Validation and Standardization of the Dutch Trauma Symptom Checklist for Young Children in a Normative and Clinical Sample

ABSTRACT

The study aimed to describe the psychometric properties of the Dutch version of the Trauma Symptom Checklist for Young Children (TSCYC) in normative and clinical populations in the Netherlands. Caregivers’ ratings on the TSCYC were obtained for 1,802 children from the normal population, and for 515 children from a clinical population of traumatized children. In the clinical sample, additional measures were taken. The internal consistency and test-retest reliability of TSCYC scales were adequate. Confirmatory factor analysis showed acceptable fit on the putative scale structure. Regarding criterion validity, the clinical sample scored significantly higher on all clinical scales when compared to the normal population sample. Within the clinical sample, significant associations were found between TSCYC scales and convergent scales of other instruments. The Posttraumatic Stress-Total subscale demonstrated excellent discriminative ability between traumatized children and children from the normal population. The Dutch version of the TSCYC proved a valid and reliable instrument to measure trauma symptoms in young children through caregiver report, similar to the original American version. Further comparisons with diagnostic interviews are warranted.

KEYWORDS:

• Assessment
• children
• domestic violence
• measurement
• sexual abuse
• trauma

Throughout their childhood, children may be exposed to different traumatic events including but not limited to child maltreatment, witnessing domestic violence, violence by peers, community and school violence, severe diseases, traffic accidents, and natural catastrophes. Children’s exposure to such threatening life events, and even less overwhelming experiences as divorce or (mental) illness of a family member, is associated with a wide range of possible mental health outcomes (Osofsky, 1995). These outcomes include low self-esteem and other self-cognitions (Evans, Davies, & DiLillo, 2008), anxiety and depression (Springer, Sheridan, Kuo, & Carnes, 2007), post-traumatic stress and dissociation (Stuber, Shemesh, & Saxe, 2003), problematic behavior or conduct disorders (Chan & Yeung, 2009), (psycho)somatic problems (Lamers-Winkelman, De Schipper, & Oosterman, 2012; Springer et al., 2007), suicidal and self-mutilating behavior (Cicchetti, Rogosch, Sturge-Apple, & Toth, 2010), anger and aggression (Litrownik, Newton, Hunter, English, & Everson, 2003), and, especially in the case of sexual abuse, age inappropriate sexual behavior and coercive sexual behavior (Kellogg, 2010). Due to the large number of children victimized by any form of violence and the many negative psychological consequences, early identification of trauma-related problems in children is needed.

Before the Trauma Symptom Checklist for Young Children (TSCYC) was introduced by Briere et al. in 2001, there were no standardized broad spectrum trauma questionnaires available for young children, beside a nonspecific symptom questionnaire for children, the Child Behavior Checklist (CBCL; Achenbach, 1991), and the Child Sexual Behavior Inventory (CSBI; Friedrich, 1997) to assess symptoms related to sexual abuse of children. The TSCYC is a caregiver report for young children, aged 3–12 years. This instrument showed adequate psychometric properties in clinical populations (Pollio, Glover-Orr, & Wherry, 2008) and in the normal population (Radford, Corral, Bradley, & Fisher, 2013). So far, there have been few studies in which the TSCYC was translated into different languages, and psychometric properties were tested. Recent studies into a Swedish version of the TSCYC (Nilsson, Gustafsson, & Svedin, 2012) and a Korean version (Bae, Jeong, Lee, & Chung, 2015) showed promising results. In both studies the scale structure was confirmed, the internal consistency of the total scale was high (between .92 and .96), and the internal consistencies of the subscales were adequate. Furthermore, the ability to discriminate between traumatized children and children from the normal population, was evident in both studies.

The TSCYC was first translated into Dutch by Lamers-Winkelman in 2007, for use in the Haarlem Trauma Center for Children in the Netherlands. The 90 items were translated into Dutch and back translated by an American native speaker (experienced trauma specialist) who resided in the Netherlands for 20 years. Based on the differences between the back translation and the original version, and to improve comprehensibility, small adjustments to some of the items were made. The final version was sent for acceptance to the publisher of the original scale. This version was tested in this study. The aim of this study was to investigate the psychometric properties concerning gender and age and sex group differences, internal consistency, test-retest reliability, construct validity, criterion validity, convergent and divergent validity, and predictive validity, of the Dutch version of the TSCYC in a normative and clinical population in the Netherlands.

