Emerging adulthood (i.e., 18–25 years old; see Arnett 2000, 2011) is characterized by an extensive moratorium, provided to young people for the purpose of immersing themselves in various experiences from which they might extract a coherent life story (Bauer and McAdams 2004; Pals 1999) and establish a clearly delineated, essentially positive sense of self (Manczak et al. 2014; Shulman et al. 2006). However, securing such a sense of self is often compromised by the various life stressors that characterize emerging adulthood, namely, relocation (Dubas and Petersen 1996), occupational transitions (Rindfuss 1991), romantic setbacks (Davila et al. 2004), and other forms of stress. Another potentially complicating factor in the establishment of a clear sense of self and identity is psychological distress. Contrary to previous depictions of emerging adulthood as relatively tranquil (e.g., Coyne 1994), a serious increase in psychological distress has now been documented during this developmental period (American College Health Association, 2000, 2006, 2009; Blanco et al. 2008; Buchanan 2012; Hunt and Eisenberg 2010; Kessler et al. 2005; Rohde et al. 2013; Schwartz 2006; Zivin et al. 2009).

Increased ambiguity is also a characteristic of emerging adulthood. It is manifested, for example, in the fact that when asked, emerging adults have trouble defining the exact stage in life that they are in (Arnett 2000). They feel as if they have left adolescence, but have not yet entered adulthood, and therefore, have no name to describe the period of life that they are in (Arnett 2000). This sense of bewilderment also accompanies the delicate process of identity exploration, which is prominent in this period of life. Emerging adults are preoccupied with identity issues in three main areas: love, work, and worldview. Exploration in these three areas often becomes intense, especially when commencing around a time of frequent change (Arnett 2011). Disappointments in relationships and failures in work and school might project onto these individuals’ self-concept, leading to an injured sense of self. In this study, we focus on emerging adults’ self-concept because we believe it is a key factor of the ability of these individuals to complete highly complex developmental tasks that they are faced with.

The purpose of the present study was to examine the scarring effect of exposure to terror-related stress and psychological distress as complicating factors in Israeli emerging adults’ self-development. To this aim, we compared vulnerability and scarring models of personality and psychological distress. The vulnerability model, which is older and extensively researched, posits that personality and the self-concept lead to psychological distress either directly (the main effect vulnerability model) or by interacting with life stress (the stress-diathesis model). For example, low self-esteem was shown to prospectively predict depressive symptoms (Kernis et al. 1998; Orth et al. 2009a, b; Orth et al. 2008; Roberts and Monroe 1994; for a meta-analysis, see Sowislo and Orth 2013), and also to interact with life stress in predicting these symptoms (Abela 2002; Abela et al. 2006; Fernandez et al. 1998; Metalsky et al. 1993; Orth et al. 2009a; Ralph and Mineka 1998; Roberts 2006). However, Orth and colleagues suggest that most of the evidence is consistent with the main effect vulnerability model (Orth et al. 2009b; Orth and Robins 2013).

In the contrary, the scarring hypothesis, first proposed by Lewinsohn et al. (1981), posits that depression adversely affect cognition, personality, and the self-concept, thereby implying a reciprocal, bidirectional relationship between personality and psychological distress. This hypothesis has important implications for both basic clinical science and applied clinical psychology. From a basic science perspective, the possibility of depressive scarring runs counter to the default tendency of theorists and investigators to construe personality and the self as “prime movers,” thus propelling numerous outcomes (e.g., psychological distress) instead of appreciating the ever-changing, fluctuating nature of personality and the self (see Caspi et al. 2005; Santor et al. 1997; Shahar and Davidson 2003; Wachtel, 1997). From an applied point of view, the possibility of scarring suggests a vicious cycle involving personality vulnerability and depression (e.g., Shahar et al. 2004), thereby accounting for depression’s noxious chronicity (Pettit et al. 2006) and necessitating an immediate reduction of depressive symptoms (e.g., Gilboa-Schechtman & Shahar, 2006).

