The Child PTSD Symptom Scale: An Investigation of Its Psychometric Properties

Abstract

The current study addresses the need for accurate measurement of posttraumatic stress disorder (PTSD) symptoms in youth by investigating the psychometric properties of the Child PTSD Symptom Scale (CPSS). The factor structure, reliability, and concurrent and discriminant validity of the CPSS were investigated in a sample of 206 6th- to 12th-grade adolescents. Exploratory and confirmatory factor analysis supported a single-factor structure, which was contrary to the hypothesized three-factor structure. Scores comprising this one-factor structure were also associated with high reliability (α = .93), and tests of concurrent and discriminant validity were also strong. The implications of these findings are discussed, with particular emphasis on future directions for research on self-report measures for adolescent PTSD symptoms.

Keywords

child abuse mental health and violence history of child abuse PTSD

Increased research interest in and clinical awareness of posttraumatic reactions in children and adolescents have facilitated research examining assessment measures designed to detect the presence of posttraumatic stress disorders (PTSD) in youth. Well-established instruments are necessary to advance research that focuses on understanding how children respond to stressful events, such as violence, abuse, war, and natural disasters.

Traumatic stress experienced in childhood and adolescence (i.e., ages 0-18) has repeatedly been found to be associated with a range of negative psychological consequences (Neubauer, Deblinger, & Sieger, 2007). Children who have experienced traumatic events can show signs of depression, anxiety, and PTSD (Foa & Rothbaum, 1998), increased substance use (Kilpatrick et al., 2003), increased instances of aggressive behavior, and an increase in attention difficulties (Costello, Erkanli, Fairbank, & Angold, 2002). Childhood traumatic stress has also been shown to increase the risk of academic difficulties (Lipschitz, Rasmusson, Anyan, Cromwell, & Southwick, 2000), engagement in high-risk behaviors (Ethier, Lemelin, & Lacharite, 2004), need for medical and mental health services, and involvement with child welfare and juvenile justice systems (Felitti et al., 1998). Studies indicate that if left untreated, these difficulties may become chronic, producing negative effects that often continue into adulthood (Danielson et al., 2009; B. L. Green et al., 1994; Kilpatrick et al., 2003). These potentially serious long-term effects underscore the importance of appropriate and timely identification of children who suffer from traumatic stress reactions. Research suggests that many children with PTSD go undiagnosed (Kendall-Tackett, Williams, & Finkelhor, 1993; Makley & Falcone, 2010) and are thus not receiving needed services. Thus, there is a critical need to establish a reliable and valid method of ascertaining symptoms among children and adolescents (Silva, 2004).

Various instruments have been developed to aid in the diagnosis of PTSD in children and adolescents, including self-report measures, clinical interviews, and structured diagnostic interviews. Although structured interviews have long been considered one of the most valid forms of PTSD assessment (American Academy of Child and Adolescent Psychiatry (AACAP), 2010; Cohen, Mannarino, & Deblinger, 2006), self-report measures hold certain advantages. In contrast to clinical interviews, self-report measures are relatively concise, require minimal clinician time, and generally entail a shorter burden of administration (cf. Ebesutani, Bernstein, et al., 2012).

The Child PTSD Symptom Scale (CPSS; Foa, Johnson, Feeny, & Treadwell, 2001) is a widely used child self-report measure of posttraumatic stress. The instrument was designed to measure the severity of PTSD symptoms and functional impairment among youth ages 8 to 18 and aligned with the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association [APA], 2000). The CPSS measures the frequency of 17 PTSD symptoms (DSM-IV criteria) using a 4-point Likert-type response scale (ranging from not at all to almost always). The scale also assesses functional impairment using seven yes and no responses. The CPSS can be used as a continuous measure of symptom severity (summation of items 1-17 with possible scores ranging from 0 to 51). Items can also be scored dichotomously to provide a diagnostic status, with any symptom endorsement included as an affirmative response in this calculation. Published recommendations for clinical cutoffs using this scoring method indicate that scores above 11 are reflective of a likely PTSD diagnosis (Foa et al., 2001).

The CPSS is concise, easy to administer, available for free, and has the potential to provide a great deal of utility for both research and clinical applications. The initial validation study of the instrument (Foa et al., 2001) demonstrated good internal consistency for the total score (α = .89) and fair to good internal consistency for the subscales (r ranging between .70 and .80). The instrument demonstrated acceptable test-retest reliability over a 1- to 2-week interval (r ranging between .63 and .85 for the total score and subscales). In terms of validity, the CPSS scores correlated highly with another measure of child PTSD symptoms, the Child Posttraumatic Stress Disorder–Reaction Index (CPTSD-RI; Pynoos et al., 1987), r = .80 (Foa et al., 2001).

Nixon, Sterk, and Pearce (2012) reported comparable psychometrics of the instrument in two samples of trauma-exposed youth (ages 6-17 years). In their study, the CPSS demonstrated good internal consistency for the total score (α = .90 and α = .83) and fair to good internal consistency for the subscales (r ranging between .65 and .84). The instrument showed moderate correlation with PTSD diagnosis (.51), as measured with the Clinician Administered PTSD Scale–for Children and Adolescents (CAPS-CA; Nader et al., 1998). Although the CPSS has been used in numerous studies since its development, including child PTSD treatment research (e.g., Nixon et al., 2012; Smith et al., 2007), there has been limited evaluation of its psychometric properties. In particular, the factor structure of the instrument has yet to be evaluated.

