Psychological Trauma: Theory, Research, Practice, and Policy In the public domain

2015, Vol. 7, No. 4, 372–381 http://dx.doi.org/10.1037/tra0000011

Differentiating Between Appraisal Process and Product in Cognitive

Theories of Posttraumatic Stress

John T. Nanney Joseph I. Constans

Southeastern Louisiana Veterans Health Care System and South Southeastern Louisiana Veterans Health Care System, South
Central VA Mental Illness Research, Education, Central VA Mental Illness Research, Education, and Clinical
and Clinical Center, New Orleans, Louisiana Center, New Orleans, Louisiana and Tulane University

Timothy A. Kimbrell Teresa L. Kramer

Central Arkansas Veterans Health Care System, South Central University of Arkansas for Medical Sciences
VA Mental Illness Research, Education, and Clinical Center,
Little Rock, Arkansas and University of Arkansas for
Medical Sciences

Jeffrey M. Pyne
Central Arkansas Veterans Health Care System, South Central VA Mental Illness Research, Education, and Clinical Center, Little
Rock, Arkansas, and University of Arkansas for Medical Sciences

Biased appraisal is central to cognitive theories of posttraumatic stress, but little research has examined
the potentially distinct meanings of the term. The ongoing process of appraising social information and
the beliefs that emerge as products of that process can be distinguished conceptually. This study sought
to examine whether these 2 meanings are empirically distinct as well, and if so, to begin exploring
potential relations between these appraisal constructs and posttraumatic stress symptoms. Soldiers (N ⫽
424) preparing for deployment to Iraq or Afghanistan were administered measures of each construct.
Results of confirmatory factor analysis suggest that the appraisal process and the products of that process
(i.e., beliefs) are indeed distinct. Structural equation models are consistent with cognitive bias and social
information processing literatures, which posit that a biased appraisal process may contribute to the
development of dysfunctional beliefs and posttraumatic stress symptoms following trauma. The potential
utility of distinctly conceptualizing and measuring the appraisal process in both clinical and research
settings is discussed.

Keywords: trauma, cognitive bias, appraisal, PTSD

Cognitive theories of posttraumatic stress posit that negatively appraisal can refer to both a process and the product of that
biased appraisals contribute to the causation and maintenance of process.
psychopathology following trauma (e.g., Brewin & Holmes, 2003; The majority of the current literature emphasizes appraisal as
Ehlers & Clark, 2000). Despite the centrality of negative appraisal product. Here, appraisal is used synonymously with beliefs, with
in contemporary trauma theory, its precise meaning is unclear. appraisal conceptualized—at least implicitly—as stable, trait-like
This imprecision appears to be rooted in the dual sense of the word interpretations about the self, the world, and others that may
“appraisal.” Appraisal can refer both to “the act of judging value” emerge following trauma (e.g., Foa & Rothbaum, 1998; Resick &
and to “a judgment of value” (Merriam-Webster, 2003). That is, Schnicke, 1992). Appraisal as product is operationalized and mea-

This article was published Online First December 22, 2014. Medicine, University of Arkansas for Medical Sciences; Teresa L. Kramer,
John T. Nanney, Southeastern Louisiana Veterans Health Care System Division of Health Services Research, Department of Psychiatry, College
and South Central VA Mental Illness Research, Education, and Clinical of Medicine, University of Arkansas for Medical Sciences.
Center, New Orleans, Louisiana; Joseph I. Constans, Southeastern Loui- This research was supported by Grant PT074626 from the Department of
siana Veterans Health Care System, South Central VA Mental Illness Defense and National Institute of Health Grant P20 GM103425-09. The
Research, Education, and Clinical Center, and Department of Psychiatry contents of this report do not represent the views of the Department of
and Behavioral Sciences, Tulane University; Timothy A. Kimbrell and Veterans Affairs or the United States government.
Jeffrey M. Pyne, Health Services Research and Development, Central Correspondence concerning this article should be addressed to John T.
Arkansas Veterans Health Care System, South Central VA Mental Illness Nanney, who is now at Department of Psychological Sciences, University
Research, Education, and Clinical Center, Little Rock, Arkansas and Di- of Missouri-Saint Louis, Stadler Hall Room 236, One University Boule-
vision of Health Services Research, Department of Psychiatry, College of vard, Saint Louis, MO 63121. E-mail: [email protected]