Method

Participants

The Dutch version of the TSCYC was tested in a standardization sample originating from the PanelClix panel. This internet panel consists of 150,000 members who participate in various surveys for a small fee (PanelClix online market research, 2016). To test for differences between the original age group and gender combinations in the American validation sample (Briere, 2005), a sample of caregivers of 300 boys and 300 girls in each of three age groups: 3–4, 5–9, and 10–12 years, was generated within PanelClix, sampling to a total of 1,802 children. The PanelClix algorithm selected potential respondents based on equal numbers of children (boys and girls) in the age groups. Other factors PanelClix used to generate a representative sample were geographic region (i.e., Conurbation of Western Netherlands, other Western, Northern, Eastern, and Southern Netherlands), parental education level, and ethnicity. This stratification by demographic variables addresses self-selection bias from respondents from different demographic subgroups.

Selected participants in the panel were invited by email to participate in the panel wave. First, respondents filled in the selection questions to establish that they belonged to the target group (families with children between 3–12 years old), before they were automatically assigned to the online questionnaire. Next, respondents were invited for this study with a focused invitation providing full informed consent. Participation was voluntary. Based on the answers, the respondents were assigned to six groups defined by gender and age category. Respondents who did not meet the criteria or who were no longer needed, were led to the end of the questionnaire after the initial screening questions.

Table 1 reports demographic data of the standardization sample in comparison with the general population. Of the 1,802 children, 50% were male. Within each age group, each year was evenly distributed. The average age of the caregivers was 38.3 years (range 18 to 67) and 43% of the caregivers were male.

Table 1. Demographical Characteristics of the Standardization Sample and Population.

Characteristic	Sample^a (%)	Population^b (%)
Geographical origin
Conurbation of Western Netherlands	14.1	16.6
Other Western Netherlands	29.5	29.2
Northern Netherlands	11.7	10.0
Eastern Netherlands	22.9	20.9
Southern Netherlands	21.9	23.3
Ethnic background caregiver
Dutch	77.1	77.3
Non-Western 1st generation	8.3	10.1
Non-Western 2nd generation	5.7	3.2
Western 1st generation	3.1	4.4
Western 2nd generation	5.8	5.0
Education level caregiver
Low	9.4	18.9
Middle	55.6	52.1
High	35.0	29.0

^aN = 1,802.

^bAccording to the MOA Gold standard (MOA, 2013), a calibration tool for representative sampling for national and regional samples, from census data of 2012 of Statistics Netherlands (CBS).

Clinical group

Following Briere et al. (2001), a clinical sample was assessed alongside the standardization sample. The clinical sample was composed of traumatized children referred to several childhood trauma centers in the Netherlands. The TSCYC was completed by the caregivers during the standard intake for treatment programs for traumatized children. Beside the TSCYC, the caregivers also completed the CBCL, the CSBI, the Children’s Depression Inventory (CDI; Kovacs, 1992), and the Child Dissociation Checklist (CDC; Putnam, Helmers, & Trickett, 1993). A total of 515 caregivers completed the TSCYC and provided the data anonymously for validation. The sample consisted of children of whom 50% were sexually abused, and 48% were victimized by witnessing domestic violence in the family. For 2% of the children, the form of child abuse was unknown. The mean age of the children was 7.9 years (SD = 2.4 years) and 50% were boys. Over 61% of the caregivers was of Dutch origin, 24% had another ethnic background, and of 15% of the caregivers, the ethnic origin was unknown.

Test-retest group

The test-retest group was composed of 100 randomly selected respondents from the standardization sample who agreed to complete the TSCYC a second time for the same children after 3 weeks (52 girls and 48 boys, mean age = 7.2 years, SD = 3.2).

