the distinctiveness of hardiness, positive emotionality, and negative emotionality in national guard...

5
Brief Report The distinctiveness of hardiness, positive emotionality, and negative emotionality in National Guard soldiers Christopher R. Erbes a,, Paul A. Arbisi b , Shannon M. Kehle c , Amanda G. Ferrier-Auerbach a , Robin A. Barry d , Melissa A. Polusny c a Minneapolis VA Medical Center, University of Minnesota Medical School, United States b Minneapolis VA Medical Center, Department of Psychology & University of Minnesota Medical School, University of Minnesota, United States c Center for Chronic Disease Outcome Research, Minneapolis VA Medical Center, University of Minnesota Medical School, United States d University of Maryland, Baltimore County, United States article info Article history: Available online 23 July 2011 Keywords: Hardiness Negative emotionality Positive emotionality Stress reaction PTSD Military Depression abstract Exploratory and confirmatory factor analyses were used to investigate the distinctiveness of hardiness (using the Short Hardiness Scale; Bartone, 1995) from the broader personality traits of negative emotion- ality and positive emotionality (NEM and PEM; assessed with items from the MMPI-2 PSY-5 scales); in a sample of 981 Army National Guard soldiers. Exploratory factor analyses demonstrated that hardiness items loaded on a separate factor from PEM and NEM items, and confirmatory factor analysis suggested that hardiness is not simply a sub-facet of either PEM or NEM. However, subsequent regression analyses found that hardiness did not predict symptoms of PTSD or depression beyond the effects of PEM and NEM among combat exposed soldiers. Published by Elsevier Inc. 1. Introduction Increasing national attention on the wellbeing of military ser- vice members deployed to Iraq and Afghanistan has led to a corre- sponding focus on predictors of healthy psychological functioning in the face of combat deployment stressors. Hardiness is one po- tential predictor of physical and emotional well-being in response to a range of stressors including military and combat situations (e.g., Bartone, 1999; King, King, Fairbank, Keane, & Adams, 1998). Hardiness has been conceptualized and measured as having three components: a commitment to tasks and challenges in life, a belief that one is able to control the outcomes of stressful events, and being open to change as a natural part of life (Kobasa, 1979). Research has shown that hardiness predicts measures of psycho- logical strain, including depression (Rhodewalt & Zone, 1989) and scales assessing clinical pathology on the Minnesota Multi- phasic Personality Inventory – 2 (MMPI-2; Maddi & Khoshaba, 1994; Maddi et al., 2002). Findings from military populations have also found that hardiness is related to lower levels of psychiatric symptoms including symptoms of posttraumatic stress disorder (PTSD; e.g., King et al., 1998) as well as greater perceptions of meaningful work (Bartone, 1999). Although some of the earlier concerns about the measurement of hardiness appears to have been addressed (Bartone, 1995; Funk & Houston, 1987; Maddi et al., 2002), the distinct role of hardiness in predicting mental health symptoms, beyond its association with other personality constructs, remains unclear. The present study represents an effort to position the construct of hardiness within the broader context of other well-established personality constructs as it relates to PTSD and depression in a military population. There has long been concern that hardiness may represent an expression of other higher order personality traits, including nega- tive emotionality (NEM) and positive emotionality (PEM). NEM (also referred to as neuroticism), reflects ‘‘dispositions toward neg- ative mood and emotion and a tendency towards adversarial inter- actions with others’’ (Miller, 2003, p. 375) while PEM (also referred to as extraversion) is an orthogonal personality dimension that can be defined as ‘‘the capacity to experience positive emotions and tendencies towards active involvement in the social and work environments’’ (Miller, 2003, p. 375). NEM and PEM have been found to correlate with both PTSD and depression in military and community populations (Miller, 2003). Given the conceptual and measurement overlap between hardiness and NEM and PEM, as well as the established relationships between these constructs and mental health outcomes, some have suggested that hardiness is simply a lower order expression of NEM (e.g., Funk & Houston, 1987) while others (e.g., Miller, 2003) have postulated that hardi- ness, and its relationship with psychopathology, may be more 0092-6566/$ - see front matter Published by Elsevier Inc. doi:10.1016/j.jrp.2011.07.001 Corresponding author. Address: Minneapolis VA Medical Center (116A6), Minneapolis, MN 55417, United States. E-mail address: [email protected] (C.R. Erbes). Journal of Research in Personality 45 (2011) 508–512 Contents lists available at ScienceDirect Journal of Research in Personality journal homepage: www.elsevier.com/locate/jrp

