The Centrality of Events Scale in Portuguese Adolescents: Validity Evidence Based on Internal Structure and on Relations to Other Variables

Abstract

We explored the measurement model of the adolescent version of the Centrality of Event Scale and its invariance across community (n = 1,079; 42.8% male), referred for foster care (n = 205; 58.0% male), and detained (n = 206 male) adolescent participants. Results indicated a three-factor measurement model, including all three functions that memories of significant life events may have, as a good fit to our data, particularly for male participants. This measurement model was invariant across boys taken from those different samples but not across gender. As for the short version of the instrument, a one-factor solution was the best fit to our data. It was invariant across boys taken from those different samples and across gender. Boys and girls expressed similar experiences, whereas community male adolescents reported the lowest impact of a meaningful event, in comparison with referred and with detained boys. These findings provide evidence on the validity of the scale for use with diverse adolescent samples, which may contribute for a better understanding of the impact that significant life events may have on the development of gender-specific and group-specific vulnerabilities.

Keywords

centrality of event adolescence confirmatory factor analysis externalizing behaviors group comparison psychometrics

Autobiographical memories help anchor and stabilize our self-concept and to give meaning, structure, and balance to our personal life narratives (Berntsen & Rubin, 2006). They are repeatedly recalled (or reflected on) and generally are linked to profound life-changing events, either positive or stressful/traumatic (Berntsen & Rubin, 2006; Boals, 2010). According to the centrality of event theory (Berntsen & Rubin, 2006, 2007), such memories of personally significant events may have three functions. First, those memories can become a reference point for interpretations of everyday events; that is to say they become an anchoring event for the perception and attribution of meaning to current and future experiences and expectations, thus validating current beliefs and feelings, and guiding one’s thoughts and behavior. Second, those memories can become a turning point in one’s personal life story, capable of modifying or redirecting the course of such a story and also being capable of changing beliefs about ourselves. Finally, memories of a highly relevant event can become a central component of personal identity, turning up to be an emblematic episode for the self and for self-understanding (Berntsen & Rubin, 2006, 2007).

Most individuals tend to remember considerably more positive than negative events of their lives, at the same time as highly negative life events are relatively rare in the general population (Boals, 2010). Nevertheless, such stressful experiences can function as traumatic memories that become central to one’s life, and so it is important to assess not only the presence and history of trauma but also, most important, the extent to which a memory of a stressful event represents a core topic for personal identity and for the attribution of meaning to other life experiences (Berntsen & Rubin, 2006, 2007). There are several measures to assess trauma exposure and the impact of such traumatic experiences (Rubin, Boals, & Hoyle, 2014). However, to our knowledge, there is only one measure in accordance with the centrality of event theoretical framework (i.e., considering how one event may become a central memory in one’s identity and life story): the Centrality of Event Scale (CES; Berntsen & Rubin, 2006).

The CES was designed taking into consideration the three functions of autobiographical memories, meaning the extent to which a memory for a stressful event becomes: (a) a reference point for everyday inferences, (b) a turning point in one’s life story, and (c) a central component of personal identity (Berntsen & Rubin, 2006). The authors suggest this measure to be unifactorial, though their exploratory factor analyses on the scores taken from an adult sample returned three factors with eigenvalues higher than one; this three-factor solution was not further explored. Berntsen and Rubin (2006) also suggested the use of a short version of the CES, composed of the seven items that correlated the highest with the total score of the scale. The 20-item (e.g., Groleau, Calhoun, Cann, & Tedeschi, 2013; Robinaugh & McNally, 2011; Schuettler & Boals, 2011) and the short (e.g., Boals, 2010; Rubin et al., 2014; Rubin, Hoyle, & Leary, 2012) versions of the CES were used in subsequent studies, all of which found good psychometrical proprieties for their results using them as unifactorial measures.

Alternatively, Gauer, Souza, Silveira, and Sediyama (2013) proposed the 20-item CES version to be a three-factor measure, which they found via exploratory factor analysis using the results of a sample of 195 Brazilian undergraduate students. The semantic content of the three factors were in line with the functions of central events (Berntsen & Rubin, 2006). Gauer et al. (2013) also proposed a short version of the CES, comprising the seven items that correlated the highest with the total score of the scale in their study. This short version of the CES was very similar to the original one, the only exception being the replacement of one item by another.

Due to the increasing interest by clinicians and researchers on the role that autobiographical memories may hold for psychological functioning in youth, the 20-item CES version has also been adapted to the adolescent population (CES-A; Cunha, Xavier, Matos, & Faria, 2015). Similar to what was found for adults, a one-factor solution was put forward as an adequate measurement model for this age group.

Though addressing the centrality of events with participants who are still developing their identities and life stories may seem contra intuitive, research suggests that meaning-making processes are present as early as adolescence, namely the integration of memories in the personal identity (McLean, Breen, & Fournier, 2010). Furthermore, adolescents struggle with developmental demands associated with their future adult identity and roles (Kroger, 2004), which are tasks they have to undertake while facing an intense emotional experience (Zeman, Cassano, Perry-Parrish, & Stegall, 2006) and possessing a less than effective repertory of emotion regulation strategies (Garnefski & Kraaij, 2006; McRae et al., 2012; Zimmermann & Iwanski, 2014). Achieving these developmental tasks may prove particularly challenging for those adolescents who face peculiar life circumstances that may represent a central event to their personal identity and life story. Youth with disruptive behaviors are one of such cases. So, the validation of the CES-A within externalizing samples of adolescents, which has not been accomplished before, seems paramount mainly for three reasons: (a) trauma exposure is highly prevalent in youth with disruptive disorders (Abram et al., 2004; Briggs et al., 2013; Dierkhising et al., 2013; Willis, Best, & Aalsma, 2013); (b) environmental factors (particularly early toxic experiences) seem to play a key role in the development of behavioral problems (Ribeiro da Silva, Rijo, & Salekin, 2015); and (c) it seems important to address how some life experiences can concur to mold these youth’s identities. Moreover, it would be important to investigate the differences in the results of the CES-A between community adolescents and youth with externalizing problems; to accurately grasp these differences, measurement invariance must be considered, in order to guarantee that researchers are assessing the same constructs across those groups and, thus, avoid inference problems (Chen, 2007).