There is some debate in the field regarding the factor structure of PTSD. The DSM-IV proposes a three-factor structure of PTSD, comprised of reexperiencing, avoidance-emotional numbing, and hyperarousal factors. The underlying dimensions of PTSD are currently represented in the DSM-5 (APA, 2013) by a four-factor structure, which splits the avoidance-emotional numbing factor into 2 distinct factors (avoidance and negative alterations in cognition and mood). PTSD factor analytic research with adults has generally supported one of two four-factor models comprised of (a) reexperiencing, avoidance, emotional numbing, and hyperarousal (King, Leskin, King, & Weathers, 1998) or (b) intrusions, avoidance, hyperarousal, and dysphoria (Simms, Watson, & Doebbeling, 2002). However, most studies have found that multiple structures fit the data well in adolescent samples (Ayer et al., 2011; Elhai, Ford, Ruggiero, & Frueh, 2009; Ford, Elhai, Ruggiero, & Frueh, 2009). Furthermore, in adolescent samples, there is evidence that PTSD factors are generally highly correlated (Ayer et al., 2011). The current study addresses the factor structure of the CPSS and seeks to increase the confidence in the instrument by testing the psychometric rigor of the tool. Despite the widespread use of the CPSS, a factor analysis of this instrument has yet to be reported in the published literature. Specifically, we examined the factor structure as well as the concurrent and discriminant validity of the CPSS scores. Based on the subscales as designed by the authors (Foa et al., 2001), it was hypothesized that (a) a CPSS three-factor model would be supported via exploratory factor analysis (EFA) and confirmatory factor analysis (CFA), (b) reliability of the CPSS subscales would be supported via adequate internal consistency, and (c) validity of the CPSS would be supported via adequate concurrent and discriminant validity.

Method

Participants

Participants were 206 6th- to 12th-grade students who were recruited from two public schools in rural Northern Mississippi. The mean age of the sample was 14.46 years (SD = 1.91, range = 11-18) and the group consisted of 121 (58.7%) girls. The ethnic makeup of the sample was 46.6% Caucasian (n = 96), 25.7% African American (n = 53), 21.8% Latino (n = 45), 2.9% Multiethnic (n = 6), and 1.5% other ethnicities (n = 3). An additional 1.5% (n = 3) of the sample chose not to report ethnicity. Regarding living situation, the average number of household members was 4.49 (range = 2-12).

Measures

CPSS

The psychometric properties of the CPSS (Foa et al., 2001) were detailed earlier in this article.

Child’s Reaction to Traumatic Events Scale–Revised (CRTES-R)

The CRTES-R (Jones, Fletcher, & Ribbe, 2002) is a 23-item self-report questionnaire designed to assess psychological responses to stressful life events in children ages 6 to 18 years. Children rate how often each item applies to them with items rated on a 4-point Likert-type scale from 0 (not at all) to 5 (often). The CRTES-R yields a total scale score, as well as three subscale scores: Intrusion, Avoidance, and Hyperarousal. To score the CRTES-R, the numerical ratings from the items that comprise each subscale are summed. These scores range from 0 to 115, with 0 to 14 representing low distress, 15 to 27 representing moderate distress, and 28 and higher representing high distress. A score of 28 or higher is indicative of a likely diagnosis of PTSD (Jones et al., 2002). There are no published psychometric data for the current 23-item CRTES-R; however, an earlier version of the measure, the CRTES (Jones, 1995; Jones, Ribbe, & Cunningham, 1994), exhibited fair to good internal consistency (α = .70-.85). The CRTES-R overlaps significantly with the original CRTES, with the major difference being the addition of 8 items that assess for arousal symptoms on the CRTES-R. The reliability of the total scale score of the CRTES-R in the current sample was .95.

Loneliness Questionnaire–Short Form (LQ-SF)

The LQ-SF (Ebesutani, Drescher, et al., 2012) is a 9-item revision of a widely used 24-item measure of loneliness in children and adolescents (the Loneliness Questionnaire; Asher, Hymel, & Renshaw, 1984). The LQ-SF was constructed through the application of item response theory on the original set of LQ items, which indicated that non-reverse-worded questions displayed a more coherent (one-factor) structure and superior psychometric properties in comparison with the reverse-worded items (Ebesutani, Drescher, et al., 2012). The reliability of this instrument in the current sample was α = .89.

Procedure

The current study was part of a larger school-based study of negative emotions in youth, which received institutional review board approval at the University of Mississippi. Participating schools sent passive consent forms with instructions for how to decline participation to parents 1 week prior to scheduled instrument administration. Data were collected schoolwide and all students who did not decline participation were administered self-report instruments in a group format. For students whose parents did not decline participation, verbal assent was obtained in a group format in their classroom prior to data collection. On the day of data collection, project staff distributed assessment instruments to students in their respective classrooms. Teachers were informed of study procedures and read a brief set of instructions prior to students completing surveys. Participants completed questionnaires anonymously and returned their forms in a large envelope when finished. Project staff remained on site in a central location to answer questions and collect envelopes from individual classrooms once all students completed their responses. The CPSS includes a question asking individuals to list their most distressing event. Although all youth answered CPSS questions, in order for individuals to be included in the study, the event listed had to meet DSM-IV-TR A-1 criteria of a traumatic event.

Data Analysis

Missingness

A total of 240 individuals filled out the CPSS measure. Among these individuals, 206 had no missing data, 22 had 1 missing item, 6 had 2 missing items, 4 had 3 missing items, 1 had 8 missing items, and 1 had 14 missing items. To examine missing data patterns, we conducted Little’s Missing Completely At Random test (MCAR test; Little & Rubin, 1989), for which the null hypothesis is that the data are MCAR. These analyses showed that our missing data were missing completely at random (MCAR), χ² = 381.09, df = 334, p = ns. Those with missing data and those without missing data also did not differ significantly (p > .01) on household size (F = 3.60, p = .06, ns), age (F = 2.81, p = .10, ns), number of close friends (F = 0.35, p = .56, ns), or the 25-item Revised Child Anxiety and Depression Scale - Child verseion Anxiety Total score (F = 3.28, p = .07) We therefore included only the 206 individuals with no missing CPSS data and excluded those with missing data. Due to the MCAR nature of the missing data, this inclusion approach is unlikely to lead to biased parameter estimates.