Upload: christopher-r-erbes

Post on 09-Sep-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Journal of Research in Personality 45 (2011) 508–512

Contents lists available at ScienceDirect

Journal of Research in Personality

journal homepage: www.elsevier .com/ locate/ j rp

Brief Report

The distinctiveness of hardiness, positive emotionality, and negative emotionalityin National Guard soldiers

Christopher R. Erbes a,⇑, Paul A. Arbisi b, Shannon M. Kehle c, Amanda G. Ferrier-Auerbach a, Robin A. Barry d,Melissa A. Polusny c

a Minneapolis VA Medical Center, University of Minnesota Medical School, United Statesb Minneapolis VA Medical Center, Department of Psychology & University of Minnesota Medical School, University of Minnesota, United Statesc Center for Chronic Disease Outcome Research, Minneapolis VA Medical Center, University of Minnesota Medical School, United Statesd University of Maryland, Baltimore County, United States

a r t i c l e i n f o a b s t r a c t

Article history:Available online 23 July 2011

Keywords:HardinessNegative emotionalityPositive emotionalityStress reactionPTSDMilitaryDepression

0092-6566/$ - see front matter Published by Elsevierdoi:10.1016/j.jrp.2011.07.001

⇑ Corresponding author. Address: Minneapolis VMinneapolis, MN 55417, United States.

E-mail address: [email protected] (C.R. Erb

Exploratory and confirmatory factor analyses were used to investigate the distinctiveness of hardiness(using the Short Hardiness Scale; Bartone, 1995) from the broader personality traits of negative emotion-ality and positive emotionality (NEM and PEM; assessed with items from the MMPI-2 PSY-5 scales); in asample of 981 Army National Guard soldiers. Exploratory factor analyses demonstrated that hardinessitems loaded on a separate factor from PEM and NEM items, and confirmatory factor analysis suggestedthat hardiness is not simply a sub-facet of either PEM or NEM. However, subsequent regression analysesfound that hardiness did not predict symptoms of PTSD or depression beyond the effects of PEM and NEMamong combat exposed soldiers.

Published by Elsevier Inc.

1. Introduction

Increasing national attention on the wellbeing of military ser-vice members deployed to Iraq and Afghanistan has led to a corre-sponding focus on predictors of healthy psychological functioningin the face of combat deployment stressors. Hardiness is one po-tential predictor of physical and emotional well-being in responseto a range of stressors including military and combat situations(e.g., Bartone, 1999; King, King, Fairbank, Keane, & Adams, 1998).Hardiness has been conceptualized and measured as having threecomponents: a commitment to tasks and challenges in life, a beliefthat one is able to control the outcomes of stressful events, andbeing open to change as a natural part of life (Kobasa, 1979).Research has shown that hardiness predicts measures of psycho-logical strain, including depression (Rhodewalt & Zone, 1989)and scales assessing clinical pathology on the Minnesota Multi-phasic Personality Inventory – 2 (MMPI-2; Maddi & Khoshaba,1994; Maddi et al., 2002). Findings from military populations havealso found that hardiness is related to lower levels of psychiatricsymptoms including symptoms of posttraumatic stress disorder(PTSD; e.g., King et al., 1998) as well as greater perceptions ofmeaningful work (Bartone, 1999). Although some of the earlier

Inc.

A Medical Center (116A6),

es).

concerns about the measurement of hardiness appears to havebeen addressed (Bartone, 1995; Funk & Houston, 1987; Maddiet al., 2002), the distinct role of hardiness in predicting mentalhealth symptoms, beyond its association with other personalityconstructs, remains unclear. The present study represents an effortto position the construct of hardiness within the broader context ofother well-established personality constructs as it relates to PTSDand depression in a military population.