Instruments

The TSCYC, developed by Briere et al. (2001), is a caregiver report assessment of trauma symptoms in children ages 3–12 years. The manual was published in 2005. The TSCYC consists of 90 items in the form of symptoms related to traumatic events, which parents rate on a scale from 0 (not at all) to 4 (very often) on how often it occurred in the last month. The TSCYC consists of nine clinical scales: Anxiety, Depression, Anger/Aggression, Posttraumatic Stress-Intrusion, Posttraumatic Stress-Avoidance, Posttraumatic Stress-Arousal, Posttraumatic Stress-Total, Sexual Concerns and Dissociation, and two validity scales, for underreport (Response Level) and for Atypical Response. In the original validation study, Briere (2005) found good internal reliability levels for all clinical scales in both the normative and clinical sample, ranging from α = .78 for the Anxiety scale in the normative sample to α = .93 for the Posttraumatic Stress-Total scale in the clinical sample.

The CDI (Kovacs, 1992) is one of the best studied, and most widely used scales for childhood depression (Myers & Winters, 2002). It is a self-report measure to assess depression in children (7–17 years of age), which consists of 27 items, rated on a 3-point scale (0 = not true, 1 = somewhat true, 2 = very true), reflecting the degree of symptoms over the past 2 weeks. Both a total score (from 0 to 54) and a score on five sub-scales (Negative Mood, Interpersonal Problems, Ineffectiveness, Anhedonia, and Negative Self-Esteem) are computed. Cronbach’s alpha of the total score was .88 (Kovacs, 1992). In the current study, only the total score was used.

The CSBI (Friedrich, 1997) for caregivers of children ages 2–12 years contains 38 items across the following content domains: Boundary Issues, Gender Role Behavior, Sexual Interest, Sexual Knowledge, Exhibitionism, Self-Stimulation, Sexual Intrusiveness, Voyeuristic Behavior, and Sexual Anxiety. The frequency of the behavior, as observed by the caregiver during the past 6 months, is rated on a 4-point scale ranging from 0 (never) to 3 (at least once a week). The internal consistency of the total score has been found high in both clinical and normative samples, respectively, α = .93 and α = .82 (Friedrich, 1997).

The CDC (Putnam et al., 1993) for caregivers of children ages 5–15 years contains 20 items, rated on a 3-point scale from 0 (not true) to 2 (very true), to describe the child’s behavior on a given item over the past 12 months. A total score is computed, which showed excellent reliability (α = .96) in the validation study (Putnam et al., 1993). A total score of 12 or higher gives a good indication of the presence of dissociative psychopathology (Putnam et al., 1993). The Dutch translation of the CDC was used in the present study (Hartvelt & Janssen, 1992).

The CBCL (Achenbach, 1991) is a caregiver report about competencies and behavioral problems of children from 4–18 years old. The list contains 120 items; response options range from 0 (not true) to 2 (very true or common) in the past 6 months. The list includes eight syndrome scales of which the following two were used in the present study: Anxious/Depressed (α = .76), and Aggressive Behavior (α = .76), and scales for Internalizing behavior, composed of three problem scales (α = .87) and Externalizing behavior, composed of two problem scales (α = .87), and a Total problems scale, composed of all eight problem scales (α = .94). The score on the list reveals a profile that indicates whether the child scores above the clinical cutoff or within normal range (Achenbach, 1991). There is considerable evidence for the reliability and validity of the CBCL (Achenbach, 1991) and for the Dutch translation (Verhulst, Van Der Ende, & Koot, 1996) used in the current study.

Statistical analysis

Differences in TSCYC scores between gender and age groups and the reproduction of the original gender and age groups (Briere, 2005) were tested using multivariate analysis of variance (MANOVA). Internal consistency was estimated with Cronbach’s α. Test-retest reliability was examined using the intraclass correlation coefficient (ICC). Construct validity was assessed by zero-order correlation (Pearson’s r) between TSCYC scales and by confirmatory factor analysis (CFA) for which maximum likelihood as estimation method was used (AMOS 22.0). Criterion validity was estimated by comparing differences in TSCYC scores between reference and clinical groups, using MANOVA. For this analysis, the clinical group was matched with a control group from the standardization sample. The control group consisted of 515 children with the same proportion by gender and age as in the clinical sample (257 girls and 258 boys, mean age = 7.9 years, SD = 2.4 years). Convergent and divergent validity was analyzed by zero-order correlation between TSCYC scales and corresponding scales in different instruments (CSBI, CDI, CBCL, and CDC) in the clinical sample. Predictive validity was based on receiver operating characteristics (ROC) curves for the distinction between samples on TSCYC scores.