Given the inconsistent results found when comparing scores of the CES between male and female participants, it also seems relevant to better explore these differences. Previous works using the seven-item CES version with an adult sample found that females scored significantly higher than males (Boals, 2010). According to the centrality of event theory, the differential childhood socialization practices for boys and girls probably lead to more complex and enriched cognitive representations of emotional events in females (Berntsen & Rubin, 2006; Boals, 2010; Davis, 1999). In contrast with males, females tend to more easily recall past events, develop more detailed representations of past experiences, and transform both positive and negative/traumatic emotional events into central points to their identities (Bloise & Johnson, 2007; Boals, 2010). The fact that females are more likely than males to take negative experiences (and not only positive ones) as central to their identity may help explain why women are particularly vulnerable after a stressful event (Berntsen & Rubin, 2006; Bloise & Johnson, 2007; Boals, 2010). Still some works do not concur with these gender differences, either with undergraduate (Gauer et al., 2013) or adolescent samples (Cunha et al., 2015). Thus, exploring if (and why) gender differences actually exist is still relevant, for which gender invariance must first be ascertained.

This work proposed to evaluate the psychometric proprieties of the CES-A using a combined sample of Portuguese adolescents, including community participants and participants either referred or detained for externalizing behavior. The CES-A has not been studied with such specific samples and so, before it can be used to understand the effects of traumatic events in their autobiographical memories, there must be evidence that such effects are being assessed properly. In particular, the internal structures of the 20-item and of the short seven-item versions of the CES were investigated as applied to adolescent samples. For the 20-item version of the CES, three alternative measurement models were investigated and compared: a one-factor solution as proposed by Berntsen and Rubin (2006), a three-factor empirically derived solution as proposed by Gauer et al. (2013), and a three-factor theoretically derived solution based on the contents Berntsen and Rubin (2006) attributed to each item (i.e., reference point for one’s life, turning point in one’s life story, and central to one’s personal identity). For the seven-item version of the CES, three competing measurement models were also tested: a one-factor solution composed by the items showing higher item-total correlation values within a North American sample of university students (Berntsen & Rubin, 2006), a one-factor solution constituted by the items found to have the highest item-total correlation values using a Brazilian sample of undergraduate students (Gauer et al., 2013; interestingly, only one item differed between these two short versions), and a three-factor theoretically derived solution. Scores on the resulting measures were then analyzed for internal consistency, measurement invariance, and latent mean comparison across gender and type of sample.

We expected that the same measurement structure would adequately represent the centrality of event experiences of boys and girls on one hand and of boys who exhibit behavioral problems to different degrees on the other hand. Moreover, following the results previously obtained with Portuguese adolescent samples (Cunha et al., 2015), we expected to find no significant gender differences. As for comparisons between male samples of youth with different degrees of behavioral problems, and considering that adolescents with externalizing problems have a history of more exposure to traumatic events (e.g., Willis et al., 2013), we expected that those participants would score significantly higher than community participants. Finally, validity evidence based on relation to external variables was also explored, concerning self-criticism, self-reassurance, and experiential avoidance. Like in previous studies, we expected scores on the CES-A to be positively associated with experiential avoidance (Rubin et al., 2014; Schuettler & Boals, 2011). Given that self-criticism and self-reassurance have been associated with various psychopathological symptoms (Castilho & Pinto-Gouveia, 2011), which, in turn, have been related to results on the CES-A (Berntsen & Rubin, 2006; Boals, 2010; Schuettler & Boals, 2011), the CES-A scores were expected to associate positively with self-criticism and negatively with self-reassurance in the current work.

Method

Participants and Procedures

The current work included a combined sample of adolescents recruited from school settings, adolescents referred for behavioral problems, and adolescents detained by the judicial system. In addition to institutional authorizations, all participants were informed of the goals of the study and the confidentiality and anonymity of their responses were guaranteed. After this, all participants younger than 18 years of age verbally consented to their own participation in addition to their parents/legal guardians’ written consent. In turn, participants older than 18 years provided verbal and written consent for their own participation. No adolescent in the community sample refused to participate. In the referred and detained samples, 15 and 14 adolescents declined to participate, respectively.

A total of 1,526 participants was thus recruited, 36 of whom returned evaluation protocols with missing values to between one to four items of the CES-A. Though not being missing completely at random, χ²(337) = 449.59, p < .001, these missing values represented only 0.13% of the total response pool. For this reason, and because exploring missing value patterns were beyond the scope of the current work, a listwise deletion approach was applied to handling missing values.

So, the final sample for this study included 1,490 Portuguese adolescents, aged between 12 and 21 years (i.e., combined sample). Boys and girls taken from this combined sample had similar mean ages, t(1,487) = −0.23, p = .08, and were evenly distributed by socioeconomic status (SES), χ²(2) = 3.07, p = .22. Of those participants, 1,079 adolescents were recruited from several national public schools located in the center and north regions of Portugal, after the national ethics committee and/or the executive boards of the schools approved the study (i.e., community samples). All of these students filled in the CES. Additionally, a subsample of 202 adolescents was randomly selected (4/5 subjects per class) and also filled in the Forms of Self-Criticising/Attacking & Self-Reassuring Scale (FSCRS; Subsample 1); of these, another 139 students were again randomly selected (2/3 from the initial 4/5 students per class) and also completed the Acceptance and Action Questionnaire (AAQ-II; Subsample 2). These data were collected in group (during classes) in the presence of the researchers.

Participants in this study also included 205 adolescents placed in Azores foster care institutions, which were referred for disruptive behavior (i.e., referred sample). Finally, 206 male young offenders placed in all of the eight Portuguese juvenile detention facilities due to criminal behavior were randomly selected to participate in this study (i.e., detained sample). These participants filled in only the CES-A after the study was approved by the institutions’ boards, and were evaluated individually by the authors or by psychologists from foster care/detention facilities.

Table 1 presents the demographic characteristics of the samples and subsamples.

Table 1.

Demographic Characteristics of the Samples and Subsamples.

	Gender		Age	Socioeconomic status
	Male	Female	Age	Low	Medium	High
Combined sample	787 (52.8)	703 (47.2)	16.15 (1.52)	378 (25.4)	762 (51.1)	350 (23.5)
Community sample	462 (42.8)	617 (57.2)	16.08 (1.52)	219 (20.3)	552 (51.2)	308 (28.5)
Subsample 1	84 (41.6)	118 (58.4)	16.82 (1.08)	73 (36.1)	—	129 (63.9)
Subsample 2	58 (41.7)	81 (58.3)	16.87 (1.11)	—	51 (36.7)	88 (63.3)
Referred sample	119 (58.0)	86 (42.0)	16.05 (1.66)	159 (77.6)	45 (22.0)	1 (0.5)
Detained sample	206 (100)	—	16.59 (1.28)	164 (79.6)	42 (20.4)	—

Note. Information for gender and socioeconomic status are presented as n (%); information for age is presented as M (SD). Socioeconomic status was measured by parents’ profession, taking into account the Portuguese profession classification (Instituto Nacional de Estatística, 2011). Examples of professions in the high socioeconomic status group are judges, higher education professors, or MDs; included in the medium socioeconomic status group are nurses, psychologists, or school teachers; included in the low socioeconomic group are farmers, cleaning staff, or undifferentiated workers.