There has long been concern that hardiness may represent anexpression of other higher order personality traits, including nega-tive emotionality (NEM) and positive emotionality (PEM). NEM(also referred to as neuroticism), reflects ‘‘dispositions toward neg-ative mood and emotion and a tendency towards adversarial inter-actions with others’’ (Miller, 2003, p. 375) while PEM (also referredto as extraversion) is an orthogonal personality dimension that canbe defined as ‘‘the capacity to experience positive emotions andtendencies towards active involvement in the social and workenvironments’’ (Miller, 2003, p. 375). NEM and PEM have beenfound to correlate with both PTSD and depression in military andcommunity populations (Miller, 2003). Given the conceptual andmeasurement overlap between hardiness and NEM and PEM, aswell as the established relationships between these constructsand mental health outcomes, some have suggested that hardinessis simply a lower order expression of NEM (e.g., Funk & Houston,1987) while others (e.g., Miller, 2003) have postulated that hardi-ness, and its relationship with psychopathology, may be more

C.R. Erbes et al. / Journal of Research in Personality 45 (2011) 508–512 509

parsimoniously accounted for by a combination of high PEM andlow NEM.

The few studies that have considered hardiness in relation toNEM and PEM have used relatively small samples, have not utilizedtrauma exposed samples, and have not assessed symptoms ofPTSD. Further, some studies are interpreted as supporting thedistinctiveness of hardiness (e.g., Maddi & Khoshaba, 1994; Maddiet al., 2002), while others are interpreted as failing to support thedistinction between hardiness and, in particular, NEM (e.g.,Rhodewalt & Zone, 1989). A recent meta-analysis by Eschleman,Bowling, and Alarcon (2010) concluded that hardiness providedincremental prediction of depression and poor mental health whencontrolling for either negative affectivity or other commonly iden-tified personality traits (i.e., five factor model traits). A closerexamination of their analysis, however, reveals that the studiesincluded in their analysis of depression and poor mental healthdid not include negative affectivity or five factor model traits,and that anxiety and PTSD were not considered at all. The onlystudy we are aware of that simultaneously considers hardinessand both PEM and NEM (Maddi et al., 2002) did not assess mentalhealth outcomes. Moreover, studies have not examined therelationship between hardiness and higher order personalityconstructs using confirmatory factor analysis, which can examinemultiple potential models of relationships.

The aim of the present study was to further examine the rela-tionship of hardiness to the well-established personalityconstructs of PEM and NEM, as measured by the MMPI-2 PSY-5scales. In order to evaluate hardiness in relation to these personal-ity constructs, we used exploratory and confirmatory factor analy-ses to test three competing models: one in which hardiness wasrelated to but distinct from positive and negative emotionality,one in which it was a facet of PEM, and another in which it wasa facet of NEM, in a large sample of National Guard soldiers. Wealso evaluated the ability of hardiness to predict symptoms ofdepression and PTSD over and above both PEM and NEM amongthe subset of soldiers in the sample who had been through combatdeployments to Iraq or Afghanistan.

2. Method

2.1. Participants

Participants in the study were Minnesota Army National Guardsoldiers who were either in training preparatory for a deploymentto Iraq (n = 746) or in annual training (n = 241). Soldiers recruitedpre-deployment were surveyed about 1 month prior to combatdeployments. A total of six participants from the pre-deploymentgroup did not provide usable responses for any of the variablesused in the present paper and were excluded from analyses. Thus,the total sample size for these analyses was 981 soldiers. Thesample was largely male (n = 842, 85.3%), Caucasian (n = 863,87.4%), and well educated (mean of 14.18 years of education,SD = 2.11), with a mean age of 30.19 (SD = 9.23). Almost half(n = 466, 47.2%) of the sample were married. The majority ofrespondents were enlisted rank (n = 843, 85.4%). A minority(n = 110, 11.1%) had already served on combat deployments to Iraqor Afghanistan.