Descriptive statistics

Means, standard deviations, and score ranges for each measure used in the present study were calculated for the entire sample as well as separately across gender and age. CPSS item-level descriptive statistics and intercorrelations between measures were also calculated and can be found in Table 2.

Factor analysis

Prior to analyses, we first split the full sample (n = 206) into two random subsamples to conduct EFA and CFA separately on each random subsample. Beginning with the first random subsample (n = 103), we used EFA to identify the best fitting model and most interpretable solution. Based on the model derived via EFA, we then examined model fit via CFA using the second randomly generated subsample (n = 103). Only the first 17 items of the instrument, which comprise the total symptom scale, were included in the EFA and CFA analysis. The remaining 7 items make up the functional impairment scale and are a measure of this construct rather than specific symptoms. All other analyses listed below were conducted using all data gathered (n = 206) and include analyses of both the total symptom and functional impairment scales.

EFA

Given that the CPSS factor structure had never been examined, an EFA was utilized first to better understand the structure of the data. To perform these analyses, we used Mplus version 4.21 (Muthén & Muthén, 2007) and treated the data as categorical (ordinal) due to the items being derived from a Likert-type scale (Brown, 2006). Accordingly, calculations were performed on polychoric correlation matrices (Holgado-Tello, Chacón-Moscoso, Barbero-García, & Vila-Abad, 2010) and the weighted least squares estimator (WLSMV; Muthén, du Toit, & Spisic, 1997) was used, given that these are recommended procedures when conducting EFA on categorical data (Holgado-Tello et al., 2010; Muthén et al., 1997). The Geomin (oblique) rotation method, available in Mplus, was used due to the scale items being theoretically correlated. The number of factors retained was based on the following criteria: (a) visual inspection of the scree plot showing a distinct and sudden change in slope, (b) the number of eigenvalues greater than 1.0, and (c) the interpretability of the various competing factor solutions. Factor loadings greater than .32 were considered to be an adequate loading on a factor (Tabachnick & Fidell, 2007).

CFA

Based on the results obtained in the EFA procedures outlined above, we conducted CFA using Mplus version 4.21 (Muthén & Muthén, 2007) on the second randomly generated subsample to examine how well the factor structure identified in the EFA procedures fit the CPSS data. We again used the WLSMV estimator (Muthén et al., 1997), as this estimator is recommended for use when conducting CFA with categorical data (Flora & Curran, 2004). Evaluation of model fit was based on the comparative fit index (CFI; Bentler, 1990), Tucker–Lewis index (TLI; Tucker & Lewis, 1973), and the root mean square error of approximation (RMSEA; Steiger, 1990). The CFI and TLI indices are considered suitable at .90 (Bentler, 1990). The RMSEA statistic is considered to be adequate at values lower than .08 and good below .05 (Browne & Cudeck, 1993). The fit of the one-factor model was also compared with the competing three-factor model via the χ² difference test.

Internal consistency reliability

We evaluated the reliability of the CPSS total symptom scale, as informed by the factor structures indicated in each of the factor analyses outlined above. Cronbach’s alpha coefficients were used for this examination, with .80 as the cutoff for acceptable reliability (as recommended by Nunnally & Bernstein, 1994).

Concurrent and divergent validity

To further investigate the utility of the CPSS as a screening tool for PTSD symptomatology, bivariate correlations of the CPSS total symptom score were examined with concurrent and discriminant validity criteria for the entire sample. Convergence with the validity criteria was evaluated through (a) a bivariate correlation between the CPSS total symptom score and the CRTES-R total scale score and (b) a t test comparing the mean CPSS total symptom score of participants with a “high” level of distress (as determined by scores on the CRTES-R) to the mean total symptom score of participants with “low” to “moderate” levels of distress (again, based on CRTES-R scores). Regarding discriminant validity, we computed bivariate correlations to test the hypothesis that the CPSS total symptom score would correlate significantly less with the LQ-SF total score (i.e., a less related construct with PTSD) than with the CRTES-R total scale score (i.e., a more related construct with PTSD). To test this type of divergent validity, we used Fisher’s z test for correlated correlations (Meng, Rosenthal, & Rubin, 1992) to statistically compare the magnitude of these correlations. We hypothesized that the CPSS would be significantly less correlated with the LQ-SF (a less related construct with PTSD) than with the CRTES-R total scale (a more related construct with PTSD).

PTSD diagnosis

To determine if a child met PTSD symptom criteria, given the original recommendations for using the CPSS (Foa et al., 2001), the 17-symptom items were scored dichotomously to provide a diagnostic status. Any symptom endorsement (i.e., a score above 0) was included as an affirmative response in this calculation. The instrument’s authors (Foa et al., 2001) recommend a cutoff score of 11 for PTSD diagnosis while the International Society of Traumatic Stress Studies (2012) suggested that a cutoff score of 15 is more appropriate. Therefore, we conducted analyses using both cutoff scores.

Differences across age and gender

To examine potential differences in CPSS raw score response differences across age groups, children were divided into younger and older groups based on middle school (Grades 6-8) versus high school (Grades 9-12). This categorization was used to compare CPSS raw scores across gender and age group by way of a 2 × 2 ANOVA.

Results

Descriptive Statistics

Means, standard deviations, and score ranges for each measure used in the present study were calculated for the entire sample as well as separately across gender and age. These results are located in Table 1. CPSS item-level descriptive statistics and intercorrelations between measures are provided in Table 2. There was a total of 30 participants (14.6% of the total sample) who answered “not at all” across all 17 CPSS items (representing those who had no trauma exposure at all).