Detained participants were significantly older than both community, t(322.3) = −5.03, p < .001, and referred participants, t(408) = −3.68, p < .001; alternatively, community and referred participants had similar mean ages, t(1,282) = 0.28, p = .78. Participants in these samples were not similarly distributed either by gender, χ²(2) = 229.57, p < .001, or by SES, χ²(4) = 296.51, p < .001. Girls were more frequent than expected in the community sample, whereas boys were more prevalent than expected in the referred and detained samples. In relation to what was statistically expected, more community participants descended from a high SES, more referred participants came from a low SES, and, finally, more detained participants derived from a medium SES.

Measures

Centrality of Event Scale–Adolescents

CES-A (Berntsen & Rubin, 2006; Portuguese version for adolescents by Cunha et al., 2015) is a 20-item scale that measures the extent to which a memory of a stressful or traumatic event becomes (a) a reference point for individual’s everyday references (e.g., “This event has become a reference point for the way I understand new experiences”); (b) a turning point in one’s life story (e.g., “I feel that this event has become a central part of my life story”); and (c) a central component of personal identity (e.g., “This event tells a lot about who I am”; Berntsen & Rubin, 2006). The CES-A gives the following prompt to participants: “Please think back on the most stressful or traumatic event in your life and answer the following questions”; it further presents a 5-point Likert-type scale ranging from 1 (totally disagree) to 5 (totally agree) for them to do so. The original study on the psychometric properties of the scale using an adult sample revealed a high internal consistency (α = .94) and a one-factor solution (Berntsen & Rubin, 2006). The Portuguese version was studied with 397 adolescents and showed a Cronbach’s alpha of .95 and a single-factor structure (Cunha et al., 2015), similar to the original version of the scale. Analyses of the psychometric properties of the CES-A with the current samples will be presented in the results section.

The Forms of Self-Criticising/Attacking & Self-Reassuring Scale

FSCRS (Gilbert, Clark, Hempel, Miles, & Irons, 2004; Portuguese version by Castilho & Pinto-Gouveia, 2011) is a 22-item scale that assesses two forms of self-criticism, namely (a) inadequate self, which focuses on a sense of personal inadequacy (e.g., “I am easily disappointed with myself”) and (b) hated self, which measures the desire to hurt or persecute the self (e.g., “I have become so angry with myself that I want to hurt or injure myself”). The scale also assesses self-reassurance (e.g., “I am able to care and look after myself”). Participants are asked to rate how much each item is true for them using a 5-point Likert-type scale ranging from 0 (not at all like me) to 4 (extremely like me); items portray ways in which people may have negative and self-critical thoughts and feelings or, alternatively, be supportive of themselves (Gilbert et al., 2004). The original version of the FSCRS presented good psychometric properties, with alphas of .90, for the inadequate self and .86, for both the hated self and self-reassurance (Gilbert et al., 2004). In the Portuguese version, Cronbach’s alphas ranged from .62 to .89 (Castilho & Pinto-Gouveia, 2011). In the present study, internal consistency values using the combined sample were .89, for the inadequate self, .81, for the hated self, and .88, for self-reassurance.

The Acceptance and Action Questionnaire

AAQ-II (Bond et al., 2011; Portuguese version by Pinto-Gouveia, Gregório, Dinis, & Xavier, 2012) is a seven-item scale that measures a person’s experiential avoidance and immobility, as well as acceptance and action. Participants are asked to rate how true each item is for them, using a 7-point Likert-type scale ranging from 1 (never true) to 7 (always true). Higher scores in AAQ-II are reflective of greater experiential avoidance and immobility, whereas lower scores reflect greater acceptance and action. The AAQ-II has been found to have adequate reliability and validity with college student and clinical samples (Bond et al., 2011; Pinto-Gouveia et al., 2012). In both the Portuguese validation study and the current work, internal consistency was .89.

Data Analysis

Data were analyzed with the IBM SPPS Statistic 21 and Mplus v6.0 (Muthén & Muthén, 2010) softwares. The IBM SPPS Statistic 21 software was used for calculating the Cronbach’s alpha, as representative of internal consistency, and correlation values between the measures of the CES-A and the external variables considered in the current work, as representative of validity evidence in relation to other variables. Moreover, this software was also used for computing descriptive measures considering the sum of the items composing each measure of the CES-A, as these are the mostly commonly used in clinical/applied settings.

Mplus was used for confirmatory factor analyses (CFAs), as well as for multigroup analyses and latent mean comparisons. CFAs were applied to test the competing measurement models proposed either for the complete version or for the short version of the CES-A; results from the combined sample (n = 1,490) were used for these analyses. After the best internal structure solution was defined, CFA was further applied separately to the combined male and female samples and to each one of the male samples (i.e., community, referred, and detained).

The fit of the models resulting from these CFAs was considered on the basis of a two-index approach. Following the guidelines provided by Hair, Black, Babin, and Anderson (2005), the acceptability thresholds were different according to the number of items and sample sizes under consideration. When analyzing the complete version of the CES-A (i.e., between 12 and 30 items), and using the combined and community samples (sample size >250), we considered a comparative fit index (CFI) higher than .92, combined with either a root mean square error of approximation (RMSEA) lower than .07, or a standardized root mean square residual (SRMR) lower than .08. For the referred and detained samples (n <250), we considered CFI >.95, combined with either an RMSEA <.08 or an SRMR <.08. When analyzing the short version of the CES-A (i.e., less than 12 items) with the combined or community samples, acceptability was based on CFI higher than .97, combined with RMSEA lower than .07; with the referred or detained samples, acceptability was based on CFI higher than .97, combined with RMSEA lower than .08. Additionally, when comparing competing nonnested models, the Akaike’s information criterion (AIC) value was considered; lower AIC values were considered as indicative of better fit.

Multigroup analyses were applied to gender in the combined sample and to the male community, referred, and detained groups. A forward approach as suggested by Dimitrov (2010) was used, which suggests starting with testing configural invariance, followed by metric invariance, and, lastly, scalar invariance. Configural invariance refers to the measurement model being an adequate fit for each group separately. Metric invariance adds to this the constraint that item loading values also are of similar magnitude for the groups being compared. Scalar invariance, in turn, also constrains the intercepts of the items to be similar across groups. At least partial invariance should be obtained before groups can be compared based on the factor variables, meaning that only about 20% of the parameters being estimated should be variant across groups (Dimitrov, 2010). To determine measurement invariance, differences on the same fit indicators (i.e., RMSEA, CFI, and SRMR) were considered. Following the guidelines provided by Chen (2007), metric invariance was determined based on ΔCFI ≤−.01, combined with ΔRMSEA ≤.015 or ΔSRMR ≤.03, whereas scalar invariance was determined based on ΔCFI ≤−.01, combined with ΔRMSEA ≤.015 or ΔSRMR ≤.01.