While Orth and colleagues reported null findings with respect to the effect of depression on low self-esteem (Orth et al. 2008; Orth et al. 2009; Orth et al. 2014; Sowislo and Orth 2013), Shahar and colleagues, using similar methodologies (i.e., multi-wave cross-lagged designs with structural equation modeling analyses), have repeatedly shown scarring effects with respect to low self-esteem (Shahar and Davidson 2003; Shahar and Henrich 2010; Schiller and Shahar 2013; Weinberg et al. 2012), self-criticism (Shahar et al. 2004), and other self-related constructs (Shahar et al. 2008; Shahar et al. 2013). In fact, Schiller and Shahar et al. (2013) demonstrated that scarring may ensue from a host of psychopathological symptoms (i.e., anxiety and somatization), rather than solely from depressive symptoms.

In the most recent study conducted by this group, Schiller et al. (2016) compared the vulnerability and scarring models, with a focus on four self-concept dimensions: self-esteem, self-criticism, generalized self-efficacy, and self-concept clarity. It is important to distinguish between these four self-concept dimensions. While self-esteem refers to people’s global representation of their typical or general feelings of self-worth (Kernis and Waschull 1995), self-criticism relates to a person’s emotional stance toward his self, and, namely, to the tendency to adopt a punitive stance toward the self once unrealistic standards are not met (Shahar 2001). On the other hand, self-concept clarity refers to the content of a person’s self and to the extent to which it is clearly and confidently defined, internally consistent, and temporally stable (Campbell et al. 1996). Lastly, generalized self-efficacy focuses on a person’s belief in his own capabilities (Bandura 1997; Wise 2007). It is clear that each self-concept dimension represents a different component in a person’s experience of himself in the world.

Schiller et al. (2016) followed 174 undergraduates at three time points over the course of one academic year. Consistent with the scarring model, psychological distress prospectively predicted a reduction in self-esteem and generalized self-efficacy, and an increase in various dimensions of self-criticism. The only finding in this study that was consistent with the vulnerability model was a prospective effect of self-concept clarity on psychological distress.

The Present Study

In the present study, we focused on Israeli young adults, who might be a unique example of emerging adulthood. In Israel, this developmental period actually commences around the early twenties, owing to the mandatory army service that most young people go through. Subsequent to discharge, many Israeli youth embark on a prolonged backpacking trip (Shulman et al. 2006), which frequently involves experimentation with various substances (Azaiza et al. 2009). Undergraduate studies begin at a relatively advanced stage in young adulthood, in turn leading to a heightened sense of urgency concerning the completion of developmental tasks (Shulman et al. 2006), and arguably derailing the exploration and moratorium needed for establishing an authentic identity. Furthermore, Israeli young adults, and specifically undergraduates, who study at the Ben-Gurion University campus, located in the city of Beer-Sheva (not far from the Gaza Strip), are exposed to repeated terror attacks in the form of missile strikes, which occur during Israeli military operations. These terror attacks are a major source of human-ecological stress and are likely to derail psychological development.

Exposure to terror has been massively explored and demonstrated as risk factor mainly for traumatic stress (Bleich et al. 2003; Bleich et al. 2006; Galea et al. 2002; Hobfoll et al. 2006; Hobfoll et al. 2009; Schiff et al. 2006) and to a lesser extent for depression (Hobfoll et al. 2006; Schiff et al. 2006). It is apparent that exposure to terror takes on a heavy toll, especially on individuals’ mental health, but we believe this heavy price runs much deeper and is prone to erode individuals’ (especially young adults) self-concept, and this is what our investigation intend to investigate.

In July 2014, a military operation (“Operation Protective Edge”) was launched in the Gaza Strip. The Israeli air strikes and Palestinian rocket attacks lasted for 7 weeks and resulted in the deaths of thousands of people, most of whom were Gazians. During the operation, rockets were fired daily on Israeli cities and villages. Beer-Sheva was one of the cities to suffer many attacks because of its relative proximity to the Gaza Strip. Shortly after the culmination of Operation Protective Edge, we returned to the participants who were studied in the aforementioned Schiller et al. (2016) investigation, and assessed (1) all of the self-concept dimensions measured in the previous phase (i.e., self-esteem, generalized self-efficacy, different forms of self-criticism and self-concept clarity), in addition to sense of self-reassurance, which was not examined in the previous study, (2) psychological distress, and (3) exposure to past and recent terrorism.

