .001. Contrasts further revealed that validation did not reduce distress relative to the control (M = −.14, SD = .74); t(119) = −1.20, p = .23, but that invalidation increased distress relative to the control condition; t(119) = 3.90, p < .001. We then analyzed the individual distress variables independently to determine which facets of distress were most influenced by the manipulations. Raffle and shame scores failed normality assumptions with skewness of 3.64 (SE = .22) and 1.48 (SE = .22) and kurtosis of 18.89 (SE = .44) and 1.99 (SE = .44), respectively. After adding a constant of one to the raffle scores, and logarithmically transforming both variables, the resulting skewness was .47 (SE = .22) and .65 (SE = .22) and kurtosis was.36 (SE = .44) and –.53 (SE = .44), respectively. One-way ANOVAs followed by a series of pairwise contrasts revealed that five of the seven distress variables (i.e., excluding pride and justification) followed the same pattern as the overall distress composite (see Table 2).

Figure 2.

Standardized distress scores for each validation condition (Study 2).

Table 2.

Contrasts of Invalidation, Validation, and Control Conditions on All Distress Measures (Study 2).

	Condition means			Test statistics
	Invalid	Valid	Control	ANOVA	Invalid vs. valid	Valid vs. control	Invalid vs. control
Measure	n = 41	n = 40	n = 41	F(2, 119)	t(119)	t(119)	t(119)
Raffle	1.10 (1.58)	3.35 (3.96)	2.22 (1.82)	13.34****	−5.03****	1.54	−3.51****
Guilt cognition	3.34 (0.65)	2.65 (0.60)	2.78 (0.68)	13.02****	4.78****	−0.89	3.92****
Hindsight	3.64 (0.67)	2.77 (0.66)	3.12 (0.86)	14.47****	5.34****	−2.16**	3.20***
Distress	3.10 (1.14)	2.14 (1.00)	2.31 (1.20)	8.43****	3.83****	−0.66	3.19***
Wrongdoing	2.61 (0.91)	1.99 (0.74)	1.92 (0.77)	9.04****	3.46****	0.39	3.87****
Justification	3.71 (0.94)	3.34 (0.80)	3.24 (0.94)	3.18**	1.86*	0.53	2.40**
Shame	2.23 (1.13)	1.49 (0.58)	1.69 (0.80)	7.74****	3.79***	−1.00	2.81***
Guilt	2.95 (1.20)	1.98 (0.95)	2.23 (1.00)	9.19****	4.12****	−1.05	3.09***
Pride	1.88 (0.65)	2.39 (0.71)	2.12 (0.89)	4.44**	−2.98***	1.59	−1.40

Note: Condition means for raffle and shame are presented in untransformed form, but test statistics are reported for logarithmic transformed data. Standard deviations are presented in parentheses.

p < .10. **p < .05. ***p < .01. ****p < .001.

As can be seen in Figure 2, the level of distress reported in the control condition does not follow the same pattern reported in Study 1 (see Figure 1). However, the results conformed to our initial prediction that invalidation would be more likely to influence distress than validation. We theorized that several factors would minimize baseline distress from killing in the control condition, specifically, that the study involved a subhuman target, a mechanized killing procedure, and made use of euphemisms for killing. Given these factors, participants may also have assumed validation in the control condition (i.e., assumed that others in this study would have killed the bugs and viewed this killing as morally acceptable). If this theorizing is correct, the validation perception measure should be (a) relatively high in the control condition and (b) should be correlated with distress level in the control condition. First, as reported, the validation perceptions mean in the control condition was above the midpoint on the scale (M = 4.18). This suggests that many of the participants in the control condition believed that other students would kill bugs and would have no ethical problems doing so. Second, the within-cell correlation between validation perceptions and distress in the control condition was negative and highly significant; r = −.54, p < .001. Thus, the level of distress in the control condition indeed appears to be a function of how much participants perceive the bug-killing as socially validated.

Mediation Analysis

Our theorizing predicts that validation perceptions should mediate the relationship between validation/invalidation condition and distress. In establishing mediation, it is common to establish three pathways (see Figure 3). As previously reported, the condition variables significantly predicted the dependent variable (i.e., distress), thus establishing Step 1 of mediation. We also established the second step of mediation in demonstrating that the condition variables significantly predicted the mediator (i.e., perceptions of validation). In the final step, we needed to establish the indirect effect of our condition variables on the dependent variable, through the mediator.