1 In a subsample of 255 soldiers who completed the MMPI-Restructured Form(MMPI-2-RF) at a later phase of this study, correlations between the shortened scalesused here and the MMPI-2-RF PSY-5- R scales were .93 for PEM and .94 for NEM,which compares positively to the correlations between the full MMPI-2 PSY-5 scalesand the corresponding MMPI-2-RF PSY-5 R scales reported in the technical manual(Tellegen & Ben-Porath, 2008), thus supporting the validity of these shortened scalesused here.

2.2. Procedures

Soldiers were provided with a description and overview of thestudy and were informed that their participation in the studywas voluntary and confidential. After providing written informedconsent, soldiers completed a survey in group classrooms understandardized conditions. No compensation for participation was

provided. All procedures and materials were approved by the insti-tutional review boards at the Minneapolis Veterans Affairs MedicalCenter and University of Minnesota as well as the MinnesotaNational Guard command. In addition to demographic information,the following measures were collected.

2.3. Measures

Personality Psychopathology-Five (PSY-5) scales from the Minne-sota Multiphasic Personality Inventory-2 (MMPI-2) (Harkness,McNulty, & Ben-Porath, 1995). The PSY-5 scales correspond to fivehigher-order factors of personality: Introversion/low positiveemotionality (PEM), negative emotionality (NEM), Disconstraint,Aggressiveness, and Psychoticism. Due to the time limitations incollecting data from soldiers in training settings, shortenedversions of the PSY-5 scales were administered. Items on the short-ened PSY-5 scales were selected rationally by the second author toprovide a non-redundant and balanced representation of theconstructs assessed by each scale. Abbreviated versions of thePEM and NEM PSY-5 scales included 21 of the original 34 itemsfor the Introversion Scale and 23 of the original 33 items for thenegative emotionality scale. Item polarity for the items of theIntroversion Scale was reversed so that higher scores representedhigher levels of PEM. Internal consistency, as measured byCronbach’s a, was .59 for PEM and .84 for NEM. Alpha coefficientswere somewhat lower, but comparable to reliabilities for thecomplete scales (Harkness et al., 1995).1

Hardiness Scale (Bartone, 1995). The construct of hardiness wasmeasured using the Short Hardiness Scale. The scale includes 15positively and negatively keyed items that assess the threeproposed domains of hardiness: commitment to life’s tasks, controlover life’s outcomes, and a perception of challenge in dealing withambiguous life events. The short scale provides a total score repre-senting overall hardiness and has demonstrated acceptable inter-nal validity (Cronbach’s a = .80 for the present sample) andcriterion-related validity (Bartone, 1995, 1999).

PTSD Checklist – Civilian Version (Weathers, Utz, Herman,Hus-ka, & Keane, 1993). The PTSD Checklist contains 17 items thatcorrespond to the symptoms of PTSD and are rated on a 1–5 scale.It has demonstrated excellent internal consistency reliability(Cronbach’s a = .93 in the present sample) and concurrent and cri-terion validity (e.g., Weathers et al., 1993). The Civilian version ofthe PTSD Checklist was used because not all soldiers participatinghad been exposed to combat experiences at the time of the study.Scores of 50 or above are considered indicative of possible PTSD(Weathers et al., 1993).

Beck Depression Inventory – 2 (Beck, Steer, & Brown, 1996). Thiswidely used self-report measure contains 21 items assessingsymptoms of depression. The BDI-II has shown excellent internaland test–retest reliability (Cronbach’s a = .80 in this sample) andconcurrent and criterion validity. Scores of 19 or above indicatemoderate to severe depression (Beck et al., 1996).

3. Results

Missing data (ranging from .82% to 4.28% for parcels and totalscores) was accounted for using Full Information Maximum Likeli-hood Estimation (MLE) procedures in the MPlus software program.

Table 1Bivariate correlations.

Hardiness NEM PEM PTSD Depression

Hardiness .429*** .360*** �.367*** �.470***

NEM 1 .157*** �.618*** �.592***

PEM 1 �.207*** �.296***

PTSD 1 .743***

Depression 1

Notes: N = 913 (with listwise deletion).*** p < .001.