372
 APPRAISAL PROCESS AND APPRAISAL PRODUCTS 373

sured using questionnaires, such as the Posttraumatic Cognitions termine whether such measurement development would be war-
Inventory (PTCI; Foa, Ehlers, Clark, Tolin, & Orsillo, 1999), ranted.
where respondents rate their agreement with certain negative in- Addressing this question is not merely an exercise in semantic
terpretations about the meaning of the traumatic event. Though hairsplitting. Identifying any meaningful distinctions between pro-
these appraisals could occur as momentary thoughts, they are cess and product could have important implications for predicting
typically understood to represent relatively stable underlying in- and treating PTSD. If processing bias is distinct from negative
terpretations in response to trauma. Altering appraisal products– beliefs, then measurement of processing bias could be important in
beliefs is a primary focus of interventions for trauma (Foa & determining risk for psychopathology. Indeed, social information
Rothbaum, 1998; Resick & Schnicke, 1992), and changes in these processing models of posttraumatic stress imply that biased ap-
products– beliefs are associated with symptom improvement (Foa praisal processes may contribute to the development of biased
& Rauch, 2004; Resick et al., 2008). Emphasis is particularly beliefs (Ehlers & Clark, 2000). If processing biases prove inde-
placed on the importance of negative appraisal beliefs about the pendent or perhaps even predictive of negative beliefs, this would
self in predicting posttraumatic psychopathology (e.g., Dunmore, suggest that modifying information-processing styles in addition to
Clark, & Ehlers, 2001; Ehlers, Maercker, & Boos, 2000; Ehlers & modifying underlying beliefs would be necessary to maintain
Steil, 1995; Foa & Riggs, 1993; Kleim, Ehlers, & Glucksman,
treatment gains. To our knowledge, no previous studies have
2012). Multiple studies show that negative self-appraisal is
examined these propositions empirically. In the study that most
strongly associated with posttraumatic stress symptoms cross-
closely addresses this question (Gonzalo, Kleim, Donaldson, Moo-
sectionally (Beck et al., 2004; Constans et al., 2012; Foa et al.,
rey, & Ehlers, 2012), self-report measures of attributional styles
1999; Moser, Hajcak, Simons, & Foa, 2007), particularly reexpe-
that are characteristic of depression (e.g., attributing negative
riencing and numbing symptoms (Blain, Galovski, Elwood, &
events to global, stable, and internal causes) predicted DSM–IV
Meriac, 2013), and may uniquely predict later development of
posttraumatic stress disorder (PTSD) longitudinally (Bryant & PTSD independently of appraisal beliefs measured by the PTCI.
Guthrie, 2005; Kleim et al., 2012). The use of a self-report measure of attribution may limit the
The less commonly found meaning of appraisal in the trauma probative value of these findings, however, as appraisal or attri-
literature uses the term to refer to a process. Here, appraisal refers bution processes per se were not measured. Rather, the individu-
to the active online interpretation of incoming social information. al’s general perception of this process was measured in this study.
Some individuals, for example, may be prone to interpreting Moreover, no path models between attributional style, beliefs, or
ambiguous situations in a more negative fashion than others and PTSD were tested. Therefore, our second aim was to examine the
may thus be more prone to emotional distress. This sense of independent and combined roles of processing bias measured more
appraisal as process has been described as a feature and etiological directly and negative beliefs in their association with PTSD.1
factor in a variety of emotional disorders (Gross & Thompson, To pursue these aims, U.S. soldiers preparing for deployment to
2007; Mathews & MacLeod, 2005), but it has been less empha- Iraq or Afghanistan were asked to complete self-report measures
sized in cognitive models of PTSD. Often implicit in these models, of PTSD, posttraumatic beliefs, and a non-self-report measure of
however, is the suggestion that a negatively biased process of biases in the appraisal process. Using these data, we first tested the
interpreting information (appraisal as process) contributes to the hypothesis that the appraisal process measures and the appraisal
development and maintenance of this disorder (Ehlers & Clark, product measures reflect empirically distinct constructs rather than
2000). Although more research is needed, available empirical manifestations of a single underlying appraisal construct. We then
evidence also suggests that, as with other emotional disorders, explored the relations among the constructs, hypothesizing that the
PTSD is associated with biased processing of incoming informa- appraisal process would be associated with both appraisal products
tion. Specifically, individuals with PTSD are more likely to inter- and PTSD. Consistent with general cognitive bias literature (e.g.,
pret ambiguous stimuli as being trauma related or as more threat- Mathews & Macleod, 2005), we expected that the appraisal pro-
ening compared with those without this disorder (Amir, Coles, & cess would predict PTSD symptoms; also, in accord with social
Foa, 2002; Elwood, Williams, Olatunji, & Lohr, 2007; Kimble et information-processing theories of PTSD, we hypothesized that
al., 2002). the relationships between the appraisal process and emotional
Although available evidence suggests posttraumatic psychopa- distress would be mediated by appraisal products– beliefs. That is,
thology is associated with both negatively biased appraisal beliefs we sought to test the viability of a model in which a general
and negatively biased appraisal processes, a number of questions negative appraisal processing style would be associated with de-
persist. One primary question is whether assessments of appraisal
velopment of negative appraisal beliefs and that these beliefs
processes and appraisal products actually measure separate forms
would lead to greater posttraumatic distress. Finally, as a second-
of cognition. Although there are clear conceptual distinctions
ary exploratory analysis, we examined the potential moderating
between appraisal processes and appraisal products, evidence
effects of prior combat exposure on our hypothesized models.
showing that the existing measures assess separate constructs is
lacking. It remains possible that simple self-report assessments of
beliefs may be a proxy measure for appraisal processes. Therefore, 1
Our study examined PTSD as defined in DSM–IV so there was no item
the first aim of this article is to examine whether existing measures overlap between measures of PTSD and measures negative beliefs as there
of biased interpretive processes are psychometrically distinct from would be now given DSM-5 (American Psychiatric Association, 2013)
changes to PTSD criteria. Within a DSM-5 framework, our study could be
measures of biased interpretive products. This project is not aimed conceptualized as examining the contribution of appraisal process and
to develop and evaluate an appraisal process measure. Our aim appraisal beliefs to development of additional PTSD symptoms beyond the
rather was to conduct an initial proof-of-concept exercise to de- appraisal beliefs themselves.
374 NANNEY, CONSTANS, KIMBRELL, KRAMER, AND PYNE

Method returned to base camp: (1) You lie awake filled with fear (legitimate/
negative); (2) You lie awake talking to your buddies (legitimate/
positive); (3) You do not eat because of nausea (foil/negative); (4)
Participants You do not eat because you want to relax (foil/positive).