Figure 3.

Mediation model of the indirect effect of condition (X1 and X2) on distress, through perceptions of social validation (Study 2).

This indirect effect was examined with the MEDIATE SPSS macro and procedure developed by Hayes and Preacher (2011). Because the condition variable was multicategorical, the predictor (independent variable) in the mediation model comprised two dummy-coded variables (X1 and X2 in Figure 3) created with the invalidation condition as the reference condition (i.e., invalidation = 0, 0; validation = 1, 0; control = 0, 1). This coding strategy allowed us to conduct an omnibus test of mediation, and specifically test if differences in validation perceptions mediated the distress differences found between the invalidation and validation conditions (X1) and the invalidation and control conditions (X2). The distress composite was entered as the dependent variable and the validation perceptions composite was entered as the mediator. The dummy-coded condition variables were then regressed onto distress while controlling for perceptions of validation. In support of mediation, when perceptions of validation were included in the model, these perceptions significantly predicted distress, t(118) = −5.66, p < .001, and the omnibus effect of condition on distress was reduced to nonsignificance, F(2,118) = .86, p = .43. More specifically, relative to the invalidation condition, the previously significant (p < .001) relationship between validation and distress was reduced to nonsignificance, t(118) = 1.17, p = .24, and the previously significant (p < .001) relationship between the control and distress was reduced to nonsignificance, t(118) = .51, p = .61.

The significance of the indirect effect was assessed by creating bootstrapped confidence intervals (CIs). One hundred twenty-two cases (i.e., the original sample size) were sampled with replacement from our original data file, the coefficients in the mediation model were estimated, and the indirect effects calculated. This process was repeated 5,000 times to empirically create a sampling distribution of indirect effects. We then calculated 95% CIs for this distribution. If “zero” is not contained within the upper and lower limits of the CIs, mediation is significant. Thus, we are confident that the effect of condition on distress is significantly mediated by perceptions of validation: omnibus CI = [–.26, –.12]. And more specifically, we are confident that relative to the high distress found in the invalidation condition, the low levels of distress found in the validation, CI = [–1.44, –.65], and control conditions, CI = [−.97, −.41], were mediated by perceptions of validation.^5,6

Discussion

The results of Study 2 replicated the main prediction from Study 1 in that participants reported significantly greater distress when killing was socially invalidated, than when it was validated. Distress reported in Study 2 only differed in regard to the control condition comparisons. In Study 1, validation significantly affected distress whereas invalidation did not, but in Study 2, invalidation affected distress whereas validation did not. Given our theoretical basis for expecting the Study 2 pattern and the much larger sample size of Study 2, the particular pattern of comparisons that emerged in Study 2 are likely more reliable than those in Study 1.

Our theoretical analysis consistent with the Study 2 pattern is further supported by internal analyses of the perceptions of validation questions. Participants in the control condition perceived killing to be relatively socially validated and reported low levels of distress. The mediation analyses revealed that the distress levels reported in each of our three conditions were mediated by perceptions of validation. In other words, the social invalidation manipulation caused participants to believe that other students refused to kill and viewed killing bugs as wrong. On the other hand, when participants were left to their own devices (i.e., control condition) or were subjected to the social validation manipulation, they believed other students willingly killed bugs and had no moral qualms in doing so. Recall that participants in the control condition had no access to social validation information. Perceived validation in this condition was not due to an experimental manipulation but was an assumption that participants made on their own. These different perceptions thus caused the different levels of distress in each of the respective conditions. In other words, when killing was perceived as socially validated, participants were successfully buffered against distress. It was only when participants did not believe that others also killed that killing was experienced as distressing.