Table 2Exploratory factor analysis of hardiness, PEM and NEM parcels: three factor solution.

Parcel PEM NEM Hardiness

Hardiness 1 .81Hardiness 2 .61Hardiness 3 .77Hardiness 4 .70NEM 1 .73NEM 2 .70NEM 3 .77NEM 4 .77PEM 1 .59PEM 2 .28 .60PEM 3 .53PEM 4 .51

Note: Principal factor analyses with promax rotation. Factor loadings above .40 arebolded and loadings below .20 are omitted. PEM = positive emotionality,NEM = negative emotionality.

2 An additional EFA of individual items instead of parcels, using Weighted LeastSquares with Mean and Variance Adjustments (WLSMV), also found that hardinessitems did not cross-load with NEM and PEM items.

510 C.R. Erbes et al. / Journal of Research in Personality 45 (2011) 508–512

In regression and correlation analyses in SPSS, missing data wasaddressed with list-wise deletion.

Means (and standard deviations) for variables of interest (rawscores) were as follows: NEM (17.33, SD = 14.35), PEM (14.17,SD = 2.75), hardiness (43.48, SD = 5.51), PTSD symptoms (26.26,SD = 10.41) and depressive symptoms (6.00, SD = 6.95). The major-ity of the sample did not exceed suggested cut-offs for possiblePTSD (n = 49, 5%) or depression (n = 59, 6.3%). Rates of screeningpositive for PTSD were higher among soldiers with prior combatdeployments (4.6% nondeployed vs. 9.3% deployed, v2 = 4.50,p = .034) though the same was not true for screening for depression(6.3% nondeployed vs. 5.8% deployed, v2 = .06, p = .819). Correla-tions between study variables are listed in Table 1.

3.1. Relationship of Hardiness with PEM and NEM

The relationship of hardiness with PEM and NEM was evaluatedwith exploratory (EFA) and confirmatory (CFA) factor analysesusing M-Plus software. Factor analyses were conducted using itemparcels for hardiness, PEM, and NEM which increase the reliabilityof observed variables. For each construct, parcels were created bysequentially assigning items to one of four parcels in the order ofthe magnitude of each item’s mean correlation with all other itemsfor the construct. Thus, items of high, moderate, and low inter-itemcorrelation were equally distributed across parcels. Item parcelcomposition is available from the first author upon request. Forfactor analyses, participants were randomly assigned to one oftwo samples, an Exploratory Sample and a Confirmatory Sample.The two samples did not differ statistically on any of the variablesused in analyses.

We first conducted EFA using the Exploratory Sample to exam-ine whether or not parcels for hardiness split between PEM andNEM. EFA were conducted using principal factor analyses with Pro-max rotation. Scree plot and eigenvalues were used to determinethe number of factors to be retained. We found that three factorshad eigenvalues equal to or greater than 1.0. Thus structures withone through three factors were examined. A one factor structureexplained 33.35% of the variance in the data and the hardinessand NEM parcels and one of the PEM parcels loaded above .40 onthe factor. One other PEM parcel loaded at .33, and the other twoPEM parcels evidenced weak loadings on this first factor. The twofactor structure explained 50.07% of the variance in the data. Forthis model, the NEM parcels all loaded on the first factor (.69 ofhigher) while the hardiness parcels and 3 of the PEM parcelsloaded on the second factor above .40. The final PEM parcel loadedon the second factor at .38. The three-factor structure explained60.74% of the variance in the data. As shown in Table 2, all parcelsloaded .40 or above on factors representing their target constructs,and there were no cross loadings above .35. Therefore, based on theeigenvalues, amount of variance explained by the three-factorsolution, and ease of interpretability, the three-factor solutionwas selected. Of note, the hardiness factor was substantially corre-lated with the NEM factor (r = .45) and the PEM factor (r = .50).These results suggest that at the item parcel level, hardiness ap-

pears to represent a relatively distinct, albeit related constructfrom PEM and NEM.2

Next, CFAs were conducted on the second, Confirmatory Sampleto compare three models to examine the interrelations betweenhardiness, PEM, and NEM. Based on our EFA results, each of thesemodels included separate latent variables for hardiness, PEM andNEM with parcels from each scale serving as observed indicatorsof the latent constructs.