Table 1.

Descriptive Statistics.

	M	SD
Total sample (N = 206)
CPSS symptom severity	13.81	11.72
CRTES-R total score	28.67	26.51
LQ-SF total score	12.01	3.66
Boys (n = 85)
CPSS symptom severity	11.14	11.99
CRTES-R total score	20.98	25.74
LQ-SF total score	11.87	3.99
Girls (n = 121)
CPSS symptom severity	15.68	11.20
CRTES-R total score	33.96	25.82
LQ-SF total score	12.12	3.43
Middle schoolers (Grades 6-8; n = 110)
CPSS symptom severity	12.84	12.18
CRTES-R total score	28.80	27.90
LQ-SF total score	12.16	3.79
High schoolers (Grades 9-12; N = 96)
CPSS symptom severity	14.92	11.13
CRTES-R total score	28.53	24.97
LQ-SF total score	11.84	3.51

Note. CPSS = Child PTSD Symptom Scale; PTSD = posttraumatic stress disorder; CRTES-R = Child’s Reaction to Traumatic Events Scale–Revised; LQ-SF = Loneliness Questionnaire–Short Form.

Table 2.

CPSS Item-Level Descriptive Statistics and Intercorrelations Between All Variables.

CPSS Item	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
Item-total correlation	.66	.67	.69	.69	.69	.69	.69	.56	.61	.68	.53	.6	.63	.64	.7	.66	.61
CRTES-R	.58*	.51*	.57*	.63*	.50*	.65*	.55*	.48*	.43*	.52*	.40*	.47*	.54*	.54*	.57*	.56*	.57*
LQ-SF	.33*	.33*	.28*	.29*	.28*	.30*	.27*	.19*	.33*	.46*	.22*	.41*	.34*	.33*	.33*	.41*	.34*
Full sample
M	0.87	0.77	0.86	1.05	0.71	1.02	0.71	0.59	0.72	0.76	0.57	0.71	0.7	0.91	1.06	0.97	0.82
SD	0.93	0.87	1.02	1.02	0.93	1.06	0.97	0.89	0.94	1.03	0.93	1	0.96	1.09	1.1	1.12	1.04
Maximum	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3
Boys
M	0.71	0.58	0.74	0.74	0.56	0.86	0.49	0.53	0.58	0.52	0.52	0.72	0.58	0.79	0.88	0.81	0.54
SD	0.87	0.88	1.01	0.97	0.87	1.09	0.85	0.87	0.89	0.88	0.97	1.09	0.92	1.09	1.06	1.11	0.87
Maximum	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3
Girls
M	0.98	0.91	0.94	1.26	0.82	1.13	0.87	0.63	0.83	0.93	0.6	0.7	0.79	0.99	1.19	1.07	1.02
SD	0.95	0.84	1.02	1.01	0.97	1.02	1.02	0.91	0.96	1.09	0.91	0.95	0.98	1.09	1.11	1.12	1.1
Maximum	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3
Middle school
M	0.83	0.71	0.81	1.06	0.61	0.93	0.65	0.53	0.62	0.71	0.4	0.73	0.75	0.85	1.04	0.83	0.81
SD	0.94	0.87	1.07	1.09	0.92	1.09	1.11	0.9	0.91	1.06	0.85	1.06	0.99	1.09	1.17	1.14	1.06
Maximum	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3
High school
M	0.92	0.84	0.92	1.03	0.83	1.13	0.79	0.66	0.84	0.82	0.76	0.69	0.66	0.98	1.09	1.13	0.83
SD	0.91	0.86	0.96	0.93	0.94	1.02	0.94	0.88	0.97	0.98	0.99	0.94	0.92	1.01	1.02	1.08	1.01
Maximum	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3

Note. CPSS = Child PTSD Symptom Scale; PTSD = posttraumatic stress disorder; CRTES-R = Child’s Reaction to Traumatic Events Scale–Revised; LQ-SF = Loneliness Questionnaire–Short Form.

p < .01.

EFA

The eigenvalues associated with the initial EFA were 9.64, 1.37, 1.01, 0.87, 0.78, 0.61, 0.50, 0.44, 0.36, 0.28, 0.27, 0.25, 0.20, 0.19, 0.17, 0.06, and 0.02. Although three eigenvalues were above 1.0, the first eigenvalue (of 9.64) was much greater relative to the other eigenvalues over 1.0 (i.e., 1.37 and 1.01), suggesting the presence of a strong unidimensional factor underlying the data. All items loaded significantly and highly on the one-factor model, ranging from .66 to .82. In addition to the one-factor solution being most interpretable, the two-factor and three-factor EFA models did not appear to evidence loading patterns consistent with any theoretically based two- or three-factor model. These EFA results thus converged to support a one-factor latent construct being measured by the CPSS. The factor loadings associated with the three-factor (correlated-traits) model, two-factor (correlated-traits) model, and one-factor model appear in Table 3.

Table 3.

Factor Loadings for Exploratory Factor Analysis of the CPSS Total Symptom Scale Among the Random EFA Subsample (n = 103).