Results

Gender and age group differences in the standardization sample

In MANOVA of gender by age group (2 * 10), multivariate main effects were found for sex, F(10, 1,773) = 4.11, p < .001, partial η² = .023, and age, F(90, 12,035.33) = 3.63, p < .001, partial η² = .020. In the original validation study by Briere (2005), the largest differences occurred at a division of age into three categories: 3–4, 5–9, and 10–12 years, one-way MANOVA, F(20, 1,476) = 5.12, p < .001, partial η² = .065. The same results were obtained in the current study with the same three age groups: one-way MANOVA, F(20, 3,580) = 12.11, p < .001, partial η² = .063. The interaction effect between gender and the three age groups was small but significant: MANOVA, F(20, 3,574) = 1.60, p = .044, partial η² = 0.009.

Scale structure in standardization sample and clinical group

The CFAs showed acceptable results with respect to the scale structure used in the original TSCYC, with eight clinical subscales and two validity subscales. Although the value of the relative chi-square (chi-square fit index divided by degrees of freedom) in the standardization sample was too high for an acceptable fit (6.17) and the value was statistically significant; the value of the relative chi-square in the clinical sample was at 2.24, more acceptable. Both the RMSEA and SRMR showed acceptable values in both samples (standardization sample: RMSEA = .054, SRMR = .063; clinical sample: RMSEA = .051, SRMR = .069). The difference in relative chi-square value might be caused by the difference in population size. Therefore, the same analysis was performed in subsamples of only respondents with (sub-) clinical scores on the TSCYC (standardization sample: n = 354; clinical sample: n = 431). The relative chi-square was almost equal in both groups (standardization sample: 2.05; clinical sample: 2.15) and showed an acceptable fit, as did the other values (standardization sample: RMSEA = .055, SRMR = .083; clinical sample: RMSEA = .052, SRMR = .071).

Reliability

Internal consistency

The internal consistency of the TSCYC was calculated in the standardization and clinical sample using Cronbach’s alpha. The results (see Table 2) demonstrated adequate internal consistency of the TSCYC scales in both samples, except for the Atypical Response scale.

Table 2. Internal Consistency of TSCYC Scales in Both Samples.

	Standardization sample^a	Clinical sample^b
Scales	α	α
Validity
Response Level	.80	.85
Atypical Response	.93	.37
Clinical
Anxiety	.78	.82
Depression	.85	.82
Anger/Agression	.86	.90
Posttraumatic Stress-Intrusion	.85	.85
Posttraumatic Stress-Avoidance	.87	.79
Posttraumatic Stress-Arousal	.80	.79
Posttraumatic Stress-Total	.92	.90
Dissociation	.86	.91
Sexual Concerns	.90	.82

^aN = 1,802.

^bN = 515.

Validity scales

The internal consistency of the Response Level scale was relatively high, with alphas of .80 and .85 for the standardization and for the clinical sample. However, the internal consistency of the Atypical Response scale was more variable across samples with an alpha of .93 for the standardization sample and an alpha of .37 for the clinical sample.

Clinical scales

In the standardization sample, the TSCYC clinical scales had alphas ranging from .78 for the Anxiety scale to .92 for the Posttraumatic Stress-Total scale (average of .85 for all clinical scales). In the clinical sample, the alphas for the clinical scales ranged from .79 (Posttraumatic Stress-Avoidance and Posttraumatic Stress-Arousal scale) to .91 for the Dissociation scale (average of .84 for all clinical scales).

Test-retest reliability

The ICC of the clinical scales (see Table 3) varied between .82 (Sexual Concerns scale) and .93 (Posttraumatic Stress-Total scale), and .82 for the Response Level scale and lowest with .71 for the Atypical Response scale.

Table 3. Test-Retest Reliability Coefficient (r) TSCYC Scales for the Standardization Sample.