Method

Participants and Procedure

The study was approved by the Human Subjects Committee at the Department of Psychology at Ben-Gurion University. Of the original 174 participants from Schiller et al.’s (2016) study, 124 were retained (29% attrition rate; 99 women, M = 23.11, SD = 1.27, range = 20–27). They were approached via email 3 months after Operation Protective Edge, which was 1 year and 4 months after their participation in the previous study, and completed the self-concept, psychological distress, and exposure to terror questionnaires online. Participants were reimbursed with 50 shekels (~ 13 USD) at the end of the study.

We conducted attrition analyses and compared participants who completed both assessment waves (“completers”) with those who dropped out after Time 1 (“dropouts”), with respect to Time 1 variables. For the purpose of this comparison, we conducted Students’ t tests and found dropouts to be significantly lower than completers with respect to self-esteem (t = 3.82, p < .000) and self-concept clarity (t = 7.00, p < .000). There were no significant differences in levels of psychological distress, self-criticism, and self-efficacy between the two groups.

Measures

Self-Concept Clarity (Campbell et al. 1996)

The SCC is a 12-item measure of the perceived consistency and clarity in an individual’s sense of self (e.g., “In general, I have a clear sense of who I am”). Responses are provided using a 5-point Likert scale. Reliability and validity of the SCC have been established in previous studies (e.g., Campbell 1990; Campbell et al. 1996), and, in the present study, the internal consistency was adequate (α baseline = .79, α T4 = .80).

DEQ-SC6 (Based on the DEQ; Blatt, D’Afflitti, & Quinlan, 1976)

The DEQ-SC6 is a short version of the DEQ, based on six items characterized by a strong face-validity in terms of measuring self-criticism. The DEQ-SC6 has been validated extensively in previous research (e.g., Bareket-Bojmel and Shahar 2011; Lerman et al. 2012; Rudich et al. 2008; Shahar 2015; Shahar et al. 2008; Soffer et al. 2008; Zuroff et al. 2016), and appears to be a sound measure of trait self-criticism. The internal consistency of the measure was adequate (α baseline = .73, α T4 = .72).

The Forms of Self-Criticizing/attacking and Self-Reassuring Scale (FSCRS; Gilbert et al. 2004) is a 22-item measure that reflects typical self-critical and self-reassuring thoughts and feelings that people have about themselves when things go wrong. The scale includes three components, two of which are forms of self-criticism: inadequate self (IS; e.g., “I am easily disappointed with myself”) and hated self (HS; e.g., “I have become so angry with myself that I want to hurt or injure myself”), and one that is a form of self-reassurance; reassuring self (RS; e.g., “I am able to remind myself of positive things about myself”). Responses are rated on a 5-point Likert scale ranging from 0 (not at all like me) to 4 (extremely like me). The internal consistency of the three components respectively was adequate (α baseline = .84, .69, .84, α T4 = .87, .60, .84, for IS, HS, and RS, respectively).

Generalized Self-Efficacy (Schwarzer 1993; Schwarzer and Jerusalem 1995)

The GSE is a 10-item measure of self-efficacy (e.g., “Thanks to my resourcefulness, I know how to handle unforeseen situations”). Items are rated on a 4-point Likert scale. Strong reliability and validity coefficients are reported in Schwarzer (1993), as well as in Schwarzer and Jerusalem (1995). In the present study, the internal consistency was adequate (α baseline = .83, α T4 = .76).

The Rosenberg Self-Esteem Scale (RSES; Rosenberg 1965) is an extensively validated and used 10-item measure assessing global self-esteem (e.g., “On the whole, I am satisfied with myself”). Responses are rated on a 4-point Likert scale. In the present study, the internal consistency of the RSES was adequate (α baseline = .87, α T4 = .89).

Psychological Distress

The Brief Symptom Inventory (BSI; Derogatis and Spencer 1982; Derogatis and Melisaratos 1983) is a 53-item measure, used to assess a wide range of self-reported psychopathological symptoms, including depression, somatization, obsessive-compulsive symptoms, interpersonal sensitivity, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. Participants were asked to indicate on a 5-point Likert scale ranging from 0 (not at all) to 4 (very often) how often they suffered from a symptom in the last month. A global BSI score was constituted by averaging the 53 items. The internal consistency of the BSI in this study was adequate (α baseline = .94, α T4 = .95).