Participants were Virginia Army National Guard members Participant endorsement of the inaccurate negative interpretation
scheduled for deployment to Iraq. Eligibility criteria included was used as the measure of the biased appraisal process. We
being an Army National Guard member scheduled to deploy for believed this tendency to endorse an interpretation that had no
Operation Enduring Freedom/Operation Iraqi Freedom (OEF/OIF)
factual basis in the actual text of the vignette and was likely rated
operations within the next 12 months, being age 18 – 60, and
as a viable interpretation only because of its negative valence was
willing to provide the name and phone number of at least one
the best indicator of a generalized affinity for negatively valenced
person to help locate the member for the follow-up assessments if
interpretations. Internal consistency of the items was .51.
necessary. Two battalions participated in this study: one from
Aviation and one Cavalry. Six hundred soldiers completed a writ- Appraisal products– beliefs. Appraisal products (i.e., beliefs)
ten consent process during a drill weekend soldier-readiness check were measured using the 27-item Negative Cognitions about the
prior to predeployment training, and a subset of 424 completed Self subscale of Posttraumatic Cognitions Inventory (PTCI-Self;
study measures during a subsequent 3-week predeployment train- Foa et al., 1999). The PTCI-Self scale consists of 75% of the items
ing. from the 36-item PTCI. We used only the PTCI-Self for greater
Four-hundred twenty-four soldiers participated in the study. parsimony and because of the emphasis placed on negative self-
Demographics are presented in Table 1. Participants were predom- appraisals in trauma theory and research. Participants are asked to
inantly male. The majority of participants identified as White. think about a recent distressing event. Items inquire about partic-
Black was the second most commonly endorsed racial identity. ipants’ agreement with beliefs about the self in relation to this
Regarding military rank and experience, the vast majority were distressing event on a scale of 1 (totally disagree) to 7 (totally
enlisted soldiers.2 Approximately half reported prior combat de- agree). The PTCI-Self has good internal consistency (␣ ⫽ .97) and
ployments. Of those with prior combat experience, average num- good ability to discriminate between those with and without PTSD
ber of months in combat zones was about 12 months. (Foa et al., 1999). The PTCI was designed for use in any trauma
population including combat soldiers (Edna Foa, personal commu-
Measures nication). Internal consistency in the current sample was .87.
Posttraumatic stress symptoms. Posttraumatic stress symp-
Appraisal process. The appraisal process measurement pro- toms were measured using the PTSD Checklist–Military version
tocol was developed by Mathews and Mackintosh (2000) to assess (PCL-M; Weathers, Litz, Herman, Huska, & Keane, 1993). The
changes in interpretive bias occurring in response to cognitive bias PCL-M is a widely used 17-item inventory with items consistent
modification (CBM) procedures and has been validated through with the Diagnostic and Statistical Manual of Mental Disorders
use in multiple other studies (Salemink & van den Hout, 2010; (4th ed.; DSM–IV; American Psychiatric Association, 1994) crite-
Salemink, van den Hout, & Kindt, 2007). Participants were asked
ria for PTSD. Participants indicate for each symptom how much it
to read and imagine eight randomly presented three- to four-
bothered them in the past month on a scale from 1 (not at all) to
sentence scenarios. We adapted the content of the original protocol
5 (extremely). Participants were prompted that items reflect pos-
to focus on military-related scenarios with the assistance of a
sible reactions to stressful life events. They were not instructed to
subject matter expert (OEF/OIF veteran). Scenarios were ambig-
uous with regard to the meaning of the event (e.g., “Your unit goes answer in regard to any specific event. The PCL-M demonstrates
on night patrol. You spot possible insurgents in the distance and robust psychometric properties (Weathers et al., 1993). Internal
pursue them for 45 minutes. They eventually evade your capture, consistency in the current sample was .92.
and you return to base camp. You do not go to sleep immedi-
ately.”) Following each scenario, the participant rated its vividness
Procedure
to reinforce the implication that the procedures are simply to
identify vivid and realistic combat scenarios. Following presenta- The study was approved and overseen by the Central Arkansas
tion of all scenarios, participants were asked to complete an Veterans Health Care System Institutional Review Board. Data
unexpected recognition memory test to assess the appraisal pro- were collected as part of the baseline assessment for a broader
cess. For each scenario, participants were presented with a brief longitudinal observational study with an embedded randomized
reminder of the scenario and asked to rate on a 10-point Likert- controlled trial of resilience interventions for combat soldiers. Data
type scale how well each of four statements represented the event collection occurred in person in individual data collection rooms
described in the scenario. Two of the statements reflected legiti- approximately 2 months before deployment to Iraq. Appraisal
mate interpretations of the ambiguous scenario, but one interpre- process measures were collected first followed by self-report mea-
tation was positive and the other one was negative. Two interpre-
sures.
tations did not represent events described in the scenario: one
interpretation was positive and the other was negative. For exam-
ple, participants were asked, 2
Aviation unit participants were not asked about prior military experi-
ence at baseline. Therefore military history data are reported only for the
For the scenario involving a night patrol, please rate the accuracy of Cavalry unit who represent the majority of our overall sample (n ⫽ 378;
the following statements that describe what happened when you 89.1%)
 APPRAISAL PROCESS AND APPRAISAL PRODUCTS 375

Table 1
Demographics and Military Service Data

Variable No (%) M (SD) Mdn

Gender
Male 387 (91.3)
Female 37 (8.7)
Race–ethnicity
White 287 (67.5)
Black 82 (19.3)
Hispanic 22 (5.2)
Asian–Pacific Islander 11 (2.6)
American Indian–Alaskan Native 4 (0.9)
Other 18 (4.2)
Marital status
Married–cohabitating 193 (45.5)
Never married 189 (44.6)
Separated–divorced 32 (9.9)
Education
No high school diploma 1 (0.2)
High school graduate–GED 100 (23.6)
Some college–technical school 246 (58.0)
College graduate 54 (12.7)
Some postgraduate work 10 (2.4)
Graduate degree 13 (3.1)
Income
⬍$20,000 84 (19.8)
$20,000–$39,999 129 (30.4)
$40,000–$59,999 88 (20.8)
$60,000–$79,999 62 (14.6)
⬎$80,000 61 (14.4)
Military ranka
Enlisted 345 (91.3)
Warrant officer 9 (2.4)
Officer 29 (5.0)
No response 5 (1.3)
Military service (in years)a 7.52 (7.66) 5.00
Previous combat deployment
Yes 184 (48.7)
No 240 (51.3)
Months in combatb 12.41 (8.50) 10.00
a
Members of the Aviation unit (n ⫽ 46) were not asked this question at baseline assessment and are not included
in these analyses.

Data Analysis Plan Structural equation modeling using maximum likelihood estima-
tion was used to evaluate relationships between constructs. Assuming
Confirmatory factor analysis (CFA) using maximum likelihood distinct appraisal process and product constructs, we planned to test
estimation was used to examine the hypothesis that appraisal process specifically the hypothesis that appraisal products (beliefs) mediate
and product measures tap distinct latent constructs. Item parceling the relation between appraisal process and posttraumatic stress symp-
was used to reduce model estimation bias (Little, Cunningham, Sha- toms (see Figure 1). The mediation model also included direct paths
har, & Widaman, 2002; Little, Rhemtulla, Gibson, & Schoemann, from appraisal process to PTSD to examine potential direct effects of
2013). Parceling entails combining individual items to create indicator processing style on emotional distress not mediated by appraisal
variables for use in latent-variable analyses (e.g., factor analysis and products– beliefs. The fit of the hypothesized model to the data was
structural equation modeling (SEM); Little et al., 2002, 2013). Par- compared with alternate causal models. All CFA and SEM analyses
celing items into indicators reduces model estimation bias in latent were performed in AMOS 20.0.
variable analyses because parcels typically have stronger psychomet-
ric properties than individual items and allow for the evaluation of
more parsimonious models (Little et al., 2002, 2013) For each con- Results
struct, items from the relevant measure were parceled into three
separate indicator variables (see the Appendix for a breakdown of Clinical Characteristics
items in each parcel). That is, three indicators each were created for
appraisal process, appraisal product, and PTSD. The balancing ap- Emotional distress was low (see Table 2). On the PCL-M, for
proach to parceling was used (Little et al., 2002, 2013). The balancing which the minimum score was 17, the median was 20 (range ⫽
approach uses item factor loadings to create relatively equivalent 17– 83), and the mean was 23.44 (SD ⫽ 8.42). Approximately 16%
parcels for each hypothesized latent variable. reported PTS symptoms above the cut-off score of 30 that is
376 NANNEY, CONSTANS, KIMBRELL, KRAMER, AND PYNE

Structural equation models.