	First rating^a		Second rating^a
Scale	M	SD	M	SD	ICC
Validity
RL	3.33	2.50	3.74	2.49	.82**
ATR	9.12	0.41	9.18	0.48	.71**
Clinical
ANX	11.82	2.41	11.41	2.32	.84**
DEP	10.48	2.06	10.41	2.07	.90**
ANG	11.73	2.73	11.18	2.11	.87**
PTS-I	10.21	1.71	10.05	1.56	.86**
PTS-AV	10.18	1.95	10.02	1.70	.87**
PTS-AR	12.97	3.30	12.71	3.08	.91**
PTS-TOT	33.36	5.94	32.78	5.52	.93**
DIS	12.06	2.77	11.46	2.68	.85**
SC	9.50	1.05	9.42	0.91	.82**

Note. RL = Response Level, ATR = Atypical Response, ANX = Anxiety, DEP = Depression, ANG = Anger, PTS-I = Posttraumatic Stress-Intrusion, PTS-AV = Posttraumatic Stress-Avoidance, PTS-AR = Posttraumatic Stress-Arousal, PTS-TOT = Posttraumatic Stress-Total, DIS = Dissociation, SC = Sexual Concerns, ICC = Intraclass correlation.

^an = 100.

**p < .01.

Validity

The validity of the TSCYC was evaluated by the inter-correlation between scales, the divergent and concurrent correlation with other instruments, and the diagnostic utility of the TSCYC to discriminate between the clinical and normal population.

Correlation among subscales

The inter-correlation between the TSCYC subscales (see Table 4) varied between .47 (between the Sexual Concerns scale and the Posttraumatic Stress-Arousal scale) and .91 (between the Posttraumatic Stress-Total scale and the Posttraumatic Stress-Avoidance scale). As in the U.S. and Swedish standardization samples, the validity scale Response Level correlated negatively with all other scales, ranging from −.21 with the Atypical Response scale to −.55 with the Anger/Aggression scale. The Atypical Response scale is most closely associated with the Sexual Concerns scale (.89), and the least with the Response Level scale, as mentioned above.

Table 4. Correlation Between TSCYC Scales in the Standardization Sample.

	RL	ATR	ANX	DEP	ANG	PTS-I	PTS-AV	PTS-AR	PTS-TOT	DIS	SC
RL	—
ATR	−.21**	—
ANX	−.42**	.59**	—
DEP	−.36**	.67**	.67**	—
ANG	−.55**	.53**	.58**	.59**	—
PTS-I	−.32**	.76**	.73**	.74**	.61**	—
PTS-AV	−.34**	.79**	.69**	.75**	.62**	.86**	—
PTS-AR	−.54**	.46**	.67**	.66**	.67**	.62**	.63**	—
PTS-TOT	−.46**	.73**	.78**	.80**	.71**	.90**	.91**	.87**	—
DIS	−.42**	.57**	.60**	.64**	.57**	.64**	.66**	.71**	.76**	—
SC	−.25**	.89**	.59**	.66**	.52**	.75**	.78**	.47**	.72**	.57**	—

Note. RL = Response Level, ATR = Atypical Response, ANX = Anxiety, DEP = Depression, ANG = Anger, PTS-I = Posttraumatic Stress-Intrusion, PTS-AV = Posttraumatic Stress-Avoidance, PTS-AR = Posttraumatic Stress-Arousal, PTS-TOT = Posttraumatic Stress-Total, DIS = Dissociation, SC = Sexual Concerns. N = 1802.

**p < .01.

To gauge the criterion validity of the TSCYC, Table 5 shows the differences between the clinical sample and the control group on the TSCYC clinical scales and the results of the multivariate analysis of variance (MANOVA); multivariate main effects were found for group, control or clinical, F(8, 945) = 90.17, p < .001, partial η² = .43. All scales showed significant differences between the clinical sample and the control group, with a moderate to large effect size (η² ranges from .09 to .41).

Table 5. Analysis of Variance of Clinical TSCYC Scores in the Clinical Sample and the Control Group.

