Validity evidence of the Penn State Worry Questionnaire-Children in Colombian Children

Title

Validity evidence of the Penn State Worry Questionnaire-Children in Colombian Children

Complete Reference

Barajas, H. N., & Ruiz, F. J. (2025). Validity evidence of the Penn State Worry Questionnaire-Children in Colombian Children. Revista de Psicología Clínica con Niños y Adolescentes, 12(1), 50-57. doi: 10.21134/rpcna.2025.12.1.6

Study Type

Instrument validation study: Spanish translation and analysis of psychometric properties of the Penn State Worry Questionnaire for Children (PSWQ-C) in Colombian child population.

---

BACKGROUND AND OBJECTIVES

Theoretical Background

Worry is defined as a feeling of uncertainty and a chain of negative thoughts about the possible occurrence of undesired future events. Although worry has an adaptive function, when it persists and generates recurrent distress despite the low probability of occurrence of catastrophic events, it becomes pathological. Pathological worry is the main characteristic of generalized anxiety disorder (GAD), which presents high worldwide prevalence in adults, adolescents, and children (between 5% and 20% in child populations according to various studies).

Despite the high prevalence of GAD in children, fewer psychological measurement studies exist in this population compared to adolescents and adults. The onset of worry symptoms may occur around five years of age, although there is limited information on this topic.

The Penn State Worry Questionnaire for Children (PSWQ-C) is the most extensively studied instrument for measuring pathological worry in children. It has been translated into multiple languages (French, Dutch, Danish, Italian, Romanian, Korean, and Chinese), although in some of these languages the reverse-scored items have generated problems with internal consistency and factor structure. Until now, there was no formal validation of the PSWQ-C in Spanish.

Objectives

To translate the PSWQ-C into Spanish and analyze its psychometric properties in a large sample of Colombian children aged 8 to 12 years, including analyses of internal consistency, factor structure, measurement invariance, and convergent and discriminant validity.

---

METHOD

Participants

The sample consisted of 585 participants (52.3% male) aged between 8 and 12 years (M = 10.35, SD = 1.25). Children came from low-income households and attended public schools in the department of Boyacá, Colombia. 63.34% were in elementary school and 36.75% were in middle school.

Inclusion criteria: obtaining authorization from the child and parents/legal guardians, being between 8 and 12 years old, and being able to read.

Exclusion criteria: the only exclusion criterion was a child's decision not to participate.

Recruitment procedure: Of the total 850 parents invited, 76.47% consented to their children's participation. Of these, 90% of the children participated by signing informed consent. The remaining 10% did not enter the study because they moved to another city or educational institution (8%) or did not consent to participate (2%). Evaluations were conducted in school auditoriums with collective administration of questionnaires.

Assessed Instrument: PSWQ-C

Penn State Worry Questionnaire for Children (PSWQ-C; Chorpita et al., 1997)

• Self-report measure assessing general aspects of pathological worry in children and adolescents
• Composition: 14 original items (11 direct + 3 reverse-scored)
• Response scale: 4-point Likert-type scale (0 = never true; 3 = always true)
• Reverse-scored items were: PSWQ-2 ("I don't really worry about things"), PSWQ-7 ("I find it easy to stop worrying when I want"), PSWQ-9 ("I never worry about anything")
• Previous studies demonstrated one-factor structure, good internal consistency, and discriminant and convergent validity
• The three reverse-scored items have shown discrimination problems across multiple languages

Translation procedure: Approval was obtained from the original corresponding author (B. Chorpita). Translation was conducted following guidelines of the International Test Commission for adapting instruments across cultures (Muñiz et al., 2013). An iterative procedure was implemented with two independent forward translations by bilingual individuals, reviewed by a committee consisting of the two previous translators and two experts on the topic. A small pilot test was conducted with approximately 20 Colombian children receiving psychological therapy to explore item comprehension.

Translation table: All 14 items were translated into Spanish (see Table 1 of original article).