General Discussion

The present findings demonstrate that social validation and invalidation for the act of killing are important determinants of resulting distress and guilt. Two studies demonstrated that when participants were induced to kill bugs, they experienced higher levels of distress and guilt when they perceived social invalidation for this behavior compared with when they perceived social validation. This effect was observed on several direct measures of distress such as trauma-related guilt and state shame and guilt, as well as an indirect, behavioral measure of guilt (i.e., self-punishment).⁷ Moreover, the mediation analyses in Study 2 enable us to draw the more direct conclusion that the more participants perceived social validation for their killing behavior, the lower their levels of distress and guilt. This was even the case when validation and invalidation were not directly induced (i.e., the control condition).

We draw on a long history of theorizing in the social sciences to explain these effects. This literature has viewed subjective reality as essentially a social construction maintained through a process of consensual validation (Becker, 1962/1971; Berger & Luckmann, 1966; Festinger, 1954; Kelley, 1967; Mead, 1934; Shibutani, 1961). In turn, one’s views of right and wrong should be strongly affected by social consensus. In the case of killing insects, learning that others refused to kill bugs should cast doubt on the belief that bugs are indeed worthy of destruction (“If bugs are bad, why didn’t other people kill them?”). In other words, killing bugs is now at odds with social reality, causing one to re-evaluate this behavior as inappropriate, leading to dissonance and distress. Likewise, learning that others willingly killed bugs creates the perception that killing is in line with social reality and thus acceptable, thereby reducing dissonance and distress.

We should note that although we referred to distress as being “high” under certain conditions, this is only within the context of our experimental paradigm. Across both studies, the reported guilt and distress in the “high distress” condition(s) hovered around or below the midpoint of the scales. These generally low levels of distress were likely due to several factors built into the bug-killing paradigm. As proposed in the introduction, moral disengagement strategies (Bandura, 1999) were present such as moral justification, dehumanization, and euphemistic labeling that should have enabled many participants to view the killing as somewhat benign. Moreover, participants likely engaged in dissonance reduction strategies of changing their attitudes about killing or about the victim, in an effort to justify their behavior (Festinger, 1957). Participants may also have reduced the dissonance and distress from killing by assuming some validation for their behavior, thus convincing themselves that the behavior is normal and appropriate. Indeed, the use of such a defense is broadly consistent with research on the false consensus effect, showing that individuals protect their self-image by overestimating the commonality of unsuccessful or undesirable behaviors (Marks & Miller, 1987). This possibility is supported by the negative correlation between social validation perceptions and distress that was observed for control participants in Study 2, as well as the mediation analyses.

Although it seems likely that people use a variety of defenses to deal with the distress of killing, we found high distress in the social invalidation conditions of both studies. This finding suggests that the presence of social invalidation may have constrained participants from engaging in some of these defenses, or at the very least, made these defenses less effective at reducing distress once enacted. As discussed previously, invalidation should increase the discrepancy between killing and the belief that killing is wrong, resulting in high distress and dissonance. In the face of this heightened discrepancy, the present research indicates that it may become more difficult to justify killing by assuming others would do the same (at least temporarily). The evidence does not speak directly to whether or not social invalidation also constrained participants from using one of the countless other dissonance reduction techniques, as they were not directly assessed. The “high” distress experienced in the invalidation condition, however, suggests that the presence of social invalidation may somewhat impair the effectiveness of a host of defensive techniques. Future research may wish to address this directly.

Although we found the predicted difference between the validation and invalidation conditions across both studies, some caution is needed when interpreting the results from the control conditions. In Study 1, distress in the control condition was more similar to that in the invalidation condition. In Study 2, distress in the control was more similar to the validation condition, supporting the initial prediction that invalidation would have more influence on distress under situations like the one created in the bug-killing paradigm. We consider the results of Study 2 to be more conclusive on two counts. First, the procedure for manipulating validation and invalidation utilized in Study 2 allowed for greater control of extraneous variables. Second, the sample size in Study 2 is much more robust. Indeed, the higher distress for control participants in Study 1 could have resulted from sampling error given the smaller sample size. We therefore have greater confidence that the lower levels of distress observed in the control condition of Study 2 more accurately reflects participants’ level of distress in the absence of social validation or invalidation.

Practical Implications

At the beginning of this manuscript, we posed the question, “Why do the majority of soldiers that kill at war return home in relatively good psychological health?” Based on the current findings, we would suggest that perceptions of social invalidation impair a soldier’s ability to morally disengage from the injurious actions undertaken during war, making the initial act of killing more distressing and increasing the potential for lingering trauma or PTSD. On the other hand, a soldier who killed but remained psychologically healthy would have been able to disengage, viewing killing the enemy as a legitimate means to a desirable goal, cognitions that may be driven by perceptions of social validation.