Exposure to Missile Attacks

The scales were adapted from previous research in Israel and were used successfully in the population sampled for this study (Kirschenbaum 2006).

Prior Exposure to Terror

Participants were asked whether they had ever been present during “a terror/rocket” attack, been injured in an attack, or had close family members, close friends, or acquaintances injured in an attack. In addition, participants were asked whether their house, other houses in their neighborhood, or other houses near their house were ever damaged from fired missiles. Answers were summed together to create an index of exposure to prior terror attacks on a scale of 0 to 8 (M = 1.84, SD = 1.16, median = 2).

Physical Exposure to Terror During Operation Protective Edge

Participants were asked three questions about the frequency of hearing missile alert alarms, hearing or feeling a missile hit, and being present in an actual missile hit during Operation Protective Edge. Answers were summed together to create an index of physical exposure to terror during Operation Protective Edge on a scale of 0 to 16 (M = 5.48, SD = 3.25, median = 5).

Relational Exposure to Terror

Participants were asked regarding exposure to terror of relatives, friends, or acquaintances. Fourteen participants responded that a friend, relative, or acquaintance of theirs had been exposed to terror attacks; therefore, we treated this variable binarily.

Analytic Strategy

After calculating the means, standard deviations, and intercorrelations among the study variables (see Tables 1 and 2), we tested both the vulnerability and scarring models using a set of cross-lagged structural equation modeling analysis. All SEM analyses were conducted in R (R Core Team, 2019, version 3.6.1), with the Lavaan package (Rosseel 2012, version 0.6-3), using the maximum likelihood (ML) iteration procedure. The following fit indices were selected to estimate model fit: χ2/df , the Tucker-Lewis Index (TLI, Marsh and Hocevar 1985), the Comparative Fit Index (CFI, Bentler 1990), and the Root Mean Square of Error Approximation (RMSEA, Steiger and Lind 1980). According to Hu and Bentler (1999), an adequate model fit is indicated by values lower than 2.5 on the χ2/df index, values higher than .95 on both the TLI and CFI, and values lower than .06 on the RMSEA. The traditional χ2index for model fit was not consulted due to its hypersensitivity to sample size (Hu and Bentler 1998). For each of the SEM analyses conducted, we arrived at a final, most parsimonious model, by omitting all statistically non-significant paths (Bentler and Mooijaart 1989), and report here standardized coefficients, along with the matching Wald test scores.

Table 1 Means and standard deviations
Full size table
Table 2 Correlations
Full size table

We specifically focused on two patterns of causal links, where (1) distress and terror-related stress propel changes in self-concept, and (2) where self-concept and terror-related stress propel changes in levels of distress.

Results

The means, standard deviations, and intercorrelations are presented in Tables 1 and 2. The BSI scores of our study participants were within the published Israeli norms (Gilbar and Ben-Zur 2002). Moderate to high correlations (.65–.77) were observed between baseline self-concept and psychological distress variables in the first measurement wave to self-concept and psychological distress variables at the fourth measurement wave point to the high stability of the self and psychological distress.

Cross-Lagged Effects Involving Self, Psychological Distress, and Terror-Related Stress

In Table 3, we present the values of the fit indices for the final, most parsimonious, cross-lagged SEMs obtained for each of the self-concept variables. Note that all values indicated an adequate fit. In Fig. 1, we present standardized parameters for these most parsimonious, cross-lagged SEMs.

Table 3 Model fit for the most parsimonious cross-lagged SEM analyses
Full size table
Fig. 1
figure 1

Self, psychological distress, and exposure to terror in young adulthood

Full size image

Self-Criticism (DEQ-SC6)

Synchronous associations between distress and self-criticism were statistically significant (rs = .51, .3 SEs = 0.07, 0.08, 4, ts = 7.7, 4.4, ps < .001). Both self-criticism and distress were stable over time (for self-criticism β[T1−T4] = 0.45, t = 5.5, SE = 0.08, p < .001; For distress β[T1−T4] = 0.42, SE = 0.08, t = 5.1, p < .001). Prior exposure to terror predicted elevated levels of self-criticism at T4 (β = 0.17, SE = 0.08, t = 2.1, p < .05), and also elevated levels of distress at T4 (β = 0.18, SE = 0.09, t = 2.1, p < .05). Relational exposure also predicted levels of distress at T4 (β = 0.18, SE = 0.07, t = 2.5, p < .01).