Items	One-Factor Model	Two-Factor (Correlated-Traits) Model		Three-Factor (Correlated-Traits) Model			Item Content
Items	One-Factor Model	Factor 1	Factor 2	Factor 1	Factor 2	Factor 3	Item Content
CPSS 1	.79*	.87*	.00	.94*	.07	−.39*	Having upsetting thoughts or images about the event that came into your head when you didn’t want them to
CPSS 2	.76*	.57*	.6	.83*	−.02	−.07	Having bad dreams or nightmares
CPSS 3	.78*	.96*	−.11	1.05*	.00	−.51*	Acting or feeling as if the event was happening again
CPSS 4	.82*	.59*	.31*	.73*	.19	−.12	Feeling upset when you think about or hear about the event
CPSS 5	.77*	.18	.66*	.00	.85*	.09	Having feelings in your body when you think about or hear about the event
CPSS 6	.76*	.36*	.47*	.43*	.39*	.01	Trying not to think about, talk about, or have feelings about the event
CPSS 7	.78*	−.01	.84*	.00	.70*	.33*	Trying to avoid activities, people, or places that remind you of the traumatic event
CPSS 8	.64*	.09	.60*	.07	.58*	.15	Not being able to remember an important part of the upsetting event
CPSS 9	.74*	−.11	.89*	.02	.53*	.46*	Having much less interest in doing things you used to do
CPSS 10	.79*	.03	.82*	.38	.10	.58*	Not feeling close to people around you
CPSS 11	.66*	−.05	.75*	−.02	.57*	.35*	Not being able to have strong feelings
CPSS 12	.75*	.57*	.25	.96*	−.24	.00	Feeling as if your future plans or hopes will not come true
CPSS 13	.75*	.62*	.21	1.19*	−.51*	.01	Having trouble falling or staying asleep
CPSS 14	.72*	.29*	.51*	.62*	−.01	.24	Feeling irritable or having fits of anger
CPSS 15	.78*	.17	.68*	.49*	.13	.35*	Having trouble concentrating
CPSS 16	.75*	.38*	.44*	.72*	−.05	.16	Being overly careful
CPSS 17	.69*	.75*	.01	.87*	−.01	−.30*	Being jumpy or easily startled

Note. CPSS = Child PTSD Symptom Scale; PTSD = posttraumatic stress disorder; EFA = exploratory factor analysis.

Items with significant factor loadings.

CFA

When the one-factor solution was evaluated in a CFA context, model fit was adequate (RMSEA = .08, CFI = .96, TLI = .95); however, the factors were highly intercorrelated (Factor 1 and Factor 2: r = .95, Factor 1 and Factor 3: r = .91, Factor 2 and Factor 3: r = .88), suggesting a lack of differentiation between the three factors and convergence with the findings of the initial EFA (i.e., unidimensionality of the measure). In addition, the χ² difference test comparing the one-factor and three-factor (correlated-traits) solutions was non-significant at the p = .001 level, $χ_{diff}^{2}$ (3) = 16.27, p = .0022. Thus, the increased complexity associated with the three-factor model was not justified from a statistical standpoint (despite evidencing relatively good model fit indices). Given the consistency of results for the one-factor solution, and the lack of comportment of individual items with their intended subscales in the multidimensional structure examined, the single-factor solution was chosen as the ideal structure for these data. The factor loadings of this single-factor solution may be seen in Table 4. This structure was thus used for all analyses that follow.

Table 4.

Factor Loadings for Confirmatory Factor Analysis of the One-Factor CPSS Total Symptom Scale Among the Random CFA Subsample (n = 103).

Item	Factor Loading	Item Content
CPSS 1	.79	Having upsetting thoughts or images about the event when you didn’t want to
CPSS 2	.76	Having bad dreams or nightmares
CPSS 3	.78	Acting or feeling as if the event was happening again
CPSS 4	.82	Feeling upset when you think about or hear about the event
CPSS 5	.77	Having feelings in your body when you think about or hear about the event
CPSS 6	.76	Trying not to think about, talk about, or have feelings about the event
CPSS 7	.78	Trying to avoid activities, people, or places that remind you of the traumatic event
CPSS 8	.64	Not being able to remember an important part of the upsetting event
CPSS 9	.74	Having much less interest in doing things you used to do
CPSS 10	.79	Not feeling close to people around you
CPSS 11	.66	Not being able to have strong feelings
CPSS 12	.75	Feeling as if your future plans or hopes will not come true
CPSS 13	.75	Having trouble falling or staying asleep
CPSS 14	.72	Feeling irritable or having fits of anger
CPSS 15	.78	Having trouble concentrating
CPSS 16	.75	Being overly careful
CPSS 17	.69	Being jumpy or easily startled

Note. CPSS = Child PTSD Symptom Scale; PTSD = posttraumatic stress disorder; CFA = confirmatory factor analysis.

Internal Consistency Reliability

The CPSS total symptom scale as well as the functional impairment scale demonstrated high internal consistency with the full sample, with alpha coefficients of .93 and .91, respectively. We also examined internal consistency across age group (i.e., middle school and high school) and gender. The CPSS demonstrated good internal consistency across age groups for the total symptom scale (middle school: α = .94, high school: α = .94) and functional impairment scale (middle school: α = .92, high school: α = .91). With regard to gender, the CPSS also evidenced good internal consistency for the symptom severity scale (boys: α = .95, girls: α = .92) and functional impairment scale (boys: α = .91, girls: α = .91).

Concurrent and Discriminant Validity

Concurrent validity of the CPSS was supported, as evidenced by the CPSS total symptom score correlating positively and significantly with the CRTES-R total scale score (r = .77, p < .001). In addition, the mean CPSS score of youth with high scores on the CRTES-R was 21.44 (SD = 10.03), which was significantly higher than the mean score of 6.41 (SD = 7.76) from children with low to moderate scores on the CRTES-R, t(197) = 11.89, p = .003. In terms of discriminant validity, as predicted, the CPSS was significantly less correlated with the LQ-SF (r = .46, p < .001) than with the CRTES-R (r = .77, p < .001) as determined by the results of a Fisher’s z test (Meng et al., 1992), z (205) = −5.22, p = .000.