In the current studies, we provided participants with social information that was either validating or invalidating so as to isolate these variables and examine their influence. However, in real-world applications, we understand that the process may not be as simple. Many times, the distinction between socially validated and invalidated action is not as clear, as there will likely be some form of minority opinion supportive of a war effort, even if the vast majority oppose. Some may wonder why a soldier would not simply choose to ignore the majority opinion in favor of this minority opinion, as it provides the necessary dissonance reduction. We accept this as a possibility and view it as a route many soldiers may take, at least initially. In applying the tenets of dynamic social impact theory (Latane, 1996), we predict this route would be most prevalent among soldiers who surround themselves with individuals who support the war effort and perceive those who have the most influence over them (i.e., close friends and family, other soldiers) as supporting the war effort. This could allow soldiers to reside within a “cluster” that protects their psychological equanimity. Upon leaving the service and entering civilian life, however, it may become more difficult to maintain this cluster and the influence of majority opinion may reinstate the guilt and trauma that were previously kept at bay.

Interestingly, the evidence presented within suggests an inherent paradox within the nature of war and expressing antiwar sentiment. Public attempts to socially invalidate a war typically exist as protests and rallies where people join in thousands to synchronously shout their disdain for war (or by decrying war efforts through a video that goes viral on the Internet). Indeed, many times these protests eclipse the actual fighting itself and become iconic of war and the democratic spirit (e.g., a Vietnam War protester placing a flower in the barrel of a soldier’s gun). However, this social invalidation occurs, the intent is always the same: people protest the war effort, but unquestioningly support soldiers entrenched in battle. Indeed, the motivational impetus behind war protest is often concern for soldiers’ safety and well-being.

Whereas these protesters feel they are doing soldiers a service and protecting their livelihood, the present research suggests that such efforts may be damaging to the psychological health of the very servicemen they seek to protect. In attempting to invalidate a war to bring troops home, the public may effectively remove the ability for those very troops to justify their behavior, potentially placing soldiers at greater risk for long-lasting distress and trauma. Is the solution to this paradox to disallow protest and only publicize public acts of support in an effort to protect servicemen and women? We think not. These findings should caution the public about the potential downside of protest, and this caution may hold greater weight in the current era where social media not only allows anyone with a computer the opportunity to invalidate a war effort but makes this information highly accessible to soldiers. But more importantly, this research should inform those who make decisions about war regarding the psychological consequences of an unfounded and publicly unpopular war. Wars last several years, sometimes decades. As the time at war increases, social validation of the war may wane, especially as more and more lives are lost. The current research suggests that this waning social validation could have a profound effect on the psychological health of soldiers.

Footnotes

Acknowledgements

We thank Jordan Ashworth, Branden Ayotte, Shannon Fischer, Alex Pavlick, and Ethan Sawyer for their help in conducting the research.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Aquino

Reed

Thau

Freeman

(2007). A grotesque and dark beauty: How moral identity and mechanisms of moral disengagement influence cognitive and emotional reactions to war. Journal of Experimental Social Psychology, 43, 385-392. doi:10.1016/j.jesp.2006.05.013

Archer

Gartner

(1992). Peacetime casualties: The effects of war on the violent behavior of noncombatants. In Aronson

(Ed.), Readings about the social animal (pp. 327-338). New York, NY: W. H. Freeman.

Bandura

(1999). Moral disengagement in the perpetration of inhumanities. Personality and Social Psychology Review, 3, 193-209. doi:10.1207/s15327957pspr0303_3

Baum

(2004, July 12-19). The price of valor. The New Yorker. Available from http://www.newyorker.com

Becker

(1971). The birth and death of meaning (2nd ed.). New York, NY: Free Press. (Original work published 1962)

Berger

P. L.

Luckmann

(1966). The social construction of reality: A treatise in the sociology of knowledge. New York, NY: Anchor.