Inadequate Self (FSCRS-IS)

Synchronous associations between distress and inadequate self were statistically significant (rs = .72, .51, SEs = 0.04, 0.07, ts = 16.4, 7.7, ps < .001). Both inadequate self and distress were stable over time (for inadequate self β[T1−T4] = 0.39, SE = 0.10, t = 4.1, p < .001. For distress: β[T1−T4] = 0.44, SE = 0.10, t = 4.4, p < .001). Prior exposure to terror predicted elevated levels of inadequate self at T4 (β = 0.20, SE = 0.07, t = 2.7, p < .01), and also elevated levels of distress at T4 (β = 0.23, SE = 0.10, t = 2.4, p < .05). Relational exposure to terror predicted an increase in levels of inadequate self at T4 (β = 0.15, SE = 0.07, t = 2.2, p < .05), and an increase in levels of distress at T4 (β = 0.19, SE = 0.07, t = 2.5, p < .05). Importantly, baseline distress predicted elevated levels of inadequate self at T4 (β = 0.23, SE = 0.10, t = 2.4, p < .05).

Hated Self (FSCRS-HS)

Synchronous associations between hated self and distress were statistically significant (rs = .58, .42, SEs = 0.06, 0.07, ts = 9.8, 5.7, ps < .001). Hated self and distress were stable over time (for hated self β[T1−T4] = 0.50, SE = 0.08, t = 6.5, p < .001. For distress β[T1−T4] = 0.41, SE = 0.09, t = 7.8, p < .001). Prior exposure to terror predicted elevated levels of hated self and distress at T4 (β = 0.25, SE = 0.07, t = 3.5, p < .01; β = 0.22, SE = 0.08, t = 2.8, p < .01). As well, relational exposure to terror predicted an increase in levels of hated self (β = 0.18, SE = 0.07, t = 2.7, p < .01), and distress at T4 (β = 0.19, SE = 0.07, t = 2.5, p < .05).

Reassuring Self (FSCRS-RS)

Synchronous associations were statistically significant (rs = − .43, − .30, SEs = 0.07, 0.08, ts = − 5.8, − 3.7, ps < .001). Levels of reassuring self and distress were stable over time (for RS β[T1−T4] = 0.49, SE = 0.07, t = 6.6, p < .001; for distress β[T1−T4] = 0.42, SE = 0.08, t = 5.3, p < .001). Prior exposure to terror predicted elevated levels of distress at T4 (β = 0.25, SE = 0.08, t = 3.2, p < .01). Relational exposure to terror predicted a decrease in levels of reassuring self at T4 (β = − 0.15, SE = 0.07, t = − 2.1, p < .05), and also increased distress at T4 (β = 0.19, SE = 0.07, t = 2.6, p < .01). Importantly, baseline distress predicted lower levels of reassuring self at T4 (β = − 0.19, SE = 0.08, t = − 2.4, p < .05).

Self-Esteem (RSES)

Synchronous associations were statistically significant (rs = − .49, − .18, SEs = 0.07, 0.09, ts = − 7.2, − 2.1, ps < .05). Levels of self-esteem and distress were stable over time (for RSES β[T1−T4] = 0.42, SE = 0.09, t = 4.8, p < .001. For distress β[T1−T4] = 0.39, SE = 0.08, t = 4.9, p < .001). Elevated levels of distress at T4 were predicted by prior exposure to terror (β = 0.22, SE = 0.08, t = 2.9, p < .01), and relational exposure (β = 0.18, SE = 0.07, t = 2.4, p < .05). Importantly, self-esteem at T1 predicted decreased levels of distress at T4 (β = − 0.18, SE = 0.08, t = − 2.1, p < .05).