Other Measurement Instruments

1. Depression Anxiety Stress Scale - Youth (DASS-Y; Szabo & Lovibond, 2022; Spanish version by Ruiz et al., submitted)

• Self-report measure with 21 items
• 4-point Likert scale (0 = not true; 3 = very true)
• Assesses affective state in three dimensions: depression, anxiety, and stress
• Demonstrated three-factor structure with internal consistency indicators between .77 and .81 in Colombian child population

2. Perseverative Thinking Questionnaire - Children (PTQ-C; Bijttebier et al., 2015; Spanish version by Ruiz et al., 2020)

• Self-report measure assessing Repetitive Negative Thinking (RNT)
• 15 items on 5-point Likert scale (0 = never; 4 = almost always)
• Higher scores indicate higher levels of RNT
• Demonstrated excellent internal consistency, one-factor structure invariant across gender and age group, and convergent validity

3. Revised Child Anxiety and Depression Scale - 30 (RCADS-30; Sandín et al., 2010)

• Screening measure for anxiety disorders and depression in children and adolescents
• 30 items on 4-point Likert scale (0 = never; 3 = always)
• Originally validated in Spanish by Sandín et al. (2010)
• Six subscales with five items each: panic disorder, social phobia, separation anxiety disorder, GAD, obsessive-compulsive disorder, and major depressive disorder
• Demonstrated good internal consistency, six-factor structure, and discriminant validity

Data Analysis

All analyses were conducted in JASP 0.18.2.0.

Item functioning: Corrected item-total correlations were calculated for both direct and reverse-scored items. Items with discrimination indices below .20 were eliminated.

Internal consistency: Cronbach's alpha and McDonald's omega were calculated with 95% confidence intervals.

Internal structure validity: Confirmatory factor analysis (CFA) of the one-factor model of the PSWQ-C was conducted using robust maximum likelihood (MLR). The Satorra-Bentler chi-square test was calculated and goodness-of-fit indices were reported: RMSEA (root mean square error of approximation), CFI (comparative fit index), NNFI (non-normed fit index), and SRMR (standardized root mean square residual). Criteria: RMSEA < 0.05 = good fit, < 0.08 = acceptable fit; SRMR < 0.05 = very good fit, < 0.08 = good fit; CFI and NNFI > 0.95 = good fit, > 0.90 = acceptable.

Measurement invariance: Additional CFAs were conducted to analyze measurement invariance across schools, gender (boys and girls), and age group (8-10 years and 11-12 years). Guidelines were followed to test metric, scalar, and strict invariances by analyzing whether factor loadings, item intercepts, and error variances were invariant. The relative fit of four progressively more restrictive models was compared: configural model, metric invariance (weak), scalar invariance (strong), and strict invariance. Criteria from Cheung & Rensvold (2002) and Chen (2007) were applied: ΔRMSEA < 0.01, ΔNNFI and ΔCFI ≥ 0.01.

Discriminant validity: Evaluated with respect to PTQ-C and anxiety symptoms (DASS-Y) using the heterotrait-monotrait ratio of correlations (HTMT; Henseler et al., 2015). Evidence of discriminant validity was considered when HTMT < 0.85.

Descriptive and correlational statistics: Descriptive data were provided for the total sample and separated by gender and age group. Two-way analysis of variance (ANOVA) was conducted to analyze differences in PSWQ-C scores. Pearson correlations between PSWQ-C and other scales were calculated to assess convergent validity.

---

RESULTS

Psychometric Quality of Items

Initial discriminative analysis (14 items): All items except the three reverse-scored items (items 2, 7, and 9) showed good discrimination indices.

• Cronbach's alpha: .80 (95% CI [.78, .82])
• McDonald's omega: .83 (95% CI [.81, .85])

Elimination decision: Reverse-scored items were eliminated due to deficient discrimination indices (negative corrected item-total correlations: -0.13, -0.04, -0.02 respectively).

11-item version (after elimination):

• Cronbach's alpha: .88 (95% CI [.87, .89])
• McDonald's omega: .88 (95% CI [.87, .89])

All subsequent analyses were conducted with the 11 direct items of the PSWQ-C. Corrected item-total correlations ranged from 0.48 (item 14) to 0.63 (item 13).