The first model, shown in Fig. 1, specified hardiness, PEM andNEM as three correlated latent variables. This model demonstratedan adequate fit as evidenced by a Tucker Lewis Index and Compar-ative Fit Index above .90 (TLI = .93, CFI = .94) and a Root MeanSquare Error of Approximation (RMSEA) and Standardized RootMean Residual (SRMR) below .08 (.063 and .062, respectively;Kline, 1998). For this model, v2(df = 51) was 151.29 (p < .001),and the Bayesian Information Criterion (BIC) was 17259.67. NEMwas only weakly correlated with PEM, and both NEM and PEMwere strongly correlated with hardiness. For the second modelwe specified hardiness as a lower-order indicator of NEM.Although this model and the third model cannot be directly com-pared with the first model because they are not nested, we maystill examine the fit indices. For the second model, fewer fit indiceswere within the adequate range compared to the first model andBIC was larger indicating poorer fit relative to the first model(Raftery, 1995; v2 = 208.65, df = 53; CFI = .91; TLI = .89, BIC =17303.98, RMSEA = .077, SRMR = .093). Specifically, the values forTLI and SRMR fell out of acceptable levels. The third model specifiedhardiness as a lower-order indicator of PEM. In this model none ofthe fit indices we examined were within the adequate range andagain BIC was larger than baseline (v2 = 234.52, p < .001, CFI = .90;TLI = .87; BIC = 17329.93; RMSEA = .083; SRMR = .110). Theseresults are consistent with the EFA findings in suggesting thathardiness is a distinct construct that is still related to PEM andNEM, as demonstrated by the correlations between hardiness andPEM (r = .50) and NEM (r = �.56) shown in Fig. 1.

3.2. Prediction of PTSD and depressive symptoms from hardiness, NEM,and PEM

The incremental utility of hardiness in predicting mental healthsymptoms was tested among the 110 soldiers (from both Explor-atory and Confirmatory Samples) who had experienced combatdeployments to Iraq or Afghanistan. We chose to restrict the

Fig. 1. Hardiness as distinct but related to PEN and NEM. Note. Double headed arrows represent correlations between latent variables; single headed arrows representstandardized factor loadings. NEM = negative emotionality, PEM = positive emotionality, HAR = hardiness.

C.R. Erbes et al. / Journal of Research in Personality 45 (2011) 508–512 511

analysis to this subset because hardiness is thought to be most rel-evant to those who have been through significant stressors andbecause PTSD symptoms are most relevant for those who haveexperienced trauma such as combat. Because 110 soldiers is arather small sample for structural equation modeling, latent vari-able scores for Hardiness, PEM, and NEM were exported fromMPlus to SPSS. Symptoms of PTSD and depression then served asdependent variables in regression analyses in which NEM, PEM,and hardiness were entered as predictors. The significance and sizeof the relationship between hardiness and depression and PTSDwere assessed through the resulting standardized partial regres-sion and correlation coefficients. In predicting PTSD, 45% of the var-iance was accounted for (R2 = .46, shrunken R2 = .45,F[3, 103] = 29.75, p < .001) and NEM emerged as an independentpredictor (b = �.73, partial r = �.58, t = �8.02, p < .001) while bothPEM (b = �.004, partial r = �.003, t = �.04, p = .966) and hardiness(b = .08, partial r = .05, t = .68, p = .498) did not. When predictingdepression scores, the model accounted for 35% of the variance(R2 = .37, shrunken R2 = 35, F[3, 100] = 19.24, p < .001). Once more,NEM emerged as a significant predictor (b = �.524, partial r = �.42,t = �.53, p = .001) while PEM (b = �.16, partial r = �.115, t = �1.45,p = .151) and hardiness (b = �.01, partial r < �.01, t = �.09, p = .927)did not.