Bolton

E. E.

Litz

B. T.

Glenn

D. M.

Orsillo

Roemer

(2002). The impact of homecoming reception on the adaptation of peacekeepers following deployment. Military Psychology, 14, 241-251. doi:10.1207/S15327876MP1403_4

Brailey

Vasterling

J. J.

Proctor

S. P.

Constans

J. I.

Friedman

M. J.

(2007). PTSD symptoms, life events, and unit cohesion in U.S. soldiers: Baseline findings from the neurocognition deployment health study. Journal of Traumatic Stress, 20, 495-503. doi:10.1002/jts.20234

Castano

(2008). On the perils of glorifying the in-group: Intergroup violence, in-group glorification, and moral disengagement. Social & Personality Psychology Compass, 2, 154-170. doi:0.1111/j.1751-9004.2007.00040.x

10.

Castano

Giner-Sorolla

(2006). Not quite human: Infrahumanization in response to collective responsibility for intergroup killing. Journal of Personality and Social Psychology, 90, 804-818. doi:10.1037/0022-3514.90.5.804

11.

Charuvastra

Cloitre

(2008). Social bonds and posttraumatic stress disorder. Annual Review of Psychology, 59, 301-328. doi:10.1146/annurev.psych.58.110405.085650

12.

Crutchfield

R. S.

(1955). Conformity and character. American Psychologist, 10, 195-198. doi:10.1037/h0040237

13.

De Hooge

I. E.

Zeelenberg

Breugelmans

S. M.

(2007). Moral sentiments and cooperation: Differential influences of shame and guilt. Cognition & Emotion, 21, 1025-1042. doi:10.1080/02699930600980874

14.

Deutsch

Gerard

H. B.

(1955). A study of normative and informational social influences upon individual judgment. Journal of Abnormal and Social Psychology, 51, 629-636. doi:10.1037/h0046408

15.

Fay

G. R.

(2004). AR 15-6 investigation of the Abu Ghraib detention facility and 205th military intelligence brigade. Retrieved from http://fl1.findlaw.com/news.findlaw.com/hdocs/docs/dod/fay82504rpt.pdf

16.

Festinger

(1954). A theory of social comparison processes. Human Relationships, 1, 117-140.

17.

Festinger

(1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press.

18.

Hayes

A. F.

Preacher

K. J.

(2011). Indirect and direct effects of a multicategorical causal agent in statistical mediation analysis. Manuscript submitted for publication.

19.

Kelley

H. H.

(1967). Attribution theory in social psychology. In Levine

(Ed.), Nebraska symposium on motivation (Vol. 15, pp. 192-238). Lincoln, NE: University of Nebraska Press.

20.

King

L. A.

King

D. W.

Fairbank

J. A.

Keane

T. M.

Adams

G. A.

(1998). Resilience-recovery factors in posttraumatic stress disorder among female and male Vietnam veterans: Hardiness, post-war social support, and additional stressful life events. Journal of Personality and Social Psychology, 74, 420-434. Retrieved from http://www.apa.org/pubs/journals/psp/index.aspx

21.

Koenen

K. C.

Stellman

J. M.

Stellman

S. D.

Sommer

J. F.

(2003). Risk factors for course of posttraumatic stress disorder among Vietnam veterans: A 14-year follow-up of American Legionnaires. Journal of Consulting and Clinical Psychology, 71, 980-986. doi:10.1037/0022-006X.71.6.980

22.

Kubany

E. S.

Abueg

F. R.

Owens

J. A.

Brennan

J. M.

Kaplan

A. S.

Watson

S. B.

(1995). Initial examination of a multidimensional model of trauma-related guilt: Application to combat and battered women. Journal of Psychopathology and Behavioral Assessments, 17, 353-376. doi:10.1007/BF02229056

23.

Kubany

E. S.

Haynes

S. N.

Abueg

F. R.

Manke

F. P.

Brennan

J. M.

Stahura

(1996). Development and validation of the Trauma-Related Guilt Inventory (TRGI). Psychological Assessment, 8, 428-444. doi:10.1037/1040-3590.8.4.428

24.

Latane

(1996). Dynamic social impact: The creation of culture by communication. Journal of Communication, 46, 13-25. doi:10.1111/j.1460-2466.1996.tb01501.x

25.

