Long-term post traumatic growth after moral- and victim-traumas among Norwegian UN military peacekeepers: the impact of emotional distress and leadership

Tedeschi and Calhoun (2004) identified five major dimensions of PTG development, which include enhanced personal strength, new possibilities, relating to others, appreciation of life and spiritual change. According to the tenets of PTG theory, changes in these areas occur because of the cognitive processing and rumination that takes place in the aftermath of trauma (Calhoun and Tedeschi 2013), often triggered and motivated by emotional dysregulation and emotional pain (Maitlis 2020). In the present study, we build on previous studies introducing a distinction between moral and victim traumas (Litz et al. 2009). According to this theory, guilt and shame are the primary emotional responses to moral traumas, whereas fear and anxiety may be more pronounced after victim trauma. Hence, the content of emotional distress after a traumatic experience, which can be seen as an antecedent of a transformational sense-making process and subsequent PTG, may differ. More so, we suggest that a moral trauma may activate more intense emotional distress and more strain on the self-concept, compared with a victim trauma, which in turn may activate a more thorough (and challenging) cognitive reorientation and sense-making process. We base this on two somewhat overlapping theoretical perspectives. First, several studies show that perceptions of moral transgressions are generally evaluated as more severe violations, activating stronger emotional and cognitive responses, compared with other transgressions within the social domain (Rozin et al. 1997; Baumeister et al. 2001; Folger et al. 2005). Hence, a sense of moral failure may propel more intense emotional distress and disruptive cognitive processing and restructuring compared with a victim experience. Second, according to Aquino and Reed (2002), morals represent a vital domain of social identity and an individual’s self-concept. This implies that moral behaviour, a positive view of one’s own moral character, is closely related to a positive self-image and, conversely, that moral failures may represent a particularly disturbing and emotionally distressing threat to a positive self-view. Thus, moral violations may in a particular way challenge a person on an existential level, possibly accommodated by a state of self-condemnation, which in turn may activate a particularly profound sense-making process – and subsequently PTG.

Several studies have shown that experiences of traumatic events may cause strong emotional distress (Tadeschi et al. 2018), which in turn is related to PTG. Emotional distress may challenge core beliefs and assumptions in a way that may stimulate PTG by facilitating continued efforts to make new meaning of the experience and move forward (Lowe et al. 2013; Maitlis 2020). This relationship between emotional distress and growth is supported by findings showing that higher levels of PTSD, encompassing emotional strain, is related to higher levels of PTG (Lowe et al. 2013; Maitlis 2020). Notably, some studies find this relationship to work together also over time, suggesting that PTG may not only be stimulated but also maintained by PTSD. Hence, emotional distress may trigger continued efforts to make new meanings and cognitive restructuring of the self, in line with PTG (Lowe et al. 2013; Maitlis 2020).

A trauma may challenge fundamental understandings of ourselves and the world and subsequently our sense of control and emotional security (Tedeschi et al. 2018; Maitlis 2020): A state accompanied by intrusive and invasive thoughts and memories that are difficult to control. Thus, to process traumatic experiences into a generative growth cycle, an important challenge is to manage distress and regulate severe emotions like anxiety, sadness, anger or guilt (Gross 1998; Van der Kolk 2014). Social support has been identified as a key enabler in this regulation process (Maitlis 2020). A relationship between social support and PTG is found in a range of work settings, including the military (e.g. Mark et al. 2018), and social support is suggested to stimulate both the managing of emotional distress and the process of sense-making (Maitlis 2020).

In the present study, we suggest that leaders’, given their influence on their team, may provide important social support (Bass and Bass 2009). Particularly, in the earliest period after a trauma, where emotional support is found most valuable (Schroevers et al. 2010). However, how leaders lead may play a vital role in how effectively they provide social support.