4. Discussion

When evaluating potential predictors of resilience in the face ofstressors, a careful examination of the relationship and overlapbetween various personality constructs is required. Findings fromthis sample of National Guard soldiers lend support to the distinc-tiveness of hardiness from the higher order personality constructsof positive and negative emotionality. Items from a hardinessmeasure form a distinctive dimension from NEM and PEM inexploratory factor analyses and confirmatory factor analysis sug-gesting that this dimension is related to, but not subsumed by,positive and negative emotionality. These findings run counter toassertions that hardiness may be entirely viewed as the lack ofnegative emotionality (Funk & Houston, 1987) or as a combinationof positive and negative emotionality (Miller, 2003). However, ourresults are consistent with earlier observations, and findings, thathardiness has considerable overlap with these personality dimen-sions. The latent variable for hardiness correlated .50 and �.56with PEM and NEM, respectively.

Further examination suggests that the overlap between hardi-ness, PEM, and NEM is quite important when evaluating the impactof hardiness. In contrast with earlier analyses that found therelationship between hardiness and distress or mental health

symptoms endures after controlling for negative affect (Maddi &Khoshaba, 1994), we found the relationship between hardinessand symptoms of PTSD was entirely accounted for by its overlapwith PEM and NEM among a group of soldiers who had served inIraq or Afghanistan. Further, this finding held when consideringsymptoms of both PTSD and depression. Past studies of hardinessand mental health have generally not controlled for other person-ality constructs (e.g., King et al., 1998) or have used conveniencesamples that may have had low levels of stress exposure (e.g.,college students, Maddi & Khoshaba, 1994; adult communityvolunteers, Maddi et al., 2002). To our knowledge, this is the firststudy in which both NEM and PEM are simultaneously assessed inthe context of hardiness and mental health and thus serves as amore rigorous test of the relationship between hardiness, NEM,PEM, and mental health. Likewise, large scale studies of hardinessin military populations, which have found it to be a negative pre-dictor (protective factor) of PTSD symptoms (e.g., King et al.,1998) have not accounted for either PEM or NEM.

This study had several strengths including the use of a relativelylarge sample, a more detailed evaluation of hardiness in thecontext of other personality factors, and modern statistical meth-ods that minimize the influence of error variance and allow simul-taneous estimation of inter-relationships. Although fit indices forour measurement models suggested that the 15 hardiness itemsfit well on a single latent dimension, a more extensive measurecould have allowed an examination of which components of hardi-ness relate to PEM vs. NEM or are in fact distinct. The use of itemparceling provided for continuously distributed observed indica-tors of each construct. However, the fact that items for thehardiness measure were assessed with five point likert-type ques-tions while PSY-5 scales utilize binary response formats may haveartificially inflated differences between these constructs in factoranalysis. Greater similarity in response format across measuresmay have decreased the findings of the distinctiveness of thehardiness construct, but also would have made findings of an inde-pendent effect of hardiness on PTSD symptoms even less likely. Theuse of National Guard soldiers provided a distinct sample with arange of life stressors. However, the sample was relatively homog-enous with regards to race/ethnicity, education, and gender. Thus,further replication with more demographically diverse samplesremains important.

The ability to examine incremental validity in a subset of soldierswho had experience with a prior combat deployment represents astrength of the study. Unfortunately this group of soldiers wasrelatively small. Finally, these results are cross-sectional in nature.While we can infer relationships between hardiness, PEM, NEM,and PTSD symptoms, we cannot rule out the possibility that PTSDor depressive symptoms are affecting these constructs instead of

512 C.R. Erbes et al. / Journal of Research in Personality 45 (2011) 508–512

the reverse (Miller, 2003). These limitations make it clear that furtherreplication with both larger and more diverse military samples andespecially longitudinal follow-up are needed before firmer conclu-sions, and more nuanced understandings of the inter-relationshipof these variables, can be achieved.