Validity Evidence Based on Internal Structure

Dimensionality: The one-factor model of the PSWQ-C obtained a good fit to the data:

• χ²S-B(55) = 88.02, p < .001
• RMSEA = 0.04, 90% CI [0.02, 0.05]
• CFI = 0.98
• NNFI = 0.97
• SRMR = 0.02

Standardized factor loadings ranged from 0.56 (original item 9) to 0.72 (item 11), indicating substantial relationships between each item and the latent factor.

Measurement Invariance (Table 2):

Invariance across gender:

• Metric invariance: ΔCFI = 0.001, ΔRMSEA = 0.003, ΔNNFI = 0.004 ✓
• Scalar invariance: ΔCFI = 0.001, ΔRMSEA = 0.003, ΔNNFI = 0.005 ✓
• Strict invariance: ΔCFI = -0.002, ΔRMSEA = 0.001, ΔNNFI = 0.000 ✓

Invariance across age group:

• Metric invariance: ΔCFI = -0.001, ΔRMSEA = 0.001, ΔNNFI = 0.002 ✓
• Scalar invariance: ΔCFI = -0.002, ΔRMSEA = 0.001, ΔNNFI = 0.001 ✓
• Strict invariance: ΔCFI = -0.012, ΔRMSEA = -0.006, ΔNNFI = -0.009 ✗ (ΔCFI exceeded 0.01)

Invariance across schools:

• Metric invariance: ΔCFI = -0.002, ΔRMSEA = 0.000, ΔNNFI = 0.001 ✓
• Scalar invariance: ΔCFI = -0.005, ΔRMSEA = -0.002, ΔNNFI = -0.003 ✓
• Strict invariance: ΔCFI = 0.001, ΔRMSEA = 0.003, ΔNNFI = 0.004 ✓

Metric, scalar, and strict invariance were confirmed across gender and schools. Measurement invariance across age group was supported at metric and scalar levels.

Discriminant Validity

The PSWQ-C demonstrated discriminant validity:

• PTQ-C (HTMT = 0.81) < 0.85 ✓
• DASS-Y Anxiety (HTMT = 0.69) < 0.85 ✓

Both HTMT values were below the cutoff of 0.85 suggested by Henseler et al. (2015).

Convergent Construct Validity

Correlation table with PSWQ-C:

• PTQ-C: r = .74***
• DASS-Y Total: r = .66***
• DASS-Y Depression: r = .56***
• DASS-Y Anxiety: r = .61***
• DASS-Y Stress: r = .60***
• RCADS - GAD: r = .56***
• RCADS - Depression: r = .55***
• RCADS - Social Phobia: r = .60***
• RCADS - Panic Disorder: r = .64***
• RCADS - OCD: r = .64***
• RCADS - Separation Anxiety Disorder: r = .40***

Note: *** p < .001

The PSWQ-C showed strong correlations with PTQ-C (r = .74), a measure of the similar but broader construct of Repetitive Negative Thinking. Correlations with emotional symptoms according to DASS-Y were strong but lower than with PTQ-C, ranging from .56 (depression) to .61 (anxiety). Regarding RCADS-30, all correlations were positive and strong, varying between .40 (separation anxiety disorder) and .64 (panic disorder and OCD).

Scores across Gender and Age

Descriptive data table:

Gender	Age	N	M	SD
Boys	8-10 years	159	1.06	0.71
Boys	11-12 years	147	1.09	0.71
Girls	8-10 years	151	1.25	0.76
Girls	11-12 years	128	1.39	0.69
Total	8-12 years	585	1.19	0.73

Two-way ANOVA:

• Gender effect: F(1) = 16.62, p < .001 (girls showed higher PSWQ-C scores than boys)
• Age group effect: F(1) = 2.24, p < .13 (not significant)
• Gender × age group interaction: F(1) = 0.79, p < .37 (not significant)

---

DISCUSSION AND CONCLUSIONS

Main Findings

The Spanish translation of the PSWQ-C demonstrated solid psychometric properties. Consistent with previous studies in other languages, the three reverse-scored items showed low discrimination indices and were eliminated, resulting in an 11-item version with good internal consistency (α = .88, ω = .88). CFA confirmed that the one-factor model showed a very good fit to the data, consistent with all previous psychometric analyses of the PSWQ-C (e.g., Benedetto et al., 2019; Chorpita et al., 1997; Muris et al., 2001; Păsărelu et al., 2017) and most studies of the adult PSWQ version.