PTSD Diagnosis According to CPSS Symptom Criteria

When evaluated dichotomously using the cutoff score of 11 for a diagnosis of PTSD, as recommended by the instrument’s authors (Foa et al., 2001), 41.3% of the sample reported symptoms consistent with a PTSD diagnosis. When utilizing the cutoff score of 15, as recommended by the International Society of Traumatic Stress Studies (2012) 15.5% of the sample met criteria for PTSD diagnosis.

Differences Across Age and Gender

We found no significant differences in CPSS total scores across younger and older subsamples using ANOVA, F(1, 204) = 1.62, p = .21. We did, however, find significant differences in CPSS total symptom scores across gender. Specifically, girls reported significantly greater CPSS total symptom scores (M = 15.68, SD = 11.20) compared with boys (M = 11.14, SD = 11.99), F(1, 204) = 7.72, p = .006.¹

Discussion

The goal of this study was to contribute to the body of research on the assessment of PTSD in youth by further investigating the psychometric properties of the CPSS, which was originally developed to include a total symptom scale comprising three subscales (Intrusion, Avoidance, and Hyperarousal; Foa et al., 2001). Results from EFA and CFA conducted in the present study, however, suggested substantial modification of the previously posited structure in that a single factor was supported. Fit indices for the encountered single-factor model were good, and although fit indices for the three-factor structure were also strong, the three-factor model exhibited highly correlated factors and items that did not correspond to theory in terms of their associated subscales. In addition, the χ² difference test comparing the one-factor and three-factor solutions was non-significant, indicating that the three-factor model was not associated with significantly better model fit relative to the more parsimonious one-factor model and thus did not justify increasing the complexity of the model/factor structure. Based on the χ² difference test results, in combination with the highly correlated three-factor structure, and the already good-fitting one-factor model, these CFA results support the CPSS as being a measure of a one-factor construct. The finding of a single-factor structure suggests that the items are largely reflective of a single common latent variable; thus, the most appropriate scoring procedure may be unidimensional rather than multidimensional. This finding is consistent with a burgeoning literature that has suggested similar results among other anxiety related measures (cf. Ebesutani, Reise, et al., 2012). In addition, this may also suggest that the CPSS does not adequately assess the individual domains of intrusion, avoidance/numbing, and hyperarousal as set forth by the DSM as distinct constructs. Alternatively, it may also indicate that the current taxonomic, categorical separation of PTSD symptom criteria is a function of convenience more than true differences between these domains.

Although the three-factor structure of the CPSS was not confirmed, the measure demonstrated strong psychometric results as interpreted in the present sample of school-based adolescents. Specifically, both the total symptom and functional impairment scales demonstrated excellent internal consistency. The total symptom scale also exhibited good concurrent and discriminant validity, as demonstrated by its correlations with the CRTES-R and LQ-SF.

Several limitations of the present study should be noted. First, although the CPSS was designed for use with youth ages 8 to 18, the current study included only youth ages 11 to 18. Future studies would benefit from inclusion of a wider age range commensurate with the full range of intended application of the instrument. Another limitation of the study is that it was based on DSM-IV-TR criteria. Although many of the symptom criteria for PTSD remained the same between DSM-IV-TR and DSM-V (APA, 2013), distinct differences do exist and thus future studies would benefit from evaluation of an updated version of the CPSS that would align with updated diagnostic criteria for the DSM-V.

It is also important to note that comparing probable diagnoses obtained from the CPSS with those derived from a clinical interview such as the Clinician Administered PTSD Scale for Children and Adolescents (CAPS-CA; Nader et al., 1996) would have represented a more rigorous test of the accuracy of the PTSD diagnosis obtained using the CPSS. Due to logistical difficulties, however, the administration of one-on-one interviews was not feasible for this study. Instead, the analogue method of dichotomous scoring with a cutoff score of 11 resulted in a prevalence rate that is considerably higher than findings reported in epidemiological studies of PTSD in the general child population. Although the prevalence rate encountered when using the cutoff score of 15 was markedly lower, it was still above typically reported rates (Copeland, Keeler, Angold, & Costello, 2007; Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995; Kilpatrick et al., 2003). In contrast, the utilization of T-scores as the criterion for probable diagnostic status resulted in findings that more closely align with typically reported findings (Copeland et al., 2007; Kessler et al., 1995; Kilpatrick et al., 2003). These results may suggest that dichotomous scoring could artificially inflate the percentage of children qualifying for a PTSD diagnosis. With dichotomous scoring, any endorsement is weighted equally, thus failing to incorporate severity of individual items into total scores. Alternatively, the results could also indicate that PTSD was more prevalent in our sample than in the population at large, although given the very high percentage of children meeting PTSD criteria using the lower threshold for the instrument this is not likely to constitute a plausible explanation. Given the wide ranging results based on scoring method or cutoff score used, future research would benefit from investigation of appropriate scoring for the CPSS by utilizing a Receiver Operating Curve (ROC) analysis (D. M. Green & Swets, 1966; Kraemer, 1992) or other appropriate analyses to determine optimal cutoff scores. Another limitation was that our relatively small sample size precluded our ability to conduct measurement invariance tests to examine whether the CPSS items are invariant across meaningful subgroups, such as between males and females. It will be important for future studies to examine the measurement invariance properties of the CPSS items to better understand the degree to which comparison of raw scale scores are interpretable. Our study was also based on data collected from the same assessment method (i.e., self-report), which likely introduced a non-trivial amount of shared method variance across all assessment measures in our study that should be considered when interpreting results.