Acknowledgments

This material is based upon work supported in part by grants toMelissa A. Polusny, PhD, from the Minnesota Medical Foundation(3662-9227-06) and Department of Defense Congressionally Direc-ted Medical Research Program (CDMRP; W81XWH-07-2-003). Thismaterial is the result of work supported with resources and the useof facilities at the Minneapolis VA Medical Center, Minneapolis,MN. The sponsors had no role in the design and conduct of thestudy, collection, management, analysis, and interpretation ofdata; preparation; review or approval of the manuscript. The corre-sponding author had full access to all of the data in the study andtakes responsibility for the integrity of the data and the accuracy ofthe data analysis. The views expressed in this article are those ofthe authors and do not reflect the official policy or position ofthe Department of Veterans Affairs or the Department of Defense.The authors would like to acknowledge the assistance of DarinErickson, PhD for his assistance with conceptualization andanalysis.

References

Bartone, P. T. (1995). A short hardiness scale. In Paper presented at the Americanpsychological society annual convention, New York (June).

Bartone, P. T. (1999). Hardiness protects against war-related stress in Army Reserveforces. Consulting Psychology Journal: Practice and Research, 51, 72–82.doi:10.1037/1061-4087.51.2.72.

Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck depression inventory: Manual (2ded.). Boston: Harcourt Brace.

Eschleman, K. J., Bowling, N. A., & Alarcon, G. M. (2010). A meta-analyticexamination of hardiness. International Journal of Stress Management, 17,277–307. doi: 10.10307/a0020476.

Funk, S. C., & Houston, B. K. (1987). A critical analysis of the Hardiness Scale’svalidity and utility. Journal of Personality and Social Psychology, 53(3), 572–578.doi:10.1037/0022-3514.53.3.572.

Harkness, A. R., McNulty, J. L., & Ben-Porath, Y. S. (1995). The personalitypsychopathology five (PSY-5): Constructs and MMPI-2 scales. PsychologicalAssessment, 7, 104–114. doi:10.1037/1040-3590.7.1.104.

King, L. A., King, D. W., Fairbank, J. A., Keane, T. M., & Adams, G. A. (1998).Resilience–recovery factors in post-traumatic stress disorder among female andmale Vietnam veterans: Hardiness, postwar social support, and additionalstressful life events. Journal of Personality and Social Psychology, 74, 420–434.doi:10.1037/0022-3514.74.2.420.

Kline, R. B. (1998). Principles and practice of structural equation modeling. New York,NY, US: Guilford Press.

Kobasa, S. C. (1979). Stressful life events, personality, and health: An inquiry intohardiness. Journal of Personality and Social Psychology, 37, 1–11. doi:10.1037/0022-3514.37.1.1.

Maddi, S. R., & Khoshaba, D. M. (1994). Hardiness and mental health. Journal ofPersonality Assessment, 63, 265–274. doi:10.1207/s15327752jpa6302_6.

Maddi, S. R., Khoshaba, D. M., Persico, M., Lu, J., Harvey, R., & Bleecker, F. (2002). Thepersonality construct of hardiness: II. Relationships with comprehensive test ofpersonality and psychopathology. Journal of Research in Personality, 36, 72–85.doi:10.1006/jrpe.2001.2337.

Miller, M. W. (2003). Personality and the etiology and expression of PTSD: A three-factor model perspective. Clinical Psychology: Science and Practice, 10, 373–393.doi:10.1093/clipsy/bpg040.

Raftery, A. E. (1995). Bayesian model selection in social research. SociologicalMethodology, 25, 111–163.

Rhodewalt, F., & Zone, J. B. (1989). Appraisal of life change, depression, and illness inhardy and nonhardy women. Journal of Personality and Social Psychology, 56,81–88. doi:10.1037/0022-3514.56.1.81.

Tellegen, A., & Ben-Porath, Y. S. (2008). MMPI-2-RF (Minnesota MultiphasicPersonality Inventory-2 Restructured Form): Technical manual. Minneapolis:University of Minnesota Press.

Weathers, F. W., Utz, B. T., Herman, D. S., Hus-ka, J. A., & Keane, T. M. (1993). ThePTSD checklist (PCL): Reliability, validity, and diagnostic utility. San Antonio, Tex:International Society of Traumatic Stress Studies (October).