Generalized Self-Efficacy

Synchronous associations were statistically significant (rs = − 0.20, − 0.20, SEs = 0.09, 0.09, ts = − 2.4, − 2.3, ps < .05). Generalized self-efficacy and distress were stable over time (for generalized self-efficacy β[T1−T4] = 0.39, SE = 0.07, t = 5.6, p < .001. For distress β[T1−T4] = 0.46, SE = 0.07, t = 6.7, p < .001). Elevated levels of distress at T4 were predicted by prior exposure to terror (β = 0.23, SE = 0.08, t = 3.0, p < .01, and relational exposure to terror (β = 0.18, SE = 0.07, t = 2.4, p < .05). Physical exposure to terror predicted elevated levels of generalized self-efficacy at T4 (β = 0.19, SE = 0.09, t = 2.2, p < .05). Importantly, baseline distress predicted decreased levels of generalized self-efficacy at T4 (β = − 0.30, SE = 0.07, t = − 4.2, p < .001).

Self-Concept Clarity

Synchronous associations were statistically significant (rs = − 0.46, − 0.29, SEs = 0.07, 0.08, ts = − 6.4, − 3.5, ps < .001). Self-concept clarity and distress were stable over time (for self-concept clarity, β[T1−T4] = 0.62, SE = 0.06, t = 10.4, p < .001; for distress, β[T1−T4] = 0.41, SE = 0.08, t = 5.4, p < .001). Elevated levels of distress at T4 were predicted by prior exposure to terror (β = 0.22, SE = 0.08, t = 2.9, p < .01), and relational exposure to terror (β = 0.18, SE = 0.07, t = 2.5, p < .05). Importantly, baseline distress predicted decreased levels of self-concept clarity at T4 (β = − 0.14, SE = 0.07, t = − 2.0, p < .05).

Interactions Between Exposure to Terrorism and Self/Distress

In an attempt to test the possibility that exposure to terrorism moderates the longitudinal associations between the self-concept and psychological distress, we examined interaction between either the self-concept variables or psychological distress on the one hand, and the two exposure variables—physical and relational exposure—on the other hand. Of the 16 interactions tested, only three yielded statistically significant effects, a pattern that is not robust in the face of type I error corrections. All of the three interactions involved relational exposure. Two of the three interactions yielded counterintuitive patterns: (1) psychological distress predicted an increase in inadequate self, but only among those without relational exposure; and (2) psychological distress predicts a decrease in hated self in the relational exposure group. The third interaction was consistent with theory: hated self predicted an increase in psychological distress among relationally exposed participants.

All in all, however, we consider these findings indicative of a no interaction pattern. This is because, as stated above, the number of the interactions found is not robust in the face of type I error corrections, two of the three interactions found are counterintuitive, and all interactions were found with the relational exposure binary variable, which pertains to a relatively small number of individuals actually exposed. Details on these analyses are available from the authors upon request.

Discussion

Findings concerning the vulnerability model were relatively weaker. A main effect of low self-esteem over subsequent psychological distress was found, consistent with previous studies examining the vulnerability effect (e.g., Orth et al. 2009; Orth and Robins 2013). In contrast, our findings provide robust support for the scarring model (Lewinsohn et al. 1981; Shahar and Henrich 2010). Thus, scarring was found to occur in four out of the seven self-concept dimensions: inadequate self, generalized self-efficacy, reassuring self, and self-concept clarity. As well, relational and prior exposure to terror predicted an increase in a sense of inadequate and hated self, prior exposure to terror predicted an increase in self-criticism and relational exposure to terror predicted a decrease in one’s sense of reassured self. Prior and relational exposure to terror also predicted an increase in psychological distress over time. A counterintuitive effect of physical exposure to terror on generalized self-efficacy also surfaced, in which physical exposure to terror predicted an increase in generalized self-efficacy. The only self-concept dimension that did not exhibit evidence of the scarring effect is self-esteem.

The wide-scale effect of scarring attests to the deleterious effect of psychological distress on self-development during young adulthood. Alongside the growing evidence of elevated levels of psychological distress during young adulthood (e.g., ACHA-NCHA 2000, 2006, 2009; Blanco et al. 2008; Buchanan 2012; Hunt and Eisenberg 2010; Kessler et al. 2005; Rohde et al. 2013; Schwartz 2006; Zivin et al. 2009), the obtained results illuminate the psychological fragility of this developmental period, perhaps particularly in the present era. The overall developmental-historical context of this surge of psychological distress in young adulthood might be located in extant knowledge on this “Y Generation” (i.e., comprised of individuals who were born in the 1980s and the early 1990s). It is a generation that is developing in a society characterized by crumbling structures and a bombardment of technological changes (Arnett 2000, 2011; Stranger 2005; Strauss and Howe 2000; Woodman 2015). Both chronosystemic changes (Bronfenbrenner 1977) might account, at least in part, for this generation’s plight.