Following at least partial invariance being achieved, latent mean comparisons were conducted across gender and across different male samples. The guidelines provided by Dimitrov (2006) were followed, by setting one group as the reference group against which the other groups’ latent mean would be compared.

Results

Preliminary analysis on the multivariate normality of the data indicated that the distribution was not multivariate normal (Mardia’s multivariate normality test = 5984.72, p < .001; Korkmaz, Goksuluk, & Zararsiz, 2014). So, all confirmatory and multigroup analyses were conducted using the maximum likelihood robust estimator.

Centrality of Event for Adolescents–Complete Version

Factor Structure

Taking into account that a one-factor solution had been previously validated for a Portuguese adolescent sample, we conducted a CFA on this measurement model. It did not achieve an acceptable fit for the data taken from the combined sample (see Table 2; AIC = 83376.138). A CFA was next applied to the three-factor model proposed by Gauer et al. (2013). It also did not achieve acceptable fit indicators (see Table 2; AIC = 82826.808). Finally, a third CFA was conducted on a three-factor model that reflected the diverse contents that Berntsen and Rubin (2006) refer to have intended to address when developing the items of the CES; Item 11 is not mentioned in this regard and so was not included in this analysis. Once again, the fit indicators were below the acceptability cutoff values (see Table 2; AIC = 78690.187).

Table 2.

Fit Indicators for CFA and Multigroup Configural Invariance Analyses of the CES-A by Samples.

	RMSEA	95% CI for RMSEA	CFI	SRMR
Complete version of the CES-A
One-factor model^a
Combined sample	.074	[.071, .078]	.881	.052
One-factor model^b
Combined sample	.065	[.061, .068]	.911	.050
19-Item three-factor model^c
Combined sample	.067	[.063, .070]	.911	.047
18-Item three-factor model^c
Combined sample	.063	[.059, .067]	.926	.041
Male participants	.059	[.053, .064]	.933	.040
Female participants	.071	[0.66, .077]	.911	.046
Community male sample	.069	[.062, .077]	.915	.047
Referred male sample	.065	[.045, .083]	.927	.050
Detained male sample	.061	[.048, .074]	.925	.053
Short version of the CES-A
One-factor model^a
Combined sample	.088	[.076, .100]	.949	.039
One-factor model^b	.090	[.074, .106]	.944	.042
Combined sample	.063	[.053, .075]	.974	.027
Male participants	.064	[.047, .081]	.973	.030
Female participants	.059	[.042, .078]	.979	.026
Community male sample	.063	[.041, .087]	.974	.030
Referred male sample	.000	[.000, .086]	1.000	.034
Detained male sample	.073	[.036, .110]	.964	.039
Three-factor model^c	.077	[.064, .091]	.969	.030

Note. CES-A = Centrality of Event for Adolescents; CFA = confirmatory factor analysis; RMSEA = root mean square error of approximation; CFI = comparative fit index; CI = confidence interval; SRMR = standardized root mean square residual.

Refers to the measurement models suggested by Berntsen and Rubin (2006). ^bRefers to the measurement models suggested by Gauer et al. (2013). ^cRefers to the measurement models proposed based on item content.

Of the three internal structures tested for the complete version of the CES-A, the three-factor model based on item content presented the best fit indicators and lower AIC value, and therefore seemed the best choice for further improvement. For this three-factor measurement model to achieve acceptable fit, modification indices suggested allowing the residuals of Items 1 and 2 to covariate. We could not find an ad hoc reason for allowing this correlation other than the items belonging to the same factor. So, rather than generalizing that reason to allowing errors of all items comprising that factor to covariate, we opted to exclude one of these items at a time. Excluding Item 2 produced acceptable fit indicators (see Table 2; AIC = 74570.148), and so this 18-item three-factor model will henceforth be considered when testing for measurement invariance, latent mean comparisons, and correlations with external variables.

According to the authors of the CES (Berntsen & Rubin, 2006), the first factor addresses whether the event became a reference point for the generation of expectations and attribution of meaning to other events in the person’s life; the second factor refers to whether the event is regarded as a turning point in one’s life story; and the third factor considers whether the event is perceived as a central component of personal identity. Table 3 presents the loading values for each item and the internal consistency values for each factor, which were always very good.

Table 3.

Loading and Internal Consistency Values for the Short One-Factor Model of the Centrality of Event Scale—Adolescent by Samples.

F1: Reference point		Combined sample	Community male sample	Referred male sample	Detained male sample
F1: Reference point		α = .87	α = .87	α = .86	α = .86
1	This event became a reference point ( . . . )	.58	.55	.61	.44
4	This event can be seen as a symbol ( . . . )	.71	.71	.76	.67
9	I often see connections ( . . . )	.60	.60	.59	.43
12	This event has colored the way I think ( . . . )	.73	.72	.77	.71
13	This event became a reference point ( . . . )	.72	.72	.59	.65
17	I think of the effects of this event ( . . . )	.78	.77	.72	.75
20	When I reflect on my future ( . . . )	.76	.75	.67	.76
F2: Turning point		α = .89	α = .89	α = .86	α = .87
10	I feel this event became central ( . . . )	.78	74	.75	.76
14	If I were to weave a carpet of my life ( . . . )	.76	.76	.66	.68
15	My life story can be divided in two ( . . . )	.79	.81	.79	.71
16	This event ( . . . ) changed my life	.81	.79	.76	.82
18	This event was a turning point ( . . . )	.82	.81	.76	.81
F3: Personal identity		α = .85	α = .85	α = .87	α = .79
3	I feel this event became part of my identity	.72	.73	.78	.61
5	This event is making my life different ( . . . )	.72	.74	.79	.63
6	This event became a reference point ( . . . )	.73	.73	.80	.58
7	People who have not experienced this ( . . . )	.57	.57	.49	.57
8	This event tells a ( . . . ) who I am	.70	.69	.75	.65
19	If this had not happened to me ( . . . )	.74	.71	.74	.70

Note. All loading values were significant at p < .001. Short paraphrases of the items are presented; for the full version of the items, see Appendix A of Berntsen and Rubin’s (2006) study. Loading values presented for male and female samples are taken from the configural invariance analysis.