Structural equation modeling was used to evaluate whether, as
hypothesized, a negatively biased appraisal process leads to biased
appraisal products which, in turn, lead to PTSD. The hypothesized
mediation model (see Figure 1) fit the data well. Though the
chi-square value was statistically significant, ␹2(24) ⫽ 62.97, p ⬍
.001, other fit indexes indicated close model fit (RMSEA ⫽ .062,
90% CI [.044, .081], CFI ⫽ .979, NFI ⫽ .966, IFI ⫽ .979). As
expected, there was a significant total effect of appraisal process
on PTSD with more negative appraisal process predicting greater
PTSD symptoms (␤ ⫽ .21, p ⬍ .001). Negative processing style
also predicted appraisal products– beliefs (␤ ⫽ .44, p ⬍ .001), and
Figure 1. Hypothesized mediation model. Circles represent hypothesized as expected, appraisal products– beliefs in turn predicted PTSD
latent variables. Rectangles represent measured indicator variables. Re- (␤ ⫽ .48, p ⬍ .001). The indirect effect of appraisal process on
gression weights are standardized. Error and disturbance terms are omitted. PTSD mediated through appraisal product– beliefs was significant
(standardized indirect effect ⫽ .22; bias corrected 95% CI [.13,
.34]). The residual direct effect of appraisal processing style on
PTSD was not significant (␤ ⫽ .00, p ⫽ .98), indicating this
recommended when screening at general military population (Na-
relationship is fully mediated by appraisal products.
tional Center for PTSD, 2014). Comparing those with prior combat
To rule out other causal explanations, we evaluated alternative
deployment to those without prior combat experience revealed
conceptually plausible structural models using SEM. A model in
significant differences in PTSD symptoms, F(2, 376) ⫽ 24.85, p ⬍
which PTSD mediated the relationship between appraisal process
.001.3
and product was a poorer fit with the data as was the mathemat-
ically equivalent model in which PTSD predicted both appraisal
Confirmatory Factor Models process and product: ␹2(24) ⫽ 90.6, p ⬍ .001; RMSEA ⫽ .079,
A CFA of a two-factor solution with separate appraisal process 90% CI [.062, .097]; Aikake Information Criterion (AIC) ⫽
and appraisal product factors revealed a very close fit with the 130.61 as compared with 104.97 in the hypothesized model. A
obtained data. There was no significant difference between the model in which appraisal process served as a mediator between
hypothesized and obtained covariance matrix even using the very PTSD and appraisal beliefs was a poor fit: ␹2(24) ⫽ 105.58, p ⬍
strict chi-square goodness-of-fit test, ␹2(8) ⫽ 14.91 p ⫽ .06. Other .001; RMSEA ⫽ .087, 90% CI [.070, .105]; AIC ⫽ 145.58 vs.
fit indexes also suggested good model fit: root-mean-square error 104.97. These results suggest that the hypothesized mediation
of approximation (RMSEA) ⫽ .045; 90% confidence interval model (see Figure 1) is the best fit for the obtained data.
(CI) ⫽ [.000, .080]; comparative fit index (CFI) ⫽ .991; normed As a final check of the validity of our model, we compared
fit index (NFI) ⫽ .980; IFI ⫽ .99. In this two-factor model, appraisal process and appraisal product between participants above
appraisal process parcels all loaded moderately on a single factor the clinical cut-off for significant PTSD symptoms (PCL-M ⬎ ⫽
(standardized regression weights ⫽ .46 –.53). Appraisal product 30). Consistent with expectations, those with significant PTSD
parcels loaded strongly on a separate factor (standardized regres- symptoms had significantly more negative appraisal processing,
sion weights: .81–.85). The correlation between the factors was F(2, 422) ⫽ 6.06, p ⫽ .01, and appraisal products– beliefs, F(2,
significant but only of moderate strength, r ⫽ .44, p ⬍ .01. In 422) ⫽ 64.41, p ⬍ .001.
contrast, a CFA that uses a single-factor solution (i.e., a single
appraisal factor underlying both process and product measures) Secondary analysis: Combat deployment as
indicated a poor fit: ␹2(9) ⫽ 65.79, p ⬍.001; RMSEA ⫽ .122, 90% moderator.
CI ⫽ [.095, .191], CFI ⫽ .924, NFI ⫽ .914, IFI ⫽ .925. Direct
comparison of the models using a chi-square difference test further As a secondary exploratory analysis we examined potential
confirmed the two-factor model was a significantly better fit than impact of prior combat deployment on our CFA and SEM models.4
a one-factor model, ⌬␹2(1) ⫽ 50.9, p ⬍ .001. Tests of measurement equivalence between those with and without
A confirmatory factor model including PTSD was next evalu- combat deployment revealed similar factor structure between
ated to ensure that appraisal process and appraisal product con- groups: Two group configural model fit indices, ␹2(16) ⫽ 33.8,
structs are distinct from emotional distress. A CFA including RMSEA ⫽ .054, 90% CI [.028, .080], CFI ⫽ .973, NFI ⫽ .951,
appraisal process, appraisal product, and PTSD yielded a signifi- IFI ⫽ .974. Thus, it appears that the two-factor model with
cant chi-square value, ␹2(24) ⫽ 62.97 p ⬍ .01, but other fit separate appraisal process and product factors identified in the
indexes suggested at least adequate model fit (RMSEA ⫽ .062, overall sample applies equally well to those with and without prior
CFI ⫽ .979, NFI ⫽ .966, IFI ⫽ .979). Examination of modifica- combat employment.
tion indexes revealed that overlap between indicators of the PTSD Possible moderating effects of combat deployment on our struc-
and appraisal product constructs most contributed to discrepancies tural mediation models were next evaluated using methods for
between the hypothesized model and obtained data. This finding
suggests our appraisal process measure taps a construct that is 3
Among the Cavalry unit (n ⫽ 378) for whom this is known.
empirically distinct from appraisal products and PTSD. 4
Among the Cavalry unit (n ⫽ 378) for whom this is known.
 APPRAISAL PROCESS AND APPRAISAL PRODUCTS 377