On the one side, we suggest that leaders that display authentic ethical leadership, seen as behaviour focussed on stimulating a goal-oriented cooperation through high moral awareness and internalised moral values (Olsen and Espevik 2017), are better providers of social support, given that they are perceived as honest (Kernis 2003) with high moral standards and values, which is found to stimulate trust and cooperation (May et al. 2003; Walumbwa et al. 2008; Olsen and Espevik 2017). Hence, authentic ethical leaders are likely to follow a moral imperative to care for a traumatised follower and engage in an emotionally supporting and non-blaming manner which may reduce emotional distress (Kolokotroni et al. 2014). Given the moral orientation, it is also likely that these leaders will commit to the process, show patience and endure the process over time (Olsen and Espevik 2017). More so, followers may find it easier to open about vulnerabilities to a leader perceived as empathic and fair. This may in turn stimulate an effort to actively confront and manage traumatic memories, seen instrumental in transforming distress into PTG (Sattler et al. 2014). Such openness may be particularly relevant after moral trauma – which otherwise may be suppressed and covered from others and thus left unmanaged (Litz 2009).

On the other hand, LF leadership is seen as a passive and destructive form of leadership, characterised by leaders who steer away, avoid making decisions, avoid getting involved and are absent when needed (Skogstad et al. 2007). Notably, this form of leadership is found highly prevalent in organisational settings (Aasland et al. 2010). After a traumatic experience, LF leadership would typically implicate a high degree of avoidance and lack of social support. The leader will not utilise his/her impact to support the process of managing emotional distress, disregard the severity of the situation, not activate peer support and may leave the soldier in an enduring flooded state. Given a relationship between LF leadership and poor leader-member relational quality (Buch et al. 2015) – potential interaction with traumatised personnel will most likely be superficial and of low support:

This cross-sectional survey involved Norwegian military personnel from the United Nations Interim Force in Lebanon (UNIFIL) serving in 6-months contingent between 1978 and 1998. Their mission was to oversee an Israeli withdrawal from southern Lebanon and maintain peace and security in the area – under restrictive ‘self-defence’ rules of engagement. The area was frequently a war zone, and the soldiers encountered numerous life-threatening situations, and 21 of them lost their life during the operation. The survey was administered by the Norwegian Armed Forces Joint Medical Services between September 2014 and April 2015 (Norwegian Armed Forces Joint Medical Services 2016). The main objective was to assess the mental health of these veterans, on average 27 years after completion of their service, by mapping variables like general health, mental health, addictions and quality of life. It was sent to all 20,678 veterans who had served in the mission by mail or electronically via internet. The response rate was 56.3%, with 11,633 veterans responding to the survey. A total of 913 veterans answered that they did not want to participate, and 115 answers were so incomplete that they could not be included. A total of 97.1% of these are men. The average age of survey participants was 53 years: 75 of them were between 30 years and 39 years (0.7%), 3,054 were between 40 years and 49 years (28.8%), 5,027 were between 50 years and 59 years (47. 4%), 1,775 were between 60 years and 69 years (16.7%) and 674 were older than 70 years (6.4%). Most of the sample (56.1%) had participated in a single contingent, while 31% and 12.9% of them had participated in two and three contingents, respectively. Participation in the UNIFIL contingents was evenly distributed between the years 1978–1998 in our sample, with the highest percentage being the year 1989 (6.1%) and the lowest being 1997 (2.4%). For the purpose of our investigation (the relationship between trauma and PTG), we created a subsample by extracting those who had either experienced a victim trauma (n = 5,686) or a moral trauma (n = 1,967) from the total sample. This subsample was further reduced to 4,985 and 1,821 veterans who experienced victim trauma and moral trauma, respectively, due to missing data from those excluded. For our analyses, we collected (1) self-assessment of present PTG (related to trauma experiences from their service in Lebanon), (2) experiences of traumatic events during their service (3) emotional distress reactions to these traumas (as they recalled it) and (4) leadership behaviour from direct leaders (as they remembered it). All participants provided informed consent, and the study was approved by the Norwegian Armed Forces Health Registry.

We employed a multiple regression analysis to test our first hypothesis and to examine the contributions of victim and moral trauma on PTG. The number of total contingents and time since first contingent were included as covariates in the analysis and were entered in the first step along with victim trauma. Moral trauma was subsequently entered in the second step.