Litz

B. T.

Stein

Delaney

Lebowitz

Nash

W. P.

Silva

Maguen

(2009). Moral injury and moral repair in war veterans: A preliminary model and intervention strategy. Clinical Psychology Review, 29, 695-706. doi:10.1016/j.cpr.2009.07.003

26.

MacNair

R. M.

(2002). Perpetration-induced traumatic stress in combat veterans. Peace and Conflict: Journal of Peace Psychology, 8, 63-72. doi:10.1207/S15327949PAC0801_6

27.

Maguen

Metzler

T. J.

Litz

B. T.

Seal

K. H.

Knight

S. J.

Marmar

C. R.

(2010). The impact of reported direct and indirect killing on mental health symptoms in Iraq war veterans. Journal of Traumatic Stress, 23, 86-90. doi:10.1002/jts.20434

28.

Maguen

Vogt

D. S.

King

L. A.

King

D. W.

Litz

B. T.

Knight

S. J.

Marmar

C. R.

(2011). The impact of killing on mental health symptoms in Gulf War veterans. Psychological Trauma: Theory, Research, Practice, and Policy, 3, 21-26. doi:10.1037/a0019897

29.

Maoz

McCauley

(2008). Threat, dehumanization, and support for retaliatory aggressive policies in asymmetric conflict. Journal of Conflict Resolution, 52, 93-116. doi:10.1177/0022002707308597

30.

Marks

Miller

(1987). Ten years of research on the false consensus effect: An empirical and theoretical review. Psychological Bulletin, 102, 72-90. doi:10.1037/0033-2909.102.1.72

31.

Marschall

Sanftner

Tangney

J. P.

(1994). The state shame and guilt scale. Fairfax, VA: George Mason University.

32.

Martens

Kosloff

Greenberg

Landau

M. J.

Schmader

(2007). Killing begets killing: Evidence from a bug-killing paradigm that initial killing fuels subsequent killing. Personality and Social Psychology Bulletin, 33, 1251-1264. doi:10.1177/0146167207303020

33.

Martens

Kosloff

Jackson

L. E.

(2010). Evidence that initial obedient killing fuels subsequent volitional killing beyond effects of practice. Social Psychology & Personality Science, 1, 268-273. doi:10.1177/1948550609359813

34.

McAlister

A. L.

Bandura

Owen

S. V.

(2006). Mechanisms of moral disengagement in support of military force: The impact of Sept. 11. Journal of Social & Clinical Psychology, 25, 141-165. doi:10.1521/jscp.2006.25.2.141

35.

Mead

G. H.

(1934). Mind, self, and society. Chicago, IL: University of Chicago Press.

36.

Nelissen

R. M. A.

Dijker

A. J.

De Vries

N. K.

(2007). How to turn a hawk into a dove and vice versa: Interactions between emotions and goals in a give-some dilemma game. Journal of Experimental Social Psychology, 43, 280-286. doi:10.1016/ j.jesp.2006.01.009

37.

Nelissen

R. M. A.

Zeelenberg

(2009). When guilt evokes self-punishment: Evidence for the existence of a dobby effect. Emotion, 9, 118-122. doi:10.1037/a0014540

38.

Schnurr

P. P.

Lunney

C. A.

Sengupta

(2004). Risk factors for the development versus maintenance of posttraumatic stress disorder. Journal of Traumatic Stress, 17, 85-95. doi:10.1023/B:JOTS.0000022614.21794.f4

39.

Shibutani

(1961). Society and personality: An interactionist approach to social psychology. Englewood Cliffs, NJ: Prentice Hall.

40.

Solomon

Waysman

Mikulincer

(1990). Family functioning, perceived societal support, and combat-related psychopathology: The moderating role of loneliness. Journal of Social & Clinical Psychology, 9, 456-472. doi:10.1521/jscp.1990.9.4.456

41.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063-1070. doi:10.1037/0022-3514.54.6.1063

42.

Weinberg

G. L.

(1995). Germany, Hitler, and World War II: Essays in modern German and world history. Cambridge, UK: Cambridge University Press.

43.

Zimbardo

(2007). The Lucifer effect: Understanding how good people turn evil. New York, NY: Random House.