Despite the noted limitations, the present study contributed to the psychometric knowledge regarding the CPSS and broadened the scope of investigations by including a population of school-based adolescents. This investigation highlights a number of important issues to consider in future studies. Future research will benefit from examination of the CPSS in a greater range of contexts (e.g., in other communities, other languages), investigation with more stringent concurrent validity criteria (e.g., in comparison with a structured clinical interview), and investigation of the suitability of the measurement construction in light of updated DSM-V PTSD symptom criteria. These future efforts can be informed by the results of the present study, which may allow for enhanced clinical and research applications of this widely disseminated instrument.

Footnotes

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) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Regan W. Stewart, PhD, is a predoctoral intern in the Charleston Consortium Psychology Internship Program at the Medical University of South Carolina and a student in the clinical psychology program at the University of Mississippi. Her research interests include the impact of trauma and violence on child and adolescent psychological functioning and ethnic differences in victimization.

Chad Ebesutani, PhD, is an assistant professor in the psychology department at Duksung Women’s University. His research interests focus on youth evidence-based assessment and treatment practices and systems that support their dissemination.

Christopher F. Drescher, PhD, is a postdoctoral fellow at Georgia Regents University. His research interests focus on youth development, disaster psychology, and international psychology.

John Young, PhD, is an associate professor of psychology at the University of Mississippi. His research interests focus on dissemination of evidence-based services for children and adolescents, particularly tools for scientific assessment.

References

American Academy of Child and Adolescent Psychiatry. (2010). Revised practice parameters for the assessment and treatment of posttraumatic stress disorder in children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 49, 414-430.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.). Washington, DC: Author.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Asher

S. R.

Hymel

Renshaw

P. D.

(1984). Loneliness in children. Child Development, 55, 1456-1464.

Ayer

L. A.

Cisler

J. M.

Danielson

C. K.

Amstadter

A. B.

Saunders

B. E.

Kilpatrick

D. G.

(2011). Adolescent posttraumatic stress disorder: An examination of factor structure reliability in two national samples. Journal of Anxiety Disorders, 25, 411-421.

Bentler

P. M.

(1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

Brown

T. A.

(2006). Confirmatory factor analysis for applied research. New York, NY: Guilford Press.

Browne

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136-159). Newbury Park, CA: Sage.

Cohen

J. A.

Mannarino

A. P.

Deblinger

(2006). Treating trauma and traumatic grief in children and adolescents. New York, NY: Guilford Press.

10.

Copeland

W. E.

Keeler

Angold

Costello

E. J.

(2007). Traumatic events and posttraumatic stress in childhood. Archives of General Psychiatry, 63, 577-584.

11.

Costello

E. J.

Erkanli

Fairbank

J. A.

Angold

(2002). The prevalence of potentially traumatic events in childhood and adolescence. Journal of Traumatic Stress, 15, 99-112.

12.

Danielson

C. K.

Amstadter

A. B.

Dangelmaier

R. E.

Resnick

H. S.

Saunders

B. E.

Kilpatrick

D. G.

(2009). Trauma-related risk factors for substance abuse among male versus female young adults. Addictive Behaviors, 34, 395-399.

13.

Ebesutani

C. K.

Bernstein

Chorpita

Weisz

(2012, March). A transportable assessment protocol for prescribing youth psychosocial treatments in “real world” settings: Reducing assessment burden via self-report scales. Psychological Assessment, 24(1), 141-155.

14.

Ebesutani

Drescher

C. F.

Reise

S. P.

Heiden

Hight

T. L.

Damon

J. D.

Young

(2012). The Loneliness Questionnaire-Short Version: An evaluation of reverse-worded and non-reverse-worded items via item response theory. Journal of Personality Assessment, 94, 186-195.

15.

Ebesutani

Reise

Chorpita

B. F.

Ale

Regan

Young

. . . Weisz

(2012). The Revised Child Anxiety and Depression Scale—short version: Scale reduction via exploratory bifactor modeling of the broad anxiety factor. Psychological Assessment, 24, 833-845.

16.

Elhai

J. D.

Ford

J. D.

Ruggiero

K. J.

Frueh

B. C.

(2009). Diagnostic alterations for post-traumatic stress disorder: Examining data from the National Comorbidity Survey Replication and National Survey of Adolescents. Psychological Medicine, 39, 1957-1966.

17.

Ethier

Lemelin

Lacharite

(2004). A longitudinal study of the effects of chronic maltreatment on children’s behavioral and emotional problems. Child Abuse & Neglect, 28, 1265-1278.

18.

Felitti

Anda

Nordenberg

Williamson

Spitz

Edwards

. . . Marks

(1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. American Journal of Preventative Medicine, 14, 245-258.

19.

Flora

Curran

(2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods, 9, 466-491.

20.

Foa

E. B.

Johnson

K. M.

Feeny

N. C.

Treadwell

K. R.

(2001). The Child PTSD Symptom Scale: A preliminary examination of its psychometric properties. Journal of Clinical Child Psychology, 30, 376-384.

21.

Foa

E. B.

Rothbaum

B. O.

(1998). Treating the trauma of rape: Cognitive behavioral therapy for PTSD. New York, NY: Guilford Press.

22.

Ford

J. D.

Elhai

J. D.

Ruggiero

K. J.

Frueh

B. C.

(2009). Refining posttraumatic stress disorder diagnosis: Evaluation of symptom criteria with the National Survey of Adolescents. Journal of Clinical Psychiatry, 70, 748-755.

23.

Green

B. L.

Grace

M. C.

Vary

M. G.

Kraner

T. L.

Gleser

G. C.

Leonard

A. C.

(1994). Children of disaster in the second decade: A 17 year follow-up of the Buffalo Creek Survivors. Journal of the American Academy of Child & Adolescent Psychiatry, 30, 945-951.

24.

Green

D. M.

Swets

T. A.

(1966). Signal detection theory and psychophysics. New York, NY: Wiley.

25.