Table 2
Clinical Characteristics

M (SD)
Total sample Prior combata No priora combat
Variable (N ⫽ 424) (n ⫽ 184) (n ⫽ 194) F p

Appraisal process 16.35 (6.67) 16.40 16.13 0.15 .70

Appraisal product 30.30 (11.35) 30.47 (10.74) 29.28 (11.29) 0.93 .34
PTSD 23.44 (8.42) 25.60 (10.19) 21.39 (5.73) 24.85 ⬍.001
Note. PTSD ⫽ posttraumatic stress disorder.
a
Members of the Aviation unit (n ⫽ 46) were not asked this question at baseline assessment and are not included in these analyses.

testing interactions among latent variables in SEM as described by manner in which individuals with PTSD process information. This
Little, Bovaird, and Widaman (2006). Combat deployment was is consistent with social information-processing models of PTSD
found to moderate the relationship of appraisal process on ap- (Ehlers & Clark, 2000), as individuals with a general predisposi-
praisal products (␤ ⫽ .18, p ⫽ .02) with a stronger relationship tion to endorse negative interpretations may be more likely to
among those with combat deployment (␤ ⫽ .65, p ⬍ .001) com- develop the negative self-beliefs and negative appraisals about
pared with those without prior combat experience (␤ ⫽ .33, p ⫽ both the meaning of a traumatic event and postevent emotional
.01). Combat deployment was similarly found to moderate the responses that promote the development of this disorder. Although
relationship between appraisal products and PTSD (interaction this study design was cross-sectional prohibiting definitive state-
␤ ⫽ .22, p ⫽ .01) with a stronger relation in participants with prior ments regarding causality, the findings are consistent with the
combat exposure (␤ ⫽ .61, p ⬍ .001) than without (␤ ⫽ .43, p ⫽ hypothesis that pretrauma biases in information processing par-
.002). Because of these interactions, the indirect effect of appraisal tially determine beliefs formed about the trauma. Furthermore, it is
process on PTSD through appraisal products was substantially possible that biased processing occurring posttrauma further rein-
greater among those with prior combat deployment (standardized forces negative beliefs once they emerge by continuously provid-
indirect effect ⫽ .39; bias corrected 95% CI [.21, .98]) than ing evidence of their apparent validity. Identification of appraisal
without (standardized indirect effect ⫽ .14; bias corrected 95% CI processing as a distinct predictor of both cognitive products and
[.04, .35]). In sum, the relations among appraisal process, appraisal emotional distress following trauma may be particularly useful in
product, and PTSD appear to be significantly strengthened among light of the inclusion of cognitive products– beliefs within the
those with a history of combat exposure. definition of PTSD in DSM-5 (American Psychiatric Association,
2013). As beliefs–products are established as elements of the
Discussion disorder, theory and research may benefit from shifting focus
Results suggest that biases in the appraisal process and appraisal toward the more proximal processes that determine their forma-
products– beliefs are empirically distinct. Future theory and re- tion.
search may benefit from differentiating between these meanings of There are several potential clinical implications of our results. If
appraisal, potentially providing a more precise cognitive model of biased information processing does indeed promote and maintain
trauma. Such higher definition theory may be particularly impor- biased products– beliefs, this suggests treatments for PTSD should
tant as attempts are made to integrate multiple social, cognitive, target not only the distorted products but also the biased processing
and biological levels of analysis into theories of mental disorders style from which they emerge. Cognitive therapies for PTSD
(Sanislow et al., 2010). It is feasible that appraisal processes and emphasize modification of beliefs surrounding the meaning of the
appraisal products may have distinct biological and social deter- traumatic event (Foa & Rothbaum, 1998; Resick & Schnicke,
minants, such that a biopsychosocial model of PTSD would need 1992; Resick et al., 2008) but give relatively less attention to
to distinguish between both senses of appraisal to be synchronous modifying interpretive biases per se. This study raises the possi-
and complete. bility that treatment may benefit from greater emphasis on biases
Our results are consistent with information-processing models in appraisal processes. That is, if biased appraisal processes par-
of psychopathology that emphasize biases in the active online tially determine the beliefs about traumatic events, negative ap-
processing of experience as determinants of emotional distress. As praisal products could reemerge unless the biased processes are
found in previous studies (Amir et al., 2002; Elwood et al., 2007; modified. There are now a number of studies showing the efficacy
Kimble et al., 2002) biased information processing appears to be of modifying biased cognitive processes (i.e., cognitive bias mod-
one aspect of dysfunction in the cognitive system associated with ification therapies) associated with generalized anxiety disorder
PTSD. The current findings also indicate biased appraisal process and social phobia (Amir, Beard, Burns, & Bomyea, 2009; Amir,
overlaps the biased products– beliefs that are consistently associ- Beard, Taylor, et al., 2009). The current study suggests that similar
ated with posttraumatic stress symptoms (Beck et al., 2004; Blain efforts to modify biases in interpretive processes may be helpful in
et al., 2013; Bryant & Guthrie, 2005; Foa et al., 1999; Moser, the treatment of PTSD. Results also suggest that processing bias
Hajcak, Simons, & Foa, 2007). It is interesting that the results of modification interventions may be useful in preventing the devel-
the structural models, though exploratory, suggest that biased opment of emotional distress or in preparing those for whom
posttraumatic beliefs may be a function of general biases in the trauma exposure may be imminent (e.g., combat soldiers).
378 NANNEY, CONSTANS, KIMBRELL, KRAMER, AND PYNE