Measurement Invariance

Fit indicators obtained for the male and female combined samples separately were acceptable only for the male sample (see Table 2). Thus, configural invariance across gender could not be achieved and further measurement invariance analyses were not performed. On the other hand, fit indicators obtained for the male participants in each sample were always acceptable (see Table 2), and so further metric and scalar invariance across samples were tested. When applying the 18-item three-factor measurement model to community, referred, and detained male participants, we found full metric (ΔRMSEA = −.001; ΔCFI = −.003, ΔSRMR = .009) and full scalar invariance (ΔRMSEA = .001; ΔCFI = −.007, ΔSRMR = .002).

Latent Mean Comparisons

Descriptive measures for the 18-item three-factor measurement models of the CES-A are reported in Table 4. We further computed latent mean comparisons between boys from the different samples. Community boys scored significantly less on all dimensions of the CES-A, when compared with referred boys (latent mean for reference point = .303, p = .007; latent mean for turning point = .612, p < .001; latent mean for personal identity = .292, p = .009) and detained boys (latent mean for reference point = .422, p < .001; latent mean for turning point = .583, p < .001; latent mean for personal identity = .238, p = .014). No significant differences were found between referred and detained participants for any of the three factors.

Table 4.

Descriptive Measures for the Three-Factor and Short One-Factor Model of the Centrality of Event Scale–Adolescent by Samples.

	18-Item three-factor model						7-Item one-factor model
	Reference point		Turning point		Personal identity		7-Item one-factor model
	M	SD	M	SD	M	SD	M	SD
Combined sample	18.79	6.59	11.49	5.43	15.60	5.86	16.99	7.05
Male	18.68	6.46	11.82	5.41	15.77	5.85	17.21	7.02
Female	18.86	6.76	11.13	5.44	15.41	5.89	16.76	7.09
Community male sample	17.82	6.34	10.55	4.89	15.13	5.71	15.87	6.51
Referred male sample	19.46	6.42	13.64	5.49	16.97	6.17	19.10	7.25
Detained male sample	20.17	6.43	13.62	5.67	16.52	5.80	19.16	7.30

Construct Validity in Relation to Other Variables

Positive and significant correlation values were found between the 18-item three-factor measurement model of the CES-A and experiential avoidance, the perception of an inadequate self and the perception of a hated self. Alternatively, negative and significant correlation values were found between this measurement model of the CES-A and the perception of a tranquilizing self (see Table 5).

Table 5.

Correlation Values Between the CES-A and the AAQ-II and FSCRS.

	18-Item three-factor model			7-Item one-factor model
	Reference point	Turning point	Personal identity	7-Item one-factor model
AAQ-II	.57***	.43***	.52***	.52***
FSCRS
Inadequate Self	.40***	.27***	.38***	.35***
Hated Self	.37***	.38***	.36***	.40***
Reassure Self	−.18**	−.19**	−.16*	−.20**

Note. CES-A = Centrality of Event Scale–Adolescents; AAQ-II = 7-Item Acceptance and Action Questionnaire; FSCRS = Forms of Self-Criticising/Attacking & Self-Reassuring Scale.

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

The magnitude of the correlations was similar across all CES-A dimensions, being moderate for the associations between the CES-A and the experiential avoidance and low between the CES-A and perceptions of the self. All CES-A dimensions were also correlated significantly among each other (p < .001): r = .81, between reference point and turning point, r = .85, between reference point and personal identity, and r = .74, between turning point and personal identity.

Centrality of Event for Adolescents–Short Version

Factor Structure

A CFA was applied to the seven items that Berntsen and Rubin (2006) suggested as comprising a short version of the CES. It did not achieve acceptable fit indicators (see Table 2; AIC = 29712.818). Gauer et al. (2013) also suggested a short version of the CES, which proved to be an acceptable fit to the combined sample used in the current work (see Table 2; AIC = 29415.849). Last, a three-factor measurement model allocating each of the seven items selected by Berntsen and Rubin (2006) to their intended content was also tested via CFA. It did not obtain acceptable fit indicators (see Table 2; AIC = 29606.338). So, the short version of the CES as proposed by Gauer et al. (2013) achieved the best fit when applied to the current adolescent sample and was subsequently used for measurement invariance, latent mean comparisons, and correlational analyses. The loading values for this measurement model as applied to Portuguese adolescents are presented in Table 6, along with the internal consistency values, which were always very good.

Table 6.

Loading and Internal Consistency Values for the Short One-Factor Model of the CES-A by Samples.

		Combined sample			Community male sample	Referred male sample	Detained male sample
		Total	Male	Female	Community male sample	Referred male sample	Detained male sample
		α = .90	α = .89	α = .91	α = .89	α = .89	α = .87
6	This event became a reference point ( . . . )	.62	.60	.64	.65	.76	.46
10	I feel this event became central ( . . . )	.78	.75	.80	.73	.76	.73
12	This event has colored the way I think ( . . . )	.68	.67	.71	.66	.72	.66
14	If I were to weave a carpet of my life ( . . . )	.75	.72	.79	.74	.67	.67
16	This event ( . . . ) changed my life	.81	.82	.80	.81	.78	.83
17	I think of the effects of this event ( . . . )	.79	.79	.81	.79	.71	.78
18	This event was a turning point ( . . . )	.80	.80	.80	.78	.78	.82

Note. CES-A = Centrality of Event Scale–Adolescents. All loading values were significant at p < .001. Short paraphrases of the items are presented; for the full version of the items, see Appendix A of Berntsen and Rubin’s (2006) work. Loading values presented for male and female samples are taken from the scalar invariance analysis.

Measurement Invariance

Configural gender invariance was established based on the optimal fit indicators obtained for the combined sample of male participants and for the combined sample of female participants (see Table 2 for fit indicators and Table 6 for loading and internal consistency values). Full metric invariance across gender was also found (ΔRMSEA = −.004; ΔCFI = −.002, ΔSRMR = .005), as well as full scalar invariance (ΔRMSEA = .005; ΔCFI = −.009, ΔSRMR = .004).

Concerning the diverse male samples, each of them separately achieved very good or good fit indicators (see Table 2 for fit indicators and Table 6 for loading and internal consistency values) and so we continued with metric and scalar invariance analyses. Full metric invariance was found (ΔRMSEA = −.002; ΔCFI = −.009, ΔSRMR = .026). Partial scalar invariance was also found after allowing the intercept of item nine to vary in the community group (ΔRMSEA = −.001; ΔCFI = −.008, ΔSRMR = .001).

Latent Mean Comparisons

Descriptive measures for the short one-factor measurement model of the CES-A are reported in Table 4. Based on latent mean comparisons, no significant between-gender differences were found (latent mean for girls = .019; p = .74). Also based on latent mean comparisons between male samples, community boys were found to score significantly less than referred (latent mean = .527, p < .001) and detained boys (latent mean = .519, p < .001). No significant differences were found between referred and detained boys.