Findings concerning exposure to terror are compatible to the deleterious effects of exposure to war and terrorism on psychological well-being reported in a vast number of previous studies (e.g., Bleich et al. 2003; Bleich et al. 2006; Comer and Kendall 2007; Hobfoll et al. 2006; Hobfoll et al. 2009; La Greca 2007; Laufer and Solomon 2006; Masten and Narayan 2012; Pine et al. 2005; Schiff et al. 2006; Slone and Shechner 2009; Slone and Shoshani 2014). Our study, however, extends this line of inquiry by showing that exposure to terrorism leads to self-concept deficiency (Shahar 2015).

The accumulated evidence on the existence of scarring in young adulthood calls for some applied implications, given the aforementioned surge of mental health problems in this age, as well as the fact that young adulthood is a developmental period during which individuals make fundamental choices that have bearing decades later (Arnett 2000, 2011). Moreover, mental health problems during college are associated with poor academic performance (e.g., Cranford et al. 2009; Eisenberg et al. 2009; Svanum and Zody 2001). In fact, 5% of college students drop out because of psychiatric disorders, and 4.2 million people in the USA would have graduated from college, if they had not been suffering from a psychiatric disorder (Kessler et al. 1995). Namely, scarring effects transpiring, in this developmental period, are particularly alarming, given the centrality of the self-concept in the development of subsequent problems in adolescence and young adulthood (Shahar 2015).

Thus, to the extent that our findings will be repeatedly replicated, we can envision interventions early in the freshman year of college that target students’ self-perceptions and attitudes toward themselves. We believe interventions should include the enhancement of positive self-concept dimensions (e.g., self-esteem, generalized self-efficacy, self-concept clarity), as well as eradication of negative self-concept dimensions (e.g., self-criticism, feelings of inadequacy, and feelings of hatred toward the self). Unfortunately, most of the college interventions that are discussed in the scientific literature deal with alcohol prevention programs (e.g., Borsari and Carey 2000; Carey et al. 2007), and only a few focus on general psychological health. Yeager and Walton (2011) reviewed several, rather “small,” psychological interventions administered in college settings for different purposes, and targeted students’ thoughts, feelings, and beliefs in and about school. The interventions discussed in their review, ranging from one session to a 1-year long program, target students’ subjective experiences in school and demonstrate considerable long-term gains, which are depicted mostly in improved academic performance. We deem these patterns promising and in need of further applied and scientific work.

The limitations of this study should be noted. First, the study is limited in its scope, since it does not attend to the links between self, psychological distress, and exposure to terror-related stress throughout the entire young adulthood, but only among individuals between 22 and 28 years of age. The years between the ages of 18 and 22, which are the continuation of adolescence, are crucial years for young adults, especially in Israel for reasons mentioned in the introduction (Shulman et al. 2006). Likewise, the years between the ages of 28 and 35 are comprised of other life tasks, such as getting married, starting a family, and establishing a profession. In order to add to the theoretical literature on young adulthood as a developmental period, future research should examine the scarring and vulnerability models during different stages of young adulthood. Secondly, some features of the sample limit generalizability to other populations (e.g., to non-college-attending participants). Similarly, attrition analyses revealed that “dropouts” had greater self-concept impairment than “completers,” which might have attenuated the results in unknown ways. An additional limitation is our reliance on self-report measures to assess psychological distress and self-concept dimensions that are limited in their ability to capture individuals’ whole experiences in the world.

Despite these limitations, this study is—to the best of our knowledge—the first to involve a longitudinal examination of two models linking the self, psychological distress, and traumatic stress during young adulthood, while espousing a multidimensional assessment of the self-concept. It thus paves the way for progress in the scientific and applied field of young adulthood, in general, and in developmental health, in particular.