These results may also open up new measurement strategies for ison of certain mathematically equivalent models. In the present
assessing biased appraisal both before and following trauma. The study, a structural model in which products– beliefs cause both the
emphasis on appraisal beliefs in trauma theory weds research to processing biases as well as emotional distress is mathematically
the use of self-report measures commonly used to measure such indistinguishable from the hypothesized mediation model (pro-
constructs. Self-report measures of trauma-related beliefs, though cess ⱖ product ⱖ distress). Though information-processing theo-
informative, are limited in that they only assess semantically ries suggest process leads to products– beliefs as hypothesized,
encoded attitudes of which the individual is introspectively aware. cognitive theory would support this alternative model in which
They are not able to directly examine information processing beliefs are the source of both processing biases and emotional
biases as such, biases which may operate outside of awareness. A distress. We cannot compare these models using cross-sectional
behavioral performance based measure of cognitive bias could SEM data. Future longitudinal or experimental research will be
provide a useful supplement in tapping aspects of appraisal bias needed to test fully the causal hypotheses suggested by the current
that are outside of introspective awareness. By focusing more results.
closely on the immediate interpretation of day-to-day scenarios, Appraisal process measures have not been through an extensive,
such measures may also be more representative of current func- formal test development process. The internal consistency of the items
tioning than self-report measures of underlying beliefs or emo- is relatively poor, which is of less concern for analyses like SEM for
tional distress and thus may serve as particularly useful tools for which measurement error is accounted. At this time, however, use of
tracking patient progress in treatment. Finally, by focusing on these items in other analyses that do not account for measurement
quotidian social information, appraisal process measures could be error (e.g., linear regression) cannot be recommended because of poor
used to determine risk for posttrauma psychopathology before internal consistency. Formal test development and demonstration of
trauma occurs. Current self-report measures assess beliefs about stronger psychometric properties would be needed before this specific
the meaning of a prior trauma and posttrauma symptoms. An measure could be used in such analyses or in clinical assessment.
appraisal-processing measure focused on daily events does not Certain characteristics of the sample are also limiting. It is un-
assume trauma has already occurred and could potentially be used known how our findings from this National Guard sample would
prior to trauma exposure to identify those with an appraisal style generalize to the general military or to civilian populations. It is also
that would predispose them to form negative beliefs or emotional noteworthy that the data collection was performed at relatively short
disorders were they to experience a traumatic event. duration before deployment such that some soldiers may have denied
The moderating effects of combat deployment on our models difficulties as a way of coping. Alternatively, impending deployment
are particularly noteworthy. Appraisal process, appraisal product, could exacerbate emotional distress in some participants. Longitudi-
and PTSD were more strongly related among those with prior nal studies with multiple time points before deployment may be
combat exposure. This finding is consistent with cognitive theories needed to identify the impact of impending deployment on emotional
of PTSD and general diathesis-stress models that posit that vul- distress. Finally, emotional distress in the sample was relatively low.
nerabilities, like appraisal processing biases or negative beliefs, It is not clear how our models would apply in a more severely
may not necessarily be associated with emotional distress until distressed sample. Restriction of range in emotional distress within
activated by traumatic stress. The stronger relation between ap- this sample may lead to an underestimation of relationship strength
praisal process and appraisal products– beliefs among those with such that more robust findings would be present in samples with
combat exposure suggests that appraisal process may similarly higher levels of distress. Comparisons between those with and without
serve as a prepotent vulnerability that only results in the formation prior combat suggest this may be the case, as at least the relationship
of such negative self-beliefs (and, in turn, PTSD) following between appraisal product and PTSD was stronger among a combat-
trauma. deployed subsample that had higher and more variable PTSD scores.
Indeed, it is possible that greater variability in PTSD scores contrib-
uted to significant interactions between combat deployment and pre-
Limitations
dictors of PTSD. That combat deployment moderated the relation
Measurement method and theoretical constructs are confounded between appraisal product and appraisal process-variables with sim-
in our study. We used only behavioral measures of appraisal ilar variance between combat-deployed and not-deployed groups sug-
process and self-report measures of appraisal products– beliefs. gests that combat deployment interactions are not solely the result of
The apparent distinction between the appraisal process and ap- greater variability–less restriction of range in the combat experienced
praisal product constructs may at least in part reflect the use of subsample.
distinct methods to measure them. Because beliefs and emotional
distress were both measured solely through self-report, shared
Summary
method variance may in part account for the stronger relation to
emotional distress found for appraisal products– beliefs compared The notion of biased appraisal is central to cognitive theories of
with appraisal process. Future research examining these constructs posttraumatic stress, but little research has examined the potentially
using a multitrait–multimethod approach would be necessary to distinct meanings of the term. Our results suggest that the appraisal
confirm our findings. process and the products of that process (i.e., beliefs) are empirically
Our structural equation models (SEM) only examined cross- distinguishable. Results are also consistent with cognitive bias and
sectional data so no conclusions regarding causation can be made. social information-processing literatures, which suggest that a biased
Though SEM compares the relative fit of various causal models to appraisal process may contribute to the development of dysfunctional
the observed data, our correlational design precludes inferences beliefs and PTSD following trauma. Though additional longitudinal
about causation. Moreover, SEM does not allow for the compar- research is needed to confirm these implications, our findings high-
 APPRAISAL PROCESS AND APPRAISAL PRODUCTS 379