Construct Validity in Relation to Other Variables

Correlation values found between the short version of the CES-A and the external variables under consideration were consistent with those found for every dimension of the 18-item three-factor model of the CES-A (see Table 5). The short CES-A also correlated highly and significantly (p < .001) with those three dimensions: r = .91, with reference point, r = .95, with turning point, and r = .86, with personal identity.

Discussion

The main goal of this study was to assess the psychometric properties of the CES using a large and diverse sample of adolescents. Particularly, the internal structure of the complete and short versions of the scale was investigated, along with analyses of the internal consistency, measurement invariance and latent mean comparison across gender and youth with behavioral problems to different degrees, and validity evidence in relation to others variables, namely self-criticism and self-reassurance and experiential avoidance. We intended to further contribute to the validation of the CES-A by examining alternative models to its internal structure and ascertaining their applicability to diverse adolescent samples (i.e., community and referred boys and girls and detained boys).

Concerning the internal structure of the complete version of the CES-A, we found evidence for a three-factor theoretically derived measurement model as the best solution for the data taken from the combined sample. This measurement model was directly taken from the contents of the original dimensions proposed by Berntsen and Rubin (2006), thus, providing evidence for the construct validity of the Portuguese version of the CES-A with Portuguese adolescents, based on internal structure. These contents are the event representing a reference point for one’s life, having become a turning point for one’s life story, and having become part of the personal identity. Alternatively, a one-factor solution, though previously advocated either with adults (Berntsen & Rubin, 2006) and adolescents (Cunha et al., 2015), did not achieve acceptable fit for this sample. The three-factor statistically derived solution proposed by Gauer et al. (2013) using an undergraduate sample was also not a good fit for this sample; though proposed as addressing the same contents, the constitution of these three factors did not fully overlap with the constructs Berntsen and Rubin (2006) stated to have had in mind when phrasing the items.

The theoretically derived measurement model was not invariant across gender. In fact, it only represented a good fit for the male participants, indicating that boys may more easily split the negative effect of a stressful experience in each of its components. On the other hand, girls may have more difficulties in marking out the different effects that traumatic events may have had on their life stories. This may explain why girls, although tending to have more complex and enriched cognitive representations of emotional events (Berntsen & Rubin, 2006; Boals, 2010; Davis, 1999), are more vulnerable when experiencing a traumatic/stressful event compared with boys (Berntsen & Rubin, 2006; Bloise & Johnson, 2007; Boals, 2010; MacMillan et al., 2014). Future research should address this issue in trying to understand the role of the three functions of memories of a traumatic event, assessed with the CES, for both males and females across the life span. For instance, it seems relevant to investigate if the vulnerability of females after experiencing a traumatic event is somehow related to the fact that they become overwhelmed in all functions of memories, whereas males more easily differentiate the harm related to a traumatic event in each of these functions and, thus, present diminished trauma vulnerability.

The CES-A had not been examined with particular groups, namely adolescents with disruptive behaviors. Our findings with two samples of youth with externalizing symptoms (one referred and one detained) provides evidence for the accuracy and invariability of a three-factor measurement model for the complete version of the CES-A to be used with such samples and with community adolescent samples. Referred and detained boys reported similar perceptions on the traumatic effect of the event, and both reported them more intensely than boys from the community. This is in line with the literature, which reports a higher prevalence (frequency and diversity) of trauma exposure in referred and detained samples, when compared with community samples (Abram et al., 2004; Briggs et al., 2013; Dierkhising et al., 2013; Willis et al., 2013). Nonetheless, these differences were not as pronounced for the personal identity factor as for the reference point and the turning point factors. This may be due to the fact that adolescents are still developing their sense of identity (Kroger, 2004; McLean et al., 2010).

Concerning the short version of the CES-A, a one-factor solution, as proposed by Gauer et al. (2013), was the best fitting model to the data taken from the combined sample in comparison with a previously proposed one-factor solution (Berntsen & Rubin, 2006) and to a three-factor theoretically derived solution. The short version of the CES-A had not previously been investigated and these results may represent the fact that the items included in this version of the scale are not sufficient (either because of their content or because of their number) to discriminate between related constructs. Given their particular intense emotional experience (Zeman et al., 2006) combined with restricted repertory of emotion regulation strategies (Garnefski & Kraaij, 2006; McRae et al., 2012; Zimmermann & Iwanski, 2014), adolescents may need more incisive items to be able to adequately discriminate their experience of emotionally intense traumatic experiences.

This short version of the CES-A proved to be gender invariant, allowing for gender comparisons. No significant gender differences were found. These results are contrary to the ones found by Boals (2010), where adult females were found to have a greater tendency to construe negative events as central to their identity when compared with adult males, but are in line with previous findings with Portuguese adolescents (Cunha et al., 2015). This may be explained by sample features: adolescents versus adults. The emotional experience of adolescents is especially intense, with adolescents more frequently experiencing intense emotions, more so than younger or older individuals (Kroger, 2004; Zeman et al., 2006). Moreover, their ability to use cognitive reappraisal and other cognitive strategies to regulate emotions increases linearly with age, with a general trend of increasing adaptive emotion regulation in adulthood (Garnefski & Kraaij, 2006; McRae et al., 2012; Zimmermann & Iwanski, 2014). These differences in emotional experience and emotional regulation may help understand our findings on adolescent gender comparisons. Future research should address this issue in trying to understand the emotional development from infancy through adolescence and adulthood, considering also gender. Specifically, it would be of major interest to test if the construction of a negative event as central to one’s identity differs throughout the life span and if these differences are consistent for males and females in each developmental period.

Overall, our findings suggested that a three-factor model might be considered when comparing diverse samples based on externalizing characteristics, whereas a short version of the CES-A might be a better option when intending to also analyze gender differences across diverse samples. We did not consider other sociodemographic or clinical variables that may further affect the measurement models of the CES-A. Developmental stage could be one of them, once it is known that childhood trauma is considered to be one of the strongest risk factors for psychopathology and could predict deficits in emotion regulation (Marusak, Martin, Etkin, & Thomason, 2015). However, this could be compensated by the fact that emotion regulation strategies (Garnefski & Kraaij, 2006; McRae et al., 2012; Zimmermann & Iwanski, 2014) and the integration of memories into one’s personal identity (McLean et al., 2010) tend to increase linearly with age. These nuances could help understand age differences in the effects of a traumatic event in autobiographical memory development. As a relevant clinical variable, the age at the time of the trauma exposure seems to be particularly important, as well as the age of the assessment of effects caused by the traumatic event (McCutcheon et al., 2010; McLean et al., 2010).