Two different mediation models were computed to test the relationships proposed in our second hypothesis. In the first model, emotional distress in reaction to victim trauma was evaluated as a mediating variable in the relationship between victim trauma and PTG. In the second model, emotional distress in reaction to moral trauma was evaluated as a mediating variable in the relationship between moral trauma and PTG. Inferential judgements about the indirect effects are based on bootstrap confidence intervals (CIs). Bootstrap CIs can be considered as empirically derived sampling distributions for the indirect effects and are often preferred over the normal theory approach because no assumptions are made about normality in the sampling distribution of the indirect effect (Hayes 2013). From each resample, an indirect effect is computed, and then, an empirical sampling distribution is generated. From this distribution, a CI is computed that can be inspected to determine if zero is within the interval. There are several ways to compute bootstrap CIs, such as bias-corrected bootstrap intervals and percentile bootstrap intervals. Hayes and Scharkow (2013) have recommended using the former if statistical power is the major concern but the latter if type 1 error rate is the major concern. Given the substantial sample sizes for our different analyses, we therefore elected to compute percentile intervals. In our analyses, 5,000 bootstrap resamples were used to estimate 95% CIs.

Our third hypothesis proposes that the indirect effects of moral and victim trauma are conditional on levels of authentic and LF leadership. To test these conditional effects, we combined mediation and moderation in what is commonly referred to as second stage moderation (Edwards and Lambert 2007). In second stage moderation, the effect of the mediator M on the outcome Y is allowed to vary as a function of the moderator W, while the X → M relationship is fixed to be unmoderated. We performed two different second stage moderation models: one for victim trauma and one for moral trauma. These are consequently based on two different subsamples: all reporting victim trauma (n = 4,985) and all reporting moral traumas (n = 1,821). Product terms representing the interactions between emotional distress and LF leadership and between emotional distress and authentic leadership were included simultaneously in the two models.

Inferential judgements about the conditional indirect effects are based on computation of the index of moderated mediation described by Hayes (2015). A second stage moderation model with a single moderator W can be described by two equations: 1 M=iM+aX+eM \[M={{i}_{M}}+aX+{{e}_{M}}\] 2 Y=iY+c′X+b1M+b2W+b3MW+eY. \[Y={{i}_{Y}}+{c}'X+{{b}_{1}}M+{{b}_{2}}W+{{b}_{3}}MW+{{e}_{Y}}.\]

The indirect effect (referred to as ω below) of X on Y through M in this model is the product of the effect of X on Y from Eq. (1) and the conditional effect of M on Y from Eq. (2): 3 ω=a(b1+b3W), \[\text{ }\!\!\omega\!\!\text{ =a}\left( {{b}_{1}}+{{b}_{3}}W \right),\] which can be expressed in the form of an intercept and a slope for the indirect effect 4 ω=ab1+ab3W. \[\text{ }\!\!\omega\!\!\text{ }=a{{b}_{1}}+a{{b}_{3}}\text{W}\text{.}\]

The slope for W in Eq. 4, ab₃, quantifies the effect of W on the indirect effect of X on Y through M and is referred to as the index of moderated mediation by Hayes (2015).

We computed indices of moderated mediation for the indirect effect of victim trauma on PTG conditional on LF and authentic leadership, as well as for the indirect effect of moral trauma on PTG conditional on LF and authentic leadership. Like the unconditional mediation analyses tested in hypothesis 2, bootstrap CIs were used to make inferential judgements about the moderation of the indirect effects.

All analyses were conducted using Stata 17.0 (StataCorp. 2021.) and the regression and SEM packages.

Of the total sample, 5,686 reported having experienced at least one victim trauma incident and 1,967 reported having experienced at least one moral trauma incident. The proportions of participants experiencing each incident are presented in Table X. The subsequent mediational analyses are based on these two samples, but due to missing information on other variables, the samples were reduced to n = 4,985 for the analyses involving victim trauma and n = 1,821 for the analyses involving moral trauma.