light the potential use of distinctly conceptualizing and measuring the and their relation to symptom development. Journal of Anxiety Disor-
appraisal process in both clinical and research settings. Differentiating ders, 21, 554 –567. http://dx.doi.org/10.1016/j.janxdis.2006.08.006
between appraisal process and product may provide for higher defi- Foa, E. B., Ehlers, A., Clark, D. M., Tolin, D. F., & Orsillo, S. M. (1999).
nition theories of posttraumatic emotional distress while expanding The posttraumatic cognitions inventory (PTCI): Development and vali-
approaches to its assessment and treatment. dation. Psychological Assessment, 11, 303–314. http://dx.doi.org/
10.1037/1040-3590.11.3.303
Foa, E. B., & Rauch, S. A. (2004). Cognitive changes during prolonged
References exposure versus prolonged exposure plus cognitive restructuring in
female assault survivors with posttraumatic stress disorder. Journal of
American Psychiatric Association. (1994). Diagnostic and statistical man- Consulting and Clinical Psychology, 72, 879 – 884. http://dx.doi.org/
ual of mental disorders (4th ed.). Washington, DC: Author. 10.1037/0022-006X.72.5.879
American Psychiatric Association. (2013). Diagnostic and statistical man- Foa, E. B., & Riggs, D. S. (1993). Posttraumatic stress disorder in rape
ual of mental disorders (5th ed.). Arlington, VA: Author. victims. In J. M. Oldham, M. B. Riba, & A. Tasman (Eds.), Annual
Amir, N., Beard, C., Burns, M., & Bomyea, J. (2009). Attention modifi- review of psychiatry (pp. 273–303). Washington, DC: American Psy-
cation program in individuals with generalized anxiety disorder. Journal chiatric Association.
of Abnormal Psychology, 118, 28 –33. http://dx.doi.org/10.1037/ Foa, E. B., & Rothbaum, B. O. (1998). Treating the trauma of rape:
a0012589 Cognitive-behavioral therapy for PTSD. New York, NY: Guilford Press.
Amir, N., Beard, C., Taylor, C. T., Klumpp, H., Elias, J., Burns, M., & Gonzalo, D., Kleim, B., Donaldson, C., Moorey, S., & Ehlers, A. (2012).
Chen, X. (2009). Attention training in individuals with generalized How disorder-specific are depressive attributions? A comparison of
social phobia: A randomized controlled trial. Journal of Consulting and
individuals with depression, post- traumatic stress disorder and healthy
Clinical Psychology, 77, 961–973. http://dx.doi.org/10.1037/a0016685
controls. Cognitive Therapy and Research, 36, 731–739. http://dx.doi
Amir, N., Coles, M. E., & Foa, E. B. (2002). Automatic and strategic
.org/10.1007/s10608-011-9429-0
activation and inhibition of threat-relevant information in posttraumatic
Gross, J. J., & Thompson, R. A. (2007). Emotion regulation: Conceptual
stress disorder. Cognitive Therapy and Research, 26, 645– 655. http://
foundations. In J. J. Gross (Ed.), Handbook of emotion regulation (pp.
dx.doi.org/10.1023/A:1020309326976
3–24). New York, NY: Guilford Press.
Beck, J. G., Coffey, S. F., Palyo, S. A., Gudmundsdottir, B., Miller, L. M.,
Kimble, M. O., Kaufman, M. L., Leonard, L. L., Nestor, P. G., Riggs,
& Colder, C. R. (2004). Psychometric properties of the Posttraumatic
D. S., Kaloupek, D. G., & Bachrach, P. (2002). Sentence completion test
Cognitions Inventory (PTCI): A replication with motor vehicle accident
in combat veterans with and without PTSD: Preliminary findings. Psy-
survivors. Psychological Assessment, 16, 289 –298. http://dx.doi.org/
chiatry Research, 113, 303–307. http://dx.doi.org/10.1016/S0165-
10.1037/1040-3590.16.3.289
1781(02)00229-9
Blain, L. M., Galovski, T. E., Elwood, L. S., & Meriac, J. P. (2013). How
Kleim, B., Ehlers, A., & Glucksman, E. (2012). Investigating cognitive
does the Posttraumatic Cognitions Inventory fit in a four-factor post-
pathways to psychopathology: Predicting depression and posttraumatic
traumatic stress disorder world? An initial analysis. Psychological Trau-
stress disorder from early responses after assault. Psychological Trau-
ma: Theory, Research, Practice, and Policy, 5, 513–520. http://dx.doi
ma: Theory, Research, Practice, and Policy, 4, 527–537. http://dx.doi
.org/10.1037/a0030251
Brewin, C. R., & Holmes, E. A. (2003). Psychological theories of post- .org/10.1037/a0027006
traumatic stress disorder. Clinical Psychology Review, 23, 339 –376. Little, T. D., Bovaird, J. A., & Widaman, K. F. (2006). On the merits of
http://dx.doi.org/10.1016/S0272-7358(03)00033-3 orthoganalizing powered and product terms: Implications for modeling
Bryant, R. A., & Guthrie, R. M. (2005). Maladaptive appraisals as a risk interactions among latent variables. Structural Equation Modeling, 13,
factor for posttraumatic stress: A study of trainee firefighters. Psycho- 497–513. http://dx.doi.org/10.1207/s15328007sem1304_1
logical Science, 16, 749 –752. http://dx.doi.org/10.1111/j.1467-9280 Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002).
.2005.01608.x To parcel or not to parcel: Exploring the question, weighing the merits.
Constans, J. I., Vasterling, J. J., Deitch, E., Xiaotong, H., Tharp, A. L. T., Structural Equation Modeling, 9, 151–173. http://dx.doi.org/10.1207/
Davis, T. D., & Sullivan, G. (2012). Pre-Katrina mental illness, postdi- S15328007SEM0902_1
saster negative cognitions, and PTSD symptoms in male veterans fol- Little, T. D., Rhemtulla, M., Gibson, K., & Schoemann, A. M. (2013). Why
lowing Hurricane Katrina. Psychological Trauma: Theory, Research, the items versus parcels controversy needn’t be one. Psychological
Practice, and Policy, 4, 568 –577. http://dx.doi.org/10.1037/a0027487 Methods, 18, 285–300. http://dx.doi.org/10.1037/a0033266
Dunmore, E., Clark, D. M., & Ehlers, A. (2001). A prospective investiga- Mathews, A., & Mackintosh, B. (2000). Induced emotional interpretation
tion of the role of cognitive factors in persistent posttraumatic stress bias and anxiety. Journal of Abnormal Psychology, 109, 602– 615.
disorder (PTSD) after physical or sexual assault. Behaviour Research http://dx.doi.org/10.1037/0021-843X.109.4.602
and Therapy, 39, 1063–1084. http://dx.doi.org/10.1016/S0005- Mathews, A., & MacLeod, C. (2005). Cognitive vulnerability to emotional
7967(00)00088-7 disorders. Annual Review of Clinical Psychology, 1, 167–195. http://dx
Ehlers, A., & Clark, D. M. (2000). A cognitive model of posttraumatic .doi.org/10.1146/annurev.clinpsy.1.102803.143916
stress disorder. Behaviour Research and Therapy, 38, 319 –345. http:// Merriam-Webster. (2003). The Merriam Webster’s Collegiate Dictionary
dx.doi.org/10.1016/S0005-7967(99)00123-0 (11th ed.). Springfield, MA: Author.
Ehlers, A., Maercker, A., & Boos, A. (2000). Posttraumatic stress disorder Moser, J. S., Hajcak, G., Simons, R. F., & Foa, E. B. (2007). Posttraumatic
following political imprisonment: The role of mental defeat, alienation, stress disorder symptoms in trauma-exposed college students: The role of
and permanent change. Journal of Abnormal Psychology, 109, 45–55. trauma-related cognitions, gender, and negative affect. Journal of Anxiety
http://dx.doi.org/10.1037/0021-843X.109.1.45 Disorders, 21, 1039 –1049. http://dx.doi.org/10.1016/j.janxdis.2006.10.009
Ehlers, A., & Steil, R. (1995). Maintenance of intrusive memories in National Center for PTSD. (2014). Using the PTSC Checklist for DSM–IV
posttraumatic stress disorder: A cognitive approach. Behavioural Re- (PCL). Retrieved from http://www.ptsd.va.gov/professional/pages/
search and Cognitive Psychotherapy, 23, 217–249. http://dx.doi.org/ assessments/assessment-pdf/PCL-handout.pdf
10.1017/S135246580001585X Resick, P. A., Galovski, T. E., Uhlmansiek, M. O., Scher, C. D., Clum,
Elwood, L. S., Williams, N. L., Olatunji, B. O., & Lohr, J. M. (2007). G. A., & Young-Xu, Y. (2008). A randomized clinical trial to dismantle
Interpretation biases in victims and non-victims of interpersonal trauma components of cognitive processing therapy for posttraumatic stress
380 NANNEY, CONSTANS, KIMBRELL, KRAMER, AND PYNE