Both the short version of the CES-A and the three factors of its longer version (i.e., 18 items) showed validity in relation to other external variables. In particular, it seems that the more a traumatic event has become central in one’s life (both generally or, alternatively, becoming a reference and turning point in one’s life crucial to defining the personal identity), the more the individual engages in experiential avoidance and self-criticism. In contrast, the less the centrality of this event, the more he or she reassures himself or herself when coping with daily experiences. This is in line with findings that a traumatic memory, particularly a shaming one, can become central to one’s identity, placing the individual (adolescent or adult) at risk for psychopathology (Cunha, Matos, Faria, & Zagalo, 2012; Gilbert, Baldwin, Irons, Baccus, & Palmer, 2006; Matos & Pinto-Gouveia, 2010; Pinto-Gouveia & Matos, 2011).

These associations between the centrality of a traumatic event and psychopathology may be mediated (at least partially) by the adoption of unsuccessful ways of coping with daily hardships. The higher the prevalence of traumatic experiences in one’s life story, the more individuals tend to cope with them by avoiding the integration of their effects into their identity (Rubin et al., 2014; Schuettler & Boals, 2011) and possibly by adopting a self-critical posture toward themselves, instead of a self-reassuring one. This may represent a way of self-protection, though only partially successful, because it prevents the individual of working through and psychologically resolve these traumatic experiences. Alternatively, coping with the event and oneself in a more compassionate way may represent a healthier alternative (Gilbert et al., 2006).

This study is not without limitations, namely the fact that it included only detained boys. Gender comparisons may have been biased by the fact that only the male group included participants who are known to have had a more traumatic life story (Abram et al., 2004). Besides, we can only advocate for the representativeness of the current community adolescent sample but not of the referred (due to it being gathered only in one region of Portugal) or the detained (due to it including only boys) samples; the detained sample stands, nevertheless, representative of the male detained Portuguese youth and it should also be noted that girls represent only 14% of the youth detained population in Portugal (Direção-Geral de Reinserção e Serviços Prisionais, 2014). Though the results presented in this work may, thus, be seen as relatively circumscribed to the Portuguese context, they still represent a valuable stepping-stone for further works investigating larger, more diverse, and more representative samples.

Overall, this study contributed to validating the CES-A for use with adolescents, either with or without behavioral problems. Due to the fact that adolescents with behavioral problems are known to having been exposed to a greater prevalence of traumatic experiences (Abram et al., 2004; Briggs et al., 2013; Dierkhising et al., 2013; Willis et al., 2013), such as shameful ones, it is important to develop credible and valid instruments that may accurately address these experiences and their impact in the adolescents’ life; such had not been done previously with the CES-A. So, the current study thoroughly explored alternative measurement models of the CES-A in a diverse adolescent sample, both in its complete and short versions. We offer two alternatives (i.e., a three-factor measurement model for the complete version of the CES-A and a one-factor measurement model for the short version of the CES-A). One allows for a more thorough report on the impact of a significant life event onto one’s life story, as given by adolescents both exhibiting or not exhibiting externalizing problems; the other provides a faster but more generic assessment of the significance of such an event by which boys and girls can be compared, as well as adolescents presenting diverse degrees of behavioral problems. It will be up to the researcher or clinician to choose at each time which version of the instrument best suits his or her goals.

Footnotes

Acknowledgements

The authors would like to thank psychologists from foster care and juvenile detention facilities for collaborating in this research, as well as Marta Capinha and Maria do Natal Sousa for helping in the assessment in the Azores Island. The authors would also like to thank the several undergraduate students at the University of Coimbra who contributed to data collection and Ana Luísa Rosa for proofreading the article.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article is part of the project “Changeability of Psychopathic Traits in Young Offenders: Outcomes From a Compassion-Based Psychotherapeutic Intervention” (PTDC/MHC-PCL/2189/2014), sponsored by the Portuguese Foundation for Science and Technology and the European Social Fund (POPH/FSE).

References

Abram

K. M.

Teplin

L. A.

Charles

D. R.

Longworth

S. L.

McClelland

G. M.

Dulcan

M. K.

(2004). Posttraumatic stress disorder and trauma in youth in juvenile detention. Archives of General Psychiatry, 61, 403-410.

Berntsen

Rubin

D. C.

(2006). The Centrality of Event Scale: A measure of integrating a trauma into one’s identity and its relation to post-traumatic stress disorder symptoms. Behaviour Research and Therapy, 44, 219-231.

Berntsen

Rubin

D. C.

(2007). When a trauma becomes a key to identity: Enhanced integration of trauma memories predicts posttraumatic stress disorder symptoms. Applied Cognitive Psychology, 21, 417-431.

Bloise

S. M.

Johnson

M. K.

(2007). Memory for emotional and neutral information: Gender and individual differences in emotional sensitivity. Memory, 15, 192-204.

Boals

(2010). Events that have become central to identity: Gender differences in the centrality of events scale for positive and negative events. Applied Cognitive Psychology, 24, 107-121.

Bond

F. W.

Hayes

S. C.

Baer

R. A.

Carpenter

K. M.

Guenole

Orcutt

H. K.

. . . Zettle

R. D.

(2011). Preliminary psychometric properties of the Acceptance and Action Questionnaire–II: A revised measure of psychological inflexibility and experiential avoidance. Behavior Therapy, 42, 676-688.

Briggs

E. C.

Fairbank

J. A.

Greeson

J. K. P.

Steinberg

A. M.

Amaya-Jackson

L. M.

Ostrowski

S. A.

. . . Pynoos

R. S.

(2013). Links between child and adolescent trauma exposure and service use histories in a national clinic-referred sample. Psychological Trauma, 5, 101-109.

Castilho

Pinto-Gouveia

(2011). Auto-Criticismo: Estudo de validação da versão portuguesa da Escala das Formas do Auto-Criticismo e Auto-Tranquilização (FSCRS) e da Escala das Funções do Auto-Criticismo e Auto-Ataque (FSCS) [Self-criticism: Validation study of the Portuguese version of The Forms of Self-Criticising/Attacking & Self-Reassuring Scale (FSCRS) and Functions of Self-Criticizing Scale (FSCS)]. Psychologica, 54, 63-86.

Chen

F. F.

(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling, 14, 464-504.

10.

Cunha

Matos

Faria

Zagalo

(2012). Shame memories and psychopathology in adolescence: The mediator effect of shame. International Journal of Psychology and Psychological Therapy, 12, 203-218.

11.