Results from the tests of the indirect effects of moral and victim trauma are presented in Figure 2. Our first mediation model showed that victim trauma had a positive and statistically significant relationship with emotional distress (a-path = 0.24, 95% CI: 0.22; 0.26). Emotional distress, in turn, had a positive and statistically significant relationship with PTG (b-path = 0.08, 95% CI: 0.05; 0.10). The indirect effect of victim trauma on PTG via emotional distress is the product of the a and b paths and was ab = 0.02, with a 95% bootstrap CI of 0.13–0.03. Since the CI does not contain zero within its boundaries, we can reject the null hypothesis and be confident that the indirect effect is in fact different from zero. The remaining direct effect of victim trauma on PTG (c′-path) was still statistically significant (p < 0.001).

Fig. 2:

Our second mediation model showed that moral trauma also had a positive and statistically significant relationship with emotional distress (a-path = 0.29, 95% CI: 0.26; 0.33) and that emotional distress had a positive and statistically significant relationship with PTG (b-path = 0.05, 95% CI: 0.01; 0.09). The indirect effect of moral trauma on PTG was ab = 0.01, with a 95% bootstrap CI of 0.002–0.03.

The mediation models described above was next repeated with interaction terms between emotional distress and LF and authentic leadership added. Results from the first second stage moderation model involving victim trauma did not provide evidence that the indirect effect of victim trauma on PTG through emotional distress was a function of leadership. The index of moderated mediation for authentic leadership was 0.002, with a 95% bootstrap CI including zero (−0.003; 0.008), and for LF leadership was −0.002, with a 95% bootstrap CI also including zero (−0.009; 0.003). Likewise, the results from the model involving moral trauma did not provide any evidence for conditional indirect effects: The index of moderated mediation was 0.01 (95% bootstrap CI: −0.003; 0.022) for authentic leadership and −0.003 (95% bootstrap CI: −0.015; 0.009) for LF leadership.

In line with H2, we found the direct effect between trauma and PTG is partially mediated by emotional distress. This is in keeping with previous studies pointing at dysregulated emotions as a prerequisite for a PTG cycle (Maitlis 2020) – initiating a process of sense-making and challenging of fundamental assumptions about the world and oneself as a human being. This suggests that traumatic events by themselves are not accurate indicators of a PTG cycle. The emotional reaction to the experience is also an important factor in this process. An upside to this, seen from an applied perspective, may be hope in a difficult situation of trauma and emotional imbalance. Caregivers, leaders and family members, as well as those struggling, might find some comfort in the fact that a situation like this may represent an opportunity for growth, if managed properly, and not only a pathway to mental disorders and a ruined life. From a clinical perspective, this suggests that emotional imbalance should be appreciated as a significant resource in recovery processes, working through sense-making processes.

An important part of this study was the inclusion of leadership as a potential resource in the PTG cycle (H3) – due to an expected impact of leaders as caregivers and sources of social support. The lack of evidence of this relationship in the present study (no moderation of the mediation) may have several explanations. First, as a source of social support, it might have been more relevant to measure the contributions from peers and colleagues, given previous studies showing that soldiers talk and seek support mostly from these sources during challenging experiences and less from professional health workers and possibly leaders (Greenberg et al. 2003). We could further speculate that fear of repatriation from the mission, due to mental problems, may have obstructed an open and frequent interaction between traumatised soldiers and their leaders – reducing a potential effect of leadership as a source of support. Second, it is possible that other forms of leadership, distinct from those embedded in the present study, might have been better strategies aiding a PTG process. In the present study, only the moral dimension of authentic leadership and LF was included (Walumbwa et al. 2008). In future studies, a better model should include all four dimensions of authentic leadership. For example, transparency, seen as the leaders’ ability to share genuine emotions and thoughts with followers, might reduce fear based on not knowing what the leader thinks and, thus, stimulate a more open interaction between the follower and leader. Transformational leadership may also be highly relevant in these situations, given its’ demonstrated ability to create hope among followers in the face of hardship, combined with its demonstrated ability to stimulate intrinsic motivation for change (Bass and Bass 2009). This leadership may also provide a more secure basis for exploring the difficulties and provide energy and motivation to approach and work through the difficulties. Finally, leader-follower exchange theory (LMX) may also be a relevant perspective here (Graen and Uhl-Bien 1995). This theory views high-quality leader-follower relationships as the key objective in leadership and emphasises the importance of leaders’ constantly working on increasing the relationships with all members of one’s working group. This effort may again antecede an open and secure interaction between leader and a struggling follower, increasing the support dimensions in a growth process.