disorder in female victims of interpersonal violence. Journal of Con- Experimental Psychiatry, 38, 212–224. http://dx.doi.org/10.1016/j.jbtep
sulting and Clinical Psychology, 76, 243–258. http://dx.doi.org/10.1037/ .2006.10.010
0022-006X.76.2.243 Sanislow, C. A., Pine, D. S., Quinn, K. J., Kozak, M. J., Garvey, M. A.,
Resick, P. A., & Schnicke, M. K. (1992). Cognitive processing therapy for Heinssen, R. K., . . . Cuthbert, B. N. (2010). Developing constructs for
sexual assault victims. Journal of Consulting and Clinical Psychology, psychopathology research: Research domain criteria. Journal of Ab-
60, 748 –756. http://dx.doi.org/10.1037/0022-006X.60.5.748 normal Psychology, 119, 631– 639. http://dx.doi.org/10.1037/
Salemink, E., & van den Hout, M. (2010). Trained interpretive bias a0020909
survives mood change. Journal of Behavior Therapy and Experimental Weathers, F., Litz, B., Herman, D., Huska, J., & Keane, T. (1993, October).
Psychiatry, 41, 310 –315. http://dx.doi.org/10.1016/j.jbtep.2010.02.010 The PTSD Checklist (PCL): Reliability, validity, and diagnostic utility.
Salemink, E., van den Hout, M., & Kindt, M. (2007). Trained interpretive Paper presented at the Annual Convention of the International Society
bias: Validity and effects on anxiety. Journal of Behavior Therapy and for Traumatic Stress Studies, San Antonio, TX.

Appendix

Item Parcels

Factor
Parcel and item contenta loading

Process Parcel 1 .46

1. You are standing on post and think you see someone in the distance.
6. You are on patrol and see something on the side of the road.
Process Parcel 2 .53
2. You see soldiers removed from a helicopter.
4. Your commanding officer approaches you and says your orders have changed.
8. You hear that senior officials were gunned down.
Process Parcel 3 .51
3. You experience negative thoughts and feelings about the Iraqi people.
5. You are ordered to assist in rescue and recovery efforts.
7. You see one of your friends fall and realize he has been hit.
Product–belief Parcel 1 .83
3. I am a weak person.
5. I can’t deal with even the slightest upset.
9. I feel dead inside.
20. I have permanently changed for the worse.
26. I can’t stop bad things from happening to me.
30. My reactions since the event show that I am a lousy coper.
33. I feel like I don’t know myself anymore.
Product–belief Parcel 2 .85
4. I will not be able to control my anger and will do something terrible.
6. I used to be a happy person but now I am always miserable.
12. I am inadequate.
17. I will never be able to feel normal emotions again.
21. I feel like an object, not like a person.
24. I feel isolated and set apart from others.
28. My life has been destroyed by the trauma.
Product–belief Parcel 3 .81
2. I can’t trust that I will do the right thing.
14. If I think about the event I will not be able to handle it.
16. My reactions since the event mean that I am going crazy.
25. I have no future.
29. There is something wrong with me as a person.
36. Nothing good can happen to me anymore.
PTSD Parcel 1 .92
1. Repeated, disturbing memories, thoughts, or images of a stressful military experience
2. Repeated, disturbing dreams of a stressful military experience
3. Suddenly acting or feeling as if a stressful military experience were happening again (as if you were
reliving it)
12. Feeling as if your future will somehow be cut short
14. Feeling irritable or having angry outbursts
16. Being “super alert” or watchful on guard

(Appendix continues)
 APPRAISAL PROCESS AND APPRAISAL PRODUCTS 381

Appendix (continued)

Factor
Parcel and item contenta loading

PTSD Parcel 2 .88

6. Avoid thinking about or talking about a stressful military experience or avoid having feelings related to
it?
7. Avoid activities or talking about a stressful military experience or avoid having feelings related to it
10. Feeling distant or cut off from other people
11. Feeling emotionally numb or being unable to have loving feelings for those close to you?
13. Trouble falling or staying asleep
15. Having difficulty concentrating
PTSD Parcel 3 .89
4. Feeling very upset when something reminded you of a stressful military experience
5. Having physical reactions (e.g., heart pounding, trouble breathing, or sweating) when something reminded
you of a stressful military experience
8. Trouble remembering important parts of a stressful military experience
9. Loss of interest in things that you used to enjoy
17. Feeling jumpy or easily startled
Note. PTSD ⫽ posttraumatic stress disorder.
a
Appraisal process item content consists of general summaries of vignettes.

Received February 4, 2014

Revision received July 14, 2014
Accepted September 16, 2014 䡲