	Control group		Clinical sample
	(n = 476)		(n = 478)
	M	SD	M	SD	F^a	η^2
ANX	11.42	2.52	16.09	5.23	308.72**	.24
DEP	10.56	2.33	14.83	4.71	314.13**	.25
ANG	11.51	2.83	15.60	5.59	203.11**	.18
PTS-I	10.22	1.93	15.12	5.00	396.61**	.29
PTS-AV	10.02	1.88	16.22	5.14	610.24**	.39
PTS-AR	12.42	3.19	18.24	5.29	422.45**	.31
PTS-TOT	32.67	6.10	49.58	13.13	649.83**	.41
DIS	11.27	2.95	13.99	5.37	93.83**	.09
SC	9.58	1.51	11.32	3.56	96.45**	.09

Note. ANX = Anxiety, DEP = Depression, ANG = Anger, PTS-I = Posttraumatic Stress-Intrusion, PTS-AV = Posttraumatic Stress-Avoidance, PTS-AR = Posttraumatic Stress-Arousal, PTS-TOT = Posttraumatic Stress-Total, DIS = Dissociation, SC = Sexual Concerns.

^adf = 1, 952.

**p < 01.

The sensitivity and specificity of the Posttraumatic Stress-Total subscale score was determined based on a receiver operating characteristics (ROC) analysis, in which the percentage of true-positives (sensitivity) was plotted against the false-positive rate (1—specificity) in Figure 1. The ROC curve is a measure of the ability to discriminate on the basis of the Posttraumatic Stress-Total score between children with a traumatic experience (the clinical sample) and a sample from the population sample (the control group). The area under the ROC curve amounted to .90, with a 95% confidence interval of .88 to .92.

Figure 1. ROC Curve of the posttraumatic stress-total score in the clinical sample and the control group.

Construct validity

Table 6 reports significant convergent associations between the TSCYC and the CBCL, with the highest correlation between the CBCL Aggressive Behavior scale and the TSCYC Anger/Aggression scale (.81), and CBCL Externalizing scale and the TSCYC Anger/Aggression scale (.73). The CBCL Anxious/Depressed scale correlated highest with the TSCYC Depression scale and the Posttraumatic Stress-Total scale (.70 and .62, respectively). The CBCL Internalizing scale showed the highest correlation with the TSCYC Depression (.65) and Posttraumatic Stress-Total scale (.64). The CBCL Total Problems scale correlated highest with three TSCYC scales, Posttraumatic Stress-Arousal, Posttraumatic Stress-Total and Depression (.67, .65 and .62, respectively).

Table 6. Correlation Between TSCYC Subscales, CBCL Subscales, CSBI, CDI, and CDC.

	CBCL Externalizing (n = 444)	CBCL Internalizing (n = 444)	CBCL Aggressive Behavior (n = 444)	CBCL Anxious/ Depressed (n = 444)	CBCL Total Problems (n = 444)	CSBI (n = 432)	CDI (n = 186)	CDC (n = 454)
ANX	.27**	.54**	.30**	.56**	.50**	.20**	.25**	.44**
DEP	.44**	.65**	.47**	.70**	.62**	.29**	.26**	.49**
ANG	.73**	.38**	.81**	.37**	.59**	.41**	.10	.49**
PTS-I	.30**	.52**	.30**	.52**	.48**	.22**	.17*	.41**
PTS-AV	.32**	.51**	.32**	.48**	.48**	.27**	.18*	.44**
PTS-AR	.51**	.58**	.54**	.59**	.67**	.33**	.19**	.59**
PTS-TOT	.45**	.64**	.47**	.62**	.65**	.33**	.21**	.56**
DIS	.31**	.46**	.30**	.47**	.49**	.26**	.14	.54**
SC	.37**	.28**	.37**	.33**	.37**	.59**	−.02	.44**

Note: Convergent coefficients are in boldface. ANX = Anxiety, DEP = Depression, ANG = Anger, PTS-I = Posttraumatic Stress-Intrusion, PTS-AV = Posttraumatic Stress-Avoidance, PTS-AR = Posttraumatic Stress-Arousal, PTS-TOT = Posttraumatic Stress-Total, DIS = Dissociation, SC = Sexual Concerns.

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

The CSBI (total) and the CDC scale were moderately associated with the equivalent TSCYC Sexual Concerns and Dissociation scale (.59 and .54, respectively). The CDI, however, was weakly associated with the TSCYC Depression scale. The clinical TSCYC scales were generally modestly associated (around .30) with scales that operationalized divergent symptom domains, which also demonstrated divergent validity.