It should be noted that from a clinical perspective, strategies and mechanisms derived from the leadership literature like this may show fruitful, providing new perspectives on how to stimulate growth and a positive transformation of the self in times of crisis (Bass and Bass 2009). Thus, an inclusion of leadership theory and measures in future studies of PTG is recommended.

According to Jacobson et al. (2021), a major flaw of the PTG-SF measure utilised in the present study, asking respondents to indicate the degree of positive change as a result of their experiences in Lebanon, is high risk of retrospective biases and attribution biases – like consistency biases, desirability biases and faulty memories (Podsakoff et al. 2003).

Thus, the independent and dependent variables are conflated in each of the 10 questions, and the respondents are challenged to remember back on traumatic experiences, which may not effectively capture actual changes in positive coping (Park and Sinnot 2018) but differences in memory or other influences. Conversely, Jacobson et al. (2021) promote longitudinal designs, measuring PTG as a current standing situation at T1 and T2 and growth as a change index between these measuring points, in order to capture genuine growth, conversely to measures of what they see as possibly self-protective mechanisms or ‘self-illusionary coping’ (Zoellner and Maercker 2006).

We recognise this as an important critique to our study. However, in line with Johnson and Boals (2015), it could be argued that due to ‘event centrality’ of the traumas experienced in Lebanon by the soldiers (i.e., the extent to which an individual interprets a stressful event as part of their core identity), found to reduce the difference between PTGI scores and ‘actual growth’, our study might still have captured genuine growth and thus a contribution to validation of the PTG model by Maitlis (2020). It could also be argued, due to the positive correlation we found between so called ‘illusions of growth’ and trauma, that these cognitions may have a protective role to play, representing a source of cognitive resilience against detrimental effects of trauma. Thus, more research is needed to disentangle the distinctions between so called ‘self-illusionary coping’ (Zoellner and Maercker (2006) and ‘genuine growth’ measures.

This study offers some practical applications. First, it replicated previous studies showing a positive relationship between severe trauma and a positive change in life. Thus, is provides basis for hope. Second, it suggests that differentiated clinical interventions aiming at stimulating PTG, contingent on trauma, may be of limited use, given that moral trauma did not explain variance in PTG beyond the effect of victim trauma. Finally, it identifies cognitions in terms of growth attributions that may increase resilience after trauma and possibly serve as a focus in posttrauma interventions.

As already noted, the research designs risk retrospective and attribution biases (Jacobson et al. 2021), and more relevant leadership variables should be considered. More so, a thorough translation procedure of the measures from English to Norwegian, in line with the recommendations of the World Health Organization (2022), could have been applied. An inclusion of a reference category with soldiers without trauma experiences could have provided a better control for other influences post-deployment that might have influenced PTG. Also, the reduction in items from the authentic leadership and LF leadership measures could impair the psychometric properties of the scales. This limitation may also have an impact on the results finding no moderation effect of leadership on the mediation of emotional distress on the trauma vs. PTG relationship. Further studies should thus address this relationship.

Despite these limitations, also some unique strengths should be noted. First, the distinction between victim and moral trauma is a theoretical and empirical contribution that should be investigated further in future studies. Particularly, a search for unique trajectories (and different clinical interventions) related to these different trauma experiences is interesting. Second, it is the first study, to our knowledge, that included leadership in the model. This is a theoretical contribution that should be built on in future studies. Third, the research model is advanced, including mechanisms in terms of mediation and moderation. A model that should be continued in future studies – preferably in prospective designs. Fourth, the sample of UN peacekeepers is an important focus, due to the somewhat underappreciated status of these missions in military contexts. Finally, it is also worth noting that the time span from service in Lebanon and data collection may provide unique insights into the duration of traumatic experiences in military personnel.