Cunha

Xavier

Matos

Faria

(2015). O impacto das memórias de vergonha na adolescência: A Escala da Centralidade do Acontecimento (CES) [The impact of shame memories in adolescence: The Centrality of Event Scale (CES)]. Análise Psicológica, 3, 1-14.

12.

Davis

P. J.

(1999). Gender differences in autobiographical memory for childhood emotional experiences. Journal of Personality and Social Psychology, 76, 498-510.

13.

Dierkhising

C. B.

S. J.

Woods-Jaeger

Briggs

E. C.

Lee

Pynoos

R. S.

(2013). Trauma histories among justice-involved youth: Findings from the National Child Traumatic Stress Network. European Journal of Psychotraumatology, 4, 1-12.

14.

Dimitrov

D. M.

(2006). Comparing groups on latent variables: A structural equation modeling approach. Work, 26, 429-436.

15.

Dimitrov

D. M.

(2010). Testing for factorial invariance in the context of construct validation. Measurement and Evaluation in Counseling and Development, 43, 121-149.

16.

Direção-Geral de Reinserção e Serviços Prisionais. (2014). Relatório estatístico anual [Annual statistics report]. Portugal: Ministério da Justiça, Governo de Portugal.

17.

Garnefski

Kraaij

(2006). Relationships between cognitive emotion regulation strategies and depressive symptoms: A comparative study of five specific samples. Personality and Individual Differences, 40, 1659-1669.

18.

Gauer

Souza

J. A.

Silveira

A. M.

Sediyama

C. Y. N.

(2013). Stressful events in autobiographical memory processing: Brazilian version of the Centrality of Event Scale. Psicologia, Reflexão e Critica, 26, 98-105.

19.

Gilbert

Baldwin

M. W.

Irons

Baccus

J. R.

Palmer

(2006). Self-criticism and self-warmth: An imagery study exploring their relation to depression. Journal of Cognitive Psychotherapy, 20, 183-200.

20.

Gilbert

Clark

Hempel

Miles

J. N. V.

Irons

(2004). Criticising and reassuring oneself: An exploration of forms, styles and reasons in female students. British Journal of Clinical Psychology, 43, 31-50.

21.

Groleau

J. M.

Calhoun

L. G.

Cann

Tedeschi

R. G.

(2013). The role of centrality of events in posttraumatic distress and posttraumatic growth. Psychological Trauma, 5, 477-483.

22.

Hair

J. F.

Jr. Black

W. C.

Babin

B. J.

Anderson

R. E.

(2005). Multivariate data analysis (7th ed.). Upper Saddle River, NJ: Prentice Hall.

23.

Instituto Nacional de Estatística. (2011). Classificação Portuguesa das profissões 2010 [2010 Portuguese classification of professions]. Lisboa, Portugal: Instituto Nacional de Estatística.

24.

Korkmaz

Goksuluk

Zararsiz

(2014). An R package for assessing multivariate normality. R Journal, 6, 151-162.

25.

Kroger

(2004). Identity in adolescence: The balance between self and other (3rd ed.). London, England: Routledge.

26.

MacMillan

H. L.

Fleming

J. E.

Streiner

D. L.

Lin

Boyle

M. H.

Jamieson

. . . Beardslee

W. R.

(2014). Childhood abuse and lifetime psychopathology in a community sample. American Journal of Psychiatry, 158, 1878-1883.

27.

Marusak

H. A.

Martin

K. R.

Etkin

Thomason

M. E.

(2015). Childhood trauma exposure disrupts the automatic regulation of emotional processing. Neuropsychopharmacology, 40, 1250-1258.

28.

Matos

Pinto-Gouveia

(2010). Shame as a traumatic memory. Clinical Psychology & Psychotherapy, 17, 299-312.

29.

McCutcheon

V. V.

Sartor

C. E.

Pommer

N. E.

Bucholz

K. K.

Nelson

E. C.

Madden

P. A. F.

Heath

A. C.

(2010). Age at trauma exposure and PTSD risk in a young adult female sample. Journal of Traumatic Stress, 23, 811-814.

30.

McLean

K. C.

Breen

A. V.

Fournier

M. A.

(2010). Constructing the self in early, middle, and late adolescent boys: Narrative identity, individuation, and well-being. Journal of Research on Adolescence, 20, 166-187.

31.

McRae

Gross

J. J.

Weber

Robertson

E. R.

Sokol-Hessner

Ray

R. D.

. . . Ochsner

K. N.

(2012). The development of emotion regulation: An fMRI study of cognitive reappraisal in children, adolescents and young adults. Social Cognitive and Affective Neuroscience, 7, 11-22.

32.

Muthén

L. K.

Muthén

B. O.

(2010). Mplus user’s guide (6th ed.). Los Angeles, CA: Author.

33.

Pinto-Gouveia

Gregório

Dinis

Xavier

(2012). Experiential avoidance in clinical and non-clinical samples: AAQ-II Portuguese version. International Journal of Psychology and Psychological Therapy, 12, 139-156.

34.

Pinto-Gouveia

Matos

(2011). Can shame memories become a key to identity? The centrality of shame memories predicts psychopathology. Applied Cognitive Psychology, 25, 281-290.

35.

Ribeiro

Silva

Rijo

Salekin

R. T.

(2015). The evolutionary roots of psychopathy. Aggression and Violent Behavior, 21, 85-96.

36.

Robinaugh

D. J.

McNally

R. J.

(2011). Trauma centrality and PTSD symptom severity in adult survivors of childhood sexual abuse. Journal of Traumatic Stress, 24, 483-486.

37.

Rubin

D. C.

Boals

Hoyle

R. H.

(2014). Narrative centrality and negative affectivity: Independent and interactive contributors to stress reactions. Journal of Experimental Psychology: General, 143, 1159-1170.

38.

Rubin

D. C.

Hoyle

R. H.

Leary

M. R.

(2012). Differential predictability of four dimensions of affect intensity. Cognition and Emotion, 26, 25-41.

39.

Schuettler

Boals

(2011). The path to posttraumatic growth versus PTSD: Contributions of event centrality and coping. Journal of Loss & Trauma, 16, 180-194.

40.

Willis

Best

Aalsma

(2013, March). Trauma experiences among detained youth: Predictors of recidivism. Paper presented at the 2013 Society for Adolescent Health and Medicine Annual Meeting, Atlanta, GA.

41.

Zeman

Cassano

Perry-Parrish

Stegall

(2006). Emotion regulation in children and adolescents. Journal of Developmental & Behavioral Pediatrics, 27, 155-168.

42.

Zimmermann

Iwanski

(2014). Emotion regulation from early adolescence to emerging adulthood and middle adulthood: Age differences, gender differences, and emotion-specific developmental variations. International Journal of Behavioral Development, 38, 182-194.