A significant contribution was the comprehensive analysis of measurement invariance of the one-factor model across gender, school, and age group. Metric, scalar, and strict invariances were confirmed across gender and school. Regarding age group, invariance was found at metric and scalar levels. These findings permit comparisons of means across groups of participants and add evidence beyond that found by Păsărelu et al. (2017) in Romanian samples.

Another contribution was providing evidence of PSWQ-C discriminant validity with respect to Repetitive Negative Thinking (measured by PTQ-C) and anxiety (measured by DASS-Anxiety). According to the authors, this is the first study conducting this analysis on a nonclinical PSWQ-C sample.

Regarding convergent validity, the PSWQ-C showed strong correlation with PTQ-C (r = .74), which measures the similar but broader construct of RNT. It also showed strong correlations with emotional symptoms according to DASS-Y (r values between .56 depression and .61 anxiety) and with disorders evaluated by RCADS-30 (r values between .40 separation anxiety disorders and .64 panic disorder and OCD).

Notably, the correlation between PSWQ-C and the GAD subscale of RCADS-30 was strong (r = .56) but not the strongest among RCADS subscales. The authors noted that RCADS GAD items focus on worry but are considerably simpler than PSWQ-C items, have more general content, and do not specify problematic aspects of worry or its duration. In contrast, the PSWQ-C was designed considering worry as a natural coping strategy that becomes pathological when perceived as uncontrollable.

Boys showed significantly lower PSWQ-C scores than girls (F(1) = 16.62, p < .001), while no effects were found for age group or gender × age group interaction.

Recognized Limitations

The authors mention several limitations:

1. PSWQ-C functioning was analyzed in only two Colombian public schools. Future studies should analyze psychometric properties in more diverse Colombian contexts.

2. No clinical sample was recruited to test PSWQ-C psychometric properties. Subsequent studies in Colombian clinical samples might confirm previous findings regarding good internal consistency, test-retest reliability, and one-factor structure in nonclinical samples.

3. Treatment sensitivity of the PSWQ-C was not analyzed. Future studies should confirm that the Spanish version of the PSWQ-C shows treatment sensitivity as the original version.

4. Correlations between PSWQ-C and remaining instruments might be artificially inflated because the same measurement procedure was used for all instruments in this study. Future studies might analyze PSWQ-C correlation with behavioral measures of worry.

Conclusion

This study represents the first psychometric analysis of the Spanish version of the PSWQ-C. The instrument demonstrated good internal consistency, the expected one-factor structure, at least scalar invariance across gender, school, and age group, discriminant validity with respect to measures of RNT and anxiety, and convergent validity in a large sample of Colombian children aged 8 to 12 years. Future studies might analyze additional psychometric properties of the Spanish version that remain unexplored and include samples from other Spanish-speaking countries.

---

Significance and contribution

The Spanish adaptation of the PSWQ-C contributes significantly to the availability of reliable instruments for assessing pathological worry in Spanish-speaking child populations. The findings demonstrate that the Spanish version of the PSWQ-C possesses solid psychometric properties, including acceptable internal consistency, appropriate one-factor structure, measurement invariance across gender, school, and age group, and discriminant and convergent validity. These results enable its use in empirical research and clinical assessment, facilitating a deeper understanding of pathological worry in children and contributing to the advancement of standardized measurement in Latin American contexts.

---

---

Verification Checklist / Lista de verificación

• Summary generated from PDF (not from memory)
• Accuracy of numerical data not re-verified in this repair pass
• Bilingual format present (Spanish + English)
• Format standardized in repair pass 2026-03-30