Holgado-Tello

Chacón-Moscoso

Barbero-García

Vila-Abad

(2010). Polychoric versus Pearson correlations in exploratory and confirmatory factor analysis of ordinal variables. Quality & Quantity, 44, 153-166.

26.

International Society of Traumatic Stress Studies. (2012). Child PTSD Symptom Scale. Retrieved from www.istss.org/assessing-trauma/child-ptsd-symptom-scale.aspx

27.

Jones

R. T.

(1995). Review of child’s reaction to Traumatic Events Scale (CRTES). In Stamm

B. H.

(Ed.), Measurement of stress, trauma and adaptation (pp. 78-80). Lutherville, MD: Sidran Press.

28.

Jones

R. T.

Fletcher

Ribbe

D. R.

(2002). Child’s Reaction to Traumatic Events Scale-Revised (CRTES-Revised). Psychological Medicine, 46, 216-219.

29.

Jones

R. T.

Ribbe

D. P.

Cunningham

(1994). Psychosocial correlates of fire disaster among children and adolescents. Journal of Traumatic Stress, 7, 117-122.

30.

Kendall-Tackett

K. A.

Williams

L. M.

Finkelhor

(1993). Impact of sexual abuse on children: A review and synthesis of recent empirical studies. Psychological Bulletin, 113, 164-180.

31.

Kessler

R. C.

Sonnega

Bromet

Hughes

Nelson

C. B.

(1995). Posttraumatic stress disorder in the National Comorbidity Survey. Archives of General Psychiatry, 52, 1048-1060.

32.

Kilpatrick

D. G.

Rugierro

K. J.

Acierno

Saunders

B. E.

Resnick

H. S.

Best

C. L.

(2003). Violence and risk of PTSD, major depression, substance abuse/dependence, and comorbidity: Results from the National Survey of Adolescents. Journal of Consulting and Clinical Psychology, 71, 692-700.

33.

King

D. W.

Leskin

G. A.

King

L. A.

Weathers

F. A.

(1998). Confirmatory factor analysis of the clinician-administered PTSD Scale: Evidence for the dimensionality of posttraumatic stress disorder. Psychological Assessment, 10, 90-96.

34.

Kraemer

H. C.

(1992). Evaluating medical tests. Newbury Park, CA: Sage.

35.

Lipschitz

D. S.

Rasmusson

Anyan

Cromwell

Southwick

(2000). Clinical and functional correlates of posttraumatic stress disorder in urban adolescent girls at a primary care clinic. Journal of the American Academy of Child & Adolescent Psychiatry, 39, 1104-1111.

36.

Little

R. J.

Rubin

D. B.

(1989). The analysis of social science data with missing values. Sociological Methods & Research, 18, 292-326.

37.

Makley

Falcone

(2010). Posttraumatic stress disorder in the pediatric trauma patient. Seminars in Pediatric Surgery, 19, 292-299.

38.

Meng

X. L.

Rosenthal

Rubin

D. B.

(1992). Comparing correlated correlation coefficients. Psychological Bulletin, 111, 172-175.

39.

Muthén

du Toit

S. H. C.

Spisic

(1997). Robust inference using weighted least squares and quadratic estimating equations in latent variable modeling with categorical and continuous outcomes. Psychometrika, 75, 1-45.

40.

Muthén

(2007). Mplus 4.21. Los Angeles, CA: Author.

41.

Nader

K. O.

Kriegler

J. A.

Blake

D. D.

Pynoos

R. S.

Newman

Weather

F. W.

(1996). Clinician administered PTSD scale, child and adolescent version. White River Junction, VT: National Center for PTSD.

42.

Nader

K. O.

Newman

Weathers

F. W.

Kaloupek

D. G.

Kriegler

J. A.

Blake

D. D.

Pynoos

R. S.

(1998). Clinician-administered PTSD scale for children and adolescents (CAPS–CA). Boston, MA: National Center for Posttraumatic Stress Disorder.

43.

Neubauer

Deblinger

Sieger

(2007). Trauma-focused cognitive-behavioral therapy for child sexual abuse and exposure to violence: Case of Mary, age 6. In Webb

(Ed.), Play therapy with children in crisis: Individual, group, and family treatment (3rd ed., pp. 107-132). New York, NY: Guilford Press.

44.

Nixon

R. D. V.

Sterk

Pearce

(2012). A randomized trial of cognitive behaviour therapy and cognitive therapy for children with posttraumatic stress disorder following single-incident trauma. Journal of Abnormal Child Psychology, 40, 327-337.

45.

Nunnally

Bernstein

(1994). Psychometric theory (3rd ed.). New York, NY: McGraw-Hill.

46.

Pynoos

R. S.

Frederick

Nader

Arroyo

Steinberg

Eth

. . . Fairbanks

(1987). Life threat and posttraumatic stress in school-age children. Archives of General Psychiatry, 44, 1057-1063.

47.

Silva

(2004). Post traumatic stress disorders in children and adolescents. New York, NY: W.W. Norman.

48.

Simms

L. J.

Watson

Doebbeling

B. N.

(2002). Confirmatory factor analysis of posttraumatic stress symptoms in deployed and nondeployed veterans of the Gulf War. Journal of Abnormal Psychology, 111, 637-647.

49.

Smith

Yule

Perrin

Tranah

Dalgleish

Clark

D. M.

(2007). Cognitive-behavioral therapy for PTSD in children and adolescents: A preliminary randomized controlled trial. Journal of the American Academy of Child & Adolescent Psychiatry, 46, 1051-1061.

50.

Steiger

J. H.

(1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173-180.

51.

Tabachnick

B. G.

Fidell

L. S.

(2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon.

52.

Tucker

L. R.

Lewis

(1973). A reliability coefficient for maximum likelihood factor analysis. Psychometrika, 38, 1-10.