Discussion

The findings of this study support the reliability and validity of the Dutch version of the Trauma Symptom Checklist for Young Children. The scale structure based on the U.S. original validation study showed an acceptable fit, with good discriminative and convergent validity, the clinical scales showed adequate to excellent internal reliability and good to excellent temporal stability, and the criterion validity of the Posttraumatic Stress-Total scale was excellent. The age and gender differences were congruent with those observed in the original version. In contrast with the original standardization study (Briere, 2005) and the Swedish standardization study (Nilsson et al., 2012), the proportion of men in the Dutch standardization sample was much higher (43%), although gender of the caregiver was not significantly associated with the study variables.

The test-retest reliability showed good to excellent temporal stability of the TSCYC clinical scales and the internal consistencies of all the scales were adequate, except the Atypical Response scale in the standardization sample, where a high alpha was found when a low alpha was expected. This phenomenon was also described in the original TSCYC Manual (Briere, 2005). In general, atypical response scales represent a tendency to respond to items which are rarely scored. Consequently, the internal consistency of such rare responses is usually low due to the limited variation in the scores. In the standardization sample a much higher standard deviation was found in the Atypical Response scores (SD = 2.1) in comparison with the clinical sample (SD = 1.0). The larger variance in the Atypical Response scores had the effect of increasing the overall variability and, as such, contributed to a higher alpha of the Atypical Response scale in the standardization sample.

The confirmatory factor analysis replicated the original ten-factor structure with acceptable Goodness of Fit values in both the standardization and the clinical sample. Based on the correlation among CBCL subscales, CSBI, CDI, and CDC with equivalent TSCYC subscales, the TSCYC showed good discriminative and convergent validity, although not all subscales correlated exactly as expected. The Depression subscale correlated highly with the CBCL Internalizing and CBCL Anxious/Depressed subscale, but lowly with the CDI. This might indicate an issue with content validity of the Depression subscale.

The comparison between the subscale scores of the standardization sample and the scores of the clinical sample showed that the clinical sample scored significantly higher on all clinical subscales, which indicates adequate criterion validity and confirmed that sexually abused children and children exposed to domestic violence were measurably more traumatized than children from the normal population. The ROC analysis of Posttraumatic Stress-Total subscale scores indicated that the Posttraumatic Stress-Total scale has an excellent ability to discriminate between children who have experienced a traumatic experience and children from the normal population. An important limitation in this study concerns the fact that the participants in the standardization sample were not questioned about possible traumatic events their children had experienced. Such a line of questioning would allow to find associations between exposure to traumatic events and TSCYC scores in the normative sample, and so contribute to criterion validity.

In his original study, Briere proposed to use T-scores, with a mean of 50 and a standard deviation of 10, for all scales based on the differences between age and gender groups and the distribution of raw scores (Briere, 2005). As the meaning of the values of T-scores are the same in most tests, the use of T-scores makes the meaning of the values more comprehensible. The same differences between the same age and gender groups were found in the Dutch study compared to Briere’s original study (2005), which would suggest the same procedure to calculate T-scores for the Dutch population.

Briere (2005) calculated standardized cutoff points for “possible concerns” and for “clinical significance.” An instrument like the TSCYC that monitors the impact of trauma on children can only be regarded as valid if it can discriminate between children who experienced traumatic events and children who have not, so the TSCYC could be used for a tentative posttraumatic stress disorder diagnosis. Future research should evaluate this function in the Dutch version in a comparison between TSCYC scores and a diagnostic interview like the Clinician-Administered PTSD Scale, Child and Adolescent version (CAPS-CA; Nader et al., 2004) or the Anxiety Disorder Interview Schedule (ADIS; Silverman & Albano, 1996). In all, the Dutch TSCYC has shown to have the characteristics required for assessing trauma symptoms in children for research, as well as screening purposes.

Funding

This article is based on research supported by a grant from the Rotterdam-Rijnmond Public Health Service in The Netherlands (PGZ/AN/mvh/2009/83/1791, Principal Investigator: Bas Tierolf).

Additional information

Funding

This article is based on research supported by a grant from the Rotterdam-Rijnmond Public Health Service in The Netherlands (PGZ/AN/mvh/2009/83/1791, Principal Investigator: Bas Tierolf).